Difference between revisions of "Book - The brain of the tiger salamander 2"

From Embryology
Line 361: Line 361:
metabolites between brain tissue and cerebrospinal fluid. Massive  
metabolites between brain tissue and cerebrospinal fluid. Massive  
thickenings of the brain wall occur in many fishes and in amniote  
thickenings of the brain wall occur in many fishes and in amniote  
vertebrates, but not in mudfishes and urodeles.  
vertebrates, but not in mudfishes and urodeles.
==Chapter III Histological Structure==
IN AMPHIBIAN brains the histological texture is generalized,
exhibiting some embryonic features; and it is at so primitive a
level of organization as to make comparison with mammals difficult.
Most of the nerve cells are small, with scanty and relatively undifferentiated cytoplasm. There are some notable exceptions, such as
the two giant Mauthner's cells of the medulla oblongata and related
elements of the nucleus motorius tegmenti. With the exceptions just
noted, Nissl bodies are absent or small and dispersed.
Almost all bodies of the neurons are crowded close to the ventricle
in a dense central gray layer, with thick dendrites directed outward
to arborize in the overlying white substance (figs. 9, 99). The axon
usually arises from the base of the dendritic arborization, rarely from
its tip, and sometimes from the cell body; it may be short and much
branched or very long, with or without collateral branches. The
ramifications of the short axons and of collaterals and terminals of
the longer fibers interweave with dendritic arborizations to form a
more or less dense neuropil, which permeates the entire substance of
the brain and is a synaptic field. Some of the nerve fibers are
myelinated, more in the peripheral nerves, spinal cord, and medulla
oblongata than in higher levels of the brain. Both myelinated and
unmyelinated fibers may be assembled in definite tracts, or they may
be so dispersed in the neuropil as to make analysis difficult. The arrangement of recognizable tracts conforms with that of higher brains,
so that homologies with human tracts are in most cases clear. These
tracts and the gray areas with which they are connected provide the
most useful landmarks in the analysis of this enigmatic tissue.
In the gray substance there are few sharply defined nuclei like
those of mammals, but the precursors of many of these can be recognized as local specializations of the elements or by the connections of
the related nerve fibers. In most cases the cells of these primordial
nuclei have long dendrites, which arborize widely into surrounding
fields (figs. 9, 24, 61, 66), so that the functional specificity of the nucleus is, at best, incomplete. This arrangement facilitates mass
movements of "total-pattern" type, but local differentiations serving
"partial patterns" of action (Coghill) are incipient. Localized reflex
arcs are recognizable, though in most cases these are pathways of
preferential discharge within a more dispersed system of conductors
(chap. vi).
Tissue differentiation is more advanced in the white substance
than in the gray. The most important and diversely specialized
synaptic fields are in the alba, and this local specialization is correlated with differences in the physiological properties of the nervous
elements represented. This means, as I see it, that functional factors
must be taken into account in both ontogenetic and phylogenetic
differentiation and that in the long view the problems of morphogenesis are essentially physiological, that is, they resolve into questions of
adaptation of organism to environment (chap. viii). This is the reason
why in this work the histological analysis is made in terms of physiological criteria, even though these criteria are, in the main, based on
indirect evidence, namely, the linkage of structures in functional
systems of conductors.
The nonnervous components of this tissue comprise the blood vessels, ependyma, and a small number of cells of uncertain relationships
which are regarded as undifferentiated free glial cells or transitional
elements ('34, p. 94; '336, p. 17). The ependymal elements everywhere span the entire thickness of the brain wall with much free
arborization. They assume various forms in different regions, and
their arrangement suggests that they are not merely passive supporting structures, though if they have other specific functions these are
still to be discovered. For illustrations see figures 63, 64, 70, 79, and
More detailed descriptions of the histological structure of urodele
brains may be found in earlier papers ('14a, p. 381; '17, pp. 232,
279 ff.; '335, pp. 16, 268; '33c; '33cf; '34; '34a,- '346; '42, p. 195; '44a).
In the olfactory bulbs of Necturus ('31) and Amblystoma ('246) we
find an interesting series of transitional cells between apparently
primitive nonpolarized elements and typical neurons, as described
on page 54.
In the generalized brains here under consideration the neuropil is
so abundant and so widely spread that it evidently plays a major
role in all central adjustments, thus meriting detailed description.
Only the coarser features of this tissue are open to inspection with
presently available histological technique. In my experience its texture is best revealed by Golgi preparations, and very many of them,
for the erratic incidence of these impregnations may select in different specimens now one, now another, of the component tissues —
blood vessels, ependyma, dendrites, or axons. In each area of neuropil
these components are independent variables, and in most of these
areas axons from many sources are so intricately interwoven that the
tissue can be resolved only where fortunate elective impregnations
pick out one or another of the several systems of fibers in different
specimens. It is difficult to picture the neuropil either photographically or with the pen, and the crude drawings in this book and in the
literature give inadequate representations of the intricacy and delicacy of its texture.
A survey of the neuropil of adult Amblystoma as a whole has led
me to subdivide it for descriptive purposes and somewhat arbitrarily
into four layers ('42, p. 202). From within outward, these are as
1. The periventricular neuropil pervades the central gray so that
every cell body is enmeshed within a fabric of interwoven slender
axons (figs. 106, 107). This persists in some parts of the mammalian
brain as subependymal and periventricular systems of fibers.
2. The deep neuropil of the alba at the boundary between gray and
white substance knits the periventricular and intermediate neuropil
together, and it also contains many long fibers coursing parallel with
the surface of the gray. The latter are chiefly efferent fibers directed
toward lower motor fields (fig. 93, layer 5; '42, figs. 18-21, 24, 29-45,
3. The intermediate neuropil in the middle depth of the alba contains the largest and most complicated fields of this tissue. It is very
unevenly developed, in some places scarcely recognizable and in
others of wide extent and thickness. Its characteristics are especially
well seen in the corpus striatum (figs. 98, 99, 108, 109), thalamus
('396, fig. 81; '42, figs. 71, 81), and tectum opticum (figs. 93, layer 2,
101; '42, figs. 26, 30, 32, 79-83). Many of the long tracts lie within
this layer and have been differentiated from it. Most of the specific
nuclei of higher animals, including the outer gray layers of the
tectum, have been formed by migration of neuroblasts from the central gray outward into this layer. Here we find much of the apparatus of local reflexes and their organization into the larger, innate
patterns of behavior.
4. The superficial neuropil is a subpial sheet of dendritic and axonal
terminals, in some places absent, in others very elaborately organized. Here are some of the most highly specialized mechanisms of
correlation in the amphibian brain, from which specific nuclei of
higher brains have been developed. Notable examples are seen in the
interpeduncular neuropil (chap, xiv) and the ventrolateral neuropil
of the cerebral peduncle described in the next section. This neuropil
seems to be a more sensitive medium for strictly individual adjustments (conditioning) than the deeper neuropil, but of this there is no
experimental evidence. This hypothesis is supported by the fact that
in higher animals cerebral cortex develops within this layer and
apparently by neurobiotactic influence emanating from it.
In the first synapses observed in embryogenesis numerous axonic
terminals converge to activate a single final common path (Coghill,
'29, p. 13), This is the first step in the elaboration of neuropil. As
differentiation advances, neurons are segregated to serve the several
modalities of sense and the several systems of synergic muscles, and
these systems are interconnected by central correlating elements. In
no case are these connections made by an isolated «hain of neurons in
one-to-one contact between receptor and effector. The central terminals of afferent fibers from different sense organs are widely spread
and intermingled. Dendrites of the correlating cells branch widely in
this common receptive field, and the axons of some of them again
branch widely in a motor field, thus activating neurons of the several
motor systems. This arrangement is perfectly adapted to evoke mass
movement of the entire musculature from any kind of sensory stimulation, and this is, indeed, the only activity observed in early
embryonic stages.
It is the rare exception rather than the rule for a peripheral sensory
fiber to effect functional connection directly with a peripheral motor
neuron. One or more correlating elements are interpolated; and, as
differentiation advances, the number of these correlating neurons is
enormously increased in both sensory and motor zones and also in
the intervening intermediate zone. The axons of these intrinsic elements ramify widely in all directions, and, as a rule, they or collaterals from them interweave to form the closely knit fabric which
pervades both gray and white substance everywhere. This is the axonic neuropil, within the meshes of which dendrites of neurons
ramify widely. These axons are unmyelinated, and every contact
with a dendrite or a cell body is a synaptic junction. Eveiy axon contacts many dendrites, and every dendrite has contact with many
axons, and these may come from near or from very remote parts.
All neuropil is a synaptic field, and, since in these amphibian brains
it is an almost continuous fabric spread throughout the bram, its
action is fundamentally integrative. But it is more than this. It is
germinative tissue, the matrix from which much specialized structure
of higher brains has been differentiated. It is activated locally or
diffusely by every nervous impulse that passes through the substance
of the brain, and these impulses may irradiate for longer or shorter
distances in directions determined by the trend of the nerve fibers of
which it is composed. This web of conductors is relatively undifferentiated, but it is by no means homogeneous or equipotential, for each
area of neuropil has characteristic structure and its own pattern of
peripheral and central connections. Many of these fibers, which take
long courses, may connect particular areas of gray substance either
in dispersed arrangement or assembled as recognizable tracts. In
fact, we find all gradations between an almost homogeneous web of
neuropil and long well-fasciculated tracts of unmyelinated or myelinated fibers. Many of these long fibers have collateral connections
throughout their length; others are well-insulated conductors between origin and termination.
The web of neuropil shows remarkable diversity in different regions. In some places it is greatly reduced, as, for instance, in the
ventral funiculi, where the alba is densely filled with myelinated
fibers; in other places there are local concentrations of dendrites and
finest axons so dense that in ordinary haematoxylin or carmine preparations they appear as darker fields, the Punksuhstanz of the early
histologists and the glomeruli of the olfactory bulb and interpeduncular nucleus (chap. xiv). When some of these fields are analyzed, it is found that their fibrous connections conform with those of
specific "nuclei" of higher brains. In Ambly stoma such a field is not a
"nucleus," for it contains no cell bodies; but examination of the corresponding area in anurans and reptiles may show all stages in the
differentiation of a true nucleus by migration of cell bodies from the
central gray outward into the alba ('27, p. 232). Such a series of
phylogenetic changes can be readily followed in the corpus striatum,
the geniculate bodies, the interpeduncular nucleus, and many other places and particularly in the tectum opticum and the pallial fields
of the cerebral hemispheres.
In phylogeny the long, well-organized tracts seem to have been
formed by a concentration of the fibers of the neuropil. The diffuse
neuropil is probably the primordial form, going back to the earliest
evolutionary stages of nervous differentiation (coelenterates) . Local
reflex arcs and specific associational tracts have been gradually differentiated within it, that is, integration precedes local specialization, the total pattern antedates the partial patterns. In ontogeny,
especially of higher animals, this history may not be recapitulated,
and tracts serving local reflexes may appear very early; but in
Amblystoma, even in the adult stage, there are few tracts which are
compactly fasciculated and free from functional connection with the
surrounding neuropil. Most of the long, well-fasciculated tracts have
some myelinated fibers that seem to have functional connections
only at their ends; but these are accompanied by others, which are
without myelin and are provided with numberless collaterals tied
into the enveloping neuropil. There is, accordingly, a seepage of
nervous influence along the entire length of these tracts.
From the primordial diffuse neuropil, differentiation advanced in
two divergent directions. One of these, as just pointed out, led to the
elaboration of the stable architectural framework of nuclei and tracts,
the description of which comprises the larger part of current neuroanatomy. This IS the heritable structure, which determines the basic
patterns of those components of behavior which are common to all
members of the species. The second derivative of the primordial
neuropil is the apparatus of individually modifiable behavior — conditioning, learning, and ultimately the highly specialized associational tissues of the cerebral cortex. In both phylogeny and ontogeny,
differentiation of the first type precedes that of the second. Primitive
animals and younger developmental stages exhibit more stable and
predictable patterns of behavior; and the more labile patterns are
acquired later. In Amblystoma both these types of differentiation are
at low levels, but they are sufficiently advanced to be clearly recognizable. Tissue differentiation is further advanced in the white substance than in the underlying gray.
During the course of this progressive specialization of tissue, the
primordial integrative function of the neuropil is preserved and
elaborated. The mechanism employed is seen in its most generalized
form in the deep periventricular neuropil, layer 1 of the preceding analysis. This is a close-meshed web of finest axons, within which all
cell bodies are imbedded. It is everywhere present, providing continuous activation (or potential activation) of every neuron, summation, reinforcement, or inhibition of whatever activities may be going
on in the more superficial layers and affecting the general excitatory
state of the whole central nervous system. It receives fibers from all
sensory and motor fields and seems to be the basic apparatus of integration. In man this type of tissue survives in the periventricular
gray of the diencephalon, and, as suggested by Wallenberg ('31), it
probably plays an important part in determining the disposition and
temperament of the individual ('34rt, pp. 241, 245). This unspecialized tissue may serve the most general totalizing function. From it
there have been derived the complicated mechanisms for synthesizing the separate experiences and organizing them in adaptive patterns — a process of differentiation which culminates in the human
cerebral cortex. The relations of neuropil to reflex arcs are discussed
in chapter vi.
The relative abundance of myelinated fibers is a rough indicator of
the relation between stable and labile types of performance. Thus
the myelinated white substance is relatively greater in the spinal
cord than in the brain, and the ratio of myelinated to unmyelinated
tissue diminishes as we pass forward in the brain, as is well illustrated
by a published series of Weigert sections drawn from a single specimen (no. IIC) from the olfactory bulb to the spinal cord ('10, figs.
8-21; '25, figs. 2-9; '44&, figs. 1-6). This is because the myelinated
fibers, most of which are long conductors of through traffic, tend to
be compactly arranged, with relatively scanty collateral connections
with the neuropil.
In the gray substance there is a similar increment m relative
amount of neuropil as we pass forward from spinal cord to hemispheres, the higher levels being specialized for correlation, association, conditioning, and integration and the lower levels for stabilized
total activities and reflexes. In the phylogenetic series this principle
takes a form which can be quantitatively expressed as Economo's
coefiicient of the ratio between total gray substance and the mass of
the nerve cells contained within it. The lower the animal species in
the scale, the greater is the mass of the cells compared with the gray
substance. This law may be expressed in the converse form: Higher
animals have a larger proportion of neuropil in the gray substance,
thus giving them capacity for conditioning and other individually acquired patterns of behavior (Economo, '26, '29). No mathematical
precision should be claimed for these laws, but they do express the
general trend of our experience. For additional data and critical
comment see von Bonin on page 64 of the monograph by Bucy and
others ('44).
The properties of amphibian neuropil have been discussed in several places, in addition to the summary already cited ('42, p. 199).
In one of these papers ('33c?) the geniculate neuropil of Necturus was
described, and later ('42, p. 280) the quite different arrangement of
Anibly stoma (compare the corresponding structure of the frog, '25),
together with a full account of the neuropil of the optic tectum and
its connections (here, again, comparison with the more differentiated
structure of the frog is instructive). In connection with a general
survey of the neuropil of Necturus, its peculiar relations in the pallial
part of the hemisphere were discussed in four papers ('336, p. 176;
'336?, '34, '34a). Amblystoma is similar ('27), though the details have
not been fully explored. In the pallial field the four layers of neuropil
tend to merge into a single apparatus of association. This is corticogenetic tissue within which the earliest phases of incipient differentiation of laminated cortex can be recognized in the hippocampal area,
as described in chapter vii.
Some samples of the appearance of elective Golgi impregnations of
amphibian neuropil are shown in the accompanying illustrations, and
many others are in the literature. The related morphological and
physiological problems can best be presented in the form of illustrative examples, one of which is the striatal neuropil (chap, vii), another the interpeduncular neuropil described in chapter xiv, and still
another in the superficial ventrolateral neuropil of the peduncle,
the area ventrolateralis pedunculi, which will next be described.
The "peduncle" in the restricted sense as here defined, together with the adjoining
tegmentum, is the chief central motor pool of the skeletal muscles. Efferent fibers
from it are among the first to appear in the upper brain stem in embryogenesis,
descending in the primary motor path, which is the precursor of the fasciculus
longitudinalis medialis, and activating the musculature of the trunk. A second
efferent path, which appears very early, goes out directly to the periphery through
the oculomotor nerve. The first sensory influence to act upon this pool comes from
the optic tectum through the posterior commissure in the S-reaction stage. In subsequent stages it is entered by fibers from many other sources, including the basal
optic tract (fig. 14) and the olfacto-peduncular and strio-peduncular tracts (fig. 6).
In the adult animal, motor impulses issuing from this pool are probably concerned primarily with synergic activation of large masses of muscle, notably those concerned with locomotion and conjugate movements of the eyes. Control of the
movements directly involved in seizing and swallowing food is believed to be
chiefly in the isthmic tegmentum, and this apparatus matures later n ontogeny.
The chief synaptic connections of the great motor pool of the peduncle are in the
intermediate, deep, and periventricular layers of neuropil. External to these is a
ventrolateral band of dense neuropil extending from the root of the III nerve forward along the entire length of the peduncle. In former papers this has been termed
"area lateralis tegmenti" and "nucleus ectomamillaris," but both these names now
seem to me inappropriate ('4'-2, p. 233). This band receives at the anterior end all
the terminals of the large basal optic tract (figs. U, 94) and, at the posterior end,
terminals of the secondary and tertiary ascending ^'isceral sensory and gustatory
tracts (figs. 8, 23). The terminals of these two systems of fibers are intimately interlaced, and among them are dendritic arborizations of tlie underlying cells of the
peduncle into which optic, olfactory, and visceral systems of nervous impulses converge and through which the combined effect is transmitted in the outgoing motor
pathways. These seem to be the primary components of this neuropil, and to them
are added axonic terminals from a wide variety of sources, the most notal)le of which
are sketched in figure '23 (compare Necturus, '34, p. 103 and fig. 4).
The peduncular dendrites, which arborize within this neuropil, include some
from the nucleus of the oculomotor nerve (fig. 24; '42, p. 275), so that here fibers of
the basal optic tract coming directly from the retina may synapse with peripheral
motor neurons — one of the rare examples of a two-neuron arc witli only one synapse
between the peripheral receptor and the effector, though even here there are at
least two additional synapses in this arc within the retina. Other peduncular
neurons may transmit retinal excitations downward through the ventral tegmental
fascicles to all lower motor fields.
At its anterior end this neuropil is connected by fibers conducting in both directions with the dorsal (mamillary) part of tlie hypothalamus, as in Necturus
('346, fig. 3), and also with its ventral (infundibular) part (fig. 23; "42, pp. 226, 227).
This is an extension of the visceral-gustatory tract into the hypothalamus, and it
may be that some of the visceral fibers pass without interruption through the
peduncular neuropil to the hypothalamus, though this has not been satisfactorily
demonstrated. In some fishes such a direct connection is evident and large; in
mammals the course of the ascending visceral-gustatory path is still uncertain.
Other connections of this neuropil, as shown in figure 23, include terminals of the
olfacto-peduncular tract from the anterior olfactory nucleus, probably the nervus
terminalis (observed in Necturus) fibers from the tectum, pretectal nucleus, dorsal
and ventral thalamus, and terminals of the fasciculus retroflexus (p. 202 and fig. 20).
These terminals of fibers from surprisingly diverse sources are all closely interwoven
with one anotlier and with the terminal dendrites of peduncular neurons, a unique
arrangement occurring, so far as known, only in Amphibia. I have indulged in the
following speculations upon its possible physiological significance.
In the first place, it is clear that this curious tissue is either an undifferentiated
primordium of a number of structures which are separately differentiated in more
specialized brains, or else it is a retrograde fusion of several such structures. The
former supposition seems more probable, for the phylogenetic history of one of its
components is easily read ('25, pp. 443-49). In Necturus there are no cell bodies
directly associated with this neuropil; in Amblystoma a few cells have migrated
out of the gray layer to its border (fig. 24) ; and in the frog the optic component of the
neuropil is separate and surrounded by a spherical shell of cell bodies, which is a true
basal optic nucleus (Gaupp, '99, p. 54, fig. 19). This nucleus attains large size and complexity in some reptiles, as described by Slianklin ('3.'3) and others. In the mammals it retains its individuality as the chief terminus of basal optic fibers, though
other fibers of this system are rather widely spreatl in the surrounding tegmentum
(Gillilan, '41). The history of the visceral and other components of the amphibian
ventrolateral neuropil has not yet been written.
In my recent description ('42) of the optic system of Amblystoma, special attention was given to the central distribution of thick and thin fibers from the retina,
and this was followed ('42, p. 295) by some speculations about its physiological significance. In development the thick fibers appear first, and the number of thin
fibers is enormously increased in later stages, particularly at the time of metamorphosis. Thick fibers conduct more rapidly than thin fibers, and this time factor
may play an important part in the central analysis of mixed retinal excitation.
Both thick and thin fibers end in the thalamus, pretectal nucleus, and tectum;
and in each of these fields their terminals are mingled, not segregated. The optic
fibers to the basal peduncular neuropil are moderately thick; upon retinal stimulation this field, accordingly, is the first to be activated, and its excitation of the underlying peduncular neurons may precede any influence upon these cells from the longer
paths by way of the tectum and thalamus. Figure 22 shows in continuous lines the
major optic tracts and the motor paths from the peduncle, and in broken lines some
internuncial connections. The thick fibers which descend from the tectum and
thalamus are myelinated; some of them are crossed in the posterior commissure
and the commissure of the tuberculum posterius, and some are uncrossed. They
connect primarily with the descending pathway in the ventral tegmental fascicles
(f.r.t.). The thinner correlating fibers take other pathways, and they may make
synaptic connection with peduncular neurons in any one or all of the four layers of
the neuropil. The efl'erent path may be to low er motor centers through the ventral
tegmental fascicles or to the muscles of the eyeball through the III nerve.
Figures 24 and 93 are diagrammatic transverse sections at the level of the III
nuclei and the middle of the tectum, illustrating some of the tecto-peduncular connections. These fibers are all of medium or thin caliber and, for the most part,
unmyelinated. The dendrites of the peduncular neurons ramify widely throughout
the entire thickness of the brain wall, and few, if any, of them are in physiological
relation with any single one of the afferent systems of fibers. Each of them may be
activated by any or all of these systems. The only significant localized specialized
tissue here seems to be in the neuropil, where the texture is different in the four layers
and where all peduncular neurons spread their dendrites in all these layers. Since
the movements which are activated from this motor pool are not disorderly convulsions, it is evident that the discriminative and well co-ordinated responses which
follow its excitation are not ordered primarily by the arrangement in space of the
motor elements. There are differences in the structural arrangements of the synaptic
junctions, though these are not so pronounced as in most other animals, and there
may be chemical and other factors yet to be determined. As I have elsewhere
pointed out, a time factor can be recognized by physiological experiment, and its
structural indicator can be identified histologically because large fibers have a faster
rate of conduction than small fibers and the interpolation of synaptic junctions in a
nervous pathway retards transmission.
In the structural setup before us it may be inferred that the first result of a retinal
excitation is the activation of the entire tectum, pretectal nucleus, and dorsal thalamus through the tliick myelinated fibers of the optic nerve and tract and also of the
ventrolateral neuropil of the peduncle through the basal optic tract. This is presumably a generalized nonspecific effect, and it will come to motor expression,
first, through the basal tract, for this is the shortest path. The resting state is changed to a state of excitation in both the peduncular gray and the peripheral
musculature with which it is connected. This is immediately followed by volleys
from the tectum, pretectal nucleus, and thalamus tlirough the myelinated tectoand thalamo-peduncular tracts; and this may contribute a spatial factor determined
by the position of the exciting object in the visual field and the sector of the tectum
upon which this local stimulus is projected. The first overt movement, accordingly, is
an orientation of the body and the eyeballs with reference to the source of stimulus.
After an appreciable time the smaller fibers from the retina deliver their volleys, and
the small fibers of the correlating tracts are activated. These deliver to the peduncle,
not unmixed or purely visual impulses but discharges, fired or inhibited, as the case
may be, by the existing excitatory state of the correlating apparatus: and this, in
its turn, is determined by numberless nonvisual features of the total situation,
present and past. If, for instance, one of the visual or nonvisual components is
fatigued, this will affect the pattern of tectal discharge.
In salamanders the delay between the preliminary orientation of the body and the
consummation of the reaction may be, and commonly is, very long — a period of
tension, which in a man we would call "attention" and which Coghill called the
"regarding" reaction (p. 78; Coghill, '33, Paper XI, p. 334; '36). During this
period an inconceivably complicated resolution of forces is in process within the
central apparatus of adjustment. Inhibition plays an important role here, and it may
be that the chief function of the ventrolateral superficial neuropil of the peduncle is
control of the inhibitory phase of these activities comparable with tliat suggested in
chapter xiv for the specific interpeduncular neuropil. Of the details of these adjustments our knowledge is scanty, but some hints may be gathered from urther examination of the structure involved; and here, to simplify the problem, we shall confine tJie discussion to the superficial peduncular neuropil.
This neuropil is interpolated "in series" in the visual-motor path of the basal
optic tract, and this short circuit is connected "in parallel" with the longer visualmotor circuits by way of the tectum and thalamus. The latter are very much larger
and more complicated, and they evidently comprise the major part of the apparatus by which behavior is regulated by visual experience. The basal optic system
seems to be ancillary to the major system and to be related with it in two quite
different ways, first, as a nonspecific primary activator, as already explained, and,
second (and subsequently), as a device for modifying or "inflecting" (to borrow
Arnold Gesell's expression) the standard patterns of behavior by intercurrent influence of present activity in other fields, including, perhaps, conditioning and the
individuation of hitherto unaccustomed types of response. This second feature involves further consideration of the nonvisual components. These have already been
listed, and some of them are shown in figure 23.
The largest components of this peculiar neuropil are the optic and tlie visceralgustatory systems, and their fibers end exclusively here. This is why these are regarded as the primary components; all other systems of entering fibers are spread
more or less widely also in the deeper layers of peduncular neuropil. In the visceralgustatory system this neuropil is inserted in series in the ascending pathway
toward the hypothalamus, and there is probably a parallel system of conduction
with no synapse in the peduncle (fig. 8). As seen in tliis figure, peduncular neurons
activated from this neuropil send axons around the tuberculum posterius into the
hypothalamus; and there is a return path from both dorsal (mamillary) and
ventral (infundibular) parts of the hypothalamus to this neuropil and contiguous
parts of the peduncle (figs. 18, 'TIS). All visceral-gustatory influences which reach
the hypothalamus are, accordingly, previously modified or "inflected" in the superior visceral nucleus or in the peduncle. The excitatory state of the hypothalamus, in
turn, affects all activity in the peduncle.
This sort of circular activity is everywhere present in these urodele brains,
working through the diffuse periventricular and deep neuropil and in many places
also through specially differentiated tracts (p. 76). This is the primordial apparatus of integration, and in its more differentiated form it is part of the apparatus
by which individuated partial patterns of local reflexes are kept in appropriate
relationship with the larger total patterns of behavior.
In a recent survey ('44o) of tlie optic and visceral nervous circuits here under
consideration, several different patterns of linkage of the component units were
listed. These include direct activation of eye muscles through the basal optic tract,
similar activation of the skeletal musculature, indirect activation of either or both
of these sj^stems of muscles by visual stimuli through the tectum and thalamus,
direct activation of either or both systems of muscles through the visceral-gustatory
path, indirect visceral -gustatory action by way of the superior visceral nucleus,
with or without intercurrent influence of the tectum upon this nucleus by way of
the tecto-bulbar tract, and the variable effects of concurrent discharge of many
other systems of fibers into each of the centers of adjustment involved. Since eacli
of these patterns of linkage is complex and the structural units themselves are not
simple, it is evident that in actual performance the number of ways in which the
known units may be combined and recombined is practically unlimited. The simplest
possible activity of stimulus-response type involves a central resolution of forces in
an equilibrated dynamic system of inconceivable complexity. Oversimplification of the problem will not hasten its solution.
The preceding description illustrates the relations of a specific field of specialized
neuropil to other parts of the brain with known functions. These visible connections
exhibit a structural arrangement which can be interpreted as putative pathways of
transmission of the several components of a complex action system, some concerned with stable reflex patterns and some with more labile individually acquired components. The relations of these two classes of components of the action
system to each other and of local units to the integrated whole present the major
problems of neurology.
SINCE the brain of Amblystoma presents a generalized structure
which is probably close to the ancestral type from which all
more highly specialized vertebrate brains have been derived, the
salient features of internal organization are here summarized in
schematic outline. Tlie accompanying diagrammatic figures 1-24
give the necessary topographic orientation, and the details may be
filled in by reference to the corresponding sections of Part 11. What
is here described may be regarded as the basic organization of the
brains of all vertebrates above fishes, that is, the point of departure
from which various specialized derivatives have been differentiated.
Amblystoma possesses the equipment of sensory and motor organs
typical for vertebrates at a rather low level of specialization and in
evenly balanced relations. All the usual systems are present, and none
shows unusual size or aberrant features. The great lateral-line system
of sense organs so characteristic of fishes is preserved, though somewhat reduced after metamorphosis. On the motor side the organs of
locomotion and respiration have advanced from the fishlike to the
quadrupedal form, but in very simple patterns. In early phylogeny
the specialization of the motor systems seems to lag behind that of
the sensory systems because the aquatic environment of primitive
forms is more homogeneous than that of terrestrial animals, and,
accordingly, fewer and simpler patterns of behavior are needed.
Our search in this inquiry is for origins of human structures and
for an outline of the history of their evolution. From this standpoint
it is evident that in the central nervous systems of all vertebrates
there is a fundamental and primary difference between the cerebrum
above and the rhombencephalon (and spinal cord) below a transverse
plane at the posterior border of the midbrain (for further discussion
of this see chaps, viii and xiii).
The spinal cord and rhombic brain contain the central adjustors of
the basic vital functions — respiration, nutrition, circulation, reproduction, locomotion, among others. This apparatus is elaborately
organized in the most primitive living vertebrates, as also no doubt
it must have been in their extinct ancestors. The cerebrum, on the other hand, except for the olfactory component, is a hiter acquisition.
This is suggested by what is seen in Amphioxus and by the retai-ded
development of the cerebrum in all vertebrate embryos, as illustrated
especially clearly in the early fetal development of the opossum.
At an early (and unknown) period of vertebrate ancestry a pair of
eyes was differentiated. These and the olfactory organs are the leading distance receptors, and as such they gave to the vertebrate
ancestors more information about their surroundings and hence
greater safety in moving about freely. The nose and eyes, with the
associated oculomotor apparatus, early assumed the dominant role
in the recognition of food, mates, and enemies, and their cerebral
adjustors were enlarged accordingly. The contact receptors are adequate for sedentary, crawling, or burrowing ancestors, and here the
response to stimulation follows immediately. But, as Sherrington
long ago pointed out, in a free-swimming animal there is a time lag
between reception of the stimulus from a distant object and the consummation of the response. The pregnant interval between the anticipatory and consummatory phases of the reaction gives the clue to an
understanding of the entire history of forebrain evolution. During
this interval there is a central resolution of forces, which eventuates
in appropriate behavior; and, with increasing complication of patterns of behavior, this central apparatus of adjustment assumes more
and more structural complexity and physiological dominance over
the entire bodily economy (chap. vii).
