AM. ZOOLOGIST, 2:79-96 (1962).

ORGANIZATION OF NEUROPIL
DONALD M. MAYNARD
Department of Zoology, The University of Michigan

Most complex nervous systems consist of cell bodies into the neuropil, from the neu-
central ganglia connected with outlying ropil to the origin of peripheral nerves, and
sensory organs and effectors by peripheral from neuropil to neuropil as connectives or
nerves. Information moves to and from the commissures. The fibers of the tracts are
ganglia in the nerves in coded form, as a either axons or extended, unbranched por-
spatial-temporal pattern of all-or-none im- tions of the dendritic arborizations. 4) The

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

pulses. Decoding of the incoming message fourth and final division is the central neu-
and recoding of the appropriate outgoing ropil, the neuron feltwork. In many cases
response occur somewhere in the ganglia. it represents the major portion of the gan-
Many of the fundamental problems of glion.
neurophysiology revolve around the related The term neuropil, however, has beef,
questions of the functional significance of used in different ways by a number of au-
impulse patterns in the central ganglia thors, and is not a precisely defined concept
(codes), and of the mechanisms of central (see Herrick, 1948; Dempsey and Luse,
neural integration (coding and decoding). 1958; Friede, 1960). Classically it is an ana-
Information about impulse patterns and tomical term referring to regions of the cen-
the properties of integrative systems can be tral nervous system composed of tangles or
derived from input-output analyses (Wiers- meshworks of nerve processes and non-neu-
raa, 1962), but problems regarding detailed ral elements such as glia and blood vessels.
mechanism eventually require penetration By extension it also refers to such tangles of
into the ganglion and dissection of the in- collaterals, dendrites, and axon termina-
tegrating system itself. tions in peripheral ganglia. Neuropil has
Let us look, then, at the organization of also been considered synonymous with syn-
a central invertebrate ganglion. The em- aptic field, however, and thus has acquired
phasis will be on arthropod material, for I functional connotations as the primary site
am most familiar with this group; most of of integrative neuron activity (Herrick,
the comments should apply with little 1948). The structural limits of neuropil are
modification to many of the annelids, mol- apparent in forms such as the crayfish or
luscs, and lower vertebrates. lobster where the majority of the centra'
neurons are monopolar with cell bodies lo-
NEUROPIL
cated in the periphery of the ganglion. As
In the central ganglia of most higher in- one moves either to such non-vertebrates as
vertebrates four histological divisions are the Coelenterata or Platyhelminthes, how-
evident (Fig. 1): 1) An outermost sheath ever, or to the higher vertebrates, bipolar or
of non-neural connective tissue cells or glial multipolar neurons in the central nervous
elements separates the central neural tissue system increase and the cell body becomes
from general body cavities. 2) Just within increasingly mixed with dendritic fields.
the sheath tissue are the cell bodies of the Finally the original distinction between cell
monopolar neurons. 3) Tracts of fibers body regions and fibrous regions is lost;
form the third division. These run from there is a continual gradation from the clas-
sical neuropil, a synaptic field which in-
Original work on Panulirus reported in this paper cluded processes only, to the synaptic field
was carried out at the Bermuda Biological Station. which places the ce.ll body in the center and
It was supported by Rackham Grants from the Uni- surrounds it with dendrites and terminal
versity of Michigan, an ONR grant to the Biological arborizations, as for example in the verte-
Station, and U.S.P.H.S. grant B 3271. Miss C. Good-
rich collaborated on the experiments with the sto- brate olivary nucleus (Fig. 2). In organisms
matogastric ganglion. with the latter kind of synaptic field the
(79)
80 DONALD M. MAYNARD

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

BIG. 1. Brain of craylish, Astacus fliwiatilis, as seen dian nerve, optic tract, tegumentary nerve, antennal
from above and behind. Methylene blue stain show- nerve, and circumesophageal connectives. (From
ing peripheral cell bodies, fiber tracts, and neuropil. Retzius, 1890, Biol. Untersuch. N.F.)
Peripheral nerves from top to bottom: anterior me-
cell body seems more intimately involved Many authors have been impressed with
with electrical activity of the neuron, and the apparent structural disorder of neuro-
it is reasonable to consider the entire inte- pil (Dempsey and Luse, 1958; Horridge,
grating portion of the nervous system as 1961). It is indeed heterogeneous and com-
functional neuropil. plex, and because of this has resisted most
attempts at a systematic analysis. It has
been called the terra incognita of neuro-
anatomy and neurophysiology. Neverthe-
less, neuropil often seems to conform to
some underlying structural pattern, even
when showing random connectivity. We
shall begin this discussion, therefore, with
a preliminary structural classification of
neuropil. The recurrence of similar pat-
terns in regions subserving similar func-
tions in widely divergent organisms sug-
gests that much of the structure may have
unique functional significance. The direct
measurement of neural activity within neu-
ropil has just begun, but electrical record-
l i d . 2. (ells of inferior olivary nucleus of new ings have revealed several functional prop-
born human, a, axon: b, collateral. (From Cajal,
1904, Textura del sistcma nervioso del hombre y rle erties that in some cases correlate closely
los vertebrados. II. Part 1.) with neuropil structure.
 ORGANIZATION OF NEUROPIL M
tic junctions seem to be lacking, and the
nerve net has been considered syncytial
(Mackie, 1960).
Plexiform neuropil may form the major
portion of the nervous system in primitive,
sessile, or burrowing organisms. In the Coe-
lenterata, for example, the entire nervous
system is a two-dimensional nerve net (or
nets) and cell bodies are continuous with
processes. In Balanoglossids, on the other
hand, the cell bodies are separated from a

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

thicker plexiform fiber mat (Bullock,
1945a). Plexiform neuropil also forms lim-
ited regions of more complex nervous sys-
FIG. 3. Plexiform neuropil. Nerve net of bipolar
tems or serves as a component of structured
cells in mesentary of Metridium. Whole mount, neuropil. It is the common pattern of in-
Holmes' silver. (From Pantin, 1952, Proc. Roy. Soc. nervation for hollow organs such as the di-
London, B.) gestive tract.
Diffuse neuropil presents the classical pic-
STRUCTURE OF NEUROPIL ture of "a tangled confusion" (Horridge,
1961) (Fig. 4). Neurons contributing to it
Microscopic Morphology
in most invertebrates are either monopolar
On the basis of gross fiber configurations, elements of average or large size, or peri-
all neuropil can be placed in either of two pheral bipolar sense cells whose axon ter-
general classes, structured or unstructured minations ramify centrally. Many of the
neuropil. These in turn can be subdivided: monopolar cells are motor neurons or inter-
the unstructured into plexiform and diffuse neurons connecting several ganglia or seg-
neuropil, the structured into glomerular ments of the organism (Fig. 5). Processes
and stratified neuropil. The groupings are are characteristically tortuous, extensively
not sharply bounded or exhaustive, but
they do emphasize significant differences in
organization and, without undue stretch-
ing, seem to include most kinds of neuropil.
Unstructured neuropil
Unstructured neuropil is characterized
by ill-defined fiber configurations. Areas of
fiber interaction are not sharply differen-
tiated within the general fiber network, and
the domain of a single neuron may be ex-
tensive. The neuropil may take either of
two forms.
Plexiform neuropil is homogeneous and
net-like (Fig. 3). The neurons forming it
are usually simple mono-, bi- or multipolar
structures. The processes are often un-
branched, straight, and of constant diam-
eter. Synaptic junctions appear simple, of
few varieties, and may occur at almost any FIG. 4. Diffuse neuropil. Central region of brain
of spiny lobster, Panulirus argus. Thin section,
point on the process. Many appear to be Masson's trichrome stain. Ramifying dendrites and
unpolarized, and transmit impulses in fiber tracts form a "haystack" with regions of vary-
either direction. In some forms, the synap- ing texture. Width of field, about 1.5 millimeters.
82 DONALD M. MAYNARD

