Cortical Circuits by Edward White and Asaf Keller

area reviews

Part I by Edward White

"The theoretical framework upon this and subsequent chapters are based derives from the proposition that the processing of information by the nervous system is effected through sunapses, and that the anatomical organization of synaptic connnections, that is, the type of synapse and the identities of the neurons involved, is largely stable within the individual adult animal and relatively constant from one species to the next. In contrast, the physioligy of synapses is plastic, allowing for variations on a scale of milliseconds in the level f electrical activity within synaptic pathways. ... The notion that the cerebral cortex of different species is built according to some common, basic plan is woven throughout the following treatise. ... The principal focus of this work is the cerebral cortex because, perhaps more than any other area of the nervous system, it is the cerebral cortex that sets apart mammals from other animals and man from other mammals."

Chapter 1. General organization of the cerebral cortex

striate cortex = where the stripe (stria) of Gennari is = primary visual cortex

"Topographically organized pathways project from sensory organs, through thalamic nuclei and on to cytoarchitecturally distinct areas of the crerbral cortex"

p. 6 "The use of the word 'secondary' =is somewhat unfortunate because of the implication that the secondary areas are in some way subordinate to the primary ones. The fact is that the secondary areas were so called simply because their discovery followed that of the primary aears (e.g. White, 1979, 1987)."

p 7 "The manner in which multiple representations cooperate in the processing of cortical information is unclear. One possibility is ... hierarchical... whereby input... is passed in stepwise fashion through a series of cortical areas... Alternatively, the multiple representations may act in parallel to process separate aspects of input from the periphery (e.g. Ballard et al, 1983), a contention supported by results showing that different cortical areas receive input from different regions of the thalamus (e.g. for visual areas, Dreher et al, 1980; for somatosensory areas, Dykes, 1983)."

p 7 "There is no general consensus as to what constitutes a cortical area or what criteria should be applied to identify and to distinguish between different cortical areas (see Van Essen, 1985)."

p 8 "Correlations of structure and function in cortical areas are now being reassessed, however, principally because of the realization that single cytoarchitectonic areas may contain multiple representations of the periphery."

Table "Patterns of lamination of the striate area of man and various subhuman primates according to various authors"

p 8 "Today, in line with the system employed by Brodmann (1906, 1909), the cerebral cortex in all mammalian species is divided into six basic layers" "The basic arrangement of six layers is clearest in areas of the neocortex referred to by Brodmann (1909) as homotypic; the six-layered arrangement is somewhat modified in areas he labeled as heterotypic."

the Van Essen hierarchy between visual cortex areas is founded on noting that 'lower' areas project to layer 4 of 'higher' areas (from superficial layers). 'Higher' areas project back to layers other than 4 of lower areas (from deep as well as superficial layers).

Vertical Arrangements

There are cortical columns. The barrels within part of mouse primary somatosensory cortex are cortical columns. However in some other places p 13 "clear structural correlates for functional columns have yet to be identified" and "One possibility for the genereal lack of correspondence between anatomical structure and functional columns probably has to do with the likelihood that certain functional columns are rather ephemeral, owing their existence to the continued presence of a specific set of stimulus conditions (cf. Crawford, 1985; see also Part III, Chapter 7)."

Neurotransmitters and receptors in the Cerebral Cortex

GABA is inhibitory. Glutmate and apartate are thought to be excitatory.

There's also acetylcholine and the monoamines including norepinephrine, dopamine, and serotonin. There's also glycine, thought to be inhibitory, histimine, and peptides including (CCK, NPY, VIP, SP, somatostatin, opiods) which are thought to be modulatory.

p 16 "However, the identification of a receptor site does not necessarily imply the the juxtaposition of an axon terminal containing the neurotransmitter related to that receptor. This raises the possibility that certain neuroactive substances may be involved in extrasynaptic activation, reaching their receptor sites by diffusion through the cortical neuropil."

Many neurotransmitters have multiple receptor subtypes.