The details of these internal connections are not relevant here. It
suffices to present two summaries, one in this chapter in topographic
arrangement by regions from spinal cord to olfactory bulbs as conventionally described and one in the next chapter on a different plan,
i.e., an arrangement in longitudinal zones which are functionally defined. For the present purpose it is convenient to recognize seventeen
subdivisions of the central nervous system, each of which is characterized by special physiological activities, though these activities
are not localized here exclusively. This subdivision might be carried
further into detail indefinitely. Numbers 2-6 in the following paragraphs are in the rhombic brain; the others are in the cerebrum.
The spinal cord is not described in this report except for some features closely related to the brain, to which reference is made in the
next paragraph. The cord segments are organized for the regulation of local reflexes of the limbs and the integration of these reflexes with
one another and with the action of the trunk musculature, as in
ordinary locomotion.
The sector of the bulbo-spinal junction includes the upper segments of the spinal cord and the lower part of the medulla oblongata.
It is the first center of correlation to become functional in embryonic
development (Coghill, '14, Paper I). Its dorsal part around the
calamus scriptorius receives fibers from the entire sensory zone of the
bulb and cord, so that this gray of the funicular and commissural
nuclei is a general clearing-house for all exteroceptive, proprioceptive, and visceral functions of the body except vision and olfaction.
Here these functions are integrated in the interest of control of posture, locomotion, visceral activity, and other basic components of
mass-movement type. Some of these connections are shown diagrammatically in figures 3, 7, 8, 87; for details and discussion see chapter
ix and a recent paper ('446).
Efferent fibers from the dorsal nuclei are directed spinalward and
forward. Most of the latter connect with motor nuclei of the medulla
oblongata; some go farther forward to the cerebellum, tectum, and
thalamus; and there is a strong, ascending visceral-gustatory tract
to the isthmus and peduncle. The motor zone of this sector is occupied chiefly by fibers of passage. The moderately developed gray substance includes motor neurons for muscles of the neck region, for the
tongue muscles, and for special visceral motor elements of the accessorius component of the vagus.
The medulla oblongata, or bulb, includes all the stem between the
isthmus and the calamus scriptorius except the cerebellum, there
being no pons. Its dorsal field receives all sensory fibers from the
head except the optic and olfactory, fibers from lateral-line organs
widely distributed over the body, and general visceral sensory fibers
chiefly by way of the vagus. The general visceral sensory and gustatory root fibers are segregated from the others in the fasciculus
solitarius; and this group has its own system of secondary fibers,
which converge into the visceral motor nuclei of the medulla oblongata and spinal cord. There is also a strong ascending secondary
visceral tract {tr.v.a.) to the isthmus, peduncle, and hypothalamus,
through which all cerebral activities may be influenced by visceral
and gustatory functions (fig, 8 and chaps, x, xi).
U'lie other afferent fibers of the V to X cranial nerves, upon entrance
into the brain, are fascicuhitetl according to the functional systems
represented, as outlined in the next chapter and shown in figures 7, 9,
88, 89, 90. The general cutaneous, lateral-line, and vestibular fibers
are arranged in a series of fascicles bordering the external surface;
the visceral sensory and gustatory fibers are assembled in a single
deeper bundle, the fasciculus solitarius. The marginal fascicles of
root fibers are arranged from ventral to dorsal in the following order :
general somatic sensory (chiefly cutaneous), vestibular, and, dorsally
of these, five or six fascicles of fibers of the lateral-line roots of the
VII and X nerves. The details of this arrangement are variable within the species and from species to species of urodeles. The fascicles of
vestibular and lateral-line roots, together with the underlying gray
and the intervening neuropil, comprise the area acusticolateralis.
The dorsal part of this, which receives only roots of the lateral-line
nerves, is lobulated on the ventricular side.
Most of these root fibers divide into ascending and descending
branches, and each fascicle spans the entire length of the medulla
oblongata. Some of the general cutaneous and vestibular fibers extend far down into the spinal cord and upward into the cerebellum
(figs. 3, 7). Terminals and collaterals of all these fibers end in a common pool of neuropil, from which secondary fibers go out to effect
local connections in the medulla oblongata, to enter the cerebellum,
to descend to the spinal cord, and to ascend in the lemniscus systems
to the tectum and thalamus. The physiological specificity of the root
fibers i.s largely, though not entirely, obliterated at the first synapse
in the neuropil of the sensory field, in sharp contrast with the mammalian arrangement (chap. xi).
From this arrangement of the sensory systems of fibers and their
central secondary connections it is clear that the bulbar structure is
so organized as to facilitate mass movements of total-pattern type,
which may be activated by any one of the exteroceptive or proprioceptive systems or by any conibination of these. There is some provision here for local reflexes of the muscles of the head, but the structure indicates that most of these are patterned from higher centers.
The central receptive field of the visceral-gustatory system is well
segregated from that of the general cutaneous and acousticolateral
systems; and this is the structural expression of the fundamental distinction between the visceral and the somatic sensory physiological
systems, to which further reference is made on pages G7 and 83.
Otherwise, there is httle histological evidence of precise localization
of function in the medulla oblongata. The visceral and somatic sensory fields are cross-connected within the sensory zone, and they
converge into a common sensory field at the bulbo-spinal junction.
Proprioceptive adjustments are made throughout the spinal cord,
medulla oblongata, cerebellum, and tectum; and each of these regions evidently plays a different role in the adjustments. Arcuate
fibers connect all parts of the sensory zone with the motor zone of the
same and the opposite side, and many of these divide into longdescending and ascending branches, thus activating extensive areas
of the intermediate and motor zones.
The motor field of the medulla oblongata and the intimately related reticular formation contain the complicated apparatus by
which the nuclei of the motor nerves are so interconnected as to act
in groups, each of which may execute a series of co-ordinated actions
in patterns determined by these connections. The tissues of the motor
tegmentum, which effect this analysis of motor performance, are so
intricately interwoven that it has not been possible to recognize the
components of the several functional systems, and further analysis
of this field is desirable.
The cerebellum is small and very simply organized, but the chief
structural features of the mammalian cerebellum are present except
the pontile system, which is totally lacking. The urodele cerebellum
consists of three major parts: (1) the median body, activated from
the spinal cord, trigeminal nerve, and tectum (figs. 1, 3, 10); (2) the
lateral auricles, which are enlargements of the anterior ends of the
sensory zones of the medulla oblongata (figs. 7, 91); and (3), ventrally of the body of the cerebellum, a nucleus cerebelli, which is the
primordium of the deep cerebellar nuclei of mammals (figs. 10, 32,
This analysis of cerebellar structure is based on the comprehensive
studies of Larsell ('20-'47) and Dow ('42), whose descriptions of
Amblystoma, published in 1920 and 1932, I have confirmed in all
respects. It should be noted that my definition of the amphibian
auricle includes more than Larsell's, for he includes in this structure
only the vestibular and lateral-line components. I find that these
terminals and the related field of neuropil are so intimately related
with the terminals of the trigeminus, the visceral-gustatory system,
and lemniscus fibers that their segregation is not practical anatomically. The auricle, accordingly, is here regarded as the common
primordium of several structures which in higher animals are diversely specialized for different functions. The most notable of these
are the superior or pontile nucleus of the trigeminus and the floccular
part of the flocculonodular lobe of the cerebellum. The primordia of
these structures are clearly evident, and the history of their further
differentiation in higher animals has been written.
Efferent fibers of tractus cerebello-tegmentalis leave all parts of the
cerebellar complex for the underlying gray; and one fascicle of
these — the brachium conjunctivum — passes forward to a ventral
decussation and distributes its fibers to the isthmic tegmentum of
both sides (figs. 10, 71). No primordium of the nucleus ruber or of the
inferior olive has been recognized.
This primitive cerebellum exhibits the typical vertebrate pattern
in very instructive form, with localization of the vestibular system
laterally and the other systems medially. It is an appendage added
to the basic sensori-motor systems; it supplements them, not as an
aid in determining the pattern of performance, but to insure efficient
execution. In species in which it is greatly enlarged, it contains
enormous reserves of potential nervous energy, which is released during motor activity to reinforce and stabilize the operation of the
effectors. For additional details see chapters x and xii.
The isthmus is unusually large in urodeles and is clearly circumscribed from surrounding parts. Dorsally it is small, containing in
and near the superior medullary velum a special segment of the
mesencephalic V nucleus and probably other peripheral connections
through the nerves of the chorioid plexuses and meninges. Below this
there is the superior visceral-gustatory nucleus (figs. 2B, 8, 23, 34).
The nucleus isthmi, which is large in the frog, is here undifferentiated.
The ventral part of the isthmus is very large, containing the nucleus
of the IV nerve and a mass of tegmental cells. This isthmic tegmentum is interpolated between the primary sensori-motor systems of the
medulla oblongata and midbrain, and it serves as an intermediary
between them. There is a large central nucleus of small cells which
receives fibers from a wide variety of correlation centers of intermediate-zone type. These enter by all the dorsal tegmental fascicles
and by several other paths (figs. 16, 17, 21). This nucleus is enveloped
by a group of larger cells, which is continuous posteriorly with similar
tegmental cells of the region of the trigeminus (figs. 29, 30, 91). The complex as a whole is believed to have two chief functions: (1) Here
are organized the patterns of the local reflexes of the musculature of
the head, particularly those concerned with feeding. (2) The smallcelled central nucleus is a special differentiation of the periventricular
gray, which serves, in addition to the specific functions just mentioned, a more general, nonspecific, totalizing function; that is, it is a
part of that integrating apparatus which appears in mammals as the
dorsal tegmental nucleus and the related fasciculus longitudinalis
dorsalis of Schiitz (p. 208). The details of structure are given in
chapter xiii.
The isthmic tegmentum occupies a strategic position between the
primitive bulbo-spinal mechanisms and the higher cerebral adjustors;
it plays a major role both in the patterning of local reflexes and in the
integration of all bodily activities. This mass of tissue, which in
urodeles is at a low level of differentiation, in higher animals is split
up and distributed so that in mammals the identity of the isthmic
tegmentum as an anatomical entity is lost in the adult brain, though
the isthmic sector is plainly marked in the early embryonic stages.
The interpeduncular nucleus also is unusually large in urodeles.
It is not interpeduncular but interisthmic, extending from the fovea
isthmi back to the level of the V nerve roots. The histological texture
is extraordinary. A well-defined, trough-shaped column of cells
borders the ventral angle of the ventricle, with dendrites extending
downward through the ventral commissure, to arborize with tufted
endings in a ventromedian band of neuropil (figs. 65, 66, 82, 83, 91).
The axonic components of this interpeduncular neuropil come from
various sources: (1) terminals of the fasciculus retroflexus, which
take the form of a flattened spiral (fig. 50); (2) terminals of tr.
tegmento-interpeduncularis from small cells of the overlying tegmentum with tufted endings, which join with the dendritic tufts of
the interpeduncular nucleus to form small glomeruli (figs. 60-66, 84) ;
(3) collaterals of thick fibers of tr. tegmento-bulbaris from the large
cells of the tegmentum with similar tufted endings in the glomeruli
(fig. 68); (4) collaterals of tr. interpedunculo-bulbaris, which also
enter glomeruli (figs. 83, 84) ; (5) terminals of tr. mamillo-interpeduncularis with dispersed free endings (figs. 60, 61); (6) similar
terminals of tr. olfacto-peduncularis (fig. 59); (7) less numerous
terminals from several other sources. The slender, unmyelinated
axons of the interpeduncular cells branch freely in the interpeduncular neuropil and continue from the nucleus in two strands (figs. 83,
84). The ventral interpedunculo-bulbar tract descends beyond the
nucleus for an undetermined distance in the lip of the ventral fissure.
The dorsal tract descends dorsally of the fasciculus longitudinalis
medialis and turns laterally to end in wide arborizations in the
tegmentum as far back as the IX nerve roots. Associated with these
dorsal fibers are interpedunculo-tegmental fibers, which end in the
neuropil of the isthmic tegmentum. The dorsal and interpedunculotegmental fibers are regarded as comparable with the isthmic and
bulbar parts of the mammalian f. longitudinalis dorsalis of Schiitz.
The physiological problems suggested by this peculiar structure
are puzzling. In the light of such scanty experimental evidence as we
possess, I have ventured to suggest that the interpeduncular complex
provides both activating and inhibitory components of reflex and
general integrative activities, the actual patterns of which are elsewhere determined.
Topographically, this nucleus lies in the motor zone, but its functions clearly are of intermediate-zone type. It is present in all vertebrates at the anteroventral border of the isthmus, that is, at the
boundary between cerebrum and rhombencephalon. Most of its
afferent fibers come from the cerebrum, and evidently it serves chiefly
as an intermediary adjustor between the sensory and intermediate
zones of higher levels and the motor zone of the rhombic brain (for
details see chap. xiv).
At this point in our analysis we. cross the boundary between
rhombencephalon and cerebrum. The radical differences in structure
and physiological properties of these two chief parts of the brain are
masked and in large measure overruled, especially in higher animals,
by ascending and descending connectives, of which the interpeduncular system is a typical illustration.
The tectum and the pretectal nucleus, as sectors of the sensory
field, together with the dorsal thalamus, form a physiological unit
within which all exteroceptive sensory systems are integrated in the
interest of cerebral control of all lower sensori-motor systems involved in the operation of the skeletal musculature. This unit is
intimately related with the cerebral peduncle and ventral thalamus.
In the most primitive vertebrates and in early embryonic stages of
all vertebrates, these structures might appropriately be united as a
middle subdivision of the brain, which serves as the dominant center of cerebral control of all somatic activities. But in the adult animal
the parts of this natural subdivision have so many distinctive connections and physiological properties that it seems preferable to
treat them separately.
In vertebrates below the mammals the tectum opticum is the chief
central end-station of the optic nerve; and, since the eyes are the
chief distance receptors in most of these animals, fibers of correlation
of all other sensory systems concerned with external adjustment
naturally converge to this station. The tectum, accordingly, becomes
the dominant adjustor of all exteroceptive systems. The tectum
mesencephali of Aml^lystoma has a larger optic part — the superior
colliculus — and a small, poorly differentiated nucleus posterior — the
primordial inferior colliculus. The latter is interpolated between the
tectum opticum and the cerebellum, and its connections suggest that
its most primitive functions are proprioceptive. It receives a small
primordial lateral lemniscus and evidently also serves such generalized auditory functions as this animal possesses (chap. xv). The
development of the optic nerve and adult tectal structure and connections have been described in detail ('41, '42). Chapter xvi is devoted to the visual system ; for the arrangement of the mesencephalic
nucleus and root of the V nerve see page 140 and figure 13.
Optic and lemniscus tracts and smaller numbers of fibers from
various other sources all terminate in a broad sheet of intermediate
neuropil, which is spread through the entire tectum and is nearly
homogeneous in texture (figs. 11, 93). The tectum is not definitely
laminated, though separation of the layers, which are conspicuous in
the frog, is incipient. Fibers diverge from it in all directions (figs. 12,
18, 21-24, 93). It is inferred from this structure that movements
activated directly from the tectum are of total-pattern type. Such
local visual reflexes as this animal possesses are probably patterned
elsewhere — in the thalamus and dorsal and isthmic tegmentum. Conditioning of reflexes is probably effected in these areas and perhaps
also in the ventrolateral peduncular neuropil (p. 38). Experiments
upon Triturus and anurans (Stone and Zauer, '40; Sperry, '43, '44,
'45, '456) demonstrate anatomical projection of retinal loci upon the
tectum opticum. This is true also in Amblystoma (Stone, '44; Stone
and Ellison, '45), though the nervous apparatus employed has not
been described.
The pretectal nucleus (figs. 2B, 35, 36, nuc.pt.) receives fibers directly from the retina and from the tectum, habenula, and cerebral hemisphere. Its efferent fibers go to the tectum, thalamus, hypothalamus, and cerebral peduncle (figs. 11, 12, 14, 15, 16, 22, 23).
Its functions are unknown, but, by analogy with mammals, this may
be part of the apparatus for regulation of the intrinsic musculature of
the eyeball. Doubtless other functions are represented also. This area
is the probable precursor of the mammalian pulvinar and neighboring
The thalamus receives many fibers from the retina, and it is broadly connected with the tectum by uncrossed fibers passing in both
directions in the brachia of the superior and inferior colliculi (figs. 11,
12). There are also systems of tecto-thalamic and hypothalamic and
thalamo-hypothalamic tracts which decussate in the postoptic commissure ; some of these crossed fibers take longer courses to reach the
peduncle and isthmic tegmentum (figs. 12, 15). This intimate
thalamo-tectal relationship is radically changed in higher animals,
where the thalamo-cortical connections are highly elaborated.
I have separated the dorsal thalamus into three sectors: (1) anteriorly, the small nucleus of Bellonci of uncertain relationships ; (2) a
well-defined middle part, an undifferentiated nucleus sensitivus,
which is the primordium of most of the sensory nuclei of the mammalian thalamus; and (3) a vaguely delimited posterior sector, which
apparently contains the undifferentiated primordium of both lateral
and medial geniculate bodies (chap. xvii).
The middle and posterior sectors receive numerous terminals and
collaterals of the optic tracts, terminals of the general bulbar and
spinal lemnisci, and, through the brachia of the superior and inferior
colliculi, these sectors are broadly connected with the tectum by
fibers running in both directions. There is a similar, but much
smaller, connection with the habenula.
From the middle sector a small, well-defined tr. thalamo-frontalis
goes forward to the hemisphere (figs. 15, 71, 72, 95, 101, tr.th.f.);
this is the common primordium of all the thalamo-cortical projection
systems of mammals, though here few, if any, of its fibers reach the
pallial area. Other efferent fibers go to the ventral thalamus, hypothalamus, peduncle, and tegmentum. These thalamic reflex connections antedate in phylogeny the thalamo-cortical connections,
and they persist in mammals as an intrinsic paleothalamic apparatus,
an important part of which is the periventricular thalamic contribution to the f. longitudinalis dorsalis of Schiitz. The largest pathways of efferent discharge from the dorsal thalamus go backward to the
peduncle and tegmentum by both crossed and uncrossed tracts (figs.
15, 18, 21, 23). The peduncular connection puts all the primary systems of total-pattern type under some measure of thalamic control.
The connections with the dorsal and isthmic tegmentum probably
co-operate in the patterning of local reflexes, particularly supplying
the visual component of the feeding reactions.
The "peduncle" described here is not the equivalent of the human
cerebral peduncle (p. 21). The intimate relations of this field with
the overlying tecto-thalamic field have been commented upon in the
preceding paragraphs. This ventral field is a well-defined column of
cells, differentiated at the anterior end of the basal plate of the
embryonic neural tube. It is the head of the primary motor column
(of Coghill), which in all vertebrates, from early embryonic stages to
the adult, contains the nucleus of the oculomotor nerve and a much
larger mass of nervous tissue, which activates the primitive mass
movements of locomotion. It maintains cerebral control of the lower
bulbo-spinal segments of the latter systems, and some other motor
functions also are represented here. Into it fibers converge from all
other parts of the cerebrum (figs. 12, 14, 15, 17, 18, 20-24), and from
it efferent fibers go out in four groups: (1) Ventromedial tracts go
to the medulla oblongata and spinal cord. The longest of these fibers
are in the f. longitudinalis medialis (fig. 6). (2) The oculomotor nerve
supplies intrinsic and extrinsic muscles of the eyeball (figs. 22, 24).
Associated with these peripheral fibers are central connections with
the nuclei of the IV and VI nerves, so arranged as to execute conjugate movements of the eyes. The details of the apparatus employed
are unknown. (3) Visceral sensory and gustatory fibers enter the
peduncle (fig. 8), and with these are related efferent fibers to the hypothalamus and to lower levels of the motor zone, The pathways
taken by the latter in the amphibian brain have not been clarified.
(4) From both ventral thalamus and peduncle, fibers diverge to
various surrounding parts, notably to the hypothalamus and isthmic
tegmentum. These probably provide for co-ordination of various
local reflex activities with the basic peduncular functions.
At the ventrolateral border of the peduncle there is an area of
superficial neuropil, which is the terminus of the basal optic tract,
large secondary and tertiary visceral-gustatory tracts, some fibers of
the f. retroflexus, and fibers from several other sources (figs. 22, 23, 24). This is an undifferentiated primordium of the basal optic nucleus
and some other structures of the mammalian brain (pp. 35, 221).
It is related with the olfacto-visceral functions of the hypothalamus
and probably also with conditioning of the fundamental peduncular
There are anterior and posterior sectors of the ventral thalamus
which differ in embryological origin (p. 239) and in certain connections of intermediate-zone type. Both sectors are here included in the
motor zone because their chief efferent connections resemble those of
the "peduncle," of which the posterior part is physiologically an
anterior extension. The ventral thalamus is the primordium of the
motor field of the mammalian subthalamus. The anterior sector contains a nucleus specifically related to the stria medullaris and
amygdala and, above this, the eminentia thalami, which is a bednucleus of tracts related to the primordium hippocampi (chap, xviii;
figs. 16, 17, 19, 20, 96).
The ventral thalamus and peduncle of urodeles form a single massive column, which is anatomically well defined. The specialized
structures derived from it in mammals are dispersed among large
masses of tissue of more recent phylogenetic origin; but in the human
brain this region still retains cerebral control of the primordial coordinated movements of the musculature of the eyeballs and of the
trunk and limbs.
In early embryonic stages the retina is part of the brain, and, as
development advances, it absorbs much of the diencephalic sector of
the early neural tube. This precociously accelerated development results from the dominance of vision in exteroceptive adjustment from
the time that the larva begins to feed. For further details of this
development and of the organization of the visual-motor apparatus
see chapter xvi.
12. IIABEN(jL.\
As described in chapter xviii, this specialized part of the epithalamus receives fibers from almost all parts of the telencephalon and
diencephalon and from the tectum (fig. 20). The habenular commissure connects the two habenulae, and it also contains two commissures of pallial parts of the hemispheres — commissura pallii posterior
and com. superior telencephali. The chief efferent path from the habenula is the f. retroflexus (chap, xviii), which terminates in the
cerebral peduncle and interpeduncular nucleus. In the brains of lower
vertebrates the habenular complex is one of the most widely spread
and physiologically important members of the central correlating
apparatus. Its primary function seems to be to integrate the activities of all parts of the brain that are under olfactory influence with
the exteroceptive functions of the tectum and thalamus in the interest of higher cerebral control of the feeding reactions of the skeletal
In the large preoptic nucleus and hypothalamus, olfactory connections dominate the picture, as they do in the habenular system; but
here the nonolfactory functions represented are interoceptive instead
of exteroceptive. All parts of the cerebral hemisphere are connected
with the hypothalamus by fibers passing in both directions in the
medial forebrain bundles (p. 273), stria terminalis (p. '^55), and
fornix (p. 254) systems. The visceral-gustatory afferent paths are
shown in figure 8. Large tracts from the thalamus and tectum also
end here, so that all kinds of sensory experience of which the animal
is capable are represented in the hypothalamus. This experience is
here organized in terms of visceral responses. The efferent tracts go
to the peduncle, tegmentum, interpeduncular nucleus, and descending fibers in the deep neuropil which are precursors of the f. longitudinalis dorsalis of Schiitz. There is a large tract to the hypophysis
for nervous control of endocrine activity. There is also evidence that
some neuro-endocrine functions are performed in the hypothalamus
itself (Scharrer and Scharrer, '40). The structure of the hypothalamus has been described ('21a, '27. ''^5a, '36, '42, and in the embryological papers, '37-'41). It is similar to that of Necturus, of which more
detailed descriptions have been published ('336, '346). For the composition of the postoptic commissure see chapter xxi.
The primordial corpus striatum occupies the thickened ventrolateral wall of the cerebral hemisphere and, like all the rest of the
hemisphere, is under olfactory influence. This is stronger at its anterior and posterior ends. Anteriorly, it is divided by a striatal sulcus
into dorsal and ventral parts (fig. 99), and posteriorly it is much enlarged as the amygdala (figs. 1, 96, 97), which has the typical mammalian connections (fig. 19).
The ventral sector of the anterior olfactory nucleus is interpolated
between the olfactory bulb and the corpus striatum, as in Necturus
(figs. 6, 86). This is the primordium of the tuberculum olfactorium
('27, p. 290), which is enormously enlarged in the lungfishes (Rudebeck, '45). Posteriorly of this nucleus is a poorly defined field, which
embraces the ventral angle of the lateral ventricle and is regarded as
the probable precursor of the head of the caudate nucleus (fig. 99)
It is intimately connected with the rest of the striatum and the
septum. The chief efferent path is the olfacto-peduncular tract to the
dorsal hypothalamus, ventral border of the peduncle, and interpeduncular nucleus.
The middle sector of this complex is the undifferentiated primordium of the mammalian lentiform nucleus, as shown by its
fibrous connections. It is characterized by very dense, sharply circumscribed neuropil in the white substance (p. 96; figs. 98, 99, 108,
109) and has the typical striatal connections with the overlying pallium and the thalamus, including a small sensory projection tract
from the dorsal thalamus (fig. 15, tr.th.f.). The chief efferent path is
the lateral forebrain bundle (f. lateralis telencephali, /.Za^./.), which
contains strio-thalamic, strio-peduncular, and strio-tegmental fibers
comparable with the corresponding components of the mammalian
extrapyramidal system. There is also a strio-tectal and strio-pretectal
connection (figs. 11, 14, 101). The separation of the lentiform nucleus
into globus pallidus and putamen is incipient (p. 97).
The septum complex occupies the ventromedial wall of the hemisphere between the anterior olfactory nucleus and the lamina terminalis and hippocampal area (figs. 4, 98, 99) . Its position and connections are similar to those of mammals. It is directly connected
with the olfactory organ by the nervus terminalis, and it receives
fibers from the olfactory bulb, anterior olfactory nucleus, pallium,
and hypothalamus. The chief efferent paths are by the medial forebrain bundle (f . medialis telencephali, f.med.t.) and to the overlying
pallium by the f. olfactorius septi ('27, p. 291). There is also a broad
connection across the ventral aspect of the hemisphere with the
amygdala and the piriform area, the diagonal band of Broca (figs. 96,
97, 98, d.b.), and a connection with the habenula by tr. olfactohabenularis anterior and tr. septo-habenularis (chap, xviii).
The olfactory bulb is very large, embracing the anterior end of the
lateral ventricle and extending back in the lateral wall for about half
the length of the hemisphere (figs. 1, 3, 4, 85, 100, 105, 110; '246, '27).
All peripheral olfactory fibers end in the glomeruli of the bulb.
Fibers of the second order pass in large numbers to the anterior
olfactory nucleus, and they enter longer olfactory tracts with wide
distribution (fig. 6), The olfactory tracts are mixtures of fibers from
the bulb and the anterior nucleus, as in Necturus ('336, figs. 6 16).
They reach all parts of the cerebral hemisphere, the habenula, and
the hypothalamus. Some of these decussate in the ventral part of the
anterior commissure and some in the habenular commissure (com.
superior telencephali).
The histological texture of the olfactory bulb is more differentiated
than that of Necturus ('31), but more generalized than that of higher
vertebrates. I have contrasted this with the mammalian pattern and
added a theoretic interpretation of probable differences in physiologic
properties of the tissue ('246). In brief, this tissue is interpreted as
illustrating several transitional stages in the differentiation of polarized nervous elements from an unpolarized or incompletely polarized
matrix. In Necturus ('31) the transitional character of this tissue is
still more clearly evident. The granule cells, in particular, give no
structural evidence of physiological polarity, i.e., of differentiation
of dendrites from axon, though the connections of these cells in
Amblystoma suggest that they have a transient and reversible polarity. In connection with this description ('246, pp. 385-95) there are
some speculations regarding possible phylogenetic stages in the differentiation of permanently polarized neurons from an unspecialized
nonsynaptic nerve net or neuropil.
In Amblystoma there is a moderately developed accessory olfactory bulb, but no other evidence of local specialization in the primary
olfactory center. (There are hints of this in some mammals, e.g., the
mink, Jeserich, '45, and references there cited). In 1921, I described
the peripheral and central connections of the accessory bulb of
Amblystoma and compared these with the more specialized structures of the frog. The anatomical connections there described are, I
believe, correct, but the theoretic interpretation of the relationships
in vertebrates generally between the vomeronasal organ, accessory bulb, and amygdala is less secure and awaits confirmation or correction.
The anterior olfactory nucleus is a zone of relatively undifferentiated cells interpolated between the bulbar formation and the more
specialized areas posteriorly of it (figs. 6, 86B and C, 105, 109; '27,
p. 288). In higher animals much of this tissue seems to be specialized
and added to the adjoining fields ('24(/). A very large proportion of
the fibers of the olfactory tracts, arising from both the bulbar formation and the anterior nucleus, are assembled in a dense superficial
sheet of fibers in the medial sector of the anterior olfactory nucleus,
which I have named the "fasciculus postolfactorius" (figs. 5, 100,
105, 110, /./JO.). Here these fibers take a vertical course and then are
distributed to all the olfactory tracts. In chapters vii and xix there
is further discussion of the significance of the olfactory system in the
morphogenesis of the hemisphere.
The pallial part of the hemisphere can be distinguished from the
stem part, though there is no laminated cortex. There are three sectors (figs. 96-99) — the dorsomedial primordium hippocampi {p.hip.),
the dorsolateral primordium piriforme {p.pir., or nucleus olfactorius
dorsolateralis, nuc.oLd.L), and between these a dorsal sector of uncertain relationships (p.p.d.). The gray, as elsewhere in these brains,
is confined to a thick periventricular layer except in the hippocampal
sector, where the cell bodies are dispersed through the entire thickness of the wall and are imbedded in dense neuropil. This is evidently
a first step toward differentiation of superficial cortex. The homologies of the hippocampal and piriform sectors with those of mammals
are clear, as shown by substantially similar nervous connections.
Further discussion will be found in chapter vii and the references
there given.
Throughout the length of the central nervous system all parts of
the two sides are broadly connected by systems of commissural and
decussating fibers. These are in two series, dorsally and vent rally of
the ventricles. Their composition is summarized in chapter xxi, with
references to more detailed descriptions. In the aggregate they make
provision for the co-ordinated action of the motor organs on right and
left sides of the body.