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

FIG. 5. Diffuse neuropil. Last thoracic and ab- motor unit in right lobe of thoracic ganglion.
dominal ganglia of Carcinus maenas. Whole mount, (From Jiethe, 1895, Arch. Mikroscop. Anat. u. Ent-
methylene blue stain. Note extensive branching of witklungsinech.)
branched, and of widely varying diameters. tions. Areas of fiber interaction are usually
They spread to varying degrees throughout clearly distinguished from surrounding
the neuropil so that it seldom presents a fiber networks. The domain of a single neu-
homogeneous appearance. Each neuron, ron may be quite limited, but it often
however, tends to have its own domain of makes contact with other elements whose
arborizations. Synaptic junctions are not domain is extensive. The rigidity and pre-
easily recognized with light microscopy, but ciseness of the fiber patterns is impressive.
are probably more heterogeneous than in They usually seem to represent a redupli-
plexiform neuropil. There is no good ana- cating system; a complex structure built up
tomical information concerning the exist- of large numbers of relatively few kinds of
ence of unpolarized synapses or syncytial units which in turn are grouped into re-
interneuron bridges. peating sub-structures. Such precise but re-
Diffuse neuropil is characteristic of the dundant structural fabric probably occurs
more posterior central ganglia of most where stability, accuracy, and temporal or
higher invertebrates, and usually contrib- spatial complexity are functional prerequi-
utes to the cerebral ganglia. It is also found sites. It seems unlikely that functional la-
in the brainstem of lower vertebrates and bility and the potentiality for "learning"
in many peripheral invertebrate ganglia; predominate in the more excessively struc-
the crustacean stomatogastric ganglion is tured neuropil.
an example. Structured neuropil is most evident in
anterior ganglia of complex organisms. It
Structured neuropil
may be directly associated with complex
Structured neuropil is characterized by sensory organs. There are many geometri-
regular, often repeating fiber configura- cal patterns of structured neuropil, but
 ORGANIZATION OF NEUROPIL 83

The internal structure of glomeruli is

largely unknown (see Bodian, 1952). Silver
and methylene blue stains suggest that dif-
ferences exist between the arborization pat-
terns of pre- and post-units.
The neurons whose processes form glo-
meruli range widely in function and anat-
omy. They may be motor, but more com-
monly are sensory or internuncial elements.
Among the latter, the dendrites of small
globular neurons of the invertebrates (Han-

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

strom, 1928) and granule cells of the verte-
brate cerebellum characteristically form glo-
merular junctions. It is evident that glome-
riG. 6. Glomerular neuropil. Nucleus rolundus of rular neuropil does not represent a ho-
sea robin, l'rionolus evolans. Thin section, upper mogeneous class, but further subdivision is
part drawn as it appears after silver impregnation undesirable until more information on
according to Bodian. Lower part, reconstruction
from Golgi preparations. Ch., commissura horizon- structure and function is available.
talis of Kritsch; Gl., glomeruli; Tr.r.-l., tractus ro- Stratified neuropil (Fig. 7) is a regular,
tundo-lohaiis. (I'rom Scharrer, 1945, J. Coinp. Neu- three-dimensional fiber lattice formed from
rol.) precisely oriented neuron processes. The
anatomy presumably reflects the functional
most seem to fall into one or the other of connectivity. The basic pattern consists of
two groups, glomerular and stratified. vertically or radially oriented fibers that
Glomerular neuropil is characterized by pass at right angles through horizontal or
tight, profuse ramifications of pre- and post- tangential layers of ramifying processes. Syn-
synaptic elements at the site of junction aptic junctions occur between the vertical
(Fig. 6). These ramifications form the dis- and horizontal fibers.
crete, knot-like clusters of fibers that give The arthropod optic ganglia represent a
this kind of neuropil its name. The glo- typical, highly structured example of strati-
meruli themselves rarely occur in precise fied neuropil (Fig. 7). The vertical ele-
relationship with each other; they may be ments are quite different in form from the
in layers or apparently unstructured clumps. horizontal fibers. Some of the vertical ele-
The size and form of the glomerulus and ments have collaterals and junctions in only
the number and pattern of elements con- one layer, others send collaterals into sever-
tributing to it may vary greatly. For ex- al layers. Often the collateral structure dif-
ample, an afferent fiber in the vertebrate fers according to the layer or the particular
olfactory bulb or the insect antennal lobe vertical neuron; presumably it reflects dif-
terminates in but one glomerulus, while ferences in the functional properties of the
neurons of the insect corpora pedunculata junction. The lateral extent of the synap-
or mitral cells of bird olfactory lobes may tic field of a vertical unit is usually lim-
have branching processes which enter a ited. Consequently most vertical units prob-
number of glomeruli. The minimal num- ably synapse with relatively few horizontal
ber of neurons necessary to form a glome- fibers in any one layer. In contrast, the
rulus is not clear, but the olfactory glome- horizontal fibers branch diffusely. They
ruli of the rabbit must represent a near often travel in preferred directions, but ap-
maximum. Each glomerulus is claimed to parently form junctions with a large num-
receive as many as 26,000 incoming fibers ber of vertical units in their level. Within
and has about 90 outgoing fibers (Allison, a layer they are probably the major link be-
1953). This is a particularly clear example tween vertical units, just as the vertical
of the convergence that often seems to oc- units appear to be the major link between
cur in glomeruli. layers.
84 DONALD M. MAYNARD

areas of cerebral ganglia. The cortex of the

vertebrate or possibly the central body of
the arthropod are examples. These regions,
however, often lend to merge in form with
diffuse, unstructured neuropil, and the
fiber domain of a single neuron appears less
rigidly defined than in the optic systems de-
scribed above.
Hislochemistry and Sub-microscopic
Anatomy