Chapter 2. Cell types

asymmetrical synapses are thought to be excitatory, and symmetrical ones inhibitory

p. 19 "Cortical neurons are classified into two broad morphological categories: pyramidal, and stellate or nonpyramidal (e.g. Peters and Jones, 1984". bayle: judging by the way the chapter is laid out, i think he meant (pyramidal or (stellate or nonpyramidal)), but i would say ((pyramidal and stellate) or nonpyramidal)

Pyramidal cells

p. 19 "Pyramidal cells include a variety of morphological types that have the following features in common: A single, dominant apical dendrite, larger in diameter than the other dendrites, which usually extends vertically from the cell body toward the pial surface, and several basal dendrites, which radiate more or less horizontally from the base of the cell body (Firgure 2.1). All the dendrites of the pyramidal cells bear spines that tend to occur with greatest frequency in the middle regions of the dendrite (Globus and Scheibel, 1967).... The axon of a pyramidal cell originates typically from the base of the cell body, or less frequently from the proximal portion of a basal dendrite, and projects into the white matter, giving off collateral branches on the way."

p 19 "...the somata of pyramidal neurons exhibit a broad spectrum of shapes..." (e.g. not just 'pyramidal' shaped "...This is especially true of pyramidal neurons in layer IV where the variety of pyramidal cell shapes no doubt contributed to the naming of this layer as polymorphic or multiform. In fact, either perikaryal size nor shape is reliable as the sole criterion for differentiating pyramidal from nonpyramidal cells (Feldman and Peters, 1978). Nevertheless..." ppl call them 'pyramidal cells' and the name stuck... "For an account of pyramidal neurons that have been distinguished by morphological features including size (e.g. Betz cells), orientation (e.g. inverted pyramids), and laminar location (e.g. Meynert cells), the reader is referred to Feldman's review (1984, pp. 176-182).

A distinguishing aspect of pyramidal cell bodies is that they receive only symmetrical synapses; of the nonpyramidal cell types, only the spiny stellate neurons share this feature (e.g. Peters and Kaiserman-Abramof, 1970; White and Rock, 1980). The cell bodies of all other neuronal types receive both asymmetrical and symmetrical synapses (e.g. Peters and St. Marie, 1984, p. 435). The distribution of synaptic types onto dendrites is also similar for pyramidal and spiny stellate cells: Most synapses onto the dendritic shafts (axodendritic) of these neurons are symmetrical, whereas synapses onto their spines (axospinous) are usually of the asymmetrical type (e.g. Hersch and White, 1981b; White and Rock, 1980). In general, asymmetrical synapses form a comparatively large proportion of the axodendritic synapses onto other neuronal types (e.g. White et al, 1984). "

there is a sense that the laminar location of the parent pyramidal cell correlates to a lot of other differences, such as where its dendrites and axon branches go. There is a sense that for an individual cell, the axons and dendrites are set to ramify in specific layers. Afferents to the cortex also ramify mostly in specific layers.

axons may travel horizontally for quite aways, e.g. an example from Gilbert and Wiesel (1979) gives one that travelled horizontally 7 mm.

we say "axon collateral" but in many cases you may as well consider the cell to have a branched axon with no branch being dominant. more useful terminology is 'local axon aollateral' (near cell body) vs 'projection axon' (far).

p 23 "For instance, pyramidal neurons typically have local axon collaterals that ramify extensively within close proximity to the parent cell body..."

bayle: these pages have a bunch of potentially interesting cited examples of specific axon connections that i am omitting here, e.g. "..most of the pyramids in cat visual cortex that project to the LGN lack horizontal collateral branches; their axons ascend and arborize within layers V and IV."

p 24 "Finally, the discovery by DeFelipe? et al (1986a) that terminal branches of widely separated parts of the same collateral and even of separate collaterals from the same cell can converge on a single focus of terminations is consistent with the view that the distribution of axon collaterals is orderly and with the notion that the cerebral cortex is a highly ordered structure."

p 25 "Pyramidal neurons in layers II and III have axons that project to other regions of the crerbral cortex; pyramidal neurons in hte deeper layers project, in addition, to subcortical regions." ".. the old idea that the superficial layers are receptive or associative but that the deeper layers are specialized for output (e.g. Bolton 1910) would seem to possess some degree of validity."

neurotransmitters: probably glutamate and aspartate.

p 28 "Apical dendrites typically cross several layers of cortex..."