The preceding outline of a regional analysis is framed in very general terms. The evidence upon which it is based is assembled in Part
II of this work and references there given. This evidence, though far
from complete, is regarded as adequate for the anatomical arrangements described. The physiological inferences drawn from these arrangements and the general theory expressed in the following chapters rest on a less secure basis. The correctness of these conclusions
can be tested experimentally, and the hope that this will be done has
motivated the labor expended upon this program of histological
THE brain of Amblystoma performs three general classes of functions, with corresponding local differentiation of structure. We recognize, accordingly, three zones on each side — a dorsal receptive
or sensory zone; a ventral emissive or motor zone; and, between
these and infiltrating them, an intermediate zone of correlation and
Figures 4 and 5 are sketches of longitudinal sections of the cerebrum of adult Amblystoma tigrinum to illustrate the areas included
in the motor and sensory zones as here arbitrarily defined. The zones
of the medulla oblongata as seen in transverse section are shown in
figure 9. The sensory zone includes those parts of the brain which
receive afferent fibers from the periphery, together with more or less
closely related tissue of correlation. The motor zone includes those
parts from which efferent fibers go out to the periphery, together
with related apparatus of motor co-ordination. Both these zones
contain some areas which, though not directly connected with the
periphery, are nevertheless primarily concerned with specific types
of sensory or motor adjustment. What is left over is assigned to the
intermediate zone, and whether a particular area will be included
here depends on one's estimate of its preponderant physiological
character. The body of the cerebellum and the pallial part of the
cerebral hemispheres are excluded from the zones as supra-segmental
The lines drawn in this analysis are frankly arbitrary, chosen primarily for convenience of description; but, as will appear, this functional analysis contributes to an understanding of the meaning of the
structure, and, moreover, it has morphological justification as well.
These zones are not autonomous units when viewed either structurally or physiologically. Their interconnections are intimate and
complicated. The more important of these connections are shown in
a number of diagrams, some in this contribution and some in previous
This analysis of the more obvious structural features of an amphibian brain in terms of physiological criteria is an artificial schematization of a complicated fabric, the several parts of which are so
intimately connected that there is an over-all integration of their
activities. The main highways of through traffic have been mapped,
with signboards pointing the way at the crossroads. Selected examples of some of the lines of fore-and-aft through traffic are illustrated in the diagrams ; but this kind of analysis does not take us to
our destination. It does not tell us how mixed traffic is actually
sorted out and so reorganized as to give the body as a whole efficient
adjustment to the momentarily changing exigencies of common experience. These problems are discussed in subsequent chapters, but,
first, the schematic outline will be summarized here.
Each zone is structurally diversified, and many of these local differentiations are directly correlated on the sensory side with the
modalities of sense represented in the end-organs with which they
are connected and on the motor side in a similar way with synergic
systems of muscles. Each sensory and motor system of nerves has a
local primary central station in direct connection with the periphery,
and each of these stations has widely spread connections within its
own zone and with other zones, thus insuring efficient correlation of
sensory data, co-ordination of motor responses, and integration of
the action system as a whole. In this summary the sensory systems
are given special attention because these are the most useful guides
in the analysis of the structure of this brain.
The sensory zone is defined as those parts of the brain that receive
fibers from peripheral organs of sense. In some fields the number of
such fibers is large, in others it is very small; and some parts of the
brain, like the cerebral peduncle, have both sensory and motor
peripheral connections. Within the sensory zone there is a complicated apparatus of correlation, and in lower vertebrates the receptive
areas and surrounding tissues dominated by them are so large that
most of the mass of the brain can be blocked into fields appropriately
designated "nosebrain," "eyebrain," and so on ('31a, p. 129). This
feature is due to the fact that in these animals the sense organs are
well developed and highly specialized, but the motor apparatus is more simply organized, chiefly for co-ordinated mass movements.
The sensory zone with its own apparatus of correlation, accordingly,
bulks larger than the motor zone.
Though the peripheral sensory apparatus in Amblystoma differs
from the human in many details, yet the general principles of its
organization are similar; the structural organization of the central
apparatus of adjustment, on the contrary, is so radically different
that comparisons are difficult. Here the several functional systems of
peripheral nerve fibers enter the brain in fascicles of the nerve roots,
which are physiologically as specific as are those of mammals ; but at
the first synapse this specificity may almost completely disappear, in
so far as it has visible structural expression. The root fibers of all
sensory systems (except, perhaps, the olfactory) terminate by wide
arborizations in a few common fields of neuropil, in each of which
several of these systems are inextricably mingled. This neuropil is a
common synaptic pool for all entering systems. The several pools are
interconnected with one another and with similar pools in other
parts of the brain stem, and there is no supreme cortical regulator.
The translation of sensory experience into adaptive behavior and the
integration of this behavior are somehow accomplished within this
interplay of the local activities of the brain stem.
The sensory zone is continuous from the dorsal gray colunm of the
spinal cord to the olfactory field, comprising the dorsolateral part of
the medulla oblongata, the auricle in the cerebellar region, the anterior medullary velum and a small amount of contiguous tissue, the
tectum of the midbrain, pretectal nucleus, dorsal thalamus, olfactory
bulb with the adjoining anterior olfactory nucleus, and optionally the
septum and some other parts of the hemisphere, a portion of the
hypothalamus, and the ventrolateral neuropil of the peduncle. The
fields optionally included receive terminals of the nervus terminalis
(p. 267); the hypothalamus has a small but significant connection
with the optic nerve, the basal root of which also connects with the
peduncle. In some vertebrates the epithalamus receives fibers from
the parietal eye, but in Amblystoma these have not been seen, and in
this animal the predominant functions of the "optional" areas are of
intermediate-zone type. The body of the cerebellum and the pallial
part of the cerebral hemisphere might be assigned to the sensory zone
as here defined anatomically; yet, as previously mentioned, they are
excluded from this zone because of their distinctive physiological
characteristics and their remarkable specialization in higher animals.
In the sensory zone of the medulla oblongata there are two elongated synaptic pools of neuropil, into which terminals of the sensory
root fibers converge (chap. xi). One of these receives terminals of all
somatic sensory systems; the other lies more ventrally and internally
in the visceral lobe and receives terminals of the visceral sensory and
gustatory systems (figs. 9, 89). The secondary fibers which emerge
from these pools are distributed locally to the motor zone of the
medulla oblongata, downward to the spinal cord, and upward to
higher levels. The last take different courses, some to the cerebellum,
some to the tectum and thalamus, and some to the hypothalamus.
Each of these pathways discharges into a higher synaptic pool of
neuropil, where its terminals are in physiological relation with terminals of other related sensory systems. The relations to which reference has just been made are in terms of the types of response to be
evoked. Thus the tectum becomes the dominant regulator of somatic
adjustments to exteroceptive stimulation, the hypothalamus becomes the regulator of visceral responses to olfacto-visceral stimulation, and the cerebellum provides regulatory control of the action of
the skeletal muscles. The dorsal thalamus is ancillary to the tectum
and shows a very early stage in the evolution of the ascending sensory projection systems to the cerebral hemispheres.
These local differentiations, each with characteristic structure and
connections, are receptive fields for the several systems of peripheral
sensory fibers, though some of them receive few peripheral fibers and
are concerned chiefly with sensory correlation.
The motor zone as here defined includes the peripheral motor
neurons and those areas of the brain stem concerned with the organization of motor impulses in patterns of synergic action. It includes the following histologically different parts: (1) corpus striatum
(paleostriatum); (2) anterior part of the ventral thalamus; (3) posterior part of the ventral thalamus; (4) nucleus of the tuberculum
posterius ("peduncle" in the restricted sense); (3) isthmic tegmentum; (6) trigeminal tegmentum; (7) a poorly defined tegmental field
extending farther posteriorly through the length of the medulla
oblongata into continuity with the ventral gray column of the spinal
The floor plate of the embryonic neural tube probably ends anteriorly at the fovea isthmi (fig. 2B, f.i.), and the adjacent basal plate, which is the primordial motor zone, extends forward of this to
include the whole of the peduncle and probably more or less of the
hypothalamus and ventral thalamus. This primordial zone contains
not only nervous elements with peripheral connections, like the
sensory zone, but also an elaborate apparatus of central co-ordination
of the neuromotor systems.
Anteriorly of the peduncle the motor zone has no peripheral connections, but the apparatus of motor co-ordination extends forward
through the thalamus into the lateral wall of the hemisphere. Since
the present analysis is based primarily on physiological criteria, this
anterior extension of the motor field is included in the motor zone.
The anterior boundaries of this zone are, of course, arbitrarily drawn ;
that they are artificial is emphasized by the fact that the large basal
optic root terminates in the peduncle, which is in the motor zone as
here defined. Efferent fibers have been described as leaving the brain
in many places outside the motor zone, even as here broadly defined.
Vasomotor and other visceral efferent fibers have been reported in
various animals associated with the nervus terminalis and the olfactory and optic nerves and in other places for distribution to meninges
and chorioid plexuses. We have nothing new to report about Amblystoma in this connection.
In the spinal cord and medulla oblongata the peripheral motor
neurons are so mingled with the co-ordinating neurons of the tegmentum and reticular formation and they are so similar in form that it is
often impossible to distinguish the peripheral neurons except in cases
where their axons are seen to enter the nerve roots. The cells of the
nuclei of the eye-muscle nerves are fairly clearly segregated, and in
some reduced silver preparations they react specifically to the chemical treatment (fig, 104); but even here their dendrites are widely
spread and intertwined with those of tegmental cells, so that both
kinds of neurons would appear to be similarly activated by the
neuropil within which they are imbedded. The cell bodies are locally
segregated; but their dendrites, where most of the synaptic contacts
are made, are not segregated.
In the medulla oblongata the motor tegmentum contains small and
large cells in an endless variety of forms, but these elements are not
segregated in accordance with either size or morphological type. It
is true that the arrangement of their cell bodies may show some
rather ill-defined local segregation, but their dendrites and axons are
so intimately intertwined in the neuropil that nothing comparable with the localized nuclei of higher brains can be recognized. Farther
forward in the isthmus and peduncle the tegmental tissue of coordination is much increased in amount and somewhat more differentiated. In some of my former papers (e.g., '30, p. 76) the term
"nucleus motorius tegmenti" was used loosely (and inaccurately)
to include a tegmental zone defined topographically. This seemed to
be justified in the case of Necturus by the lack of localization of the
large motor elements which characterize this region; but this justification is inadequate, both factually and morphologically — see the
discussion by Ariens Kappers, Huber, and Crosby ('36, pp. 653, 666).
It is obvious that most of the tissue of the motor zone is concerned
with co-ordination of the action of the peripheral elements, so that
synergic groups of muscles are activated in appropriate sequence;
but, with the technic available, it has not been possible to analyze
this complex so as to reveal the mechanism employed. In the medulla
oblongata this organization is chiefly for local control of bulbar and
spinal reflexes, the intermediate zone participating. In the isthmus
and peduncle the number of peripheral elements is relatively small
and the co-ordinating apparatus larger, giving these areas control
over all motor fields spinal ward of them. This intrinsic motor apparatus is supplemented by a segregated band of correlating tissue in
the intermediate zone, the subtectal dorsal tegmentum. In mammals
both these zones are further specialized into separate nuclei distributed in the tegmentum.
The primary patterns of somatic movements are predetermined by
the course of central differentiation within the motor and intermediate zones in premotile stages of development. After connection
with the peripheral musculature is established, each of these muscles
seems to exert some sort of distinctive reciprocal influence upon that
motor field of the central nervous system from which its innervation
is derived. The nature of this influence is unknown, but its reality is
well attested by experiments of Paul Weiss ('36, '41) and colleagues
upon "myotypic response" and "modulation."
In later stages the primary motor patterns may be modified, or
"inflected," by sensory experience and practice. Influence of use or
some other functional factors seem to be essential for maintenance of
motor efficiency, as graphically shown by Detwiler's observations
('45, p. 115; '46) on the behavior of decerebrate larvae of Amblystoma (to which further reference is made on p. 118). In young
larvae of stage 37, swimming movements may be perfectly executed after transection immediately below the auditory vesicle under control of the lower medulla oblongata and spinal cord (Coghill, '26,
Paper VI, p. Ill; '29, p. 15); but, subsequent to Harrison's stage 40,
Detwiler finds that sustained motor activities, including swimming,
fail rapidly if the influence of the midbrain is blocked in prefunctional stages, though feeding reactions are preserved after complete
ablation of hemispheres and visual organs. The midbrain evidently
supplies a factor essential for maintenance of motor efficiency.
The motor field of this brain is smaller and more simply organized
than the sensory field because most of the activities are mass movements of total-pattern type. Within this larger frame of total behavior, the partial patterns of local reflexes are individuated with
more or less capacity for autonomous action. The number of these
local partial patterns is smaller than in higher animals, and all of
them are far more closelj^ bound to the total patterns of which they
are parts. The segments of each limb, for instance, may, upon appropriate stimulation, move independently; but in ordinary locomotion they move in a sequence related to the action of the entire limb,
the other limbs, and the musculature of the trunk.
The peripheral motor nerves (omitting the general visceral components of preganglionic type not here considered) are in three
groups: (1) the spinal nerves; (2) the eye-muscle nerves. III, IV, and
VI pairs of cranial nerves, which are somatic motor; and (3) the
special visceral motor nerves of the V, VII, IX, and X pairs, innervating the striated musculature of the visceral skeleton of the
head. The primary movements of trunk and limbs are organized for
locomotion in the motor zone of the spinal cord. This organization is
under exteroceptive and proprioceptive control locally throughout the
length of the cord and more especial!}^ at the bulbo-spinal junction;
it is under additional proprioceptive control from the labyrinth and
the cerebellum; there is further control from the cerebrum — optic,
olfactory, and the related apparatus of higher correlation. The bulbar
group of special visceral motor nerves is primarily concerned with
movements of the head, notably those of respiration and feeding.
The feeding reactions are under visual, olfactory, somesthetic, gustatory, and general visceral afferent control, and the pattern of performance seems to be organized in the large isthmic tegmentum.
The very large and complicated interpeduncular nucleus is an isthmic
structure which is physiologically of intermediate-zone type (chap.
xiv) . Details of the structure and connections of the various parts of
the motor zone are in Part II, and the pecuhar features of its forward
extension in the cerebral hemisphere are discussed in chapter vii.
The characteristics of this zone are imphcit in the preceding account of the sensory and motor zones. It is more elaborately developed and its boundaries are more clearly defined in parts of the
cerebrum than in the rhombencephalon. These boundaries are necessarily arbitrary, for all parts of the brain are involved in correlation
and integration of bodily activities ; but throughout the length of the
spinal cord and brain there is a band of tissue between the sensory
and motor zones primarily concerned with these adjustments. At
lower levels I have termed this tissue the "reticular formation," and
here it infiltrates the other zones with no clear boundaries (for details
see chap. xi). At higher levels it increases in amount and is more
clearly segregated. It would be appropriate to include in this zone
most of the diencephalon and telencephalon except the specific optic
and olfactory terminals ; but, for reasons mentioned above, a different
subdivision has been adopted, primarily for convenience of description. The dorsal tegmentum, or subtectal area, is a typical representative of this zone in position and physiological connections. In the
isthmic and bulbar tegmentum the characteristics of the intermediate and motor zones are inextricably mingled. The habenula, hypothalamus, and interpeduncular nucleus, as elsewhere described,
clearly belong physiologically to the intermediate zone; and the
whole cerebral hemisphere, except the olfactory bulb, might appropriately be so classified in all Ichthyopsida.
In the most primitive vertebrates the intermediate zone is scarcely
recognizable as an anatomical entity. As the action system becomes
more complicated in higher animals, this zone shows corresponding
differentiation. This specialization is more directly dependent upon
complication of the peripheral motor apparatus than upon sensory
differentiation, for, so long as the action system is largely confined to
mass movements, the patterning of these total activities is effected in
the sensory and motor zones. In tetrapods and birds more complex
central adjustors are required, and these are differentiated between
the two primary zones and anteriorly of them. With the appearance
of more autonomy of the local reflex systems, more efficient apparatus of integration is demanded. The final result is that in the human
brain the apparatus of intermediate-zone type has increased so much that it comprises more than half the total weight of the brain, for
both cerebral and cerebellar cortices are derivatives of this primordial matrix, as will appear in the ensuing discussions.
The preceding physiological analysis of the brain obviously rests
upon the peripheral relations of its several parts. The two primary
functions of the nervous system are, first, the maintenance of the
integrity of the individual, with efficient co-operation of parts among
themselves and with the total organization, and, second, the analysis
of experience and the translation of the sensory data into appropriate
behavior. The peripheral nerves are key factors in both these domains. Our knowledge of the functional analysis of the cranial nerves
has been greatly increased during the last fifty years, largely by the
work of the so-called American school of comparative neurologists,
which I have recently reviewed ('43).
Before I discuss the components of these nerves, a few definitions
are in order. In the attempt to envisage the nervous system from the
operational standpoint, distinctions have been drawn between sensory correlation, motor co-ordination, and those central processes
that provide integration, and some measure of spontaneity of action
which might be grouped under the name "association" ('24c, p. 235;
'31a, p. 35). This classification is necessarily artificial, for all these
processes are interrelated. They interpenetrate, and they are not
sharply localized in the structural fabric. Nevertheless, these several
types of action are recognizable components of the unitary dynamic
system, and there are local differentiations of the structural organization correlated with preponderance of one or another of them, more
clearly so in higher vertebrates than in lower.
Sensory correlation, as the term is here employed, refers to interaction of afl^erent impulses within the sensory zone, that is, within the
field reached by terminals oi peripheral sensory fibers. The interplay
of these diverse afi'erent impulses takes two forms: (1) in fields of
undifferentiated neuropil, the activation of which results in alterations of the central excitatory state or in mass movements of large
numbers of synergic muscles; (2) in more restricted areas (nuclei),
which activate the neuromotor apparatus of local reflexes. The members of both groups are interconnected by systems of internuclear
fibers like the lemniscus systems, all within the sensory zone, so that
all activities of this zone interact one with another. These internuncials are so arranged that functional systems of afferents, which
normally co-operate to effect a particular type of motor response, are
more intimately associated. Thus the tectum opticum receives most
of the lemniscus fibers of all somesthetic systems and minimum numbers of olfactory and visceral systems. This basic pattern as seen in
Ambly stoma is changed in mammals, where higher associational centers have taken over most of the functions of correlation.
Motor co-ordination is effected primarily in the motor zone, which
is so organized as to activate synergic groups of muscles in appropriate sequence with inhibition of their antagonists. This grouping
may be adapted for mass movements or for local reflexes. Internuclear tracts connect the various parts of the sensory zone directly
with appropriate parts of the motor zone. More refined analysis and
conditioning of motor responses are effected through the intermediate zone, and the tissues of the latter group are greatly enlarged and
complicated in higher brains.
The activities of stimulus-response type which have just been considered are so interconnected with internuclear tracts and the interstitial neuropil as to facilitate the integration of all local activities in
the interest of the requirements of the body as a whole. Every local
part of the brain is a component of the apparatus of general integration, and some of these parts have this association as their dominant
function. In Ambly stoma most of this suprasensory and supramotor
tissue is dispersed as interstitial neuropil. In mammals, higher types
of associational tissue have been differentiated locally, notably in the
cerebral cortex and its dependencies, with corresponding enhancement of those synthetic functions which are manifested as conditioning, educability, and reasoning. Parallel with these changes there is
an enormous increase in accumulated reserves of potential nervous
energy, which come to expression as spontaneity, memory, and creative imagination.
A survey of the nerve fibers of Amblystoma as a whole in view of
the principles just expressed shows that they may be classified in four
groups: (1) the peripheral afferent systems and associated internuclear correlating tracts within the sensory zone (lemniscus systems, etc.) ; (2) the peripheral efferent systems and the related coordinating fibers of the motor zone; (3) the central internuclear systems intercalated between the preceding two and so interconnected
as to yield appropriate responses to ordinary recurring stimuli; and
(4) infiltrating these mechanisms of stimulus-response type, a different sort of adjusting apparatus which insures general integration of
these systems, with provision for conditioning of reflexes and other
forms of individually acquired behavior and for release of accumulated reserves of nervous potential as needed. These four groups
intergrade but, in general, are recognizable.
The peripheral fibers are grouped in functional systems, each of
which is defined as comprising all nerve fibers and related endorgans, which are so arranged as to respond to excitation in a common mode, either sensory or motor. These functional systems are
convenient anatomical units also, for all fibers of each sensory system, regardless of variations in the peripheral distribution of their
end-organs and regardless of the particular nerve trunks or roots
through which they connect with the brain, are segregated internally
and converge into local areas or zones. In higher vertebrates (but less
so in lower) the secondary connections of these terminal stations tend
to retain their physiological specificity. From this it follows that the
peripheral systems of sensory analyzers are extended into the brain
as far as related central pathways are separately localized — in the
human brain it may be even up to the projection areas of the cerebral
cortex. Accordingly, we include in the sensory zone as here defined
not only the terminal nuclei of peripheral sensory nerves but also
their related nervous connections, so far as these are with other parts
of the sensory zone and not directly with the motor zone. The neuromotor apparatus can be similarly analyzed into functional systems, each of which is concerned with the synergic activation of some
particular group of muscles.
This anatomical segregation of the functional systems is not carried to perfection, even in the human nervous system. The various
modalities of cutaneous and deep sensibility, for instance, are not
completely segregated and localized either peripherally or centrally.
Yet this differentiation has gone so far that it provides our most useful guide in the analysis of the structure of the brain.
The activities of the body may be divided into two major groups;
(1) those concerned with adjustment to environment, the somatic
functions, and (2) those concerned with the maintenance and reproduction of the body itself, visceral functions. These, of course, are not
independent of each other; nutrition, for instance, involves somatic
activity in the search and capture of food and visceral activity in its
digestion and assimilation. Nonetheless, these types of function are
so different, especially in the responses evoked, that this strictly physiological criterion marks also the most fundamental structural
analysis of the nervous system. Anatomically, the somatic systems
of peripheral organs and nerves and central adjustors, including the
proprioceptors, are, in general, rather sharply distinguished from the
visceral. The systems are cross-connected by internuclear tracts, and
some sensory systems, like the olfactory, may serve, on occasion,
either somatic or visceral adjustments.
A phylogenetic survey of these systems reveals remarkable plasticity in their interrelationships. Thus, taste buds, which in most vertebrates are typical interoceptors, may in some fishes be spread over
the external surface of the body, where they serve exteroceptive functions, with corresponding changes in the anatomical pattern of the
central apparatus of adjustment (chap, x; '446). On the motor side
the apparatus of feeding and respiration exhibits still more remarkable transformations. In fishes this musculature is connected with the
visceral skeleton — jaws, hyoid, and gill arches — and the functions
performed are obviously visceral, though the larger part of this
musculature is striated. The related parts of the nervous system are
classified as special visceral motor. With suppression of the gills in
higher animals, some of these muscles undergo remarkable transformations. Those which are elaborated to form the mimetic facial
musculature of mammals become physiologically somatic ('22, '43).
The classification of peripheral end-organs and their related nerves
which has proved most useful grew out of the analysis of these nerves
into their functional components, to which reference has just been
made, by histological methods. Serial sections through the entire
bodies of small vertebrates differentially stained for nerve fibers
enable the investigator to reconstruct not only the courses of the
nerves but also the arrangement of the functional systems represented in each of them and to follow these components to their
peripheral and central terminals, a result that cannot be achieved by
ever so skilful dissection. The first successful application of this
method was Strong's analysis of the nerve components of the tadpole
of the frog in 1895, a fundamental research which provided the generalized pattern which prevails, with endless modifications of details,
throughout the vertebrate series, as has been abundantly demonstrated by numerous subsequent studies by many workers.
This was followed in 1899 by my Doctor's dissertation on the nerve
components of the highly specialized teleost, Menidia. From these and subsequent studies the peripheral nervous system of the head
was analyzed into functional systems as follows:
1. Somatic sensory fibers of two groups. — (1) Exteroceptive systems, including (a) the specialized olfactory (in part), optic, auditory, and lateral-line nerves with differentiated end-organs, and (6)
the nerves of general cutaneous and deep sensibility with simple free
endings, these entering chiefly in the V nerve root with some in the
VII, IX, and X roots. (2) Proprioceptive fibers from specialized endorgans of the internal ear and (probabljO lateral-line organs and also
fibers from muscles, tendons, and other deep tissues. Here belongs
also the peculiar mesencephalic root of the V nerve. See chapter x for
further comments on the proprioceptive system.
2. Visceral sensory fibers of two types. — (1) With specialized endorgans, viz., the olfactory organ (in part) and the taste buds, the
latter entering by the VII, IX, and X nerve roots. (2) Fibers of general visceral sensibility with free endings, entering in the same roots
as the preceding and mingled with them.
3. Somatic motor fibers. — Somatic motor fibers which supply striated muscles derived from the embryonic somites, those in Amblystoma being limited to the nerves of the extrinsic muscles of the eyeball in the III, IV, and VI nerves.
4. Visceral efferent fibers of two types. — (1) Special visceral motor
fibers of cranial nerves supplying striated muscles, not of somitic
origin, related with the visceral skeleton, jaws, hyoid, branchial
arches, and their drivatives (in the V, VII, IX, and X roots).
(2) Preganglionic fibers of the general visceral (autonomic) system,
terminating in sympathetic and parasympathetic ganglia, where they
activate postganglionic fibers distributed to unstriated and cardiac
muscles and glands (in the III, VII, IX, and X roots). The last system is not further considered in this work. For application of this
analysis to the human nervous system see my Introduction to Neurology ('31a, chaps, v and ix).
This analysis has yielded our most useful clues for resolution of the
complexity of both peripheral nerves and brain. Descriptions of the
peripheral end-organs and the courses of the nerves do not lie within
the scope of this work. Some of these details which are significant for
understanding their central connections are included in chapter x.
IN A primitive brain like that of Amblystoma the stable framework
of localized centers and tracts performs functions that are primarily analytic. The sense organs are analyzers, each attuned to
respond to some particular kind of energy. The sensory systems of
peripheral nerves and the related internal sensory tracts are parts
of the analytic apparatus, in so far as their functional continuity
with the peripheral organs of the several modalities of sense can be
On the motor side similar conditions prevail. The neuromotor
apparatus is organized in functional systems, each of which is
adapted to call forth the appropriate sequence of action in a particular group of synergic muscles. These units are as truly analyzers
as are those of the sensory systems, though in an inverse sense. Out
of the total repertoire of possible movements, those, and only those,
are selected which give the appropriate action. The efferent fibers are
grouped, the members of each group being so bound together by
central internuclear connections that they act as a functional unit
adapted for the execution of some particular component of behavior,
such as locomotion, conjugate movements of the eyeballs, seizing and
swallowing food, and so on.
The several sensory systems are so interconnected within the sensory zone as to react mutually with one another. They form a
dynamic system so organized that all discharges from this zone are
resultants of this interaction. This interplay has pattern. The various
modalities of sense are not discharged into a single common pool of
equipotential tissue. The sensory components of the nerves are segregated, more or less completely, so that related systems converge into
dominant centers of adjustment— exteroceptors in the tectum, proprioceptors in the cerebellum, olfacto-visceral systems in the hypothalamus, olfacto-somatic systems in the habenula, and so on.
A review of the internal architecture of the adult brain of Amblystoma suggests that the specifications of the general plan are drawn in terms of current action. The elaboration of the analytic apparatus
on the sensory side is carried only so far as is requisite to facilitate
responses to any combination of sensory stimuli in patterns determined by the appropriate use of such motor equipment as the animal
possesses. In species with simpler action systems the central analytic
apparatus is less elaborate; in species endowed with more complicated motor organs the central architecture is more elaborate. In all
species the peripheral sensory equipment determines the architectural plan of the primary centers of the sensory zone; internally of
this level the details of the plan are shaped by two additional factors :
first, the motor equipment available and, second, the amount and
quality of the apparatus of correlation and integration required for
the most efficient use of such sensory excitations as the animal experiences. The cross-connections between the sensory and motor
zones are quite direct and simple in early embryological stages, so
arranged as to provide uniform stereotyped responses to oft recurring
situations. But as development advances these connections become
more and more complicated, an intermediate apparatus of correlation
is interpolated, and, correlated with this, the behavior becomes more
diversified and unpredictable.
In the sequence of development of behavior patterns this change
can be accurately dated. For instance, in Amblystoma between the
early swimming and early feeding stages, at about Harrison's stage
40, the swimming movements, which in younger stages are perfectly
co-ordinated by the bulbo-spinal central apparatus alone, lose this
autonomy, and participation of the midbrain is essential for the
maintenance of efficient swimming, as was mentioned on page 62 in
describing experiments by Detwiler ('45, '46). It is during this period
that tecto-bulbar and tecto-spinal connections of essentially adult
pattern are established ('39, p. 112). In human fetal development
there is a similar critical period at about 14 weeks of menstrual age
(Hooker, '44, p. 29). At this time the upper levels of the cerebrum
acquire functional connections with the lower brain stem, and the
behavior shows a corresponding change. "The fetus is no longer
marionette-like or mechanical in the character of its movements,
which are now graceful and fluid, as they are in the new-born."
Synthesis and integration may be effected in various ways. The
most evident nervous structures employed here are the internuclear
tracts which form a complicated web of conductors, which interconnect the analytic units with one another so that the entire complex forms an integrated equilibrated system. This is the apparatus
of the stable heritable components of the action system — the reflexes
and instincts. A second integrating apparatus is found in the allpervasive neuropil, and a third in specialized derivatives of the latter,
the associational tissues locally differentiated in the brain stem and
reaching maximal development in the cerebral cortex.
The total behavior of neuromuscular type emerged within a preexisting bodily organization, which maintained the unity of the individual by nonnervous apparatus. The nervous system is from its first
appearance a totalizing apparatus. Local differentiations of tissue
for the analysis of sensory experience and of motor responses arise
within this integrated structure, and local reflexes similarly emerged
within a total neuromuscular pattern of action adapted to maintain
the unity of the organization. As development advanced, the mechanisms of the local reflexes acquired increasing autonomy, but they
are never completely emancipated from some control in the interest
of the behavior of the body as a whole. The organic unity of the
whole is preserved while local specificity is in process of development,
and this unitary control is never lost during the normal life of the
The stimulus-response formula has wide application and great
usefulness as a basic concept in physiology and psychology, but its
apparent simplicity is illusory and has tended to divert attention
from essential features of even the simplest patterns of behavior.
This I have illustrated ('44a) by an examination of the simplest reflex
connection known in Amblystoma — from retina to ocular muscles by
way of the basal optic tract.