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

The above discussion gives a general
view of fiber patterns in neuropil, but it is
necessarily superficial. In addition to the
obvious limitations resulting from this su-
perficiality, there are at least two serious
drawbacks in the methods themselves which
reduce the functional significance of the
findings. On the one hand, the usual pre-
parative techniques of classical micro-anat-
omy do not reveal chemical differences
among fibers of the neuropil, and on the
other, light microscopy does not have suffi-
cient resolution to show the detailed intra-
cellular structure of fibers and synapses.
Both of these aspects are particularly rele-
vant in a consideration of the functional
anatomy of neuropil.
Histochemislry
FIG. 7. Stratified neuropil. Diagram of retina and There is considerable evidence for several
optic ganglia of honey bee, Apis melifica L. I, ret- neurotransmitters in the nervous system
ina; II, lamina ganglionaris; III, medulla externa; (Florey, 1962). A detailed map showing
IV, medulla interna. (From Cajal and Sanchez, the localization of these, or substances in-
1915, Investig. Biol.)
volved in their production or break-down
Both vertical and horizontal fiber groups could prove of aid in determining the func-
contain afferent and efferent elements. In tional properties of a given neuropil. Many
the case of optic ganglia, however, the pri- of the transmitters seem to be specific for
mary input fibers from the visual receptors particular kinds of neurons.
are vertical rather than horizontal. In con- We have been interested in the localiza-
trast, only horizontal fibers seem to carry tion of cholinesterases in crustacean gan-
information from the optic ganglion com- glia (Fig. 8, Maynard and Maynard, 1960a).
plex into more central areas of the brain. There are high concentrations of the en-
The retina and superior colliculus in the zyme in peripheral glial tissue (see also,
vertebrate and the retina profundus of the Wigglesworth, 1958), but it also occurs in
octopus are formed from stratified neuropil some neuron somata and particularly in
essentially like that of the arthropod. Such certain regions of the neuropil. Experi-
a configuration may be important in deal- ments with more peripheral structures in
ing with complex spatial patterning, and the lobster (Maynard and Maynard, 1960b)
will be considered in more detail below. suggest that concentrations of the enzyme
Less precise layering is evident in other may correlate with physiological character-
 ORGANIZATION OF NEUROIML

dendrite and axon in the neuropil is lack-

ing. The peripheral cell bodies are devoid
of terminal contacts, and fiber junctions
are confined to the neuropil. Three kinds
of junction are recognized: 1) cross con-
tacts, 2) longitudinal contacts, and 3) end-
knobs (Fig. 9). In all, neuron membranes
come into direct contact with each other,
but there is no fusion of neuroplasm. Al-
though membranes are often flattened or
indented at the point of contact, invagina-

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

tions as found in the synapse between giant
fibers in the crayfish (Robertson, 1953) are
apparently absent. Only the end-knob junc-
tion contains the accumulations of micro-
vesicles and mitochondria often considered
characteristic of synapses (De Robertis,
1958; Palay, 1958). Microvesicles, however,
also occur along fibers in non-junctional
areas.
ol

FIG. 8. Distribution of cholinestcrase in brain of

lobster, Homarus americmuis. Thin section, acelyl-
thiocholine substrate. Al, accessory lobe, a glome-
rular neuropil; gc, globular cells, two different
groups; ol, olfactory lobe, another glomerular neu-
ropil. Enzyme concentration to right of accessory
lobe represents a fiber tract; light areas surrounded
by a ring of enzyme are large neuron somata; other
areas of high enzyme concentration are in diffuse
neuropil. (Courtesy of E. A. Maynaid)

istics, but this remains to be established in

the central ganglia. Perhaps the final in-
terpretation o£ such chemoarchitectonics
must await clarification of the role of ace-
tylcholine esterase in neural activity.
Sub-microscopic anatomy
A number of scattered observations on
the fine structure of invertebrate nervous
systems exists, but a systematic, comprehen-
sive investigation of neuropil has not been
published. One of the closer approxima-
tions to such a needed work is a preliminary
investigation of caterpillar ganglia (Tru-
jillo-Cenoz, 1959). FIG. 9. End-knob in neuropil of caterpillar, Pholus
Most fibers in the (diffuse?) neuropil of labruscoe L. Electron micrograph. E.K., end-knob;
a caterpillar ganglion are naked and lack M, mitochondria; S.M., synaptic membranes; T.F.,
thin fibers (post-synaptic?). Microvesicles accumu-
glial sheaths. In this they differ from the late to left of mitochondria. Width of field, about
sheathed fibers in interganglionic connec- 2.3 microns. (From Trujillo-Cenoz, 1959, Zeit. f.
tives. Fine structural distinction between Zellforsch.)
86 DONALD M. MAYNARD

Several important conclusions can be dense in neuropil areas. In fish, for ex-
drawn from these observations. 1) Neuron ample, the glomeruli of the olfactory lobe
interaction in caterpillar ganglia, if depend- and nucleus rotundus are heavily vascular-
ent upon close anatomical contact, seems to ized. This often correlates with concentra-
be confined to the neuropil. 2) Some junc- tions of mitochondria or oxidative enzymes
tions in the neuropil are similar to the clas- (Scharrer, 1945; Friede, I960) suggesting
sical synapse; one may therefore anticipate that neuropil is a region of unusually high
similar properties of synaptic transmission. physiological activity. Even in the lobster
3) Several forms of fiber junction occur, and crayfish, which are generally consid-
suggesting the possibility of more than one ered to have an open circulatory system, the

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

mechanism of neuron interaction. It should central nervous system, particularly areas
be clear, however, that full benefit of the of structured neuropil, is laced with a capil-
increased resolution of electron microscopy lary network. In both vertebrates and lob-
will be apparent in neuropil analysis only sters, the higher correlative functions of the
when correlations between fine structure nervous system are very sensitive to oxygen
and specific microscopic structures are pos- deprivation.
sible.
FUNCTIONAL PROPERTIES OF NEUROPIL
Glia and Capillaries
Forms of Electrical Activity
Thus far we have spoken of neuropil as
though it were composed solely of neuron Electrical changes provide one of the
processes. This is misleading, for both glial most sensitive measures o£ neural activity.
cells and vascular elements are also present. For the past two years, in collaboration
As mentioned earlier, glial elements usu- with J. Clarridge, V. Shoemaker, R. Ste-
ally form an outer sheath enveloping the phens, and M. J. Cohen, we have been
entire nervous system. They may also sur- using intracellular microelectrodes to probe
round cell bodies and individual fibers in into parts of neurons in the perfused spiny
tracts and connectives (Wigglesworth, 1959). lobster brain. One of our purposes was to
Just how far glial processes interpenetrate answer the simple question: What kind of
the denser fiber tangles of invertebrate neu- electrical activity normally occurs in com-
ropil is not known (Trujillo-Cenoz, 1959). plex central ganglia? Are there propagated
In the mammalian grey matter, of course, impulses? Are there slow potentials? Do
glia and neuron processes intermingle, and all parts of the neuron act alike?
glia-neuron junctions have been described
(Scheibel and Scheibel, 1958). General remarks
The physiology of glia is of current inter- First, we find that in fibers in tracts only
est. In the past glia have been considered full-sized action potentials occur. These are
supporting or nutritive elements (Wiggles- indistinguishable from those found in any
worth, 1960), but they also have been im- peripheral nerve.
plicated in the control of extracellular en- Second, electrodes placed in peripheral
vironment in the neuropil (Hoyle, 1953; cell bodies record only attenuated, slow de-
Twarog and Roeder, 1956; Treherne, 1961). polarizations in our preparations. The so-
In addition, there are reports of electrical ma membrane seems to be inactive, and the
activity in vertebrate glia (Tasaki and recorded potentials presumably represent
Chang, 1958), and current speculations sug- passive, electrotonic spread from distant
gest that they may be involved in integra- sites of activity in the neuropil (Hagiwara
tion in the nervous system (Galambos, and Bullock, 1957; Preston and Kennedy,
1961). This topic is beyond the scope of the 1960). Such a finding perhaps is not sur-
present paper, but obviously glia cannot be prising in the light of observations made by
completely dismissed from a final func- Bethe (1897) and Hardy (1929) that reflex
tional analysis of the nervous system. activity can continue for short periods after
Capillaries. Capillary beds are usually surgical removal of cell bodies.
 ORCAN'IZATIO.N OF NF.UROPIL 87

have more than one size spike potential, im-

plying that one neuron may have several
different impulse sites, each capable of
maintaining an independent action poten-
tial (Bullock and Terzuolo, 1957; Preston
and Kennedy, 1960; Spencer and Kandel,
FIG. 10. Synaptic noise recorded with intracellular 1961).
electrode from a nerve process in brain of spiny lob- 3. Generator potentials are slowly grow-
ster. Read from right to left. One spike potential ing depolarizations that usually give rise to
rises from unusually large synaptic potential. Both
epsp and ipsp probably contribute to subthreshold propagated action potentials. (They are