The late G. E. Coghill, during a productive period of foi'ty years,
studied the development of the action system of Amblystoma and
the correlated processes of bodily growth. These researches have
demonstrated beyond question that in this animal the neuromuscular
system is so organized in prefunctional stages that, when first activated from the sensory zone, the resulting movement is a total response of all the musculature that is mature enough to respond to
nervous excitation. These "total patterns" of activity are not disorderly, and they become progressively more complicated while the
apparatus of local reflexes ("partial patterns") is slowly differentiated within the larger frame of the total pattern. The development
of both the total pattern and the partial patterns is initiated centrally, and throughout Hfe all of them are under some measure of unified central control so that the body acts as an integrated whole with
diverse specialization of its parts (Coghill, '29; Herrick, '29). Coghill's contributions of factual observations and the principles derived
from them have been critically reviewed by the writer in a book
('48), to which the reader is referred.
The patterning of these orderly movements is determined by the
intrinsic structure of the nervous system. This structural pattern is
not built up during early development under the influence of sensory
excitations, for in the embryo the motor and sensory systems attain
functional capacity independently of each other; and when central
connection between the sensory zone and the motor zone is made, the
first motor responses to excitation exhibit an orderly sequence, the
pattern of which is predetermined by the inherited organization then
matured (Coghill '29, p. 87; '30, Paper IX, p. 345; '31, Paper X,
pp. 158, 166). This early structural differentiation goes on independently of any stimulus-response type of activity, though the
latter may modify the pattern of subsequent development. This is a
principle of wide import, applicable not only in lower vertebrates but
in higher forms as well (Coghill, '40), including man (Hooker, '44).
The stimulus-response mechanism is not a primary factor in embryogenesis; it is a secondary acquisition.
It has been pointed out that the functions of the sensory and
motor zones are fundamentally analytic — analysis of environmental
influences and analysis of performance in adjustment to those influences. How the units of the analytic apparatus are actually related so
as to insure the appropriate correlated action of the separate parts is
the key problem, which must be resolved before animal (and human)
behavior can be approached scientifically in other than a descriptive
way. Good progress has been registered. The sensory and motor
analytic apparatus has been exhaustively studied and well described;
and this was the appropriate place to begin, for these organs are most
accessible to observation and experiment. Because these systems of
peripheral end-organs and the related pathways of conduction and
centers of control are, in the human nervous system, obviously interconnected in stable and definitely localized patterns, it was natural
to use this structural framework as the point of departure in the
elaboration of the hypothetical superstructure of current doctrines of reflexology. But reflexes can be conditioned, and this name for a
well-known physiological fact is for the neurologist scarcely more
than a symbol of complete ignorance of the mechanisms actually
The several reflexes have been so closely colligated with specific
details of central architecture that the reflex arc came to be regarded
as the primary unit of nervous organization, and it was assumed that
the increasing complexity of the upper levels of the brain in higher
vertebrates has been brought about by progressively more intricate
interconnections among these elementary units. The integrative action of the nervous system was conceived in terms of the definition of
mathematical integration — "the making up or composition of a
whole by adding together or combining the separate parts or elements." This conception leaves unexplained how any additive process of this sort can result in such a unique centrally controlled unitary
organization as we actually observe, capable of conditioning the
reflexes in terms of individual experience (learning), of abstracting
some common features of mixed experience and synthesizing these
into quite original patterns of response, and of maintaining some
measure of "spontaneous" or self-determined directive control.
A far more serious charge against traditional doctrines of reflexology is the observed fact that in the development of Ambly stoma
the early responses to external stimulation are not local reflexes but
total movements of the entire available musculature. The integrated
total pattern precedes in time the appearance of the partial patterns.
These are individuated within the total pattern; they are integral
parts of it, and for an appreciable time they are subordinate to it.
Even in the adult animal the local partial patterns are not completely
emancipated from control by the body as a whole. It is, indeed, impossible to find in this brain any sharply defined, well-insulated reflex
What happens during the emergence of specific reflexes from the
total reactions is, first, the development of an increasing number of
collateral branches of the primary axons and the central linkage of
sensory and motor pathways in ever more complicated patterns.
Then, second, in the adjusting centers additional neurons are differentiated, the axons of which take longer or shorter courses, branching freely and participating in the formation of a nervous feltwork of
extraordinary complexity. These neurons are not concerned primarily with specific reflexes but with the co-ordination and Integration of all movements. Some parts of this intricate fabric, generally
witli thicker fibers, more or less well fasciculated, activate mass
movements of primitive type, and other parts control local reflexes
as these are individuated. But these systems of fibers are not segregated in comjjletely insulated reflex arcs. They are interconnected by
collateral branches with one another and with the interstitial neuropil. There are lines of preferential discharge, but whether any one
of them is actually fired depends on numberless factors of peripheral
stimulation and central excitatory state.
The phylogenetic history is parallel. The further down we go toward the primitive ancestral vertebrates, the less clear evidence do
we find of definitely localized reflex arcs, and the overt behavior
tends more toward mass movements of total-pattern type.
It must be borne in mind that the development of the individual
does not exactly recapitulate the phylogenetic development (Hooker,
'44, pp. 15, 33). The pattern of the sequence of structural changes'
which take place during prefunctional stages of growth is determined
by the organization of the germ plasm and the interaction of the
genetic factors with one another. This organization, in turn, has been
determined during preceding evolutionary history in adaptation to
the environment and habitus of the species in question. In broad
lines the history of ancestral development is repeated in the growth
of the embryo, but cenogenetic modifications of it may appear in
adaptation to changing conditions, as illustrated, for instance, by the
appearance of some local reflexes earlier in mammals than in amphibians.
The structural organization of the brain sets off in sharp relief a
few important general physiological principles. First, it is to be noted
that the "resting" nervous system is not inert. The body acts before
it reacts. There is always some spontaneous — that is, centrally excited — activity, and the importance of this factor increases as we
ascend the phylogenetic scale. There is always intrinsic activity, as
demonstrated, for instance, by the Berger rhythms, and it is always
acted upon by numberless extrinsic agencies. When an excitation is
received from the periphery, there results a change in the central
excitatory state both locally and diffusely, which involves both activation and inhibition.
Another general principle may be mentioned here. The flow of
nervous impulses from receptor to effector is not one-way traffic.
The excitation of a peripheral sense organ may be followed by an
efferent discharge back to the receptor. An instructive illustration of
this is seen in the auditory apparatus of mammals. Excitation of the
cochlea is followed by an efferent return to the tensor tympani and
stapedius muscles and also to the cochlea itself (the latter pathway
recently demonstrated by Rasmussen, '46). Almost all contracting
muscles report back to the center by a system of proprioceptive
fibers. The central nervous system is full of similar reciprocating systems. Many of the fasciculated tracts of Ambly stoma are two-way
conductors, transmitting in both directions, and there are numberless illustrations of a circular type of connection, efferent fibers of one
center activating another, which has a return path, perhaps by a
devious route, back to the first center. A neuropil may be interpolated in any of these types of circuit. The thalamo-cortical connections of the human brain are of this sort, exhibiting what Campion
and Elliot Smith ('34) have aptly named a "thalamo-cortical circulation," a circulation not of blood but of nervous transmission. All
parts of the cerebral hemispheres are in similar reciprocal interconnection, as has recently been emphasized and illustrated by Papez
Here reference may be made to Dewey's ('96) stimulating analysis
of the reflex-arc concept or, as he prefers to say, the "organic circuit"
concept. This he elaborated in terms of psychology, and nearly
twenty years later I made this comment about it ('13«) :
"Let us see how it may be applied to biological behavior. The simple reflex is
commonly regarded as a causal sequence: given the gun (a physiologically adaptive
structure), load the gun (the constructive metabolic process), aim, pull the trigger
(application of the stimulus), discharge the projectile (physiological response), hit
the mark (satisfaction of the organic need). All of the factors may be related as
members of a simple mechanical causal sequence except the aim. For this in our
illustration a glance backward is necessary. An adaptive simple reflex is adaptive
because of a pre-established series of functional sequences which have been biologically determined by natural selection or some other evolutionary process. This gives
the reaction a definite aim or objective purpose. In short, the aim, like the gun, is
provided by biological evolution and the whole process is implicit in the structurefunction organization which is characteristic of the species and whose nature and
origin we need not here further inquire into The aim (biological purpose) is
so inwrought into the course of the process that it cannot be dissociated. Each step
is an integral part of a unitary adaptive process to serve a definite biological end, and
the animal's motor acts are not satisfying to him unless they follow this predetermined sequence, though he himself may have no clear idea of the aim. These reactions are typically organic circuits Always the process is not a simple sequence of distinct elements, but rather a series of reactions, each of which is shaped
by the interactions of external stimuli and a preformed or innate structure which has been adapted by biological factors to modify the response to the stimuli in
accordance with a purpose, which from the standpoint of an outside observer is
teleological, i.e., adapted to conserve the welfare of the species."
This apparent teleology is commented upon in chapter viii. Since
the passage just quoted was written, control of gunfire by radar has
been perfected, thus reinforcing our analogy at one weak spot. In
the reflex the "aim" does not precede the stimulus that pulls the
trigger; it is automatically adjusted to the stimulus as in radar. But
this automatism in both cases is dependent upon the presence of a
preformed structure adapted to provide it.
Our analysis of the adult structure of the brain of Amblystoma
confirms and supplements the conclusions reached by Coghill from
his study of the development of the same species. His major contribution, as I see it, was the demonstration of the primacy of the
integrative factors in the development of behavior patterns and of
some of the features of structural growth during the individuation of
local partial patterns within the larger total pattern. The adult
structure of the brain of Amblystoma is in perfect conformity with
the conclusions to which he was led. One of these conclusions should
receive special emphasis here, for it clarifies our conception of what
the reflex is in general, and in particular it helps us over some hard
places in our attempt to discover the actual mechanisms involved in
the individuation of local reflex patterns within the frame of the
total pattern.
In the central resolution of forces which eventuates in some particular pattern of overt movement there is always an inhibitory factor (Coghill, '36, '43). In discussing the individuation of partial patterns (local reflexes) from the total pattern, he wrote ('40, p. 45):
"Individuation is obviously the result of organized inhibition
The major division of the total pattern must be under inhibition
when a part acquires independence of action, and the same part can
be inhibited while the major segment of the total pattern acts. So
that the whole individual probably acts in every response, either in
an excitatory or inhibitor}^ way," This he generalized in the following
statement ('30, p. 639):
"For an appreciable period before a particular receptor field acquires specificity
in relation to an appropriate local reflex its stimulation inhibits the total reaction.
Inhibition, accordingly, through stimulation of the exteroceptive field, begins as a
total pattern. It is then in a field of total inhibition that the local reflex emerges.
The reflex may, therefore, be regarded as a total behavior pattern which consists of two components, one overt or excitatory, the other covert or inhibitory. The essential anatomical basis for this is (1) in the mechanism of the total pattern of action,
or primary motor system, and (2) in the mechanism of the local reflex, or secondary
motor system; the mechanism of the total pattern being inhibited and that of the
reflex excited. But since inhibition is not a static condition but a mode of action,
the mechanism of the total pattern must be regarded as participating in every local
This conception of the reflex as involving a factor of inhibition of
the total pattern Hnked with excitation of the partial pattern is
fruitful. Total inhibition plays a more obvious role in the overt behavior of amphibians than in most other animals, not only in embryogenesis of behavior but also in the adult. This was emphasized
by Whitman ('99) in his classic description of the behavior of Necturus. In my manuscript notes of a conference with Dr. Coghill on
January 1, 1929, I find a record of his remarks which is here transcribed.
"The first neurons to differentiate in Amljlystoma are in the floor-plate. These
and others adjacent form the primary motor column, the dominant function of
which is activation of muscles of the same side for mass movement of the trunk and
limbs and inhibition of the musculature of the opposite side which is in the same
phase of locomotor movement. In later stages, when mechanisms of specific local
reflexes emerge, residual neurons in the region of the floorplate maintain their functional importance for mass movements as activators of the whole somatic motor
apparatus. They may prime this neuromotor system, putting it into a subliminal
excitatory state in advance of its patterned activation.
"At an age which immediately precedes the first feeding reactions and before it is
possible to open the mouth and swallow, the larva will react to a moving object in
front of the eyes by a total reaction, a leap forward. It cannot seize the object. The
general activator mechanism here comes to overt expression before the specific local
reflex patterns are mature enough to function. After feeding activities have matured
there is a similar general activation, accompanied by inhibition, as illustrated by the
'regarding' reaction [p. 38]. A larva which had been feeding for several days was
stimulated by moving a hair slowly across the field of vision. The animal responded
by moving the head slowly following the hair. The head is bent to the side, with
rotation of the eyes, movement of the fore limbs, and adduction of both hind limbs.
When the hair was not too far distant, the animal finally, at the end of this 'regarding' reaction, jumped after it. Here there is a clear distinction between what Sherrington calls the anticipatory phase and the consummatory phase of the reaction,
and evidently in the anticipatory phase inhibition plays the major role. This is
obvious also in almost all adult behavior of these animals."
The mechanism of central inhibition is still obscure. There is some
evidence that a nervous impulse impinging upon a central neuron
may, on occasion, activate the element, or under other conditions of
central excitatory state, strength, or timing of the afferent flow it
may inhibit activity ui process. Whether or not this is true, it is well
known that a central neuron may exhibit a large variety of types of synaptic junctions, differing in histological structure, electrical properties, and perhaps also in chemical reactivity. These afferent fibers
may come from widely separated regions with diverse functions, and
the impulses delivered may differ in intensity and temporal rhythm.
Bodian's description ('37, '4'2) of axon endings on Mauthner's cell
of the medulla oblongata shows four main types of synaptic contact
which vary from 0.5 to 7/i in extent, with a wide variety of arrangements. There are between four and five hundred of these endings on a
single cell, and the presumption is legitimate that these diverse
structures are correlated with significant differences in electrical and
chemical properties, including the timing of the pulses of transmission. It has been suggested that some of the influences transmitted
across the synaptic junctions are excitatory and that others are inhibitory. Synaptic junctions on dendrites are in some cases structurally different from those on the axon hillock or axon, and they
may be activated from different sources. Some observers believe that
excitation of dendrites is excitatory and of axons is inhibitory, a
supposition supported with physiological evidence by Gesell and
Hansen ('45, p. 156). In their theory of the electronic mechanism of
activation and inhibition, these functions are viewed as basically
similar, activation being associated with an increasing, and inhibition
with a decreasing, intensity of the electronic current. The connections of horizontal cells of the retina as described by Polyak ('41, p.
385) suggest to him a different inhibitory apparatus. The horizontal
cells may exert an inhibitory influence upon the synapses between
the rods and cones and the bipolar cells, that is, the synapses of the
horizontal cells may function as "countersynapses" to the photoreceptor-bipolar synapses.
Whatever may be the mechanism employed in central inhibition,
it is clear that in some parts of the brain excitatory functions predominate, in other parts inhibitory functions. Noteworthy examples
of the latter are (1) the head of the caudate nucleus (Fulton, '43, p.
456) ; (2) a region in the reticular formation of the medulla oblongata
explored by Magoun ('44) ; and (3) certain specific zones of the cerebral cortex (areas 4^, ^s, 19s, and some others) which are known as
"suppressor bands." In all these cases, excitatory and inhibitory
fields are intimately related physiologically in such a way as to secure
appropriate balance of activation and inhibition of the members of
synergic systems of muscles in proper sequence.
The role of general inhibition in the patterning of behavior has been under investigation for several years by Beritoff and his colleagues. The first half of the fifth volume of the Transactions of his
institute is devoted to studies on the nature of general inhibition and
its role in the co-ordination of cortical activity and reflex reactions
of the spinal cord. Beritoff believes that the neuropil possesses an inhibitory function — slow changes in voltage, expressing the active
state of the neuropil, show an anelectrotonic effect on the cellular
bodies, lowering excitability in them and weakening the excitation.
The evidence is drawn from both somatic and visceral stimulation.
He writes ('43, p. 142) : "Thus, during each reflex reaction in the
visceral organs, taking place in response to a stimulation of the interoceptors and of visceral afferent fibers, just exactly as during
somatic reflexes, the spinal cord acts as a whole, making the given
reflex local and every spinal reflex reaction entire by means of general
inhibition." This is essentially the same as Coghill's position as stated
in the preceding quotations. In other articles in the same volume the
role of the neuropil in a great variety of spontaneous and stimulated
activities of the brain is emphasized by Beritoff.
The neuropil as a whole is not, in my view, a specific inhibitor. It
may partipciate on occasion in either excitation or inhibition, and in
the inhibitory phase it acts as part of the covert component of the
reflex or of mass action, as the case may be, in Coghill's analysis as
quoted above. In my discussion of the habenular system (chap, xviii)
and the interpeduncular nucleus (chap, xiv) I have ventured to suggest a possible mechanism through which general inhibition effected
in the interpeduncular neuropil may operate in the facilitation of
both mass movement and local reflexes. On this hypothesis this local
band of neuropil must be able to act as a specific inhibitor in Beritoff's sense.
The amphibian neuropil in its various forms . is structurally
adapted for a considerable variety of functions of different grades
of specialization. There is generally a diffuse spread of terminals, so
that a single incoming fiber may activate many neurons of the second
order. If the receptive tissue is homogeneous, this provides for simple
central summation. If the receptive tissue is heterogeneous, as in
most sensoiy fields, this arrangement facilitates mass movement of
the musculature or total patterns of action. If many fibers converge
upon a single neuron, the threshold of central excitation is lowered,
as in the mitral cells of the olfactory bulb and in the *'motor pool," as this concept has been developed by Sherrington. If the activated
motor pool is large, with wide distribution of the efferent fibers, complicated integrated mass movement may result. If the pool is small,
with a single final common path, a local reflex may follow. If the outlet comprises a number of open channels with different connections
and physiological properties, there is provision for discriminative
response, the selection being made (presumably) in terms of the
central excitatory state of the components of the system ('42, p. 295).
It has been objected that the preceding comments on the limitations of current doctrines of reflexology are based upon the amphibian organization, which is aberrant and degenerate and therefore not
typical or significant in the interpretation of the behavior of higher
animals. But, even so, the Amphibia live well-ordered lives, and their
behavior conforms in basic patterns with that of other vertebrates.
We want to know how they behave as they do. Accepting the current view that Amblystoma is a retrograde descendant of some more
highly specialized ancestor now extinct and that some of its characters are aberrant, yet the evidence seems to me adequate that such
retrogression as may have occurred has been toward a generalized
form ancestral to modern amphibians and mammals.
Conclusion. — I have assembled in these pages some factual description of observed structure, together with speculative interpretations of its probable physiological significance. The organic structure here under consideration is not something vague and ill-defined.
Its anatomical distribution, histological organization, and fibrous
connections can be described with precision. Not until this has been
done can our imperfect knowledge of its functions be advanced by
experiments designed to reveal its physiological properties.
In a discussion of "localized functions and integrating functions"
more than a decade ago ('34a), the significance of neuropil in the
evolution of cerebral structure was summarized in these words :
"The neuropil is the mother tissue from which liave been derived both the specialized centers and tracts which execute the refined movements of the local reflexes and the more general web which binds these local activities together and
integrates the behavior. It retains something of embryonic plasticity and so is available as a source of raw material for two very dift'erent lines of specialization — first,
toward the structural heterogeneity requisite for the execution of localized reflex
and associational functions, and, second, toward the more generalized and dispersed apparatus of total or organismic functions of tonicity, summation, reinforcement, facilitation, inhibition, 'spontaneity,' constitutional disposition and temperament, and extra-reflex activities in general."
The gi-eat advances that have been made in the diagnosis and
treatment of nervous diseases have been due in large measure to the
more accurate mapping of the structural features of the nervous system and recognition of the specific functions of its several parts.
Before a disorder can be successfully treated we must know what it
is and where it is. The most notable triumphs in this medical field
have been registered with those diseases whose situs can be recognized and then subjected to appropriate therapy or surgery. Even a
systemic disorder like anemia has localization in blood corpuscles
and blood-forming organs; and a general infection, like poliomyelitis,
spreads in preferential paths determined by the histochemical structure of the tissue. The stable heritable tissues of the nervous system
are most accessible to this kind of inquiry, of diagnosis, and of treatment; conquest of the unlocalized disorders has been retarded.
Some kinds of disorder, particularly those of primary concern to
psychiatrists, have resisted all attempts at localization in accordance
with conventional principles, and in the field of physiology the concept of local reflex arcs has limited application. The various attempts
to elaborate a comprehensive account of animal and human behavior
in terms of conventional reflexology have broken down. These conspicuous failures have led some competent authorities to question
the over-all significance of localization in space of nervous functions
and to search for other principles in the realm of pure dynamics or
chemical interaction or some as yet unknown factors which operate
quite independently of stable structural patterns. But no nervous
tissue is structurally homogeneous or physiologically equipotential.
In this connection it is interesting to note that Lashley, the leading
advocate of the equipotentiality of the nervous tissues, has given us
clear demonstration of point-to-point projection of retinal loci upon
the lateral geniculate body and the cerebral cortex of the rat (Lashley,
'34, '34a). This is the most refined anatomical localization of function known. In a later communication ('41) he demonstrated a very
precise projection of the thalamic nuclei upon the cerebral cortex and
added: "A functional interpretation of the spatial arrangement of
the thalamo-cortical connections is not justified on anatomic grounds
alone for any sensory system."
Somewhere between the extreme views of rigid localization in spatial patterns and a labile physiological equipotentiality a practicable
working conception of the meaning of the structural configuration will be found. Clearly, the nervous system does not operate, even in
the case of the simplest known reflex, on the mechanical principles
of an automatic telephone exchange. We get only confusion by oversimplification of the problem. On the other hand, there are no disembodied functions, and the apparatus that performs functions has
locus in space and time. Our problem is, first, to observe what is done
and then to find out wliere and when it is done and how.
The observed spatial arrangements are not meaningless, and their
functional interpretation is possible and fruitful, as evidenced by
their practical utility in medical diagnosis and treatment. These
structural patterns are stable and heritable. Their phylogenetic development can be traced, and in broad outline this has been done.
But as these patterns are followed backward in the evolutionary
series they become, not more simple and sharply defined, but less so,
until in the most primitive and generalized vertebrates they tend to
disappear in a more nearly homogeneous matrix. This would seem to
support the view that localization of function is a secondary acquisition, derived from a primitive equipotentiality. With certain important qualifications, this is probably true; and if we follow in phylogeny the differentiation of local centers and their connecting tracts
in correlation with types of function performed, the significance of
localization appears. The problems of cerebral localization have
usually been attacked in mammals and especially in man, where
clinical applications are vitally important. Let us approach the subject from the other end of the phyletic series and look for the inception of localization patterns in primitive animals.
In the simplest known organisms localization of function is minimal and transient. In ameba any part of the cytoplasm may on occasion perform any function of which the organism is capable. There
is a local differentiation of nucleus from cytoplasm, but in some bacteria even this localization disappears. A surface-interior pattern is
always present, but the physical substance may shift from one to the
other of these zones. In primitive multicellular species, ectoderm and
entoderm were early differentiated — a specialization which persists
throughout the animal kingdom as manifested in the basic distinction between somatic and visceral organs, a structural differentiation
that has physiological meaning. Further specialization advanced
more rapidly in somatic organs than in visceral, and in the former
more rapidly on the sensory side than on the motor side. This again has physiological significance because the acute problems of subsistence involve adjustments to surroundings. The primitive motor responses are mass movements, but the surface of the body is exposed
to a manifold of diverse stimuli which must be analyzed, and, accordingly, diverse organs of sense were locally differentiated. The course
of this differentiation followed this rule: from the generalized and
equipotential to the special and local. Thus in some primitive eyeless forms the entire skin is sensitive to light, in others only the most
commonly exposed surfaces of it; and in the leech, Clepsine, Whitman ('92) found that a single animal exhibited all transitions from a
series of segmentally arranged sense organs of generalized function in
the posterior part of the body to well-formed eyes at the anterior end.
As I have elsewhere pointed out ('29), in the most simply organized vertebrates, the hagfishes, as described by Jansen ('30) and
Conel ('29, '31), the brain is organized around two dominant sensory
systems — olfactory and cutaneous — and the other special senses are
in various stages of arrest or degeneration in correlation with a semiparasitic habit. Without eyes, jaws, or limbs, the visible behavior is
reduced to a simple system of mass movements. Within the brain
there is little local differentiation except for the primary sensory and
motor fields directly connected with the peripheral end-organs, and
yet this brain is the adjusting mechanism of a very rigid system of
simple movements.
Larval Amblystoma is similar, though here the specialization of
tissue is further advanced, and there is progressive advancement in
representatives of later phylogenetic stages. The principle of progressive transformation from the general and dispersed to the specific and
local applies throughout phylogenetic development; it is a general
principle of embryogenesis (Weiss, '39, p. 288) ; it is clearly exemplified in human development, as has been demonstrated on the physiologic side by Hooker ('44) and on the structural side by Humphrey
('44, p. 39); and it appears in the course of conditioning reflexes
(CoghiU, '30).
In the phylogenetic history of vertebrates the basic pattern of
sensory equipment was apparently laid down very early, with no
radical changes except at the transition from aquatic to terrestrial
life. And at this period of transition from fish to tetrapod the neuromotor apparatus experienced even more radical transformation, with
elaboration of local reflexes which supplement and largely replace the
more primitive mass movements.
In amphibian development this history is recapitulated in the long
period which cidminates at metamorphosis; and during this period
the texture of the brain undergoes two divergent Hues of differentiation of tissue in correlation with the expansion of two types of activity, the analytic and the synthetic, as described in the pi-eceding
section. The structural arrangement of the analytic apparatus is, in
its main features, predetermined in the hereditary organization; it is
stable and approximately the same in all members of the species. The
intervening synthetic and integrating apparatus, on the other hand,
is less rigid and is more labile in function. The pattern of its performance will vary from moment to moment in adjustment to every
change in sensory and motor activity and every fluctuation of central excitatory state. But in even the most primitive vertebrates
some cross-connections between sensory and motor zones, which are
interrupted in the intermediate zone of correlation, are laid down in
the stable, heritable structure. These serve the standardized ("instinctive") patterns of behavior, and the arrangement of these connections is determined more by the motor equipment of the animal
than by the sensory equipment (Crosby and Woodburne, '38; Woodburne, '39).
From these considerations it follows that the concept of localization of function must not be formulated in static terms. It is localization of action, and the spatial pattern of this localization reflects
every change in the character of the action. The structural pattern
of this localization is more stable at the afferent and efferent endpoints of the system, and it becomes less so as we pass inward from
these fixed points. The apparatus of standardized behavior like reflex is more rigidly localized than is that of more labile individually
modifiable behavior.
Two types of structure which have just been contrasted may be
characterized as unspecialized or generalized, and locally differentiated in specific stable and heritable patterns. The second was probably derived phylogenetically from the first, and in higher animals
both of them exhibit progressive differentiation of structure in divergent directions. Some examples of these two types of cerebral architecture will next be cited, beginning with the second, which has been
investigated in more detail.
An early stage in the evolution of localized conductors of specific
sensory systems is illustrated by the connections of the lemniscus
systems described in chapter xi. These tracts of Ambly stoma are not well separated, and in the aggregate they comprise a rather dispersed
collection of ascending fibers loosely assembled in several tracts,
which are distinguished more by their general fields of origin and
termination than by the functions which they serve. It is to be borne
in mind that this low level of functional specificity of the secondary
tracts is not correlated with a corresponding generalized structure
and function of sense organs and related peripheral nerves. These
organs, though different from those of higher animals, are highly
specialized and as sharply localized. The correlation, on the contrary,
is with the generalized character of the motor responses evoked. In
higher animals with a wider range of motor capacities the functional
specificity of the lemniscus tracts becomes more precise.
During the process of differentiation of these more specific tracts
they retain collateral connections along the entire course, so that
they continue to perform integrative functions similar to those of the
less specialized ancestral pattern. In this connection we quote a passage from Dr. Papez ('36) :
"In tracing the central connections of any one of the main afferent systems in the
vertebrates one gains the impression that there is a progressive pliyletic tendency of
each system to enter into connections with all the important segments of the central
organ. In this way there arises a totally integrated anatomical pattern common to
all the receptorial systems in spite of the wide diversity of the receptors, their individual pathways and their interpolated centers. This central integration is not
essentially of a reflex nature and cannot be appropriately described as a chain of
reflex connections insomuch as each level has a highly individual structural organization designed primarily for the production of distinctive organic functions."
Papez appropriately emphasizes the integrating action of these
long conductors; it seems to me, however, that his conception of "a
progressive phyletic tendency of each system to enter into connections
with all the important segments" in the interest of integration is a
reversal of the actual course of phyletic history. These collateral connections are more numerous and more dispersed in lower forms than
in higher. The integration is primary, and the analysis is secondary.
It is true that the primary integration is not subordinated in the
course of phylogeny; it is accentuated; but the apparatus employed
is radically changed. Dispersed nonspecific connections are progressively replaced by localized specific structures, which are so interrelated as to work together harmoniously in the performance of
standardized patterns of behavior. And, in addition to this, higher
centers are elaborated, notably in the cortex, which progressively acquire dominant control of the total action system.
The phylogenetic history of the differentiation of the visual-motor
system also illustrates the principle just stated. In most vertebrates
the eyes are the dominant organs concerned with the orientation of
the body and its members in space. The visual apparatus within the
brain, accordingly, exhibits the most precise localization of function,
and the refinement of this localization increases progressively in the
phyletic series.
In Amblystoma the fibers of the optic tracts are widely spread in
the brain stem, in marked contrast with those of other sensory systems, which tend to converge into a single receptive field. There is
no evidence that this dispersal of fibers of retinal origin to the tectum, pretectal nucleus, thalamus, hypothalamus, and cerebral peduncle is correlated with any specificity of visual function. The explanation of this central spread of retinal fibers is to be sought on the
motor side of the arc, that is, it is determined primarily by the nature
of the response to be evoked.
In the Amphibia all these visual areas are centers of correlation,
for in all of them optic terminals are mingled with those of other systems. Within this class, however, as we pass from generalized urodeles to specialized anurans, there is a conspicuous trend toward
segregation of some terminals of the optic nerve in the tectum and
lateral geniculate body of the thalamus. This trend culminates in
mammals and is correlated with the differentiation of the visual area
of the cerebral cortex, until in primates, as pointed out by Clark
('43), the retinal-geniculate-cortical pathway provides a very precise point-to-point projection of the visual field upon the cerebral
cortex, and "there is no possibility that these impulses can be disturbed and modified 'en route' by other, unrelated, types of nervous
impulse In other words, the cerebral cortex receives retinal
impulses in a remarkably pure and unadulterated form."
The highly specialized optic tecta of some lower vertebrates exhibit two types of specific localized structure, which differ in form
and physiological significance. There is, first, an arrangement of
sensory terminals spread superficially in mosaic pattern. This provides for point-to-point projection of retinal loci upon the tectum and
perhaps for other forms of sensory localization. In the second place,
there is a pattern of lamination at different depths from the surface.