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

activity. Calibrations, 0.1 second; 4 millivolts. not to be confused with the generator po-
tentials of receptors.) They presumably
Thirdly, ramifying processes in the un- originate spontaneously in the normal unit,
structured neuropil are the site of much but their genesis is obscure and they also
complex electrical activity. This is gratify- occur in injured axons. It is not clear how
ing, for neither the all-or-none spikes of the much of the spike activity of neuropil ele-
tracts nor the small, relatively simple slow ments results from generator potentials and
potentials of the neuron soma are the kinds how much from synaptic noise.
of activity expected in a center of complex 4. Inhibitory post-synaptic potentials (ip-
integration. Some of the neuropil elements
are silent, but many are either spontane-
ously active or subject to continuous sub-
threshold, presynaptic bombardment (Fig.
10). There are at least four types of activ-
ity:
1. Excitatory post-synaptic potentials (ep-
sp) are the most prominent. These are de-
polarizing potentials lasting several milli-
seconds. They are non-propagated, and con-
sequently of greatest amplitude at the site
of the synaptic junction. Since those re-
corded in a single cell often vary consider-
ably in amplitude, they must represent dif-
ferent junctions either at unequal distances
from the recording site, or possibly, of un-
equal effectiveness. Epsp initiated by differ-
ent presynaptic elements may sum alge-
braically, and upon reaching critical depo-
larization, initiate propagated potentials
that travel down the axon (Fig. 11).
2. Action potentials are recorded from
most neuropil processes as small, all-or-
none spikes lasting 1-2 milliseconds. These
may propagate distally, for they correlate
perfectly with impulses recorded in peri-
pheral nerves. They usually rise from gen- * • I « • * * t •

erator or synaptic potentials. In the latter FIC. 11. Epsp. Intracellular recording from unit
process in diffuse neuropil of spiny lobster brain
case, they tend to be smallest in those units (tracing). Input via homolateral antennular nerve
with the largest epsp. This suggests that with increasing stimulus strengths from bottom
they may originate some distance from the trace up. Strongest stimulus evokes an action po-
tential whose peak is not shown. Time mark, 2
site of the synaptic junction and do not in- millisecond intervals; voltage calibration, 10 milli-
vade the junctional regions. Elements may volts.
88 DONALD M. MAYNARD

leaving the neuropil depends upon the pre-

n ceding summed, integrated, non-propagated
activity of the dendritic arborizations.
4. At least some neuropil integration in-
volves typical synaptic transmission, both
excitatory and inhibitory.
FIG. 12. Ipsp. Iniracellular recording as in Fig. 11. 5. Many, if not most, of the neuropil as-
Read from right to left. Input at 50 stimuli per sociation or motor elements receive con-
second via medial bundle of heterolateral antennu- tinuous, subthreshold presynaptic bom-
lar nerve. Calibrations, 0.1 second; 10 milivolts. bardment. This is not immediately appar-
Upper trace of this and following figures taken from

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

monitoring electrodes on antennular nerve; lower ent in the output of the system, but is prob-
trace from microelectrode within ganglion. ably the basis of a variable central excita-
tory state. It seems unlikely that normal
sp) are similar to the epsp but often hyper- neuropil can ever be considered truly at
polarize rather than depolarize (see Fursh- rest or in a static state.
pan and Potter, 1959b). On occasion these
Specific remarks on single units
reduce the excitability of the unit and
either block spike initiation or reduce spon- One of the remarkable things about many
taneity (Fig. 12). Like the epsp, the ipsp neuropil elements is the extent of their ar-
originate from activity in presynaptic ele- borizations. We have suggested that func-
ments, and like them, help comprise the tional differences exist between arboriza-
continuous subthreshold synaptic noise tions and axons, but one may also ask: Are
found in most units of the neuropil. all excitatory synapses alike? How does one
T h e activity described above is well branch of a neuron affect another if it is
known from other preparations, but it is several millimeters distant? Although we
important that it be demonstrated in neu- cannot fully answer these questions, certain
ropil itself. Other kinds of electrical activ- observations on units in the lobster brain
ity are probably to be found in the ganglia are suggestive.
(see below), but the above are the most ob- First: Many of the synaptic potentials
vious, and from them we may begin to caused by primary afferent fibers from the
build a tentative picture of functional neu- antennular nerve in the spiny lobster do
ropil as a coding center of the nervous sys- not show temporal facilitation. In fact,
tem. they begin to diminish in amplitude at low
1. Integrative activity is largely confined frequencies of repetitive stimulation and
to the neuropil. Tracts only conduct im- soon fall below the spike-initiating thresh-
pulses, and cell bodies seem to be electri- old. In contrast, other epsp with longer lat-
encies do not fatigue and usually require
cally inactive.
temporal summation to reach maximum
2. Although both input and output from effectiveness in spike initiation. There are,
the ganglion, and communication between therefore, at least two kinds of epsp in cen-
neuropils within, involves propagated ac- tral as well as peripheral crustacean ganglia
tion potentials and a pulsed code, the elec- (see Terzuolo and Bullock, 1958; Furshpan
trical events concerned with actual integra- and Potter, 1959b). These give the inte-
tion may be more similar to a non-pulsed, grating neuron the potential of responding
analog system. in very different manners to similar tem-
3. Portions of the neuron concerned with poral patterns from different inputs (Fig.
reception and integration of information 13).
from other neural elements are distinct Second: Several observations suggest that
from, and have much more variable activ- a neuron can maintain independent spike
ity than portions concerned solely with im- potentials in more than one of its processes.
pulse propagation. T h e entire neuron, Although evidence is not conclusive, it is
therefore, does not a d as an all-or-none possible that these potentials represent
unit, and the eventual propagated impulse propagated impulses rather than synaptic
 ORGANIZATION OF NEUROPIL 89

RMB

I I I 1 I M 1 11
T T I r
LMB

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

0.2 sec
FIG. 13. Patterned responses from different inputs. erolateral antennular nerve (lower pair of traces).
Intracellular recording from one unit as in Fig. 11. Note different patterns of response to first stimulus
Read from right to left. Input at 10 per second via of train and differences in adaptation or fatigue.
medial bundle of homolateral antennular nerve Synaptic potentials were not present in this record.
(upper pair of traces) and via medial bundle of het-

potentials (Furshpan and Potter, 1959a). not appear as the final output of the neu-
Jf so, they provide a mechanism of intra- ron unless capable of exciting the efferent
cellular interaction in extensively branched axon itself. Our records indicate that this
neurons (Spencer and Kandel, 1961; Pres- does not always occur (Fig. 14), and, if our
ton and Kennedy, 1960). The propagated hypothesis is correct, that synaptic poten-
impulse would occur in stretches of the den- tials initiating a propagated impulse in one
drite connecting separate arborizations, and portion of a single neuron's arborizing proc-
esses may ultimately fail because of failure
of the impulse at some dendrite-axon junc-
tion. This implies that the single neuron
may be analogous to a two-neuron chain,
and the point of potential spike failure or
lowest safety factor is analogous to a synap-
tic junction. Possibly the extensively
branched neurons often found in the ar-
thropod neuropil (see Fig. 5) should not be
regarded as functional units. They may
embody in one morphological structure
two or more functional elements and may
be, in fact, similar to a synaptic chain of
FIG. 14. Various spike potentials in one unit. Re- several less ramifying neurons. One may re-
cording as in Fig. 11. Unit responds in different call that many invertebrates have remark-
manner to each of four different inputs: LLB, lat- ably few neurons in their nervous system
eral bundle of left antennular nerve: LMB, medial and yet are capable of much complex, adap-
bundle of left antennular nerve; RLB, lateral bun-
dle of right anlciinular nerve; RMB, medial bundle tive behavior.
of right antennular nerve. Note synaplic potential
following RMB stimulation and smaller spikes fol- AsyitapUc in term: I ion
lowing RLB. Under other conditions, these smaller
spikes were capable of initialing the larger spike.
The observations reported above indicate
Small spikes never appeared upon stimulation of that at least some of the integrative activity
bundles of the left antennular nerve. of the neuropil involves synaptic transmis-
90 DONALD M. MAYNARD

sion. There is evidence from other prepa- ropil. They also suggest correlations be-
rations, however, that synapses may not be tween the morphology and the physiology
the only means of significant interaction be- of specific neuropils.
tween neurons; non-synaptic, electronic in-
fluences may exist (Watanabe, 1958; Ben- Connectivity in diffuse neuropil
nett, I960). Unlike synaptic potentials, the From anatomy one might presume that
electrotonic influence does not occur as dis- in diffuse neuropil each element connects
crete events, but appears in the second unit with almost every other, in an unorganized
as an attenuated and distorted reflection of fashion. Preliminary electrical recordings
prolonged potential changes originating in from single elements in the lobster brain do

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

the first unit. This kind of interaction not entirely support such a conclusion in so
should tend to insure that a group of neigh- far as functional connections are concerned.
boring neurons remain at the same excit- It is true that many neurons form junctions
ability state. In two preparations where with a variety of presynaptic elements, but
they have been analyzed, electrotonic inter- these are not disorganized.
actions are associated with synchronized Central connections of the lobster anten-
discharge of elements of the system. Syn- nular nerve can serve as an illustration.
chronized discharges of neighboring neu- Massive stimulation produces the responses
rons as found in the cockroach corpora pe- shown in Figure 15. Variation of stimulus
dunculata (Maynard, 1956) may also result parameters leads to the conclusion that the
from such mechanisms of interaction. beginning of the initial response is pro-
Summary duced by direct action of afferent fibers.
The prolongation of the initial response
T o return to our first question, we may seems to be the result of activity in inter-
say that the electrical activity in neuropil posed excitatory internuncials, and the fol-
is complex, and that a large portion of the lowing drop, the result of inhibitory inter-
complexity appears as a property of individ- nuncials. The terminal rise in excitation
ual units rather than as the result of com- shows different properties and must result
plex connectivity patterns. The relatively either from a second population of inter-
simple model of synaptic action as we know nuncials or much more slowly conducting
it from the neuro-muscular junction, the afferent fibers. Such a patterned sequence
vertebrate motor horn cell, or the squid seems less likely to result from random cen-
giant axon synapse therefore seems less ap- tral connections than from some specific
plicable to an analysis of central integra- pattern like that illustrated. Further evi-
tion than the more complex patterns found dence for regular connectivity patterns in
in such peripheral autonomic ganglia as diffuse neuropil in lobsters comes from
the crustacean cardiac ganglion. Tt may be stimulating identical nerves on each side of
misleading, when analyzing central integra- the body while recording from one central
tion, to concentrate too exclusively on wir- element. Responses to the two inputs may
ing diagrams, functional or anatomical, and be qualitatively different. On the other
to ignore the properties of the switches and hand, stimulation of normal and hetero-
oscillators of the system. morphic appendages may cause similar re-
sponses, again suggesting specific connectiv-
Functional Connectivity ity within the diffuse neuropil. Although it
Functional connectivity patterns are im- has been implied (Horridge, 1961) that
portant, however, and much of our struc- functional connectivity patterns in unstruc-
tural analysis of neuropil is based on the tured neuropil can be explained solely by
anatomical substrate of such patterns. I physiological mechanisms such as neuro-
should like to turn about and review sev- transmittor and neuroreceptor specificity,
eral observations which seem to demon- it seems more likely that functional pat-
strate this importance in the determination terns reflect appropriate anatomical wiring
of physiological characteristic* of total neu- diagrams.
 ORGANIZATION OF NEUROPIL 91

fibers of the optic nerve. Each ommatidium

A. acts as a unit and serves as a node in the
network of nerve branches forming the
plexus. The discharge from each ommati-
dium depends upon both the incident light,
which stimulates, and the activity in the
plexus, which depresses. The plexus, there-
fore, is the site of the first stage of integra-
tion of the visual pattern received by the
receptors. Our concern is to show how geo-
B.

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

metrical parameters play a large role in the
function of this rather simple system.
Non-geometrical properties of the junc-
tions between ommatidia in the Limulus
eye are as follows: 1) All interaction be-
tween elements is inhibitory. 2) The de-
gree of inhibition is proportional to the fre-
quency of discharge of the fiber from the
PHh presynaptic ommatidium (when above a
threshold frequency). 3) The effect of one
presynaptic unit may sum with the effect of
others. 4) Interaction is reciprocal, but
need not be symmetrical with respect to ef-
fectiveness. 5) With the possible exception
l-'IG. 15. Connectivity in dill use neuropil. A. In- of ommatidia around the borders of the
tracellular recording of response to antennular eye, all seem to have similar junctions with
stimulation from unit in spiny lobster brain (trac-
ing). Calibrations, 10 milliseconds, 10 millivolts. B. the neuropil. 6) Under normal circum-
Diagram of possible connections from antennular stances, some ommatidia pairs do not inter-
nerve with schematic post-synaptic potential changes act, suggesting that not all ommatidia are
which, when summed, could produce the above re- directly connected with each other. These
sponse. Antennular nerve fibers enter from left; properties permit fairly accurate prediction
dot and vertical line represent excitatory and in-
hibitory synapses respectively. See text for further of the result of simultaneous visual stimula-
details. tion of any two receptor elements, but they
do not allow one to say how the message in
the optic nerve, resulting from patterned il-
Connectivity in plexiform neuropil lumination of the entire eye, differs from
In the above discussion we were con- the one that would obtain if interaction in
cerned primarily with formal connectivity the plexus were absent. Some additional
between units, and not with geometrical statement about geometrical connectivity is
configurations. In many neuropils, however, necessary.
geometrical considerations seem fully as An anatomical wiring diagram of the
necessary as formal connectivity diagrams subretinal plexus is not available, but its
for an understanding of function. This may apparent homogeneity suggests either of the
be illustrated by two examples from a sys- following possibilities: 1) Connectivity in
tem in which spatial or geometrical pattern- the eye is truly random, and the probabil-
ing is unusually important, the visual sys- ity of interaction between elements is com-
tem. pletely unrelated to the position of or dis-
The first example is the subretinal plexus tance between elements. 2) Connectivity in
of Liniulus as described by Hartline and the eye is biased so that the probability of
his colleagues (see Ratliff, 1961). This is a interaction between elements is a function
plexiform neuropil interposed between the of the direction and distance between ele-
receptors of the compound eye and the ments. If the first were true, one might ex-
92 DONALD M. MAYNARD