These laminae differ somewhat in their sensory connections, and the
sensory influence is stronger in the moi-e superficial members of the
series. The arrangement of the deeper layers seems to be determined
primarily by the directions taken by their efferent fibers. The mosaic
pattern is primarily in terms of sensory analysis, the lamination pattern in terms of sensory correlation superficially and of motor
analysis in the deeper layers. The cerebral cortex of mammals also
exhibits both mosaic and laminated patterns of localization and in
far more complex designs (Huber and Crosby, '33, '34).
Specific structure of this analytic type, with well-defined localization in both gray and white substance, increases in amount as we
pass from lower to higher animals in the phyletic series; and this
increment progresses from the sensory and motor periphery inward
toward the upper cerebral levels, where the apparatus of integration
and synthesis is most elaborately developed. In submammalian
brains the amount of myelin present at successive levels of the brain
stem is a rough indicator of the relative mass of tissue of this analytic type.
This type of tissue, as pointed out above, predominates in the most
primitive vertebrate brains. With advancing differentiation, we observe specialization of this tissue in three directions. (1) Some of it
retains its primitive generalized structure with little change. In urodeles this is true of a large proportion of it; in mammals it survives
in the periventricular system of cells and fibers of the diencephalon
and mesencephalon and in some other regions. (2) A progressively
larger proportion of this tissue is transformed into specifically localized structures, as just described. (3) Another large proportion of it
is transformed into the relatively unspecialized tissues of the intermediate zone of correlation and its highly elaborated derivatives in
the suprasegmental apparatus of the cerebellar and cerebral cortex.
Doubtless all parts of the body participate in the total integration
and the determination of general attitudes and types of response,
but the brain exercises dominant control over overt behavior and
orders it in the interest of the welfare of the body as a whole. The apparatus of these totalizing functions evidently includes many diverse
components, of which one of the most obvious is the neuropil, which
in primitive vertebrates pervades the entire brain, so that activity in
any part of it may affect the whole fabric, as elsewhere described.
This dispersed tissue 's not homogeneous, and it is not equipotential.
It is doubtless always active and in diverse ways in different places
at different times. Such localization of function as it exhibits can best
be conceived in dynamic terms, that is, in terms of what intercurrent nervous volleys act upon it in momentarily changing places,
rhythms, and intensities. We are dealing here with an equilibrated
dynamic system comprising many activated fields in interaction, and
this interplay is in patterns quite different from those of the stable,
locally differentiated centers and tracts. It is more labile, and the
patterns of performance are not stereotyped. Nevertheless, these
fields are not structurally identical, and each one has distinctive
physiological properties correlated with these histological differences.
Each field of neuropil differs from others in internal texture, in the
source of afferent fibers, and in the distribution of efferents. A local
field may be sharply segregated, as in the ventrolateral neuropil of
the peduncle and the ventral interpeduncular neuropil, or it may
interpenetrate tissue of the specific localized systems, as in the corpus
striatum and optic tectum. The pattern of this localization is different structurally and physiologically from that of the specific systems
of cells and fibers, and the two patterns of localization may both be
present in the same block of tissue in primitive brains. In higher
vertebrates these local differences are accentuated, the segregation
of the synthetic apparatus is carried further, and its tissue is locally
differentiated in a radically different way from that of the analytic
apparatus, as is best exhibited in the associational tissue of the human cerebral cortex.
The "field" concept has been much exploited of late in several contexts, and it is fruitful, as applied, for instance, by Weiss ('39, p. 289)
in general embryology and by Agar ('43, chap, ii) in general biology.
As applied psychologically in Gestalt it has been difficult for the
structurally minded neurologist to transfer the dynamic formulations into the biological frame of reference; but there is a structural
"ground" within which every "configuration" of experience is set,
and the primitive neuropil, together with its specialized derivatives,
is one important component of the organic substrate of the "gestalt
qualities." The "field" as here conceived is an organized living structure in action, some components of which we recognize as stable
architecture and some as fluctuations in the excitatory state of the
structural fabric. This structure has visible organization, and its
properties are open to investigation anatomically and physiologically
('34a; '42, p. 293).
The two kinds of structure which have just been considered perform functions which are localized according to different principles,
a distinction which has been generally ignored. The specialized analytic structures have stable arrangement in space, and their functions
have corresponding localization in three-dimensional mosaic patterns. The functions of the generalized structures are, in the main,
syntlu'tic rtillicr than Jiiuilylic, and they arc usually descrihod in
dynamic terms in which a fourth dimension a time factor — plays
an important pail. Yet this ef|nilil)rated dynann'c system is not disembodied, and its component parts have locus in space and time.
These loci, or fields, may have some degree of permanence with characteristic structural organization, as in the so-called "association
centers" of the cortex, or they may fluctuate as ever changing patterns of linkage of the o})erating neui-ons. As Papez ('44) expresses
it: "The anatomical structures are stable, the function is labile, depending on numbers of cells and their excitable or refractory states
at any particular time." Many of these synthesizing patterns are
repetitive, as in habit and memory. Though the actual structures
involved may not be identical in successive repetitions, the i)attern
of performance is similar, that is, it recurs in conformity with an
enduring "schema" (to employ Henry Head's term), or the engram
of conventional teiininology. Doubtless the engram has a structural
(or chemical .f*) counterpart, but we do not know what it is.
The stable localization of the structural fields is contrasted with
the evanescent localization of the i)attern of their combination at
each repetition of the schema. It is possible to find out where the
tissue is that yields these dynamic schemata and to delimit it; but
these limits cannot be circumscribed on the surface of the brain in
simple mosaic patterns. "^I'he manifestation of any schema at a particular time is always a function of a configuration of nervous elements, which has location in space. But a very similar schema may at
another time be exhibited by a different structural configuration,
whose locus in space is by no means identical with the first ('30a).
It has recently been shown by Lashley and Clark ('46) that cortical structure is variable to a degree not hitherto appreciated in
different individuals of the same species of monkey, and it is probable that the range of this variability will be found to be still greater
in any human population. They conclude that "marked local variations in cell size and density among individuals of the same species
may constitute a basis for individual difi'erence in behavior"; but
they challenge the validity of the criteria in current use for parcellation of the cortex into functionally specific areas, except for rather
large areas of projection. This is supported by the experiments of
Murphy and Gellhorn ('45) and the observations of Bailey and von
Bonin ('4(5).
THE human brain is the most complicated piece of mechanism
that we know, and the products deHvered by this thinking
machine are, for us, the most interesting. Detailed descriptions of it
and some of its operations have been written, but where and how it
fabricates its unique wares is still the basic problem of science and
philosophy. Thinking is a part of our living, and apparently we think
all over, just as we live in all parts of our bodies. Yet it is evident
that some parts of us play crucial roles in our mental life, just as other
parts do in our movements, our digestion, and so on. That the
cerebral cortex as a whole is a specific organ of much of our mental
life is as well established empirically as anything in biology, but how
thinking is done and where the critical processes are carried on are
still mysterious.
Study of origins and early stages of embryologic and phyletic development has resolved many biological and psychological problems,
but essential features of the inception of cerebral cortex remain obscure. Interest in this theme instigated my program of research upon
the amphibian nervous system. All reptiles possess well-organized
cortex, that is, superficial laminae of gray substance in the pallial
part of the cerebral hemisphere, simple in pattern but obviously comparable with, and ancestral to, the human cortical complex. They
also exhibit enormous enlargement of some subcortical parts of the
hemisphere, notably the strio-amygdaloid complex. In birds these
subcortical areas are still further enlarged and complicated, with
reduction in the amount and specialization of the cortical tissue. In
mammals the subcortical parts of the hemisphere are relatively
smaller, some of their functions apparently having been taken over
by the progressively expanding cortex ('21, p. 452; '26, p. 122).
Correlation of these structural peculiarities with the characteristic
modes of life of these several classes of animals gives some clues to
the significance of the cortex in the vital economy.
As already pointed out, structural differentiation of cortex is incipient in the hippocampal sector of the pallium in all Amphibia;
more clearly defined primordia of cortex are seen in adult lungfishes
and in embryonic stages of some other fishes (Rudebeck, '45) ; but
well-differentiated cortex of typical structure first appears in reptiles.
The problem set is : What morphogenic agencies were operative during the emergence of cortex from a noncorticated matrix?
Early in the attack upon this problem it became evident that the
key factors were to be sought, not in the pallial field, but in its
environs. What comes into this field and what goes out of it at successive stages of morphogenesis and how are these factors related
one to another in both subpallial and pallial territory.^ This requires
in the upshot a histological analysis of the entire nervous system
directed toward the physiological interpretation of all visible structure. In my Brains of Rats and Men ('26) the available evidence
regarding the origin of cerebral cortex was surveyed. Since that time
additional evidence has been recorded, and in the present work the
results of a renewed examination are summarized. The problems
centering in corticogenesis have not been solved, but some progress
has been made. The conclusions reached can, at best, be only tentative, pending physiological control; and for such experiments exact
information about the anatomical arrangements of parts is indispensable.
There is reason to believe that in the early ancestors of vertebrates
the central nervous system was a simple tubular structure comparable with that of living Amphioxus. Accompanying enlargement
of the dominant sense organs — nose and eye — the anterior part of
this tube was expanded in four places, which became cerebral hemispheres, hypothalamus, epithalamus, and tectum opticum. In most
lower vertebrates the olfactory system is very large and has played
a dominant role in the earlier stages of the morphogenesis of the
hemispheres. In fishes this differentiation took a wide variety of
forms, some of which were surveyed in two papers ('21, '22a). These
diverse forms and patterns of internal structure were shaped in
adaptation to various modes of life which employed different equipment of sensory and motor organs.
It has been mentioned that the more sluggish fishes, and especially
mudfishes living in stagnant water, have enormously enlarged and
highly vascular chorioid plexuses, the ventricles are dilated, and the
walls of the brain are thin. This insures a supply of oxygen to the
brain which is adequate for their quiescent existence. The more
active fishes living in well-aerated water lack these features and cannot survive in stagnant water. Their brains have solid masses of
tissue in great variety of forms and are sensitive to asphyxiation.
The fossil record shows that amphibians were derived from generalized fishes similar to the living lungfishes. These are mudfishes
with enlarged chorioid plexuses and thin-walled widely evaginated
hemispheres. There is geological evidence that in early Devonian
time, when the amphibians emerged from the water, there was general continental desiccation. The lakes and streams were drying up,
and over extensive areas the freshwater fishes were faced with the
alternative of adaptation to drought or extinction. The more highly
specialized species perished, but some generalized and sluggish types
of mudfishes made successful adjustment. Their extensive chorioid
plexuses and widely expanded thin-walled hemispheres had survival
value, for so they were tided over the critical period of oxygen deficiency during phyletic metamorphosis. These characteristics persist
today in all urodeles, most of which have retained the ancestral mode
of life.
In later evolutionary stages the expanded hemispheric vesicles had
the further advantage that space is available for further differentiation, and especially for spreading out the correlation tissue in thin
sheets, an arrangement which seems to be requisite for refinement of
adjustment to the spatial relations of things and for high development of labile, individually modifiable behavior as contrasted with
stable, heritable behavior of instinctive type. The interested reader
will find further details about the origin and significance of the evaginated form of cerebral hemispheres in the two papers cited ('21,
'22a) and in chapter xvi of my Neurological Foundations ('24c).
From the beginnings of differentiation of the evaginated hemispheres, their medial and lateral walls have shown striking and interesting differences. The telencephalic connections are variously arranged in the different groups of fishes; but in the Amphibia the
cerebral hemispheres have acquired the definitive form with connections that are not fundamentally changed in the higher groups.
There is no evidence that this arrangement is due to differences in
the quality of the olfactory impulses transmitted from the olfactory
bulb, except for the possibility suggested (p. 54; '21a) that specific
impulses pass from the vomeronasal organ to the accessory bulb and
amygdala. The nervus terminalis is more evidently specific, but of its
functions nothing is known. All other descending olfactory impulses
seem to be physiologically homogeneous. From this it follows that
local differentiations in the hemispheres are due mainly to differences
in the distribution of the various systems of ascending nonolfactory
fibers or to differences in the destination of descending fibers arising
from various parts. Evidently both factors are involved.
The connections between the hemispheres and the lower levels are
assembled in the basal forebrain bundles and the stria medullaris
thalami. The former are roughly comparable with the subcortical
components of the mammalian extrapyramidal systems plus large
numbers of ascending fibers, and the details of their connections are
given in chapter xx. The analysis of these systems of fibers provides
the key to the interpretation of the hemispheres of Amblystoma; and
similar analysis in reptiles, birds, and mammals is essential for an
understanding of the history of cortical evolution.
The fibers of the stria medullaris system are efferent or commissural, passing from all parts of the hemisphere to the habenula or
decussating in the habenular commissure and returning to the hemisphere. The arrangement of the components of the stria is shown in
figure 20, and this is essentially similar to that of mammals. The
stria medullaris, in fact, is remarkably similar in all vertebrates; and
these fibers evidently belong to a different category from those of the
basal bundles, which contain ascending and descending fibers in arrangements that vaiy widely from species to species as we pass from
lower to higher vertebrates. These variations seem to be especially
significant for interpretation of morphogenesis. In this section, accordingly, attention will be directed to the latter systems, and the
reader is referred to chapters xiv and xviii for the interesting details
of the habenular connections.
The basal bundles comprise three groups of fascicles, the descending components of which are shown diagrammatically in figure 6:
(1) dorsally and laterally the lateral forebrain bundles (f.lat.t.) ; (2)
ventrally and medially the medial bundles (f.med.t.) ; (3) the olfactopeduncular tract (tr.ol.ped.), lying between the two preceding and
with connections which are intermediate between them at both ends.
Fibers ascend in these bundles from two sources: (1) from the
somatic sensory field of the dorsal thalamus, by way of the thalamo
frontal tract (figs. 19, 111, tr.ih.f.) in the lateral forehrain bundle,
and (2) from the visceral field of the hypothalamus, by way of the
medial bundle (fig. 113). The first of these bundles connects with the
lateral wall of the hemisphere, the second with the medial wall. In
conformity with this, the descending path from the lateral wall goes
by way of the lateral forebrain bundle to the somatic motor field in
the peduncle, and the path from the medial wall goes by way of the
medial bundle to the visceral field in the hypothalamus (figs. 6, 11,
111, 112, 113). These connections inaugurated the different types of
differentiation seen in lateral and medial walls of the hemisphere, a
difference in type which becomes more pronounced in higher animals. The amygdala is of intermediate type, with somatic and
olfacto-visceral connections of both afferent and efferent fibers
(fig. 19). . _
All parts of the hemisphere are under olfactory influence, with
olfacto-visceral correlations effected medially and olfacto-somatic
laterally. The motor responses are radically different, and this difference is probably the basic determining factor in shaping the structural plan, not only of the olfactory connections, but of the organization of the hemisphere as a whole.
The hypothalamus is disproportionately large in lower vertebrates
as compared with higher, and the olfacto-visceral functions of the
hemisphere are correspondingly magnified. This doubtless accounts
for the fact that differentiation in the pallial field is further advanced
on the medial (hippocampal) side than elsewhere (fig. 99) and also
for the fact that efferent projection fibers from this part of the
pallium (fornix system, p. 254) appear in large numbers very early in
Fibers ascend from the dorsal thalamus to the hemisphere in fishes
and in all higher animals. In Amblystoma these thalamo-frontal
fibers arise in the generalized nucleus sensitivus, ascend in the lateral
forebrain bundle, and end in the amygdala and middle part of the
corpus striatum (figs. 19, 111, tr.th.f.). They are comparable with
thalamo-striatal fibers of mammals and are precursors of mammalian sensory projection fibers, though in urodeles none have been
seen to reach the pallial part of the hemisphere. This tract is small,
and there is no evidence that different sensory systems are separately
localized within it; its terminal nucleus, the corpus striatum, is correspondingly small and simply organized. The thalamo-frontal system is larger in reptiles and birds, where the striatal complex is magnified; but only in mammals are the several functional systems well
segregated and separately localized, and here this localization is correlated with the differentiation of the sensory projection areas of the
neopallium and the related nuclei of the dorsal thalamus. The course
of the evolutionary differentiation of the cerebral hemisphere has
been determined, in its main features, by the penetration of these
nonolfactory fibers into its medial and lateral walls and the elaboration of related centers for the reception and correlation of these sensory impulses and appropriate motor discharge.
The entire ventrolateral wall of the hemisphere of Amblystoma is
an olfacto-striatum. Its anterior end is under strong olfactory influence and is regarded as the primordium of the head of the caudate
nucleus (figs. 98, 99). Posteriorly, the large amygdala (fig. 97) also
receives many olfactory fibers. The middle sector receives fewer
olfactory fibers and is a terminal station of the thalamo-frontal tract.
From it fibers descend to the ventral thalamus and peduncle. It is,
accordingly, regarded as paleostriatum, or somatic striatum, primordium of the mammalian lentiform nucleus.
The striatal gray of the middle sector is obscurely divided into
dorsal and ventral nuclei, separated by a shallow sulcus striaticus
(fig. 99, s.st.). Small cells of putamen type and larger cells of globus
pallidus type are mingled in both nuclei, so that these structures are
not separated as in mammals; yet their connections, as described in
chapter xx, suggest that the ventral nucleus is the precursor of the
globus pallidus. Both these nuclei receive fibers from the overlying
piriform area, some of which descend into the lateral forebrain bundle
(figs. 6, 111). The ventral nucleus is continuous with and intimately
connected with the primordial caudate ('27, p. 298), and its efferent
fibers go chiefly to the cerebral peduncle. The dorsal nucleus receives
thalamo-frontal fibers and is in intimate relation with the amygdala
and the piriform area. Its descending fibers have wide distribution to
the tectum, thalamus, dorsal part of the peduncle, and dorsal,
isthmic, and bulbar tegmentum as far back as the V nerve roots (fig.
QJ.lat.t.d.). In Necturus (fig. HI; '336, p. 197; '33e) there is a complicated system of associational connections between the striatum
and the piriform area; these are present also in Amblystoma, but the
details have not been described.
Most of the white substance of the striatal field is occupied by a
very dense and sharply circumscribed intermediate neuropil of peculiar texture (p. 53 and figs. 98, 99, 108, 109, 113; '27, p. 300; '42,
p. 'iO'^Z; Necturiis, '336, p. 149). There is a web of interlaced branches
of ependymal elements and among these a still more densely woven
tangle of dendrites and slender, contorted, unmyelinated axons. The
thicker descending axons of the dorsal and ventral fascicles of the
lateral forebrain bundle are assembled within this neuropil (figs. Ill,
113). I have seen similar texture in Golgi impregnations of the head
of the caudate in the opossum ('24c?, p. 342), a structure adapted for
diffusion and summation of all nervous impulses entering it.
The urodele type of striatal structure could be transformed into
that of mammals by reduction of the olfactory component; further
differentiation of the caudate nucleus and amygdala; segregation of
the large efferent neurons in the ventral nucleus, which becomes the
globus pallidus; and segregation of the smaller elements in the putamen. In Ambly stoma these changes are only dimly foreshadowed.
The ventral fascicles of the lateral forebrain bundle connect with the
primary motor field of the ventral thalamus and peduncle and are
comparable with those of the ansa lenticularis. The connections of
the dorsal nucleus suggest relations with the reptilian neostriatum
and the mammalian putamen. Here are probably to be found the
earliest indications of those formative agencies which in later phyletic
stages led to the differentiation of neopallium. In the still more primitive hemisphere of Necturus, these indications were recognized and
discussed ('33e).
The strio-amygdaloid complex as a whole is the highest center of
dominance in the control of the skeletal musculature, a role which is
enormously enlarged in reptiles and birds. In mammals, parallel with
the elaboration of cortex, the part which the striatum plays in the
patterning of behavior is progressively reduced, but it retains important functions of co-ordinating and stabilizing motor performance.
Just as the cerebellum is added to the sensori-motor systems for
facilitation of muscular co-ordination, so in mammals the striatal
complex is interpolated in the efferent cortical systems as an accessory facilitating mechanism. ^
In the Amphibia we find a critical stage in the morphogenesis of
the cerebral hemisphere. The definitive major subdivisions are here
blocked out in recognizable form, and the pallial part is incompletely
segregated from the stem part ; yet the most distinctive feature of the
pallium — its superficial cortex — has not yet appeared. We want to
know more about the agencies which are in operation here to initiate
the separation of pallium from stem and what further changes led to the migration of the palHal gray from deep to superficial position and
its subsequent compHcations. A beginning has been made, and I have
at various times reported progress in this analysis and some discussion of its meaning ('24c, chaps, xv, xvi; 'Md, p. 354; '26; '27, p. 315;
'33a; '336; '33e; '34a). Yet much remains to be done before we can
fill in those finer details of structure which the physiologist needs to
know in order to plan crucial experiments. Frogs are probably better
adapted for such experiments than are salamanders, and to this end a
more detailed analysis of the histological structure and connections
of the forebrain of the frog is urgently needed. Sufl5cient knowledge
of this structure is now available to enable the physiologist to explore
the instrumentation of some components of the behavior pattern, as
illustrated by a recent study by Aronson and Noble ('45).
More than thirty years ago I published some reflections under the
title given to this chapter ('13a). Though parts of that paper require
revision in the light of subsequent research, yet it sketches the background of the present discussion. Attention was called to Dewey's
('93) concept of the organic circuit as a substitute for the classical
formulation of the reflex arc, as mentioned in the preceding chapter.
Some illustrations of these organic circuits were given in my article.
Here we need only to emphasize the fact that all behavior is the
resultant of their interplay, for which provision is made in the cerebral mechanisms, such as are described in this and other works devoted to neuroanatomy. These are all circular reactions between
receptor and effector organs or the related internal adjustors. In the
course of phylogeny, cerebral cortex has been differentiated as the
culminating member of a series of progressively more complicated
integrating mechanisms adapted to make more efficient use of the
preformed circuits of the brain stem in the interest of more flexible
behavior in terms of individually acquired experience, as contrasted
with the stereotyped patterns of the stem (for a convenient summary
of the human connections see Papez, '44).
In this connection two sentences may be quoted from von Bonin
('45, p. 52) : "It is of the essence of cortical organizations that sensory and motor areas become divorced more and more from each
other — are pulled farther apart as it were — as evolution proceeds As we ascend the evolutionary scale, the cortex assumes
increasingly a structure which may be interpreted as leading to increased degrees of indeterminacy."
A fundamental feature of cortical functions is that they are delayed
reactions (p. 78; '!24r, p. 271); there is, first, the arrest or inhibition
of some lower and more primitive patterns of behavior of reflex or
instinctive type. This allows time for cortical reorganization of the
component factors of the situation, conditioning of reflexes, or other
modifications of the stereotyped patterns of response. The cortex,
accordingly, is lifted up away from the lines of through traflic in the
brain stem, and the dorsal convexity of the evaginated cerebral hemisphere is conveniently located, with ample space for indefinite enlargement.
During the phylogenetic development of cortex, ascending and
descending pallial projection fibers are added to the pre-existing systems of the underlying stem. They do not entirely supplant them, for
even in mammals, where cortical projection systems are highly developed, the subpallial parts of the hemisphere retain their own
diencephalic connections.
Primitively, as in cyclostomes, the entire cerebral hemisphere is
little more than an olfactory bulb and a secondary olfactory nucleus.
In Amblystoma the hemispheric evagination is more extensive, and
there is a large increment of ascending nonolfactory fibers; yet here
the pallial part of the hemisphere receives the largest olfactory tracts,
and all of it is essentially an olfactory nucleus.
The olfactory reflexes seem to be adequately provided for in the
stem portion of the hemisphere. I have suggested ('33) that here, and
especially in the corticated mammals, the olfactory sense, lacking
any localizing function of its own, co-operates with other senses in
various ways, including a qualitative analysis of odors (desirable and
noxious) and also the activation or sensitizing of the nervous system
as a whole and of certain appropriately attuned sensori-motor systems in particular, with resulting lowered threshold of excitation for
all stimuli and differential reinforcement or inhibition of specific
types of response. The olfactory cortex (and its predecessors in lower
vertebrates) may, then, serve for nonspecific facilitation of other
activities, in addition to its own specific olfactory functions. This
facilitation may involve both general excitatory action and general
inhibition. That the latter is present is indicated by the observation
of Liggett ('28) that anosmic rats are more active than the normal
controls. The organization of the olfactory system as a whole in all animals seems consonant with this interpretation; and in Aronson
and Noble's study of the sexual behavior of frogs this facilitating
action of the olfactory field is clearly demonstrated.
In primitive vertebrates the dominance of the entire anterior end
of the brain by the olfactory apparatus implies more physiological
homogeneity than in higher brains, where this tissue is invaded by
larger numbers of nonolfactory fibers with more diverse specificities.
The case is somewhat like the invasion of a hitherto isolated continent with homogeneous and primitive population by immigrants of
numerous other races with very diverse cultural standards. When
European peoples colonized North America, in some regions the
newcomers intermarried with the natives and the two races amalgamated; in other places the indigenous population was driven farther
and farther back or exterminated altogether. Something analogous
to these processes has taken place during the invasion of the olfactory
area by nonolfactory functional systems. In some regions there is
blending of the old and the new, as in the amygdala, septum, and
olfactory tubercle; in other places the indigenous olfactory system
has more nearly retained its unmixed character, as along the margin
of the olfactory bulb; and in other extensive regions of the hemisphere the indigenous elements have been almost entirely displaced
by nonolfactory systems, as in part of the corpus striatum and the
neopallial cortex. In the Amphibia this invasion of the olfactory field
by nonolfactory systems is extensive, but the invading forces are not
sufficiently diversified and localized to invoke the differentiation of
cortical tissue in the pallial part of the hemisphere. This is probably
correlated with the fact that amphibian behavior, by and large, is
mass movement, with relatively little refined analysis into partial
The primitive differentiated cortex of reptiles has three welldefined sectors. These are spread, respectively, on the dorsomedial,
dorsolateral, and dorsal convexities of the hemisphere. The first is
archipallium, the precursor of the mammalian hippocampal formation. The second is paleopallium, or piriform cortex, represented in
man by a relatively small area at the lower border of the temporal
lobe and including the uncus and some adjoining areas. The third
sector, the dorsal cortex, is of uncertain relationships. It occupies the
position of the neopallium, which comprises the larger part of the
human cortex, and probably is its precursor, though this apparently
is not its only relationship.
In all amphibian brains these sectors of the pallial field can be
identified, though no superficial cortical gray is present in any of
them, as illustrated in figures 96-99. On the lateral aspect thepiriform
area (figs. 85, 86, 111, p.pir.) shows no evidence of cortical differentiation; it is, in fact, the chief secondary olfactory nucleus (nuc.ol.d.L);
nevertheless, its location and connections identify it unmistakably
as the primordium of the piriform cortex of reptiles and lower mammals. Its neurons are small, simple, and similar to those most commonly seen in the brain stem (figs. 98, 99, 105). On the dorsal convexity there is an undifferentiated and poorly defined field (p.p.d.),
which is the precursor of the reptilian dorsal cortex.
The medial sector — primordium hippocampi (p.hip.) — shows a
first step toward cortical differentiation, for here the compact central
gray layer is dispersed by outward migration of the cells throughout
the thickened wall, and these are imbedded in dense neuropil. These
neurons vary from small to quite large and, in general form, resemble
those of other parts of the pallium, though they are evidently more
specialized (figs. 97, 98, 99, 105). One to several thick and thorny
dendrites arise from the cell body and spread widely, some reaching
the external limiting membrane. The axon may arise from the cell
body, but usually from the base of one of the dendrites, It may
divide, sending one branch into the dorsal pallium and one to the
septal nuclei or medial forebrain bundle (fornix). Some axons are
short, branching freely within the area of spread of the dendrites
(fig. 105; '396, fig. 44), but most of them send one or more long
fibers from this arborization into tracts which leave this field. Close
to the surface are a few tangential neurons which at one time I regarded as precursors of the reptilian cortical cells. Similar cells are
found also throughout the brain stem, and in the. pallium (fig. 98)
their axons take short courses as correlation fibers. They are more
numerous in the frog (P. Ramon y Cajal, '22, figs. 6, 7). The differentiation of the hippocampal cells is further advanced in Amblystoma than in Necturus ('33a, p. 183) and less so than in anurans.
The hippocampal neuropil increases in density and complexity as we
pass from Necturus to Amblystoma and the frog.
The four layers of neuropil characteristic of the brain stem (p. 30)
are very unequally developed in the amphibian hemisphere (Necturus, '336, p. 176). The periventricular neuropil is everywhere abundant. The deep neuropil of the alba contains an elaborate system of association fibers, which has been described in detail (figs.
Ill, 113; '336, p. 194; '33e). The intermediate neuropil contains
many of the recognizable long tracts, and in the striate area it is
elaborately developed, as already described. Superficially of the striate neuropil is a strio-amygdaloid neuropil, which is continuous
dorsally with the piriform and dorsal neuropil. This sheet as a whole
is evidently the synaptic field of the pallial associations. It receives
the dendrites of the underlying gray substance but contains no cell
bodies. In higher animals this synaptic zone seems to exert a neurobiotactic influence, so that in embryonic stages all the neurons of the
pallial field migrate outward and are incorporated within it, thus
producing the laminate cortex.
In the amphibian primordium hippocampi, this movement has begun but is not consummated. The deep gray layer has been broken
up, and its elements are dispersed. The periventricular neuropil of
the grisea and the neuropil of the alba merge, so that the entire area
is pervaded by a dense entanglement of dendrites and axons, within
the meshes of which the cell bodies are imbedded. This neuropil is
denser in two places — rostrally and ventrally, where subpallial connections predominate, and dorsally, where pallial associations predominate. In the reptiles, with differentiated cortex, the corresponding two parts of the hippocampal cortex are structurally different.
In Necturus the most rostral fascicles of the strio-pallial association go far forward and dorsally above the posterior end of the
olfactory bulb to reach the dorsolateral sector of the anterior olfactory nucleus and territory adjacent to it (fig. 111). This is clearly a
secondary olfactory nucleus of subpallial type, adjacent to the olfactory bulb and traversed by the great dorsolateral olfactory tract,
from which it receives numberless terminals and collaterals. Its principal discharge is. backward into the primordium pirif orme, a pallial
area (fig. Ill, tr.ol.pal.L). These connections would, perhaps, have
no special significance in themselves, but comparison with reptiles
shows that there the corresponding region exhibits remarkable peculiarities. In urodeles the area in question is one of the least differentiated parts of the hemisphere, except for the strong fascicles of
the strio-pallial association, and perhaps this lack of specialization
favors the role which it seems to play as germinative tissue for
neopallial cortex.
Cortex of simple pattern is present in turtles in each of the three
pallial fields seen in Amphibia; and the dorsal and lateral cortex
(general cortex and piriform cortex) are related to a massive subcortical thickening of the lateral wall of the hemisphere, which was
called the "dorsal ventricular ridge" by Johnston and "hypopallium" by Elliot Smith. The thalamic radiation, comparable with the
amphibian tractus thalamo-frontalis, is large in turtles. It ends
chiefly in the rostral part of the hypopallium, but some of these fibers
pass through without synapse into the dorsal or general cortex. The
latter fibers are true thalamic sensory projection fibers with connections of neopallial type. In front of this region there is a "pallial
thickening," from which motor cortical projection fibers go out to the
cerebral peduncle — again a neopallial type of connection — and this
part of the cortex is electrically excitable (Johnston, '16).
In the alligator the topographic relations are very different. The
dorsal, or general, cortex has no contact with the hypopallium except
at its rostral end in the region of the pallial thickening of turtles,
which Crosby ('17, pp. 358, 381, figs. 5, 6) calls "primordial general
cortex." This primordium she regards as the germinal tissue or focal
point in the differentiation of the general cortex, and she gives a clear
statement of the factors which probably were operative in the differentiation of this general cortex.
The dorsolateral sector of the anterior olfactory nucleus of Necturus, together with some adjoining tissue, is just such an area of
basal, i.e., subpallial, type as Crosby postulated; it is in the exact
position with reference to other parts of the hemisphere as her primordial general cortex; and it receives especially strong fascicles of
the strio-pallial association, which turn far forward to reach it. It is
significant that in the alligator this primordium is the only region
where both projection fibers of the lateral forebrain bundle and
shorter fibers from the hypopallium and corpus striatum can connect
with the general cortex (Crosby's figs. 5-9, 12-19, 37). Bagley and
Langworthy ('26) have shown that in the alligator this area and
parts of the cortex adjoining are electrically excitable, thus furnishing experimental proof that true motor projection fibers of neopallial
type arise from it. The underlying hypopallium was tested and found
to be unexcitable. Ariens Kappers ('29, p. 140) accepts Crosby's interpretation of the reptilian primordial neopallium and states that in
the lizard, Varanus, a small number of thalamic projection fibers
ascend directly to this area. This, he says, is the source of the neopallium of mammals, not the more differentiated general cortex
of the dorsal convexity of the hemisphere.
The dorsolateral sector of the anterior olfactory nucleus and the
rostral end of the primordial piriform area of Necturus may, accordingly, be regarded as critical points in further search for the earliest
primordium of the neopallium. This region is related with the primordial general cortex of reptiles, which may be regarded as the
precursor of the subiculum and other transitional fields rather than
of neopallium, sen.su stricto. The preceding account of the probable
history of cortical evolution is drawn largely from Crosby's graphic
and discerning analysis published in 1917.
The structure of the pallial field of Amblystoma and its connections were described in 1927 and subsequently in greater detail in
several papers devoted to Necturus ('336, '336?, '34, '34a). In these
papers and some earlier publications I commented on the fact that
the first well-differentiated cortex appears in reptiles in three clearly
defined areas; and the opinion was expressed that a prerequisite for
this differentiation is the penetration into the pallial field of thalamic
projection fibers in separately localized tracts with different physiological properties. This minimal localization of function in the projection systems goes hand in hand with local differentiation in the pallium and amplification of the cortical associational connections of
these areas. This process of local cortical differentiation continues to
advance in complexity of pattern in proportion as the systems of
thalamic projection fibers are amplified and diversified.
This principle of cortical morphogenesis receives its first and clearest exemplification in the obvious difference in the subpallial connections of the medial and lateral parts of the pallial field, hypothalamic
connections predominating medially and thalamic connections laterally. In Ichthyopsida the hypothalamic influence is much stronger
than the thalamic, a relation which is strikingly reversed in higher
vertebrates. These diencephalic influences are not sufficient to cause
cortical differentiation in the amphibian pallium, though there are
some local differences in the three recognizable pallial areas and
Soderberg ('22) found clearer evidence of cortical incipience in some
early larval stages. Holmgren ('22) found evidence of cortical differentiation in developmental stages of selachians and some other
fishes, and in adult lungfishes a primitive cortex is clearly delaminated externally of the central gray (Rudebeck, '45).
In none of the fishes and amphibians do we find so well-differentiated cortex as in reptiles. Why is this? The answer seems to be that
in all Ichthyopsida the entire forebrain is dominated by the olfactory
system to such an extent as to retain a measure of physiological
homogeneity, which is not favorable for cortical differentiation. The
basic feature of cortical function is the association of diverse components of the action system with separate localization of the functional systems involved. In fishes and amphibians this localization of
function in the forebrain is incipient, but it is not sufficiently advanced to evoke cortex of definitive type. In reptiles, on the other
hand, the great increase in the system of somatic sensory exteroceptive thalamic radiations is correlated with enlargement of the corpus
striatum complex, including an extensive area quite free from olfactory and hypothalamic connections and the extension of some of the
fibers of the thalamic radiation to the dorsal pallial field without
interruption in the striatum. Thus the pallial field is subdivided into
three well-circumscribed areas, each with a physiological specificity
different from those of the others and one of which is emancipated
from dominance of the olfactory system. Now for the first time in
phylogenetic history the pallium possesses an intracortical system of
association fibers adapted for the specific cortical type of function
('26, pp. 78, 123; '27, pp. 315 ff.; '33e; '34a).
In a survey of the history of cortical evolution the birds occupy an
anomalous position. They are much more highly differentiated than
reptiles, but in an aberrant direction, with no mammalian affinities.
In most of them the olfactory system is reduced almost to the vanishing-point, and the optic system is greatly enlarged. There is extensive
local differentiation of thalamic nuclei, but not in the mammalian
pattern. The system of ascending thalamic pi-ojection fibers is larger
than in reptiles, and most of these fibers end in the enormously enlarged and complicated corpus striatum. Correlated with the latter
point is the striking fact that, despite the great increase in thalamic
projection fibers, the cortex of many birds is scarcely more extensive
than in reptiles and in some species is less well differentiated (Craigie,
'40) . Birds are more highly specialized in both structure and behavior
than are the lower mammals, and yet their cerebral cortex is rudimentary in comparison with even the most primitive mammals. The
explanation for this is that the bird's more diversified behavior is
largely stereotyped in instinctive patterns, adequately served by
subcortical apparatus, while the patterns of mammalian behavior,
even in the lowest members of the class, are in larger measure individually learned. And enhancement of learning ability goes hand in
hand with cortical differentiation.
It is impossible to define a primordial boundary between pallial
and subpallial territory ('27, p. 316; '33a). There is apparently no
primitive (palingenetic) distinction between the pars pallialis and the
pars subpallialis of the cerebral hemisphere; cortical types of structure and function may be differentiated out of such raw materials as
are available, and the locations of these indefinite boundaries will
vary from species to species. Even in the human brain the boundary
is in some places obscure and controversial.
This, in outline, seems to be the history of the origin and evolution
of the cerebral cortex. The details have not yet been filled in, and this
can be done only by experimental methods, checked and controlled
at every step by accurate histological analysis of the tissue operated
upon. When the facts about the sequences of the evolution of cortical
structure and function are colligated with experimental studies of
behavioral capacities of the animals in question, we shall have a
secure foundation upon which to build a sound comparative psychology, and this, in turn, will clarify much that is now obscure in human
Returning, now, to a general survey of the factors involved in the
differentiation and normal operation of the central nervous system,
we find that these fall into two categories. Some of them conform
perfectly with well-known laws of traditional mechanics of the inorganic realm, as formulated in the Newtonian system and its modern derivatives. Others have not been successfully fitted into this
frame of reference. From the beginning of inquiry into this problem,
there has been a tendency to set these refractory components apart
from the natural order in some mystic realm of vitalism. To the naturalist this solution is not acceptable, for the two classes of phenomena are empirically indissociable.
That the operations of nature are not bound by the man-made
rules of Newtonian mechanics is now evident. It has been shown that
the formulas of Newtonian mechanics, Euclidean geometry, and
Aristotelian logic are not universals. They are valid in a restricted
field but not in the realm opened up by current conceptions of relativity and quantum mechanics. In view of this situation, the field of
neurological inquiry is immeasurably enlarged and complicated.
In our search for the laws of growth and normal action of the
nervous S3^stem, we naturally and properly look, first, for those features which can be fitted into the conventional formulations of inorganic mechanics. Up to the present time the science of neurology
has been concerned almost exclusively with this aspect of the problem, with eminently successful results. Mechanical stress and tension, pressure and movements of fluids, local chemical action, surface
tensions and permeabilities, electrical phenomena of many kinds —
these and other physical factors now under investigation are bringing
to light many basic principles of nervous action. In a wider field
D'Arcy Thompson's great work. On Growth and Form ('44), does not
trespass beyond these boundaries, for he says (p. 15) : "When we use
physics to interpret and elucidate our biology, it is the old-fashioned
empirical physics which we endeavour, and are aloije able, to apply."
A useful general summary of Medical Physics, edited by Otto Glasser
('44), has recently been published. But this line of attack sooner or
later reaches limits beyond which it has not yet been possible to go.
In human neurology the major problem of all times has been the
relation of these physicochemical processes to the conscious experience which emerges from them. The normal subjective life is not disorderly, but the laws of this order as revealed by introspective psychology seem to be incommensurable and disparate with those of the
underlying physicochemical system as known objectively. This gap
has not been bridged by any acceptable formulation in terms of
Newtonian mechanics, and more and more of the experts in this field
are coming to believe that this cannot be done. This does not imply
any appeal to mysticism. Quantum mechanics takes its place in the
order of nature along with Newtonian mechanics, and it may well be
that the solipsistic qualities of that "private" conscious experience
which is not objectified are related to the events of the objectively
known "public" physicochemical world in accordance with principles as different from those of Newtonian mechanics as the latter
are from quantum mechanics. If so, these principles of the mind-body
relationships have not yet been formulated, and we live in hope that
some day this will be done. Just as quantum mechanics has added a
time dimension to the three Cartesian co-ordinates of space and additional dimensions beyond our range of experience in theoretic mathematical physics, so it may well be that introspective experience and
objective or extraspective experience are related in terms of dimensions not yet recognized and given scientific expression, defining a dimension as "any manifold which can be ordered," as Reiser ('46,
p. 89) does in an interesting discussion of this problem.
The theories of mind-body relationships are mentioned here because I regard mentation as a vital process as truly as are muscular
contraction and glandular secretion. The laws of operation of these
three processes are different, their products are different, and the
apparatus employed is different. All these operations have locus in
place and time. The organs which perform them have been slowly
differentiated and matured during embryonic and phyletic history.
We have reason to believe that mind is not an exclusive prerogative
of mankind. Mental capacity has developed parallel with the growth
of its organs. The genetic and phylogenetic approach to the mindbody problem has already yielded significant data; and, as this study
is carried backward toward earlier stages (embryonic and phyletic),
the unresolved problems of human psychophysics do not disappear.
Something akin to the mental as we experience it may be a common property of all living things and even of the cosmos as a whole,
as some suppose. Or it may be that, just as life emerged on our planet
from the nonliving in some as yet undiscovered way, so mind appeared as an emergent at some unknown stage of organic evolution.
If so, the naturalist must assume that the emergence in both cases
occurred in lawfully ordered ways within the frame of the natural.
In the present state of knowledge an open-minded skepticism on this
question is the only safe attitude.
More intensive study of the properties of the nervous tissues seems
to be the most promising approach to these unsolved problems. In
the past, escape from mystery has too often been sought through
verbalisms and mysticism. Rigid adherence to scientific method —
accepting as evidence not wishful thinking but verifiable experience — will avoid this pitfall. Obviously, conventional methods of
inquiry must be pushed to the limit of their availability, and in the
meantime new formulations of problems must be sought with all the
resourcefulness that scientific imagination can command, not neglecting the possibility that some of these formulations may lie outside
the frame of current Newtonian and quantum mechanics.
THROUGHOUT the preceding descriptions and interpretations,
some general morphological principles are implicit. Since these
are moot questions, it is fitting that these assumptions be explicitly
stated and clarified. Several years ago in a discussion of the morphogenesis of the brain ('33a) , some of these principles were critically
examined, and a part of what follows is condensed from that essay.
A century and a half ago the German Naturphilosophie elaborated
the mystic and poetic conception of an archetypical form, which was
popularized by Oken and Goethe and culminated with Sir Richard
Owen and Louis Agassiz. As Professor Owen wrote in 1849: "The
archetypal idea was manifested in the flesh under diverse such
modifications, upon this planet, long prior to the existence of those
animal species that actually exemplify it." These geistige Krcifte were
conceived as enduring morphogenic agencies which shape the course
of all animal and plant differentiation.
After the publication of Darwin's Origin of Species in 1859, a new
school of morphologists arose under the leadership of men like Gegenbaur, Haeckel, and T. H. Huxley, the guiding principle being a search
for phylogenetic relationships as key factors in morphogenesis. The
mystic enduement of the earlier archetype was replaced by a sound
biological principle of progressive change effected by verifiable internal and environmental agencies.
This conception of morphology as the visible record of phylogenetic history has stood the test of time. It is dynamic, not static; and
our search is for the natural agencies which have operated to produce
the observed modifications of form and correlated behavior during
the course of evolutionary change. This century of progress has, however, witnessed a curious relapse, with resurgence of the ancient predilection for rigid categories and artificial systems of logical analysis,
which yield in the end a formulation of inflexible, and therefore obstructive, morphological principles. Ancient and heritable patterns, such as metamerism, germ layers, and so on, were formalized, and the
tendency was to regard them as stable and immutable factors in
morphogenesis. The problems were thus simplified in terms of misleading logical categories, resulting in a static, rather than a dynamic,
analysis of development and evolution. These formal and rigid concepts have too often retarded, rather than facilitated, a true understanding of the structure.
The dialectic of some current "form-analytic" programs of research on morphogenesis of the brain seems to postulate a predetermined primordial pattern of the neural tube, which is preserved
throughout all stages of differentiation and which can be recognized
by the arrangement in space of cellular areas and their relations to
one another independently of their functional connections or of any
dynamic agencies in morphogenesis other than cellular proliferation.
The result is an oversimplified, logically consistent, morphological
schema uncontaminated by functional or other complicating factors;
but this has little interest for the working anatomists and physiologists, for it has no obvious relation with the structural forms with
which they are practically working. Nerve fibers are quite as important as cell bodies in cerebral organization and as elements in cerebral
forms. By what right does the morphologist ignore them in his study
of form? We have ample evidence that the growth of nerve fibers and
the migration of cells may be determined by functional requirements,
which differ from species to species in correlation with different
modes of life. Are forms which have physiological meaning of no significance in morphology, and can they safely be ignored by morphology.^
My paper of 1933 cites several illustrations from my own experience and that of others of the seductive influence of rigid categories of
morphological concepts, which simplify analysis by neglect of other
significant factors in morphogenesis. In modern morphology the
search is for genetic relationships, and homologies are defined in
terms of such relations. In all phylogenetic study we must constantly
keep in the foreground of attention two main classes of morphogenetic agencies. These are, first, the conservative factor of stable
genetic organization and, second, the more labile influence of the
specific functional requirements. Both factors are always present,
and one important task of the morphologist is the analysis of his
material so as to reveal the parts played by each of them. A sound
and fruitful morphology will take both into account. For the practical
purposes of descriptive anatomy and experimental physiology an
analysis in terms of functional efficiency is indispensable, and a morphology which ignores the dynamic factors of tissue differentiation in
terms of physiological adaptiveness lacks something and is sterile, as
I have repeatedly emphasized ('08, '10, '13, '226, '25a, '33a).
This does not imply that in either embryonic or phylogenetic development the pattern of structural differentiation is determined primarily by peripheral influences, which, acting directly on the germplasm, produce heritable changes in its structure. Whether such inheritance of acquired characters ever occurs is in controversy. Certainly it is the general rule that adaptation of structure to physiological requirements is effected by indirection — natural selection or
some other principle — but that it is acquired in some way is evident,
and these physiological requirements are key factors in morphogenesis. This is more evident in the nervous system, perhaps, than in
other organs of the body, for one of the prime functions of nervous
tissue is adjustment of the body to its environment.
The analysis of morphogenic factors is best accomplished experimentally, as is well illustrated by Holtfreter's recent study ('45).
This paper sets out in sharp relief the contrast between growth and
differentiation. The relation between intrinsic and extrinsic factors
and between nuclear and cytoplasmic influences in morphogenesis is
discussed in two recent books by W. E. Agar ('43, chap, v) and R. S.
Lillie ('45, chap. x). The reciprocal relations between genes and
cytoplasm are now under active investigation.
There is decisive evidence that in embryogenesis the pattern of
differentiation of both sensory and motor systems is determined
largely by intrinsic agencies and that it proceeds more or less autonomously up to functional capacity. These structural patterns are
laid down in the inherited organization, and that organization has
been elaborated in the course of phylogenetic development in adaptation to the physiological needs. of the species in question. Morphology
here ties in with ecology.
The last point carries with it the necessary implication that the
intrinsic agencies which initiate and shape the course of embryonic
differentiation are not strictly autonomous and that even the hereditary factors have arisen ab initio as responses of the protoplasmic
organization to environmental influences. It is this point of view
which has been emphasized by Child ('41) in his search for more
general morphogenic agencies which antedate the speciaHzed features
of ontogenetic differentiation. The old controversies about preformation versus epigenesis are outmoded and may well be ignored, for
both factors are present in all development. It can no longer be
claimed that "the basis of developmental pattern is an inherent
property of protoplasm and therefore continuously present and independent of external conditions" (Child, p. 6), for at no period in
the life of any organism is it independent of its environment. The
vital process is fundamentally an interaction between the protoplasmic organization and its surroundings— respiration, nutrition,
and all the rest. But when a given heritable pattern of internal organization has once been established, then the several components of
this organization may acquire a large measure of autonomy in both
further development and adult function. This specialization has obviously taken place; and when, in the course of development of
vertebrate neuromuscular organs, they are approaching functional
capacity, the patterns of this performance, though not independent
of internal and external environment, clearly are not directly determined by any specific influences acting upon them from receptor
organs. In this restricted sense these patterns are initiated intrinsically, and their performance is autonomous. But throughout development there is a complicated interaction of the inherited factors with
one another, and this intrinsic interplay of moiphogenic agencies is
an important feature of developmental mechanics.
The widely current belief that heritable variations sometimes occur
in progressive series set in a definite direction rather than always in
accordance with the probability curve of chance deviations around
fixed unit characters has been strongly supported by many independent lines of evidence. It has been pointed out that the progressive
senescence of tissue in both ontogenetic and phylogenetic series involves a fixation or stabilization of originally undifferentiated plastic
tissue into permanent structural patterns. So far as this differentiation is heritable and irreversible, the future course of evolution is thereby intrinsically determined, for variations will be distributed around the new pattern as a mode in accordance with a different frequency curve than would be shown if the inherited structural pattern were different. The process of differentiation is therefore itself a natural cause of limitation of the future course of evolution within boundaries set by the efficient working of the established
pattern. The nervous systems of arthropods, teleosts, reptiles, birds,
and mammals furnish illustrations of the effect of such irreversible
differentiation on the course of animal evolution. The significance of
this principle in the evolution of the nervous system was briefly discussed in my paper on orthogenesis in liHO, and more fully subsequently ('21, '22a, '24c, '26, '33rt, '48).
In the embryo, as soon as connection is made between the sensory
zone and the motor zone in the central nervous system, specific
peripheral influences begin to operate to a much greater degree, and
these may modify the subsequent course of development of the already fabricated sensory and motor apparatus and "inflect" the pattern of performance. The nature and degree of this modification by
use is, however, limited to the range permitted by the pre-existing
intrinsic organization (Coghill, '29, p. 86; Weiss, '41, p. 59). That this
range is very restricted in the Amphibia has been made clear by much
recent experimental work. The intrinsic growth factors which predetermine appropriate patterns of nervous organization in embryological development are operative also in the regeneration of peripheral nerves in the adult animal (see the experiments of Sperry cited
on p. 228).
With progressive increase in the complexity of adjustment to the
external and internal environment, there has been a corresponding
differentiation of structure. The genetic organization determines the
primary pattern, and this pattern is modified by use and the personal
experience of the individual. The first of these factors yields that
stable organization of tissue which is common to all members of the
species and which forms the subject matter of most of the literature
of neuroanatomy. The structural changes effected during growth by
the second factor are harder to recognize, and this critical field is
largely unexplored. In our examination of Ambly stoma we are
searching for primordia of both these kinds of tissue — the stable
heritable structure and the more labile, individually modifiable tissue
involved in conditioning of reflexes and other adjustments to personal experience. Both components are at a low level of differentiation in this animal, but their characteristic structure is recognizable,
and the successive steps of their further evolutionary development
can be followed.
Most species of urodeles apparently are reversals from more highly
differentiated ancestral forms. This, however, in its main features is
not a dedifferentiation but an arrest of development, so that adult
characteristics of the descendants resemble larval features of their ancestors. Such reversals and the accompanying instances of actual
degeneration doubtless are brought into being by changes in the
mechanism of heredity, involving rearrangement of the complicated
pattern of the genes (Emerson, '42, '45). Evolution is an irreversible
process, with divergent ramifications in inconceivable variety and
many instances of approximate parallelism. The apparent repetitions, when viewed in time, are not circular but spiral in form; and
apparent retrogressions are never exact reversals of the preceding
historical sequence. When a mammal returns to the water, it does not
become a fish. The amphibians present conspicuous illustrations of
recapitulation in individual development of some features of the
phylogenetic history, and most of the apparent retrogression is really
an arrest of development, which does not go back so far as the piscine
ancestry. Their limbs are not fins, and their gills are not fishlike.
There is another functional factor in morphogenesis to which too
little attention has been given— a greater or less capacity for socalled "spontaneous" activity, that is, behavior initiated internally
and manifested in patterns determined by the intrinsic structure.
All protoplasm is active as long as it is alive, and the essential vital
properties are inherent in protoplasmic organization. This behavior
is not a mere reflection of external influences, for the living stuff
transforms the energies which impinge upon it and recombines them
in original designs. Reserves are accumulated, and these are expendable on occasion in accordance with need, with or without external excitation. In the central nervous system this intrinsic "spontaneous" automaticity attains maximum potency, and it is exhibited
by even isolated fragments of it in characteristic patterns, of which
oscillographic records can be made. In the intact brain the interplay
of these intrinsic activities is always going on, and as we pass from
lower to higher animal types it becomes progressively a more and
more important factor in determining patterns of behavior. The
enormous reserves of potential nervous energy in the brain are
evoked and manifested as stabilizing influences (cerebellum, corpus
striatum, etc.) and also in that spontaneity, initiative, and inventiveness which culminate in cortically directed human behavior. These
capacities are shown in some measure by all animals, and a search for
the apparatus employed in even so lowly a creature as a salamander
may be fruitful (see '48, chap, xv, for discussion of Coghill's contributions on automatism, spontaneity, and motivation).
It has been pointed out by von Bonin ('45) that the logical foundations of the concept of morphology based on phylogenesis as developed by T. H. Huxley, Gegenbaur, and others of their time are insecure. The mathematical argument need not here be examined, for
it is based on certain restricted postulates, and in animal evolution
there are many variables not embraced by these postulates. It certainly does not follow that "the task of understanding structures on
the basis of their phylogenetic history" is "an insoluble problem."
Though neither cultural history nor phylogenetic history has been
reduced to mathematical formulation, there is general agreement
that history, judiciously interpreted, is an accredited guide for understanding the present and prognosticating the future. Phylogenetic
history is not a sealed book, and Dr. von Bonin assures me that he
would be the last to deny a positive value to the historical approach
to problems of morphogenesis and that such studies have actually
contributed much toward an understanding of human cerebral structure and function. The positive paleontological evidence regarding
the phylogeny of the brain is more extensive and illuminating than
is generally recognized. Though fossilized brains are unknown, the
very large number of casts of skull cavities, when skilfully interpreted, yield a surprising amount of reliable information about the
nervous organs which once occupied those cavities, as illustrated, for
instance, by Stensio's studies ('27) of fossil ostracoderms. It must be
freely granted, of course, that conclusions reached are tentative, to
be accepted only as checked against other lines of evidence, particularly the known sequence of evolutionary history as revealed by fossilized skeletal remains.
No single mode of attack upon problems of morphogenesis is adequate. Experimental methods yield the most decisive evidence, and
these require adequate knowledge of anatomical structure. This last
is the contribution of comparative anatomy and comparative embryology, and both of these must be functionally interpreted to be
fruitful. The anatomist should recognize the limitations of his method. His task is to lay foundations — stable and adequately broad —
and to suggest fruitful working hypotheses. The Amphibia occupy a
strategic position here for the same reason that the experimentalists
find them so useful.
Comparative study shows that some general features of structural
plan run through the vertebrate series with remarkable constancy
and that other features undergo amazing transformations. The discovery of the laws in accordance with which these transformations
are effected is the goal toward which we are working. We are confronted with a similar, but not parallel, series of problems in the
study of embryological development. In so far as we succeed in our
search for these laws, we advance our understanding of fundamental
vital processes.
The stable structural features of the nervous system are the most
useful landmarks for the comparative anatomist. They are expressions of the conservative hereditary factor in morphogenesis ; but this
stability cannot safely be interpreted as the simple manifestation of
some primordial archetypical pattern, for these features are retained
during the course of phylogenetic history only in so far as the fundamental features of the peripheral connections and their internal relationships are constant, that is, because they are parts of an apparatus
of adjustment to environment which is common to all vertebrates.
Other parts of the brain are more variable because, with complication of the behavior pattern in higher species, more elaborate and
diversified mechanisms of adjustment and integration are requisite.
In all vertebrate brains the most fundamental structural landmark is the transverse plane separating the spinal cord and rhombencephalon below from the cerebrum (as defined in the BNA)
above. In Amblystoma this plane is marked externally by the fissura
isthmi and internally by the sulcus isthmi. The zonal arrangement as
described in chapter v is well defined in the rhombic brain and the
midbrain, rostrally of which it is obscured by various secondary
modifications which become more complicated as we pass from lower
to higher members of the vertebrate series. A second important landmark is the transverse plane separating the diencephalon from the
telencephalon, marked externally by the deep stem-hemisphere
These two planes also mark the positions of two strong flexures of
the neural tube in early embryonic stages, caused by inequalities of
growth of the dorsal and ventral zones of the neural tube. The first
of these flexures to appear is a ventral bending of the neural tube in
the mesencephalic region, caused by precocious enlargement of the
tectum. This is followed by a flexure in the reverse direction at the
di-telencephalic junction (p. 212). These flexures are less obvious in
adult brains because they are somewhat straightened in later stages
and masked by growth of interstitial tissues; but the site of each of
them is a zone of transition between major divisions of the brain
with distinctive physiological characteristics.
The foundations of the current anatomical analysis of the brain were laid by
Wilhelm His in terms of human embryological development. The early neural tube
was divided into a linear series of blocks separated by transverse planes, and a
longitudinal sulcus limitans on each side marks the boundary between a dorsal
sensory alar plate and a ventral motor basal plate. The adult derivatives of this
embryonic mosaic are the primary anatomical units. The nomenclature derived from
this analysis as officially adopted (the BNA), or modifications of it, is now almost
universally employed, to the great advantage of human descriptive neurology. But
this scheme has its limitations. Some features of it are quite inapplicable to the
brains of lower vertebrates; for, though the embryonic neural tube is similar in most
of them, its adult derivatives vary so widely in adaptation to diverse modes of life
that no inflexible formula is applicable. Since the brain of Amblystoma is generalized,
few of these difficulties arise here.
There is difference of opinion about where the sulcus limitans ends anteriorly.
If the embryonic floor plate ends at the fovea isthmi (Kingsbury, '30), it is evident
that the basal plate extends farther forward to include the mesencephalic cerebral
peduncle and probably more or less of the adjoining parts of the hypothalamus and
ventral thalamus. The remainder of the mesencephalon and diencephalon (including
the retina) and the whole of the telencephalon are derived from the expanded anterior part of the alar plate and the related neural crest. The adult derivatives of
alar and basal plates include much tissue that is specifically neither sensory nor
motor; and this intercalated associational fabric crosses the boundaries of the primitive embryonic mosaic in ways which difi^er from species to species. Each species
must be analyzed in terms of its own mode of life and distinctive action system.
C. von Kupffer ('06) put special emphasis upon two deep transverse sulci in the
ventricular wall of the neural tube in early embryonic stages of a series of lower
vertebrates, including Necturus and Salamandra. These were termed sulcus intraencephalicus anterior and posterior. Study of later stages of developing urodele
brains shows that this emphasis was well placed, for these sulci mark the positions
of the two transitional sectors of the brain to which reference was made above —
the first, the diencephalic, and the second, the isthmic sector.
In our specimens of Amblystoma, von Kupffer's anterior sulcus in early motile
stages (Harrison's stages 33-36) is a sharply defined groove, which extends dorsally
from the lateral optic recess in front of the chiasma ridge to the region of the velum
transversum. Its dorsal part is varikble, but clearly the primary course is into the
posterodorsal initial evagination of the hemispheric vesicle, as described b.y von
Kupffer and by Rudebeck ('45). This is clearly the case also in A. jeffersonianum, as
shown by Baker and Graves ('32) in their five stages from 5 to 17 mm. long. My
published references to this sulcus and its adult derivatives have been successively
modified, as more material was examined ('10, pp. -119, 43:2; '£7, p. 238; '33b, p.
240; '35a, p. 252; "38, p. 212; '38b, pp. 401, 402; '39a, p. 262). These differences in
interpretation are doubtless due in part to the natural variability of the specimens
and in part to lack of a sufficiently close series of well-preserved stages to reveal the actual sequence of the changes. In our Amblystoma material the ventral part of this
sulcus seems to shift its position and to be transformed directly into the sulcus
preopticus, and it was so described ('39a, p. 262). Rudebeck finds in dipnoans,
Necturus, and Triturus that the sulcus intraencephalicus anterior passes from the
lateral optic recess dorsalward to the posterodorsal hemispheric ventricle and that
the definitive sulcus preopticus arises as a secondary outgrowth from this primary
groove. Our specimens of Amblystoma have not revealed this secondary origin of
the sulcus preopticus but resemble that of the anuran, Pelobates, as described by
Rudebeck ('45, p. 53).
In Amblystoma the posterior intraencephalic sulcus of von Kupffer persists as the
sulcus isthmi of the adult. In the 6-mm. Ammocoetes (von KupflFer's fig. 47) it
extends transversely from the plica rhombo-mesencephalica to a point in the floor a
short distance spinalward of the tuberculum posterius, i.e., to the fovea isthmi, and
it is similar in several other species figured. During larval development of x\mblystoma it and the related external fissura isthmi shift their relative positions (p. 179).
As emphasized above, this plane of separation between cerebrum and rhombencephalon, whether or not it is marked by a visible sulcus in the adult brain, in all vertebrates is the boundary between the two chief subdivisions of the brain.