Ganglion cells in the retina collect from

rods and cones via interposed bipolars. Ac-
cording to our terminology the bipolars are
radial or vertical elements in the structured
lattice and have relatively small terminal
fields, while the ramifying dendrites of the
ganglion cells in the inner plexiform layer
are horizontal elements. There are five
kinds of ganglion cells. They differ in size,
lateral extent of the dendritic tree, and the

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

stratum or strata of the plexiform layer in
0.5 mm. at the eye which the dendrites ramify. Ganglion cell
FIG. 16. Connectivity in f.itnuhis eye. Discharge axons pass from the retina directly to the
of impulses from a single receptor unit in response colliculus or optic tectum where they pre-
to a "step" pattern of illumination in various posi- sumably separate and terminate in four
tions on the retinal mosaic (see inset). The upper horizontal strata according to ganglion cell
(rectilinear) graph shows the frequency of discharge
of the test receptor, when the illumination was oc- type. There is a point-to-point representa-
cluded from the rest of the eye by a mask with a tion of the retina in each layer of the col-
small aperture, minus the frequency of discharge liculus, and these points are in vertical
elicited by a small control spot of light of constant alignment. Input to the colliculus, there-
intensity also confined to the facet of the test recep- fore, may be likened to a stack of superim-
tor. Scale of ordinate on the right. The lower
(curvilinear) graph is the frequency of discharge posed geographical maps, each showing a
from the same test receptor when the mask was re- different feature of the same countryside.
moved and the entire pattern of illumination was Output originates in dendrites of collicular
projected on the eye in various positions, minus the cells that pass vertically through the hori-
frequency of discharge elicited by a small control
spot of constant intensity confined to the facet of zontal strata.
the receptor. Scale of ordinate on the left. (From Recordings from the terminal arboriza-
Ratliff and Harlline, 1959, J. Gen. Physiol.) tions of ganglion cells in the optic tectum
pect visual patterns to be transmitted with show that the four horizontal layers corres-
relatively little distortion, but with an in- pond to five functional classes, one each in
crease in the uncertainty or noisiness of the layers 1, 2, and 4, and two (Classes 3 and 5)
message. If the second were true, and if in- in layer 3. Each class responds optimally to
teraction increased as distance decreased, a different aspect of a single complex stimu-
then one might expect intensity discon- lus. For example, convex edge detectors
tinuities in the spatial or temporal visual (Class 2) do not respond to changes in gen-
pattern to be emphasized in the afferent eral light intensity, but do discharge when
message. The latter, in fact, appears to be a small object (1-3°) with a sharp edge is
the case in Limulus (Fig. 16). It seems of moved across a lighter background. All
particular interest that insertion of a single, parameters are important, and if, say, the
rather simple geometrical bias in an appar- background moved with the object or were
ently random system transforms it from one darker than the object, the response would
which increases noise in the afferent mes- be greatly diminished or absent. These
sage to one which, among other things, re- units have also been called "fly detectors."
duces redundancy in the message without The other classes are: Class 1, sustained
great loss of information about the spatial edge detectors; Class 3, changing edge de-
pattern (Barlow, 1961). tectors; Class 4, dimming detectors; and
Class 5, dark detectors. The functional
Connectivity in stratified neurofril properties of the five classes correlate rather
The second example is the retina-collicu- well with the five anatomical ganglion cell
lar system in the frog (Maturana, et al., types. If the suggested correlation proves
I960; Lettvin, et al., 1961). The pertinent correct, then the dendrites of edge detectors
anatomy is summarized first. and of dimming detectors ramify in differ-
 ORGANIZATION OF NEUROPIL 93

ent strata of the inner plexiform layer, each required for neural integration in visual
collecting from a different set of bipolar ter- systems.
minals.
Complex Patterning in Neuropil
Only two functional classes of tectal cells
have been described, "newness" detectors The last section pointed out that many
and "sameness" detectors. It is not neces- units in afferent neuropil may act as event
sary to go into their detailed properties at detectors. Such neurons tell whether or not
this time, but only to indicate that their re- a pre-determined complex sensory input is
ceptive fields are larger than that of any present, but do not describe that input in
ganglion cell, and that their optimal stimu- the patterns of their own activity. One may

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

lus, if not more complex, at least involves ask whether the reverse occurs: Are there
new combinations of qualities not found in significant, complex, and relatively autono-
the single ganglion cells. mous patterns of activity that may be
Let us review: Complex integration oc- elicited by simple stimuli in the appropri-
curs in the layered neuropil of the retina so ate neurons? To use Wiersma's phrase
that by the second internuncial neuron, (1952), is there a "push-button" to initiate
units recognize and respond optimally to appropriate events as well as a signal light
specific temporal-spatial configurations of to indicate when events have occurred?
the complex visual pattern within their re- Several analyses of animal behavior sug-
ceptive field. These configurations, or ef- gest that many relatively complex acts are
fective attributes of the stimulus, are lim- elicited by specific stimuli and may be
ited in number and apparently correspond treated as units with all-or-nothing prop-
to the kinds of neurons involved at the per- erties. This implies that complex, preset
tinent level of integration. The effective at- activity patterns exist in certain areas of the
tributes become potentially more complex neuropil and that once initiated, these pat-
at each stage of transmission; from receptor terns are like the regenerative action poten-
to bipolar, from bipolar to ganglion cell, tial in a single nerve fiber and are inde-
from ganglion cell to collicular cell, and so pendent of the initiating stimulus. Direct
on. The layered neuropil in this system, recordings of responses meeting these re-
therefore, permits successive separation and quirements have been described in several
recombination of stimulus attributes while preparations (Wiersma, 1952; Hagiwara
retaining the original geometrical relations and Watanabe, 1956; Horridge, 1961).
of the primary sensory layer. Physical as The response of an isolated lobster sto-
well as functional separation seems neces- matogastric ganglion to repeated trains of
sary, for each layer maps a separate attrib- presynaptic stimuli emphasizes another as-
ute. Vertical elements select from these pect of such "push-button reflexes" (Fig.
layers and apparently recombine in new 17). The first train elicits no post-synaptic
configurations at later stages in the system. response; with repetition, facilitation oc-
An analogy between the retina or colliculus curs and increasing numbers of post-synap-
and a punch card system seems rather apt. tic units become active. Eventually the pat-
Both employ a geometric code and both are tern of Uie response alters altogether, shift-
concerned with cross-reference and recom- ing from an irregular continuous barrage to
bination of a limited number of attributes. periodic bursts that continue for consider-
.Since the attributes themselves can involve able time after the termination of the
geometrical configuration of the stimulus, stimulus. One is reminded in some ways of
this may be a system which permits recog- the initiation of seizure activity in the ver-
nition of universals such as "triangle" or tebrate brain. Although the mechanism is
"circle." Further information about the not yet clear, it does appear that the gan-
mechanisms of attribute recognition or sort- glion shifts in a stepwise manner from one
ing is necessary, but if the above interpreta- functional state into another, and that in
tions are correct, proper geometry as well as one of these states it produces complex,
connectivity and electrical phenomena are maintained responses to rather brief, un-
94 DONALD M. MAYNARD