The distinction between these subdivisions is conspicuous in prefunctional and
early functional stages. Coghill ('24, Paper IV, p. 97; '31, Paper X, p. 162) reports
that at all stages of development of Amblystoma from premotile to swimming the
rate of proliferation of cells and differentiation of neuroblasts is rapid in rhombencephalon and spinal cord, on the one hand, and in the cerebrum, on the other hand;
but "there is a distinct gap between fields of both differentiation and proliferation
at the isthmus. Such a gap does not appear at any other level in the brain or cord."
Diencephalon and telencephalon appear to be about equally involved in this process,
and so do rhombencephalon and cord; but during these stages growth appears to be
initiated independently in these two major divisions of the nervous system.
The experiments of Detwiler ('45, '46), to which reference has already been made
(p. 62), show that ablation of the cerebral hemispheres and visual organs of Amblystoma in prefunctional stages results in no demonstrable change in size or weight of
the medulla oblongata. He finds no evidence that the hemispheres or visual organs
exert any morphogenic influence upon the medulla oblongata up to a larval age of
48 days (33 mm. in length), and these mutilated larvae are capable of performing all
the ordinary feeding reactions.
The most primitive and fundamental patterns of total behavior
are organized in the spinal cord and lower medulla oblongata, particularly at the bulbo-spinal junction. Very early these come under
the control of the vestibular apparatus (Coghill, '30, p. 638) and
midbrain, and this control must be maintained throughout life if the
primitive mass movements are to retain their efficiency (Detwiler,
'45; '46). The isthmus, interpolated between these regions, seems to
be concerned mainly with the organization and control of local reflexes (partial patterns) of the medulla oblongata under the influence
of both ascending and descending systems of correlation fibers.
For convenience of description the rhombic h^-ain is here arbitrarily
divided into four regions, each with characteristic structure and
functions: (1) the bulbo-spinal junction, (2) medulla oblongata, (3)
cerebellum, (4) isthmus. In these regions the functionally defined
zones are unevenly differentiated, and the cerebellum as a "suprasegmental" apparatus is in process of emergence from the sensoiy
and intermediate zones. Similarly, in the forebrain the pallial field
exhibits an early prodromal phase leading toward cortical differentiation, as seen in reptiles.
The rhombic brain receives all sensory components of the cranial
nerves except the olfactory and the optic. In lower vertebrates, in
which the auditory apparatus is at a very low level of differentiation,
the sensory components of the rhombic nerves are relatively unspecialized, in sharp contrast with the highly specific optic and olfactory
systems; and the physiological dominance of the two systems last
mentioned in the control of behavior is the determining factor which
gives to the cerebrum unique properties that are in marked contrast
with those of the rhombencephalon.
The isthmus is a transitional sector, within which the patterns of
all bulbar activities are ordered and integrated. It bulks larger in
lower vertebrates than in higher, in which the cerebral cortex has
taken over the larger part of this control. Above it the cerebellum
was differentiated, not as part of the apparatus which patterns performance but as an ancillary mechanism on the efferent side of the
arc, to reinforce and regulate the execution of movements.
Within the cerebrum the two primary centers of dominance — optic
and olfactory — are separated by a similar transitional sector in the
diencephalon. This is plastic tissue, not dominated by any single
sensori-motor system; it is the meeting place of ascending and descending sensory paths. In noncorticated vertebrates we find here
the apparatus of a type of adjustment from which influences pass
forward into the hemispheres and there act as morphogenic agencies
in the elaboration of cortical structure.
It appears, then, that the loci of some characteristic features of the
vertebrate brain were fixed by their peripheral connections in the
earliest members of the series and that some of these have remained
essentially unchanged throughout the phylogenetic history. Others
have emerged very gradually from a nonspecific matrix which is diffusely spread throughout a wide field. The search for primordia of
the latter type in the lower forms as parts of a mosaic pattern with
rigidly defined boundaries cannot be successful. In each animal species the tissue requisite for successful adjustment to the mode of life
adopted is fabricated out of such raw material as is available, and nature is not bound by our formal rules of logical consistency. The
major subdivisions of the brain were thus defined very early in
vertebrate phylogeny, and they retain their general characteristics
throughout the series, but there is no apparent limit to the range of
modifications which these sectors may undergo in adaptation to
specific physiological requirements.
In premotile stages of Amblystoma, Coghill mapped several areas
in the walls of the neural tube, characterized by distinctive proliferation and differentiation. Adopting a modification of his scheme, I
gave arbitrary numbers to twenty-two such areas from the olfactory
bulb to the cerebellum ('37, p. 392), and their development can be
followed up to the adult stage. These units of the mosaic pattern of
the premotile embryo undergo remarkable shiftings of position during larval development and an equally remarkable diversity in patterns of differentiation and fibrous connections. When the comparative embryologist surveys the vertebrate series as a whole, he recognizes a striking similarity in the early stages of differentiation of the
neural tube of all of them. In later stages of development this similarity gives way to wide diversity in the progress of differentiation of
these primordial units of structure, and practically all this divergent
specialization can be seen to be directed toward adaptive modifications of structure, correlated with differences in the action systems of
the several species. Some limits to the range of this modifiability are
set by the inherent qualities of the genetic organization so that some
general principles of morphological pattern can be recognized everywhere. Yet the structure when viewed phylogenetically is remarkably
plastic, and the available materials are adapted to a wide variety of
uses in diverse combinations and interconnections in all the different
phyla. The most alluring feature of these comparative studies lies in
our ability to sort out of this apparent confusion of detail those
strong threads of ancestral influence which are interwoven in ever
changing designs under the influence of adaptive adjustment to different modes of life.
During the past half -century, morphology has seemed to be declining in favor, its problems submerged in the more attractive programs of the experimentalists. Nevertheless, activity in this field has
not abated, and now there is a renaissance, the reasons for which are
plain. Conventional methods of anatomical research have laid a secure factual foundation, but the superstructure must be designed on
radically different lines. Several centuries of diligent inquiry by
numerous competent workers have produced a vast amount of published research on the anatomy and physiology of the nervous systems of lower vertebrates; but most of this literature is meaningless
to the student of the human nervous system, and, as mentioned at
the beginning of this book, its significance for human neurology has
until recently seemed hardly commensurate with the great labor
expended upon it. The last two decades have inaugurated a radical
change, in which we recognize two factors.
In the first place, technical improvements in the instrumentation
and methods of attack have opened new fields of inquiry hitherto
inaccessible. To cite only a few illustrations, new methods for the
study of microchemistry and the physical chemistry of living substance, radical improvement in the optical efficiency of the compound microscope, the invention of the electron microscope, and the
application of the oscillograph to the study of the electrophysiology
of nervous tissue are opening new vistas in neurology, which involve
quite as radical a revolution as that experienced a few centuries
earlier when microscopy was first employed in biological research.
A second and even more significant revolution is in process in the
mental attitudes of the workers themselves toward their problems
and toward one another. A healthy skepticism regarding all traditional dogmas is liberating our minds and encouraging radical innovations in both methodology and interpretation. And, perhaps as a
result of this, the traditional isolationism and compartition of the
several academic disciplines is breaking down. The specialists are
now converging their efforts upon the same workbench, and cooperative research by anatomists, physiologists, chemists, psychologists, clinical neurologists, psychiatrists, and pathologists yields results hitherto unattainable. What is actually going on in the brain
during normal and disordered activity is slowly coming to light.
Here the comparative method comes to full fruition, and comparative morphology acquires meaning, not as an esoteric discipline dealing with abstractions but as an integral and indispensable component
of the primary task of science — to understand nature and its processes and to learn how to adjust our own lives in harmony with
natural things and events, including our own and our neighbors'
motivations and satisfactions.
The objective toward which we are directing our efforts is a better
understanding of human hfe and its instrumentation. Our mode of
hfe has been achieved througli eons of evolutionary change, during
which the conservative and rehitively stable organization of the
brain stem has been supplemented and amplified by the addition of
cortical apparatus with more labile patterns of action, resulting in
greater freedom of adjustment to the exigencies of life. In all behavior there is a substrate of innate patterns of great antiquity, and
in practical adjustments these primitive factors are manipulated and
recombined in terms of the individual's personal experience. Memory
and learning are pre-eminently cortical functions, but these cortical
capacities have not been given to us by magic, and we want to know
how they have been developed and the roots from which they have
The incentives which motivate research in comparative neurology
are the same as those of all other science, pure and applied, and of all
truly humanistic endeavor in other fields— to find out what is good
for humanity and how to get it. This implies, as I have recently exhorted ('44), that the humanistic values of science must always be
acknowledged and cultivated.
THE first part of this work includes a schematic outhne of the
organization of the amphibian brain and discussion of some
morphological and physiological principles suggested by this inquiry.
In Part II, evidence is presented in sufficient detail to document the
conclusions and principles summarized in Part I. This material is
arranged by topographic regions, as these are listed in chapter iv.
These descriptions are supplementary to those in Part I, and each
topic should be read in connection with the corresponding passages
of the preceding text.
THOUGH the spinal cord is not included in this survey, some
features of its upper segments must be considered here because
of their connections with the brain and particularly with the complicated structure at the bulbo-spinal junction. The spinal cords of
urodeles have not been adequately described, and our material is
not suitable for this purpose. Early stages of development of Amblystoma have been described by Coghill, and the cord of larval Salamandra by van Gehuchten ('97). A wide variety of experimental
studies of development involving the spinal cord have been reported
by others.
The slender dorsal gray column is continuous with the nuclei at
the bulbar junction, to be described shortly. These cells are activated
by the dorsal spinal root fibers and the spinal V and spinal vestibular
roots. A few of them in and near the mid-plane are a spinal continuation of the commissural nucleus of Cajal and are probably visceral
sensory in function. This supposition is supported by Sosa's description ('45) of similar cells in the septum dorsale of mammals and birds,
which he regards as spinal representatives of Cajal's nucleus.
Most of the dorsal root fibers immediately upon entrance bifurcate
into descending and ascending branches, and the latter in the upper
segments comprise most of the massive dorsal funiculi, with which
some other fibers are mingled, notably those of the descending
vestibular root and bulbar correlation tracts a and h of Kingsbury.
The large spinal V root lies ventrally of these funiculi, and below this
is a dorsolateral funiculus, containing fibers of the spinal lemniscus,
spino-cerebellar tract, and other fibers of spinal and bulbar correlation. Many of these dorsolateral fibers decussate, descending from
the dorsal gray as internal arcuates.
The intermediate zone is not clearly defined, its gray substance
being continuous with that of the motor zone. In the alba the
neuropil of the reticular formation is less well developed than in the
medulla oblongata.
The motor zone of the cord is continuous with that of the medulla
oblongata, with no recognizable boundary; and throughout these
regions the peripheral motor neurons are mingled with co-ordinating
neurons, and they often resemble the latter so closely that they cannot be distinguished unless their axons are seen to enter the motor
The upper spinal nerves are modified. The first usually has no
sensory root or ganglion. The arrangement of the motor roots of the
first and second pairs is exceedingly variable. The nervus accessorius
and the nervus hypoglossus are not separately differentiated. The
primordia of the former are represented in the lowermost vagal rootlets, which emerge from the lateral aspect of the medulla oblongata;
and the primordia of the latter are in the first and second spinal
nerves, the ventral roots of which emerge at the ventral surface.
In one specimen the lowest vagal root was seen to emerge at the level
of the calamus scriptorius, but usually it is at a more rostral level.
The lowest root of the first spinal nerve usually emerges in the region
of the calamus, and the first root of this nerve at Variable distances
The first spinal nerve of Salamandra as described by Francis
('34, p. 159) agrees with that of Amblystoma. On page 161 he quotes
Goodrich, who has shown that the urodele hypoglossus innervates
muscles derived from the ventral outgrowths of the second, third,
and fourth myotomes and that "the hypoglossus of Amphibia and
Amniota may certainly be considered as homologous, although not
necessarily composed of the same segmental nerves."
The neurons of the ventral horns of gray include tegmental elements, and motor cells which give rise to peripheral fibers ('44&,
fig. 10; van Gehuchten, '97). Both types may have very large, muchbranched dendrites, which in the larva ramify through almost the
entire cross-sectional area of the cord and may cross to the other side
in the ventral commissure. In the adult animal, internuncial connections within the dorsal and ventral zones and between these provide
for co-ordinated spinal reflexes; but all movements of the trunk and
limbs are subject to further control from bulbar and other higher
centers. The details of the structural apparatus by means of which
these ordered movements are effected remain obscure. At the inception of motihty in the embryo the first neuromotor responses to
stimulation are mass movements, and the apparatus of local reflexes
matures later (Coghill, '29). This implies that integrative functions
of total-pattern type mature earlier than do the partial patterns of
the local reflexes. Coghill's studies revealed a transitory system of
peripheral and central connections in the early stages of the development of motility when mass movements prevail, followed by radical
changes as the action system becomes more complicated.
Before the spinal ganglia are functionally mature, a series of transitory giant ganglion cells (Rohon-Beard cells) within the cord send
peripheral processes out to skin and myotomes and central processes,
which effect connection with the neuromotor elements. The transitory cells subsequently disappear and are replaced by the more specialized elements of the spinal ganglia.
Intramedullary cells of sensory type were observed by Humphrey
('44) in the spinal cords of human embryos. Two types of bipolar
sensory cells appear in embryos of 5 mm., one of which is transient
and is regarded as homologous with the Rohon-Beard cells. The other
type persists to functional stages, and at the beginning of motility
{'i''2.5 mm.) many of these are changing to a unipolar shape and resemble cells of the spinal ganglia. These intramedullary unipolar
cells are found in embryos of from 16 to 144 mm. in length, and are
regarded as functioning components of the dorsal roots in the early
stages of motility. Youngstrom ('44) also reports the occurrence of
sensory cells within the spinal cords of human embryos of from 19 to
63 mm. These cells are in the mantle layer and resemble those regarded by Humphrey as comparable with Rohon-Beard cells. Similar
intramedullary cells of sensory type have been seen by many others
in embryos and adults to accompany root fibers of spinal and cranial
nerves (see Pearson, '45, for illustrations) ; and it is probable that
these are all derived from the neural crest, like the mesencephalic
nucleus of the V nerve (as Piatt, '45, has demonstrated).
On the motor side of the arc two types of peripheral neurons were
described by Coghill ('26, Paper VI) and Youngstrom ('40) : (1) The
thick primary fibers appear first in ontogeny and course for long
distances in the ventral funiculus before emergence. The first ventral
root fibers arise as collaterals of these longitudinal axons. (2) Thinner
secondary fibers, which appear later, pass out from the gray of the
cord more nearlv transverselv- The Rohon-Beard cells are centrally so connected with the primary motor cells as to evoke mass movement of the musculature of the trunk in response to adequate stimulation of any kind. In subsequent stages central connections between
spinal ganglion cells and secondary motor neurons are made, and
these are regarded as provision for execution of local reflexes. The
primary motor neurons persist in adult Amblystoma. They occur in
larval anurans but disappear at metamorphosis (Youngstrom, '38).
Humphrey ('44) describes cells in the spinal cords of very young
human embryos, which she believes are surviving vestiges of primary
motor neurons of amphibian type.
lu our sections of the adult the ventral spinal roots contain fibers of
primary and secondary type. The primary root fibers are thick and
heavily myelinated centrally and peripherally. Some of the thinner
secondary fibers are well myelinated, and many of them seem to lose
their myelin as they emerge from the spinal cord. Coghill ('26, Paper
VI, p. 135) reported that in early swimming stages "a single fiber
may innervate an entire myotome, and branches of these same fibers
form the earliest nerves to limbs and tongue." At this early stage,
however, the musculature of the limb bud is still an undifferentiated
primordium. Youngstrom confirmed these observations and expressed the opinion that the limb musculature, like that of the trunk,
has a double innervation of both primary and secondary fibers; but
no details of the distribution of these fibers in the definitive limb
musculature are given. More recently, Yntema ('43a, p. 331) says of
the primary fibers in larvae of from 1''2 to 19 mm. in snout-anal
length that "typically, they supply the myotomic musculature. In
addition, fibers of this kind run to muscles of the extremities"; but
again details of their distribution in the limb are lacking. In a personal communication he adds: 'T have found evidence for the distribution of primary motor fibers to at least some muscles of the
girdles of larvae, and have seen larger fibers which appear to be primary in the limbs themselves."
In the frog, with an action system very different from that of
Amblj^stoma, the development of these nerves shows corresponding
differences, for, as mentioned above, Youngstrom ('38) found that
larval frogs have primary and secondary fibers like those of Amblystoma ; but in the adult frog the primary fibers have completely disappeared. The opinion expressed by Taylor ('44) that in frog larvae
the primary fibers do not enter the limbs may have no bearing on the
innervation of limbs in Amblvstoma because of the radical dift'erence
in the neuromuscular apparatus of these species. The primary fibers
evidently are concerned with massive movements of the trunk
musculature. The significance of the two sorts of fibers in the innervation of the limbs is still obscure. In Amblystoma the number of primary motor fibers is not markedly reduced by removal of the early
undifi^erentiated neural crest, while secondary fibers are, as a rule,
greatly reduced in number, the growth of the latter being dependent
on the presence of sheath cells and the former not (Yntema, '43a).
It will be of interest to learn whether the primary motor fibers have
functions in the embryogenesis of Amblystoma comparable with
those postulated for the "pioneer motor neurons" observed in the
bird by Hamburger and Keefe ('44, p. 237).
Little need be added here to the general description of this important region in chapter iv and to the details of structure and connections recently pubUshed ('446). The topography as seen in transverse
Weigert section is shown in figure 87. If the calamus scriptorius is
taken as the arbitrary boundary between spinal cord and brain, this
junctional region in Amblystoma may be considered to comprise the
segments of the first and second pairs of spinal nerves, the second
below the calamus and the first above. The entire length of the first
spinal segment overlaps the lower vagus region of the medulla
In the sensory zone the somatic sensory systems of the neurons of
the dorsal gray columns are somewhat enlarged to form the nucleus
of the dorsal funiculi, which extends far forward in the lower vagus
region. Medially of this is the much larger collection of compactly
arranged smaller cells of visceral-gustatory function^ — the commissural nucleus of Cajal. This nucleus extends downward from the
calamus for a distance of about one spinal segment, below which
visceral sensory function is represented by scattered cells in the
dorsal median raphe. Above the calamus the commissural nucleus
merges insensibly with the nucleus of the fasciculus solitarius.
The funicular nucleus is regarded as comparable with the external
cuneate nucleus of mammals rather than with the nuclei of the f.
gracilis and f. cuneatus, since Amblystoma has no medial lemniscus
('446, p. 318). The arrangement and connections of the commissural
nucleus are similar to those of man.
Secondary fibers from the funicular nucleus (many of them myelinated) pass downward as internal arcuate fibers to the spinal cord
and medulla oblongata, some uncrossed and some decussating in the
ventral commissure. Other crossed fibers join the tractus spinocerebellaris and the spinal lemniscus (fig. 3). The secondary fibers
from the commissural nucleus are unmyelinated. Some of them are
internal arcuates, which distribute to neighboring parts of the spinal
cord and medulla oblongata of the same and of the opposite side;
and some pass directly laterally to the pial surface, where they turn
rostrad in tr. visceralis ascendens (fig. 8; '44&, figs. 10, 11, 12, tr.v.a.)
to reach the superior visceral nucleus in the isthmus and the ventrolateral neuropil of the peduncle. Some further details about the connections of these nuclei are in the next two chapters.
The region of the calamus scriptorius is evidently an important
center of correlation and integration of general somatic and visceralgustatory sensibility of the entire body, with efferent discharge directly to the motor zone and also to higher centers of sensory correlation. Here root fibers of cutaneous and deep sensibility from the
head, trunk, and limbs; of vestibular and lateral-line sensibility; and
of gustatory and visceral sensibility converge into a common pool,
which is the first integrating center of these functional systems to
mature in ontogeny.
DETAILS of the peripheral distribution of the several systems of
nerve components have been recorded for a considerable number of amphibian species, notably in many important papers by
H. W. Norris. The first of this series was Strong's paper ('95) on
the larval frog, which was followed by Coghill's description of the
cranial nerves of Amblystoma, published in 1902. Their arrangement
here may be regarded as typical for the vertebrate phylum as a
whole, with no extreme specialization of any system. The constancy
of the arrangement of these components at the superficial origins of
the nerve roots in all vertebrates is remarkable, in view of the extreme diversity of both peripheral and central connections of their
fibers and of the enormous differences in the number of fibers represented in the several systems among the various species. Except for
the specific differences just mentioned, the chief departures from
uniformity of composition of the nerve roots are the suppression in
all Amniota of the large lateral-line components of the Ichthyopsida
and the correlated differentiation of the cochlear apparatus in the
higher classes.
The central connections of the olfactory and optic nerves and the
nervus terminalis are described in the chapters relating to the forebrain and the midbrain. The other functional systems are discussed
in chapters iv and v, and to those general statements some additional
details of their arrangement in Amblystoma are given here.
Some peculiar features of the development of the somatic motor
roots were mentioned in the preceding chapter. The development of
the visceral motor roots was described by Coghill, though many
details remain to be filled in. The early development of the sensory
systems of root fibers was studied by Coghill ('16, Paper II) and
Landacre ('21 and later papers). In Landacre's paper of 1921 the
embryos studied were identified as Plethodon glutinosus, but they
subsequently proved to be Amblystoma jeffersonianum (Landacre,
'26, p. 472). Older stages were described by Kostir ('24).
The embryological studies just mentioned were based on series of
normal embryos. The conclusions reached have been checked experimentally, extended, and in some details corrected by Stone ('22, '26)
and by Yntema ('37, '43), so that we now have very accurate information about the sources of the nerve cells of each sensory component of the ganglia of the V to X cranial nerves. The ganglion of
the trigeminus is derived chiefly from neural crest, which also contributes some cells to the ganglia of the VII, IX, and X nerves
(Landacre, '21, p. 15). Yntema ('37) found no neurons of neuralcrest origin in the facial ganglion ; but, since there is a small general
cutaneous component of this nerve in adult Amblystoma, it is probable that some cells of neural-crest origin are present, as is known to
be the case in some other animals. Part of the trigeminal ganglion
(profundus ganglion of Landacre, '21, p. 23) is delaminated from the
lateral ectoderm. The lateral-line ganglia are derived exclusively
from the dorsolateral placodes of the ectoderm, and the ganglion of
the VIII nerve from the auditory vesicle. The visceral ganglia of the
VII, IX, and X nerves arise from epibranchial placodes. Landacre
derived only the special visceral (gustatory) component of these
ganglia from these placodes, but Yntema has shown that the larger
part of the general visceral component also is of placodal origin. According to Yntema's analysis, epibranchial placodes give rise to general and special visceral components of the cranial ganglia, dorsolateral placodes to lateral-line components, the auditory placode to
the VIII ganglion, and neural crest to general cutaneous and general
visceral components. The mesencephalic nucleus of the V nerve is
derived chiefly from a portion of the neural crest which is incorporated within the neural tube, though it is not certain that this is the
exclusive source of these cells (p. 141, and Piatt, '45).
Here are included cutaneous sensibility of several modalities —
touch, temperature, pain, and, in aquatic animals, refined chemical
sensitivity. Associated with these nerves are those of deep pressure.
The nervous apparatus of these various qualities of sense has not
been successfully analyzed in lower vertebrates. Their fibers are
mingled peripherally and also centrally, except for those of the
mesencephalic V root. It is not improbable that some peripheral
fibers may serve more than one of the modahties of sense as centrally
The peripheral fibers of this system are usually described as the
general cutaneous component of the nerves, though some of them are
distributed to deeper tissues. Most of them enter the brain in the
trigeminus root and smaller numbers in roots of the VII, X, and
(probably) IX nerves. The vagal fibers of this system have wide
peripheral distribution (Coghill, '02), including the ramus auricularis
and other vagal branches and also anastomotic connections with
branches of the IX and VII nerves. The peripheral distribution of the
VII fibers has not been described. A few fibers of the sensory IX root
have been seen (rarely) to descend in the spinal V fascicles. Most
urodeles are said to lack a general cutaneous component of the IX
nerve, though there is some evidence of it in Necturus ('30, p. 22).
If the presence of these fibers is confirmed, they probably join the
general cutaneous component of the vagus peripherally.
Many of the trigeminal fibers divide immediately upon entering
the brain into the thick descending branches of the spinal V root and
thinner ascending branches (fig. 40) of the cerebellar root. The longest course that can be taken by one of these bifurcated fibers is
shown in figure 3. Some of these fibers take deeper courses, penetrating the spinal V root to enter the fasciculus solitarius (p. 148). Some
of the mesencephalic V fibers also divide near their exit from the
brain, with descending branches arborizing in the reticular formation
of the upper medulla oblongata (p. 141 and fig. 13).
The spinal V root is large and well myelinated. It can readily be
followed through the length of the medulla oblongata and for an
undetermined distance into the spinal cord (figs. 87-90). Each of its
fibers for its entire length is provided with a fringe of short collaterals
(fig. 38), which are directed inward into a neuropil which is continuous dorsally with that similarly related with the VIII and lateralline roots, the whole forming^a common pool for the reception of all
somatic sensory components (fig. 9). In the calamus region these collaterals mingle with collaterals and terminals of spinal root fibers of
the dorsal funiculus and the spinal vestibular root, bulbar correlation
tracts a and b, and the dorsolateral funiculus. This axonic neuropil is
permeated by dendrites of the nucleus of the dorsal funiculus and
commissural nucleus of Cajal.
In many of our Golgi preparations the central courses of the sensory V fibers are electively impregnated, often with no other fibers
visibk- in their vicinity. Some of these from the larva have })een ilhistrated ('14a, figs. 48-51, 54; '396, figs. 4'2, 46, 47, 57-61, 67, 77).
Figures 27-32 show V roots as seen in horizontal Cajal sections of the
adult brain. Figure 32 passes through the two motor V roots and
their nucleus; figure 40 includes three impregnated neurons of this
nucleus and the bifurcating fibers of the sensory root, which are also
shown in figure 38. Woodburne ('36, p. 451) saw thick root fibers of
the trigeminus entering the cerebellar root; but, in the absence of
Golgi sections, the thinner collaterals of the spinal root were not
The superior or cerebellar root of the trigeminus is much smaller
than the spinal root, and at the ventrolateral border of the auricle it
joins the spino-cerebellar tract (figs. 30, 31, 91). Many fibers of both
tracts end here with open arborizations in a neuropil which is the
primordium of the superior (chief, or pontile) nucleus of the mammalian trigeminus ; but some fibers of both tracts pass through this
neuropil and continue dorsomedially into the body of the cerebellum,
where they end, some on the same side and some decussating in the
commissura cerebelli (figs. 31-34, 37, 91). These commissural fibers
are joined by others arising from cells in the vicinity of the superior
trigeminal neuropil, and many of the decussating fibers, after crossing, reach the superior neuropil of the other side, thus forming an
intertrigeminal commissure. We here confirm, in the adult, Larsell's
description ('32, p. 413) of the cerebellar commissure of the larva as
composed chiefly of trigeminal and spinal components. In Amblystoma the superior sensory nucleus of the trigeminus probably is concerned chiefly with the proprioceptive aspects of cutaneous sensibility (deep proprioception being provided for in the mesencephalic V
root). This cerebellar connection persists in man, but here the chief
V nucleus has also acquired refined types of sensibility which Amblystoma lacks.
Neither the superior nor the spinal nucleus of the trigeminus has
well-defined boundaries. The central cells which engage terminals and
collaterals of the sensory V fibers may also have synaptic contacts
with terminals of all other sensory systems that enter the medulla
oblongata. There are, however, certain lines of preferential discharge
for each group of sensory systems, and the segregation of local nuclei
and secondary pathways for each functional system is incipient.
These special somatic sensory systems are closely related genetically, structurally, and physiologically, but much remains obscure
about their relationships. The labyrinthine apparatus seems to be
at the focus of these systems. It is very conservative, except for the
cochlear part, showing relatively little change in structure and function from lowest to highest vertebrates; moreover, its physiological
properties have been thoroughly explored. The lateralis system attains its maximum in fishes, persists in larval amphibians and adults
of some urodeles, and disappears entirely in all higher groups, both
embryonic and adult. Organs of hearing are poorly developed in
fishes. Auditory functions seem to be performed by the vestibular .
apparatus and also (for slow vibration frequencies) by the organs of
the lateral line, which undoubtedly have other functions also.
The peripheral end-organs of all these systems are specialized
epithelial structures, in contrast with the free nerve endings of the
general somatic system. The vestibular end-organs of the internal
ear resemble the end-organs of the lateral lines, in that in both cases
there are specialized epithelial cells which are the receptive elements.
The epithelium is thickened, and among the slender elongated supporting elements there are shorter ovoid cells with ciliated outer
ends. These specific nerves have thick myelinated fibers, the
branched unmyelinated terminals of which closely embrace the cell
bodies of the specific receptive elements (Larsell, '29; Chezar, '30;
Speidel, '46).
The lateral-line organs of Amblystonia are papillae, some of which
are depressed in pits but are not inclosed in canals as in most fishes.
Their arrangement conforms with the general pattern in fishes, with
rows above and below the eye, on the lower jaw, and extending into
the trunk as far back as the tail. The related nerves comprise one of
the largest systems of the larva, which is reduced but not lost at
metamorphosis. These thick and heavily myelinated fibers enter the
brain in two large roots spiimlward of the VIII roots and three or
four which enter dorsally and slightly rostrally of the VII roots.
They are conventionally assigned to the VII and X pairs of nerves,
though they are more properly aligned with the VIII roots.
The arrangement of these roots is shown in figures 7, 9, 89, 90.
Most of their individual fibers bifurcate immediately upon entrance
into the brain into ascending and descending branches with numerous widely spread collaterals. These root fibers are arranged in
fascicles, which span almost the entire length of the medulla oblongata (fig. 7), except for the most dorsal of the three or four
lateralis VII roots, which ends in a "dorsal island" of neuropil
("cerebellar crest" of Larsell) at the level of entrance (figs. 7, 33, 45).
Lateral-line fibers have not been seen to descend into the spinal cord.
Anteriorly, they enter the auricle and end here (fig. 91); none have
been traced into the body of the cerebellum, though secondary
lateralis fibers after synapse in the auricle enter the com. vestibulolaterahs cerebelli in company with vestibular fibers (figs. 32, 33, 34,
The exact functions served by the lateral-line organs are still imperfectly understood. The organs of the lateral lines and those of the
internal ear have many similarities in embryological development,
structure of the receptive apparatus, and central connections. They
probably have had a common evolutionary origin from a more generalized form of cutaneous sense organ similar to the so-called "sensillae" of some invertebrates. This may be the explanation of the intimate association in the human ear of sense organs of such diverse
functions as the cochlea for hearing and the semicircular canals for
equilibration, both being highly refined derivatives of primitive
tactile organs. The sense organs of the lateral lines are probably intermediate in function between tactile sensibihty of the skin and the
auditory and equilibrating functions of the internal ear. In fishes
they have been shown to be sensitive to mechanical impact, slow
vibrations, and currents in the water (Parker and Van Heusen, '17;
Parker, '18). Hoagland ('33) and Schriever ('35) have investigated
the functions of lateral-line nerves of fishes with the aid of oscillograph records of their action currents. Hoagland finds that these
organs are in a state of continuous activity and that the nervous discharge is increased by application of pressure, by ripples and currents in the water, by movements of the trunk muscles, and by
temperature changes.
In Amblystoma larvae Scharrer ('32) found evidence that the
lateral-line organs may participate in the snapping reaction when
moving prey is seized; and, subsequently, Detwiler ('45) reports that
the lateral-line organs of these larvae constitute an adequate receptor
apparatus for the detection of food in motion after extirpation of the
eyes and nasal organs. The central connections of these nerves sug
gest that they play an important part in proprioception, and this is
supported by Hoagland's experiments.
The VIII nerve of Amblystoma carries fibers from the membranous labyrinth, the structure of which resembles those of fishes
plus a recognizable rudiment of the cochlea. These fibers enter the
brain by two closely associated roots, dorsal and ventral, each of
which contains many rather fine myelinated fibers, with some very
coarse fibers mingled with them. Each fiber has a T-form division
within the brain, the branches ascending and descending through the
entire length of the medulla oblongata (figs. 7, 87-90). The dorsal
and ventral roots remain separate as far forward as the V root and
backward as far as the second root of the vagus. Beyond these limits
the two roots merge. It is evident that fibers of the ventral fascicle
take longer courses within the brain than do those of the dorsal
fascicle, but the significance of the separation of vestibular fibers
into two roots has not been determined. Some of these fibers descend
for a long and undetermined distance into the spinal cord, mingled
with the more ventral fibers of the dorsal funiculus and those of correlation tract b. The ascending fibers enter the auricle (figs. 29, 30,
31, 91), where many of them end. Others continue into the body of
the cerebellum, decussate in the vestibulo-lateral cerebellar commissure, and terminate in the vestibular and lateralis neuropil of the
auricle of the opposite side (figs. 32, 33, 34).
Within the medulla oblongata the collaterals and terminals of the
vestibular fibers arborize in the common pool of neuropil, which also
receives terminals of the V and lateral-line roots. Most of the neurons
of the second order in the acousticolateral area spread their dendrites
within this neuropil so as to engage terminals of several of these
fascicles of root fibers of different physiological nature (fig. 9).
There is ample physiological evidence that salamanders exhibit
vestibular control of posture and movement similar to that of other
animals, and this implies that there is some central apparatus that is
selective for the specialized end-organs of the internal ear. Sperry
('45a) has shown that in the case of the frog this specificity is. preserved after section of the VIII nerve and its subsequent regeneration
and that the precision of restoration of vestibular function is quite as
exact as it has been shown to be in the case of regeneration of the
optic nerve (p. 229). Since the specific functions of the several vestibular end-organs are not visibly localized in the medulla oblongata of the salamander, some other method of selection must be
employed. Sperry's experiments on frogs lead him to favor the supposition that there are physicochemical axon specificities and selective contact affinities between the different axon types and neurons
of the vestibular centers, a supposition which accords with much
other evidence (p. 79). Differences in threshold and the time factor
in the transmission rhythm may act selectively at the central
Though Amblystoma has no recognizable cochlear root of the VIII
nerve, there is a primordium of the pars basilaris cochleae, which is
better developed in the frog. This rudiment is lacking in Necturus, so
that among the Amphibia successive stages in the early differentiation of the cochlear apparatus can be observed.
The fibers of the dorsal lateral-line VII root are shorter than those
of the others, all ending in the "dorsal island" of neuropil at the
posterior border of the lateral recess of the ventricle (figs. 33, 45;
'446, fig. 14; Larsell, '3^2, fig. 57). These fibers, like those of the other
lateralis VII roots, come from lateral-line components of all three
chief peripheral branches of the lateral-line VII nerves ('14a, p. 357).
The dorsal island appears to be a remnant of the dorsal neuropil
(cerebellar crest) of the lobus lineae lateralis described by Johnston
('01) in fishes.
In very young larvae of the frog (Larsell, '34) the relations are
similar to those of urodeles, but soon a dorsal branch of the VIII
nerve (derived from the primordial cochlea) enters this area dorsally
of the dorsal lateral-line root. The dorsal island of neuropil retains its
individuality to the time of metamorphosis, meanwhile becoming
entirely surrounded by cells which proliferate from the dorsal lip of
the area acusticolateralis. The dorsal VIII root is greatly enlarged to
become the cochlear nerve; it terminates in relation with the cells
surrounding the dorsal island, which now constitute the cochlear
nucleus. After metamorphosis is complete, all lateral-line fibers degenerate, so that the gray of the larval area acusticolateralis becomes
in the adult the cochlear nucleus dorsally and the vestibular nucleus
Without here going into the further details of this differentiation,
it is evident from Larsell's studies that the dorsal gray of the area
acusticolateralis of urodeles and larval anurans is, during the metamorphosis of the frog, transformed directly into cochlear nuclei.
There is no degeneratioii of these cells and rei)lacement by others.
The same neurons which in the tadpole are activated from lateralline organs lose their lateral-line connections in the adult frog and
receive their excitations from the auditory apparatus, with a radical
change of function. That which I at one time regarded as improbable
('30, p. 60) is exactly what happens in ontogeny, and doubtless the
phylogenetic history is similar, as Ariens Kappers has long maintained. Parallel with the differentiation of the cochlear nerve and
nucleus in anurans, the related lateral lemniscus is enlarged and
specialized in its definitive form.
In Necturus, which has no recognizable cochlear primordium, the
functions of the dorsal cells of the area acusticolaterahs evidently are
related exclusively with lateral-line organs. The dorsal lateralis VII
root terminating in the dorsal island does not differ physiologically
from the other lateral-line roots of the VII nerve, so far as known.
Like them, it receives fibers from lateral-line organs distributed over
the entire head ('30, p. 21). But there is an obscure indication of a
lateral lemniscus. Why do the fibers of the dorsal lateral-line root end
in the restricted area of the dorsal island instead of extending through
the whole length of the acousticolateral area like the other lateralis
roots .^ The answer is probably to be sought in the phylogenetic history of the extinct ancestors of living urodeles. In fishes the lobus
lineae lateralis of this region is covered by a neuropil, which has been
termed the "cerebellar crest" and which extends forward into continuity with the superficial neuropil of the cerebellum. Larsell ('32,
p. 410) regards the neuropil of the dorsal island as a survival of the
cerebellar crest of fishes. This is the region within which the dorsal
cochlear nucleus of anurans has been differentiated; and, if, as is
generally believed, the living urodeles are descendants of more highly
specialized ancestors with better organs of hearing, the preservation
of their dorsal island may be regarded as a vestigial record of an
ancestral history now lost.
The amphibian auricle (pp. 20, 44) receives terminals of trigeminal, lateral-line, and vestibular fibers. The connections of these fibers
and their secondary pathways make it clear that this area contains
primordia of two quite distinct mammalian structures. One of these
is the terminal station of lateraMine and vestibular root fibers, and
this tissue in higher animals is incorporated within the cerebellum
and becomes the flocculus, as described by Larsell. The other primordium is trigeminal, and this in Amblystoma is probably concerned chiefly with proprioceptive functions of the skin and deep
tissues of the head, as indicated by its strong cerebellar connection.
This connection persists in mammals but is relatively insignificant
here because, as mentioned above, the enlarged mammalian superior
V nucleus is concerned chiefly with refined functions of the skin that
Amblystoma does not possess.
The system of the mesencephalic nucleus and root of the trigeminus is here well developed in typical relations, with some instructive
special features. Its thick, well-myelinated fibers go out with
branches of the V nerve. The details of their peripheral courses in
Amblystoma have not been described. Experiments by Piatt ('46)
indicate that the majority of the fibers of this system, which go out
from the tectum opticum, are distributed to the jaw muscles. The
more caudal cells of the mesencephalic nucleus probably have other
connections. No evidence has been found for supply of any eye
muscles from this nucleus.
Unlike other sensory systems, the cell bodies of these neurons lie
within the brain. Their arrangement and the courses of the fibers
arising from them have been described in the larva ('14a, p. 361) and
in the adult ('36, p. 345) and are shown diagrammatically in figure
13. These cells vary in size, cytological structure, and number. Most
of them are very large and of so characteristic appearance that they
are easily recognized. They are sparsely distributed throughout the
tectum in all layers of the gray substance, somewhat less numerous
anteriorly near the posterior commissure, and densely crowded within and adjoining the anterior medullary velum. Occasionally, they
are seen in the body of the cerebellum and in the nucleus cerebelli.
Ten large larvae of A. punctatum had an average of 159 of these
cells, the extremes being 76 and 208 (Piatt, '45). A subsequent count
by Piatt ('46) of the total number of these cells in ten larvae of 45
mm. gave an average of 261 cells, equally divided on right and left
sides. Of these cells, 86 on each side are in the tectum opticum and 45
in the nucleus posterior tecti and velum medullare anterius. Individual variations in numbers of cells are large, but approximate bilateral
symmetry is quite consistently present. These cells are unipolar, the
single thick processes accumulating near the outer border of the
tectal gray and here acquiring myelin sheaths. These fibers are di
rected posteroventrally in loosely arranged dorsal and ventral fascicles, which converge toward the V nerve roots.
The ovoid cell body has a smooth contour, with no processes except
the single thick fiber. It is imbedded in dense neuropil and closely
enveloped by a web of these fibers. Every contact of the fibers of the
neuropil with the cell is a synaptic junction. This is doubtless the
explanation of the wide dispersal of these cells in all parts of the
tectal gray, and they are so arranged that the entire extent of the
deep tectal neuropil may be simultaneously activated by excitation
of the mesencephalic V system.
The striking resemblance of the cells of the mesencephalic V
nucleus with those of the semilunar and spinal ganglia and with the
transitory Rohon-Beard cells of the spinal cord (Coghill, '14, Paper
I) has often been commented upon and is well illustrated by the
excellent photographs pul)lished by Piatt. They have, accordingly,
been generally regaixled as derivatives of the embryonic neural crest
that have remained within the neural tube. This hypothesis has been
tested experimentally by Piatt ('45), with the conclusion that neural
crest is at least one source of these cells, though a possible origin from
other sources is not excluded. Many observers have reported the
presence of intramedullary cells of sensory type along the courses of
roots of spinal and cranial nerves (Pearson, '45, cites instances), and
some of these cells also may be of neural-crest origin. Others may be
of autonomic type, migrating out from the brain (Jones, '45), though
this is controverted.
Just as the typical unipolar cells of the sensory ganglia of spinal
and cranial nerves have a single process, which divides into peripherally and centrally directed branches, so the mesencephalic V fibers
(or some of them) divide shortly before emergence from the brain
into peripheral and central branches (fig. 13). The central branches
descend as far as the level of the IX nerve roots. My earlier statement
('14a, p. 362) that these fibers ."arborize among the dendrites of the
motor VII neurons" is misleading, for these terminals are spread
widely in the intermediate zone between the levels of the V and the
IX roots.
This bifurcation of the root fibers and the fact that the bodies of
the cells of the mesencephalic V nucleus are in synaptic contact with
all the deep neuropil of the tectum suggest that afferent impulses
transmitted by these fibers may take either or both of two courses:
(1) They may pass upward to the tectum, where they activate the deep neuropil diffusely and here are in relation with terminals of the
optic and lemniscus systems; or ('2) they may descend into the
reticular formation of the medulla oblongata, where they act directly upon the motor nuclei and also upon the apparatus of bulbar
neuromotor co-ordination (including the cells of Mauthner). These
descending branches are accompanied by fibers of the spinal V root
and by other fibers from the tectum and tegmentum, which end in
the same field of the reticular formation (fig. 13, tr.t.h.p. and tr.teg.b.).
In larvae of early feeding stages, thick uncrossed fibers, which descend from the tectum and subtectal areas into the bulbar reticular
formation, are especially clearly seen, and also others which take
similar courses after decussation in the ventral commissure. Some of
these fibers arise from neurons of the isthmus, which are in synaptic
connection with terminals of the secondary visceral-gustatory tract.
These connections of mesencephalic V fibers seem well adapted to
facilitate the feeding reactions, a conclusion which is supported by
observations on the cat by Corbin ('40) and the literature which he
cites. In Ambly stoma the field of reticular formation within which
the movements of the mouth and pharynx are organized receives the
descending mesencephalic V fibers, collaterals of V fibers, and fibers
of correlation from the nucleus of the f. solitarius, isthmic visceralgustatory nucleus, tectum, and the underlying dorsal tegmentum.
Control of the course of muscular movement in process is insured
by a variety of sensory end-organs, including those in muscles,
tendons, joints, and the overlying skin. At the beginning of motility
in the embryogenesis of Amblystoma a single peripheral sensory
element (the transitory Rohon-Beard cells) may perform both exteroceptive and proprioceptive functions (Coghill, '14, Paper I, p. 199),
and this may be true of some spinal ganglion .cells in the adult,
though here special proprioceptive apparatus also is provided. The
Rohon-Beard cells are believed to be derived from a portion of the
neural crest which is incorporated within the neural tube; and the
mesencephalic nucleus of the trigeminus, as just described, has a
similar origin. The latter cells survive in the adults of all vertebrates,
in the service apparently of co-ordination of movements involved in
the feeding reactions.
In the head the membranous labyrinth is the dominant organ of
this system, with participation of nerves of cutaneous and deep
sensibility, and probably the lateral-line organs also. Some proprioceptive control is doubtless organized in the reticular formation of
the cord and bulb, but we have little information about how this is
done. From the entire sensory zone of these regions, proprioceptive
influence is filtered off and directed to the cerebellum, which is the
general clearing-house for these functions. Many vestibular root
fibers and a smaller number of trigeminal fibers go directly to the
cerebellum, and secondary fibers from the sensory zone enter it by
way of the spino-cerebellar tract and bulbar correlation tracts a and b
(p. 159; '44&). That exteroceptive and proprioceptive functions are
not completely segregated in these brains is shown by the fact that
many fibers of the spinal lemniscus (tractus spino-tectalis) send collaterals into the cerebellum ('14o, p. 376). Within the cerebellum the
general somatic sensory and vestibular components of the proprioceptive system are locally segregated, the former in the body of the
cerebellum and the latter in the auricle, and this localization is a
primary feature of the cerebellum in all vertebrates, as Larsell has
shown. This author ('45) has also made it clear that cerebellar function includes much more than proprioception, or else the concept of
proprioception must be redefined in more inclusive terms. The second
alternative, I think, is better, as I have suggested in an article ('47)
on the proprioceptive system, from which some of the following paragraphs are taken, by courtesy of the editor of the Journal of Nervous
and Mental Disease.
Sherrington ('06, p. 347) defines the cerebellum as the head ganglion of the proprioceptive system, taking as the basis for his classification of receptors "the type of reaction which the receptors induce."
In his exposition of this idea he makes it clear that the proprioceptive system is segregated from other sensory systems, not in terms of
the receptors involved but because the system as a whole exerts
regulatory control over the action of all skeletal muscles. The criteria
employed here are applied in the efferent, not the afferent, side of the
arc. In view of present knowledge of cerebellar function, Sherrington's original concept of proprioception should be emphasized and
It has long been recognized that in the cerebellum of lower vertebrates the sensory inflow is of two kinds, which are separately localized, viz., (1) the vestibular and lateral-line systems in the lateral
part and (2) the spinal and trigeminal systems in the median body.
The second category traditionally comprises deep sensibility of several sorts, notably that of muscle spindles, tendons, joints, and some
other internal end-organs. Current physiological research requires
radical revision and broadening of this traditional analysis. It has
been shown that in mammals different cutaneous areas, vibrissae,
audition, and vision have local representation in the cerebellum, as
do also various systems of synergic muscles. In lower vertebrates the
cerebellum has a broad connection with the hypothalamus, implying
representation in the cerebellum of olfactory sensibility also.
In brief, cerebellar control of muscular movement employs practically all modalities of sense represented in the action system of the
animal. The function of the cerebellum as the "head ganglion of the
proprioceptive system" is not to pattern the muscular response (for
these functions are localized elsewhere) but to facilitate its execution ;
and this facilitation employs all available sensory experience. Many
organs of sense perform simultaneously both exteroceptive and proprioceptive functions. Sherrington's fruitful analysis of the action
system into interoceptive, exteroceptive, and proprioceptive components was not based upon the specificities of the receptive organs,
considered either anatomically or physiologically; but, on the contrary, the distinction was drawn in terms of what the animal does in
response to sensory excitations. The interoceptive systems are defined in terms of internal adjustments, chiefly visceral. The exteroceptive systems are those which evoke adjustments of the body or its
members to events in the external world. The proprioceptive systems
are ancillary to the activities of the musculature in maintenance of
tonus, posture, and regulation of the action of synergic groups of
agonist and antagonist muscles in appropriate strength and sequence.
Proprioception, therefore, must be defined not in terms of the
modalities of sense employed but in terms of the results achieved.
Cerebellar proprioceptive control is accomplished by the application
of all relevant types of sensory inflow to specific and successive muscular activities which may be in process from moment to moment;
and the definition of "proprioception" must be formulated in terms
of the motor response rather than of the sensory systems involved.
The proprioceptive system, accordingly, includes all peripheral endorgans and nerves and all central adjustors in the spinal cord, brain
stem, cerebellum, and cerebral hemispheres that collaborate in the
co-ordination and synergizing of muscular activity in process. In a
recent conference with Dr. Larsell he suggested to me that, in view
of the inadequacy of current conceptions of the true nature of the
proprioceptive system and the faulty connotations of the term in
present usage, it might be better to avoid the word hereafter and
replace it by the more inclusive name, "proprius system."
The preceding comments on the proprioceptive system apply,
mutatis mutandis, to Sherrington's exteroceptive and interoceptive
systems also. In his original definitions of these terms, Sir Charles
was careful to insist that each component of each of these three
subdivisions of the total pattern of behavior must be viewed in its
entirety as a unitary act and that the significance of these acts can
be understood only in terms of their reciprocal relationships with one
another and with the total action system of the animal. The critical
feature of each of these acts is the end-result, the actual behavior
exhibited. The names originally given to these three classes of functions put the emphasis on the receptive organs, w^here it does not
belong. Some obscurity and confusion may be avoided if the unity of
these several components of behavior is recognized in their nomenclature. The exteroceptive systems, viewed in their entirety, are
somatic, the interoceptive systems are visceral, and the proprioceptive
systems are ancillary to all muscular activity and, accordingly, may
be termed proprius. Sherrington's terms, "exteroceptors," "interoceptors," and "proprioceptors" are suitable names for the receptive
organs, with the qualification that the same organ may, on occasion,
activate somatic, visceral, or proprius responses.
General visceral sensory fibers of wide peripheral distribution
enter the brain by the vagus roots, and the IX and VII roots contain
smaller numbers of similar fibers from the mucous surfaces of the
mouth and pharynx. Taste buds are widely distributed in these
mucous surfaces, and the gustatory fibers are indistinguishably
mingled with the general visceral fibers peripherally in the roots of
the VII, IX, and X nerves and centrally in the f . solitarius, more of
them entering the brain anteriorly than posteriorly. This mixed
group of peripheral fibers, as a whole, is quite distinct from all other
functional systems and it was termed by the earlier students of nerve
components in lower vertebrates the "communis system" because all
its fibers converge into a single central bundle, the f. communis
(Osborn, '88). This we now know is homologous with the mammalian
f. solitarius. The peripheral and central arrangements of the chemoreceptors illustrate some general principles which will next be examined.
The peripheral terminals of the sensory fibers of the V to X cranial
nerves take three forms: (1) The fibers of the general somatic sensory
and visceral systems have free nerve endings within or beneath
epithelium or widely spread in deeper tissues. (2) The end-organs of
the special somatic sensory systems are differentiated epithelial
structures of the internal ear or lateral lines with receptive hair cells,
which are shorter than the surrounding supporting cells. (3) The
chemoreceptors of the gustatory (special visceral sensory) system
are budlike epithelial structures, which resemble the naked lateralline organs but differ from them in that the specific receptive cells are
slender, elongated elements, which span the entire thickness of the
epithelium. The fibers which innervate them are generally thinner
than those of lateral-line organs and are less myelinated or unmyelinated.
Those species of fishes which have taste buds abundantly distributed in the outer skin and also naked organs of the lateral lines not
inclosed in pits or canals present both morphological and physiological problems of great diflSculty ('03, '03a, h). In the earlier literature
all these cutaneous organs were termed indiscriminately "terminal
buds," with resulting confusion which was not clarified until the
nerve fibers which supply them were found to belong to different
functional systems. The fibers supplying lateral-line organs, wherever
situated, converge centrally into the acousticolateral area, and fibers
supplying taste buds,whether in mucous surfaces or in the outer skin,
converge into the f. solitarius and its nucleus. The separation of the
gustatory from the lateral-line system of cutaneous sense organs by
the anatomical method has been confirmed by physiological experiments performed by the writer, G. H. Parker, and others.
Though this distinction is perfectly clear in some species of fishes,
in others there are transitional forms of "terminal buds," and much
remains obscure about the functions of these various types of receptors. The problem is complicated by the fact that in fishes the skin is
everywhere very sensitive to a large variety of chemical substances
(Sheldon, '09; Parker, '12, '22; Ariens Kappers, Huber, Crosby, '36,
chap, iii; for a more general discussion of the chemical senses see
Moncrieff, '44). The skin is sensitive, in general, to different substances from those which activate the olfactory organ and taste
buds, but there are some puzzling exceptions.
For instance, the gurnard fishes (Prionotus, Trigla) have three
rays of the pectoral fin which are mocUfied to serve as "feelers" in the
search for food on the floor of the sea. Somewhat similar filamentous
pelvic fins of the gourami, codfish, and several other teleosts are
abundantly supplied with taste buds with the usual functions and
nervous connections ('00, '03; Scharrer, Smith, and Palay, '47); but
the free pectoral fin rays of the gurnards have no taste buds, and yet
it has been shown (by the authors last mentioned) that these fin rays
are sensitive to the same sapid substances as are the cutaneous taste
buds of other fishes and that the reactions also are similar. These
authors, in tracing the central courses of the large nerves which supply
these free fin rays, find that these fibers have central connections
similar to those of the pectoral fins of other fishes, belonging, that is,
to the general cutaneous system. They do not enter the f. solitarius.
They present evidence also that some secondary fibers from these
general cutaneous centers connect centrally with the superior gustatory nuclei of the isthmus and hypothalamus, just as do the true
gustatory fibers arising in the nucleus of the f, solitarius. This is interpreted to mean that these nerves of general cutaneous chemical
sensibility are so specialized that they can serve typical gustatory
reactions, though they do not connect peripherally with taste buds.
These observations seem to show that some peripheral fibers of the
general cutaneous system, without specialized receptive end-organs,
may acquire functions substantially identical with those of cutaneous
taste buds and that such fibers have central connections similar to
those from taste buds. It is evident that no rigid categories can be
recognized here in terms of our conventional classification of "the
senses" or of their organs, a principle illustrated also by Whitman's
('92) observations on the cutaneous sense organs of the leech,
Clepsine, to which reference is made on page 84. Nature is not
bound by our rules of logical analysis.
Taste buds within the mouth are interoceptors, but similar buds in
the outer skin of fishes are typical exteroceptors, used in the selection
and location of food, as are also the free nerve endings of the general
cutaneous nerves that respond to chemical excitants. It is evident
that all these nerve endings co-operate with the nerves of ordinary
tactile sensibility in the normal process of finding food. That this cooperation is intimate and in some cases indispensable has been shown
by Parker ('12) in the case of the catfish, Ameiurus. In this fish the
skin has general chemical sensitivity to acid, alkali, and salt, a
sensitivity which is served by general cutaneous nerves. In the skin there are also innumerable taste buds which are innervated by fibers
which enter the f . solitarius. The general chemical sensibility is preserved if the taste buds are denervated, but the specific gustatory
function of the taste buds is lost if the general cutaneous innervation
of the surrounding skin is eliminated. A similar relation prevails with
taste buds within the mouth, for these have a double innervation;
and in man the gustatory function is abolished if the trigeminal innervation of the tongue is surgically destroyed, even though the
specific innervation of the buds remains uninjured; this loss, however, is temporary, and after a few weeks gustatory function returns
(Gushing, '03).
Amblystoma has no cutaneous taste buds, but the mouth cavity is
abundantly supplied with them. They are especially numerous on the
palate among the vomerine teeth, and these buds have a peculiar
accessory innervation — a compact skein of circumgemmal fibers of
uncertain origin ('256; Estable, '24). These fibers separate from a
plexus related with the ramus palatinus and may be derived from a
trigeminal anastomosis; but this has not been demonstrated.
The tactile, general chemical, and gustatory systems are as intimately related centrally as they are peripherally. All taste buds of
all animals, wherever found, are supplied by fibers which discharge
centrally into the nucleus of the f. solitarius or its derivatives. In
those fishes which have cutaneous taste buds with exteroceptive
functions the central connections of these buds differ from those of
buds within the mouth which have interoceptive functions. These
details need not be given here ; the interested reader is referred to a
recent paper ('446) and references there given. These differences are
explained by the fact that stimulation of interoceptive taste buds
evokes visceral responses, but excitation of exteroceptive buds is followed by somatic movements for capture of food.
In Amblystoma, as in man and all other vertebrates, all fibers from
taste buds enter the f . solitarius. Most fibers of all modalities of general cutaneous sensibility of the head enter the sensory V nucleus;
but a small number of them pass through this nucleus to enter the
f . solitarius, thus providing for integration of general somatic sensory
and gustatory sensibility. The prefacial f. solitarius carries gustatory
impulses forward into the neuropil of the superior trigeminal nucleus
in the auricle. A third and much more extensive provision for bringing general cutaneous and both general and special visceral sensi
bility into physiological relation is at the bulbo-spinal junction
(chap. ix).
The preceding analysis illustrates the intimate physiological relationship which exists among the various modalities of sense which
may be concerned with the resolution of mixed sensory experience in
the interest of securing the appropriate responses. The integrating
apparatus is spread from the peripheral end-organs throughout the
central nervous system. Within this machinery for conjoint action
there have been differentiated the specific sensory and motor systems, that is, the analyzers. The first step in this analysis is the separation in the central adjustors of the visceral from the somatic systems. Thus the fibers of taste and general visceral sensibility converge
into the f. solitarius, well separated from all the somatic sensory
systems which are assembled more superficially. This segregation
obviously has arisen because of the radical differences in the courses
taken by the efferent fibers from visceral and somatic receptive fields
to visceral and somatic effectors, respectively.
In general, the gustatory fibers tend to end near their entrance into
the brain, and the general visceral fibers to descend toward the lower
end of the system. In elasmobranchs the nucleus of the f . solitarius is
locally enlarged, with a separate lobe for each of the nerves of the
gills. These enlargements, which show as a beadlike row in the wall of
the fourth ventricle, are probably chiefly gustatory. In the carp and
some other teleosts with enormous numbers of taste buds, there are
separate enlargements of this nucleus known as facial, glossopharyngeal, and vagal lobes. These are known to be largely gustatory in
function. In other species of teleosts there are various modifications
of these arrangements. In some birds with very few taste buds the
f . solitarius is clearly double. A very slender medial bundle carries the
few gustatory fibers and the much larger lateral bundle, the general
visceral fibers (Ariens Kappers, Huber, Crosby, '36, p. 370). In
Amblystoma, as in mammals,. none of these specializations have occurred, and the visceral sensory system as a whole retains its primitive characteristics.
These roots in Amblystoma comprise only fibers for the extrinsic
muscles of the eyeball in the roots of the III, IV, and VI nerves. As
previously mentioned, all peripheral motor neurons are mingled with those of the motor tegmentum and are usually indistinguishable
from them except in cases where their axons can be followed into the
nerve roots. The cells of the nuclei of the eye-muscle nerves are
fairly clearly segregated, and in some reduced silver preparations
they react specifically to the chemical treatment (fig. 104) ; but even
here their dendrites are widely spread and intertwined with those of
tegmental cells, so that both kinds of neurons would appear to be
similarly activated by the neuropil within which they are imbedded.
The oculomotor nucleus lies in the posteroventral part of the peduncular gray (figs. 6, 18, 22, 24, 30, 31, 104). The nucleus of the IV
cranial nerve lies about midway of the longitudinal extent of the
isthmic tegmentum and far removed from the oculomotor nucleus
(figs. 61, 104). The thick IV root fibers (most of them myelinated)
ascend along the outer border of the gray to decussate in the anterior
medullary velum in the usual way.
As mentioned in chapter xiii, the sensory zone of the isthmus contains cells of the mesencephalic V nucleus and others which send
axons peripherally to meninges and chorioid plexus. Some of the
latter go out with the IV nerve roots to unknown destinations. It is
possible that some of these cells are secondarily displaced neurons of
the motor IV nucleus, similar to those described by Larsell ('476) in
cyclostomes. There is no definite evidence that this condition exists
in urodeles; and, indeed, all connections of the cells lying within and
adjoining the superior medullary velum require further study.
The floor plate of Amblystoma throughout its length contains a
special type of ependymal elements and the cell bodies of some neurons. These neurons do not invade the floor plate from the basal
plate, but they develop within this plate intrinsically, as was first
pointed out by Coghill ('24, Paper III). In the adult medulla oblongata there are few of them at any one level, but they constitute a
definite nucleus raphis, which is enlarged in some places, notably so
in the interpeduncular nucleus. Most of the cells of the nucleus of the
VI nerve are median, as in Necturus ('30, p. 14). These cells in
ordinary preparations cannot be distinguished from others of the
nucleus raphis except by observation of axons emerging in the VI
nerve root. They are distributed sparsely in the ventral raphe and
adjacent to it between the levels of the VII and IX nerve roots ('445,
fig. 2). They emerge usually by two widely separated roots, though
more rootlets are sometimes seen.
Preganglionic fibers for unstriatcd muscles and glands leave the
brain in the III, VII, IX, and X roots, probably with the IV nerve
and its environs (p. 181) and perhaps with the parietal nerve (p. 235).
In other vertebrates, fibers of this system have been described as
leaving the brain with the optic nerve, with the nervus terminalis,
and independently from other regions of the brain for the meninges.
Our material is inadequate to reveal satisfactorily either the central
connections or the peripheral courses of any of these fibers, so that
this topic remains to be clarified. The large unmyelinated hypophysial nerve belongs in this system, as described on page 244.
The striated muscles related with the visceral skeleton of the head
— jaws, hyoid, branchial arches, and their derivatives in higher animals — belong in a special category (p. 69). These muscles are visceral in phylogenetic and embryologic or