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

1'IG. 17. Complex response of isolaletl stomatogas- terruption through the ganglion, and consequently
tric ganglion of lobster, Homarus americanus, to one represent direct responses to the electrical stimuli.
second trains of presynaptic stimuli at 10 per sec- Two post-synaptic impulses occur with the 6th
ond, repealed at five second intervals. Extracellular train of stimuli, three with the 8th, 15 with the
10th, and bursts with the 12th. Bursts continued
recording electrodes placed distal to the ganglion.
for several seconds beyond the records shown, and
Numbers indicate the number of the stimulus train; the changed state of the ganglion continued still
dots on records indicate individual stimuli. There longer, as shown by test trains of stimuli. Calibra-
is no post-synaptic response in the first record, the tion, 0.5 second.
spikes shown come from fibers that pass without in-
Structured stimuli. The potentialities for ent responses would extend still further the
such activity must lie in the organization of functional potentialities of a limited num-
the ganglion neuropil itself, for extra-gan- ber of neurons with rigid, preset activity
glionic neural feed-back is absent in this patterns. The anatomical and physiological
preparation. properties of the hypothetical priming
If the above observations can be gener- mechanism remain to be specified in inver-
alized to apply to central ganglia, then we tebrates.
may answer our first question as follows:
Braiyi Waves
Yes, there are "push-button reflexes," but
activation requires two operations; first the A discussion of neuropil organization
switch must be turned on and then the but- should not be concluded without reference
ton pushed. It is not sufficient for a given to the rhythmic, often sinusoidal potential
neural system to have the structural or- oscillations called "brain waves." These are
ganization requisite for complex, semi-au- usually recorded by surface electrodes from
tonomous activity; the system must also be specific masses of nerve tissue. Although
in the proper physiological state in order to there is no complete agreement on their
respond to the specific, triggering stimulus. genesis or significance, most would concede
If much of an organism's behavior is or- that the waves are not envelopes of action
ganized around push-button responses, then potentials, that they reflect potential oscil-
such a priming or activating mechanism lations in neurons, that they appear in typi-
gives an added degree of freedom, and cal form only in masses of nerve tissue, and
should facilitate adaptive behavior. Several that they represent some synchronized ac-
functional states with corresponding differ- tivity in large numbers of units. The alpha
 ORGANIZATION OF NEUROPIL 95

rhythm of the vertebrate cortex is perhaps tegrative neural activity. The mechanisms
the best known example, but analogous si- of such interaction, however, are not lim-
nusoidal oscillations have been recorded ited to those we know from peripheral junc-
from olfactory bulb, cerebellum, optic gan- tions and nerve processes. They also involve
glia of insects, and recently, the ocelli of other properties, many as yet inadequately
some insects (see Gerard and Young, 1937; studied. Indeed, it seems likely that specific
Crescitelli and Jahn, 1942; Bullock, 1945b; forms of neuropil may have their own
Bum and Catton, 1956; Ruck, 1961). The unique functional characteristics.
structures giving rise to the above oscilla-
tions differ rather markedly in general ap-
REFERENCES

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

pearance, but they do have one factor in
common. In all cases the tissue appears to Allison, A. C. 1953. The morphology ot the olfac-
be structured neuropil. Smooth oscillations tory system in the vertebrates. Biol. Rev. 28:195-
244.
apparently have not been recorded from un- Barlow, H. B. 1961. Three points about lateral in-
structured neuropil. If the synchronization hibition, p. 782-786. In W. A. Rosenblith, (ed.),
characteristic of brain waves is mediated by Sensory communication. The M.I.T. Press and
some mechanism not involving action po- John Wiley and Sons, New York.
Bennett, M. V. L. 1960. Electrical connections be-
tentials and synapses, possibly electrotonic tween supramedullary neurons. Fed. Proc. 19:282.
potentials or chemical agents, then such Bethe, A. 1895. Studien iiber das Centralnervensys-
geometric organization may have greater tem von Carcinus maenas nebst Angaben iiber
significance than originally postulated. ein neues Verfahren der Methylenblaufixation.
Arch. Mikroskop. Anat. u. Entwicklungsmech. 44:
SUMMARY 579-622.
. 1897. Vergleichende Untersuchungen iiber
Much of the factual information and die Funktionen des Centralnervensystems der Ar-
many of the ideas presented in the above thropoden. Arch. ges. Physiol. Pfliiger's 68:449-
discussion are neither new nor original. 545.
Bodian, D. 1952. Introductory survey of neurons.
Nevertheless, they are worth considering Cold Spr. Harb. Sympos. Quant. Biol. 17:113.
for they stress several points of some impor- Bullock, T. H. 1945a. The anatomical organization
tance: oC the nervous system of Enteropneusta. Quart.
1. Although acceptable for general conver- J. Micr. Sci. 86:55-111.
sation, neuropil is unsatisfactory when un- . 1945b. Problems in the comparative study
of brain waves. Yale J. Biol. Med. 17:657-679.
modified as a technical term in that it does Bullock, T. H., and C. A. Terzuolo. 1957. Diverse
not refer to a commonly accepted, well-de- forms of activity in the somata of spontaneous
fined structure. Its original anatomical con- and integrating ganglion cells. J. Physiol. 138:
notations have broadened to include func- 341-364.
Bum, E. T., and W. T. Catton. 1956. Electrical re-
tional aspects, and thus it logically encom- sponses to visual stimulation in the optic lobes of
passes the entire portion of the nervous sys- the locust and certain other insects. J. Physiol.
tem devoted to neural integration, cell 133:68-88.
bodies, processes, and all. Cajal, S. R. 1904. Textura del sistema nervioso del
2. Neuropil structure is complex, but not hombre y de los vertebrados. II. Part 1. Nicolas
Moya, Madrid. 608 p.
haphazard. Similar patterns reappear in Cajal, S. R., and D. Sanchez. 1915. Contribution al
widely divergent organisms. Where ade- conocimiento de los cenlros nerviosos de los insec-
quate analyses are available, the structural los. Trabajos Lab. Investig. Biol., Madrid. 13:1-
pattern proves to have functional signifi- 164.
cance. A preliminary classification on the Crescitelli, F., and T. H. Jahn. 1942. Oscillatory
electrical activity from the insect compound eye.
basis of fiber configuration divides all neu- J. Cell. Comp. Physiol. 19:47-66.
ropil into four categories: unstructured Dempsey, E. W., and S. Luse. 1958. Fine structure
plexiform, unstructured diffuse, structured of the neuropil in relation to neuroglia cells, p.
glomerular, and structured stratified. Geo- 99-108. In W. E. Windle, (ed.), Biology of neu-
roglia. C. C Thomas, Springfield.
metrical configurations as well as connec- De Robertis, E. 1958. Submicroscopic morphology
tivity diagrams are important. and function of the synapse. Exp. Cell. Res.,
3. Neuropil is the site of most complex in- Suppl. 5. 347-369.
96 DONALD M. MAYNARD

Florey, E. 1962. Recent studies on synaptic trans- tem. Proc. Roy. Soc. (London), B, 140:147-168.
milters. Am. Zoologist 2:45-54. Preston, f. B., and U. Kennedy. 1960. Integrative
Friede, R. I.. I960. Histochemical investigations on synaptic mechanisms in the caudal ganglion of
succinic dehydrogenase in the central nervous sys- the crayfish. J. Gen. Physiol. 43:671-681.
leni. IV. A histochemical mapping of the cere- Ratliff, F. 1961. Inhibitory interaction and the de-
bral corlex of the guinea pig. |. Neurochem. 5: tection and enhancement of contours, p. 183-203.
156-171. /// W. A. Rosenblith, (ed.). Sensory communica-
Furshpan, E. J., and D. D. Potter. 1959a. Trans- tion. The M. I. T. Press and John Wiley and
mission at the giant motor synapses of the cray- Sons, New York.
fish. J. Physiol. 145:289-325. RatliU, I"., and H. K. Hartline. 1959. The responses
•. 1959b. Slow post-synaptic potentials re- of Limulus optic nerve fibers to patterns of illu-
corded from the giant motor fibre of the crayfish. mination on the receptor mosaic. J. Gen. Physiol.
J. Physiol. 145:326-335. 42:1241-1255.

Downloaded from https://academic.oup.com/icb/article/2/1/79/2110873 by guest on 01 May 2024

Galambos, R. 1961. A glia-neural theory of brain Retzius, G. 1890. Zur Kenntniss des Nervensystems
function. Proc. Nat. Acad. Sci. 47:129-136. der Crustaceen. Biol. Untersuch. N. F. 1:1-50.
Gerard, R. W., and J. Z. Young. 1937. Electrical Robertson, J. D. 1953. Infrastructure of two in-
activity of the central nervous system of the frog. vertebrate synapses. Proc. Soc. Exp. Biol. Med.
Proc. Roy. Soc. (London), B, 122:343-352. 82:219-223.
Hagiwara, S., and T. H. Bullock. 1957. Intraccl- Ruck, P. 1961. Electrophysiology of the insect dor-
lular potentials in pacemaker and integrative neu- sal ocellus. II. Mechanisms of generation and in-
rons of the lobster cardiac ganglion. J. Cell. hibition of impulses in the ocellar nerve of drag-
Comp. Physiol. 50:25-47. onfiies. J. Gen. Physiol. 44:629-639.
Hagiwara, S., and A. Watanabe. 1956. Discharges Scharrer, E. 1945. Capillaries and mitochondria in
in motoneurons of cicada. J. Cell. Comp. Physiol. neuropil. J. Comp. Neurol. 83:237-243.
47:415-428. Scheibel, M. E., and A. B. Scheibel. 1958. Neurons
Hanstrom, 15. 1928. Vergleichendc Anatomie des Ner- and neuroglia cells as seen with the light micro-
vensystems der wirbellosen Tiere. Julius Springer, scope, p. 5-23. In W. F. Windle, (ed.), Biology of
Berlin. 628 p. neuroglia. C. C Thomas, Springfield.
Hardy, W. B. 1929. Note on the central nervous Spencer, W. A., and E. R. Kandel. 1961. Electro-
system of the crayfish. J. Physiol. 67:166-168. physiology of hippocampal neurons. IV. Fast pre-
Herrick, C. J. 1948. The brain of the tiger sala- potentials. J. Neurophysiol. 24:272-285.
mander, Amblystoma tigrinum. The University Tasaki, I., and J. J. Chang. 1958. Electric response
of Chicago Press, Chicago. 409 p. of glia cells in cat brain. Science 128:1209-1210.
Horridgc, G. A. 1961. The centrally determined se- Terzuolo, C. A., and T. H. Bullock. 1958. Acceler-
quence of impulses initiated from a ganglion of ation and inhibition in crustacean ganglion cells.
the clam Mya. J. Physiol. 155:320-336. Arch. Ital. Biol. 96.117-134.
Hoyle, G. 1953. Potassium ions and insect nerve Treherne, J. E. 1961. Sodium and potassium fluxes
muscle. J. Exp. Biol. 30:121-135. in the abdominal nerve cord of the cockroach,
Lcttvin, J. Y., H. R. Maturana, W. H. Pitts, and Periplanela americana L. J. Exp. Biol. 38:315-
W. S. McCulloch. 1961. Two remarks on the vis- 322.
ual system of the frog, p. 757-776. In W. A. Trujillo-Cenoz, O. 1959. Study on the fine structure
Rosenblith, (ed.), Sensory communication, The of the central nervous system of Pholus labrnscoc,
M.I.T. Press and John Wiley and Sons, New L. (Lepidoptera). Zeit. f. Zellforsch. 49:432-446.
York. Twarog, B. M., and K. IX Roeder. 1956. Properties
of the connective tissue sheath of the cockroach
.Mackie, G. O. 1960. The structure of the nervous
abdominal nerve cord. Biol. Bull. 111:278-286.
system in J'elella. Quart. J. Microscop. Sci. 101:
Walanabe, A. 1958. The interaction of electrical ac-
119-132.
tivity among neurons of lobster cardiac ganglion.
Maturana, H. R., J. Y. Lcttvin, W. H. Pitts, and Jap. J. Physiol. 8:305-318.
W. S. McCulloch. 1960. Physiology and anatomy Wiersma, C. A. G. 1952. Neurons of arthropods.
of vision in the frog. J. Gen. Physiol. 43 suppl.: Cold Spr. Harb. Sympos. Quant. Biol. 17:155-163.
129-175. . 1962. The organization of arthropod cen-
Maynard, D. M. 1956. Electrical activity in the tral nervous systems. Am. Zoologist 2:67-78.
cockroach cerebrum. Nature 177:529-530. Wigglesworth, V. B. 1958. The distribution of es-
Maynard, E. A., and D. M. Maynard. 1960a. Cho- terase in the nervous system and other tissues of
linesterases in the nervous system of the lobster, the insect, Rhodnius prolixus. Quart. J. Micro-
ffomarus americanus. Anat. Rec. 137:380. scop. Sci. 99:441-450.
. 1960b. Cholinesterase in the crustacean . 1959. The histology of the nervous system
muscle receptor organ. J. Histochem. Cytochem. of an insect, Rhodnius prolixus (Hemiptera). II.
8:376-379. The central ganglia. Quart. J. Microscop. Sci.
I'alay, S. I.. 1958. The morphology of synapses in 100:299-313.
the central nervous system. Exp. Cell. Res., Suppl. . 1960. The nutrition of the central neivous
5. 275-293. s\stein in the cockroach, Periplaneta americana
Pantin, C. F. A. 1952. The elementary nervous sys- L. J. Exp. Biol. 37:500-512.