Nonpyramidal cells

White likes the classification scheme from Feldman and Peters (1978): p 29 "According to this scheme, nonpyramidal cells multipolar, bipolar, or bitufted according to the shape of their dendritic tree and as spiny, sparsely spiny, or smooth (nonspiny)... This classification scheme is sufficient for differentiating nonpyramidal cell types according to the shapes of their dendritic trees. However, a more comprehensive scheme can be achieved by taking into accoount the form and distribution of the axonal ramifications..."

recommends Fairen et al (1984) for a good comprehensive review of nonpyramidal neurons; this is in Vol 1 of The Cerebral Cortex, and he likes the other chapters too.

with some exceptions, p 30 "...the axonal ramifications of nonpyramidal cells are local and do not leave the area of cortex in which their parent cell body is situated."

Chandelier cells

Named because their p 30 "vertically oriented arrays of axon terminals" look like "te branches and candles of a chandelier".

p 31 "...The occurrence of chandelier cells in additional areas of the neocortex and in nearly every cortical layer...suggests that chandelier cells are an integral component of the cerebral cortex in all mammalian species....Their dendrites, which may span one or several layers of the cortex, have few branches and bear only occasional spines. Typically, the axons.. form a profuse plexus in the vicinity of the cell body, coextensive with or somewhat above or below the distirbution of the cell's dendritic tree. Some chandelier cells in the upper layers of the cortex may also posses a second, less extensive axonal ramification in the deeper layers of the cortex (e.g. Fairen and Valverde, 1980). Vertical rows of boutons belonging to chandelier cell axons form only symmetrical synapses, and these are presynaptic only to the axons initial segments of pyramidal cells (citeations). However, as Lund (1987) observed, the presence of chandelier cell axonal boutons in layer IVC of the macaque visual cortex, where pyramidal neurons are scarce, suggests that spiny stellate cells may also receive synapses from chandelier cells."

GABAergic. b/c of axonal position on postsynaptic initial axon segment, probably exerts a powerful inhibition.

Basket cells

p 32 "As with chandelier cells, basket cells are recognized primarily by the distribution of their axonal branches, which form nests or baskets around cell bodies and proximal dendrites belonging to pyramidal cells (Fairen et al 1984).... Cajal (1909-1911) noted that each is composed of branches from more than one axon collateral and that each axon collateral contributes to several pericellular nests." "smooth or sparsely spiny dendrites that often extend well above and below the layer containing the parent cell body" "(Marin-Padilla (1969, 1970, 1972)...occur mainly in layers III and V" "dendrites...radiate in all directions, but the vertical ones prodiminate such that some cells have a bitufted appearance." "...axons...initially ascend or descend upon leaving the cell body and then run horizontally for as much as a millimeter while contributing to pericellular nests; the presence of a long, horizontal axon is contsidered typical for basket cells of higher animals (Jones and Hendry, 1984, pp. 316-420)"


you often can't see the baskets if you only stain one cells, as multiple basket cells' axons form baskets together

p 33 "...we propose that it may be useful to include as basket cells other nonpyramidal cell types whose axons contact pyramidal cell somata whether or not true axonal baskets are formed"

p 33 "The initial descriptions... emphasized the association of the 'basket' with pyramidal neurons, and from this the impressino might be gotten that basket cells synapse exclusively with the cell bodies and proximal dendrites of pyramidal neurons. However, recent evidence suggests" otherwise, "For example, counts of 241 elements postsynaptic to three basket cells in the cat visual cortex show less than half to synapse onto the cell bodies and proximal dendrites of pyramidal cells. THe permainder synapse with the cell bodies and dendrites of non-pyramidal neurons and with spines, axons, and distal dendrites belonging to pyramidal neurons or to cells of undetermined origin (Freund et al, 1986; Somogyi et al, 1983a)."

p 34 "Because the axons terminals belonging to basket cells form a large proportion of their synapses with neuronal cell bodies and proximal dendrites, it can be assumed that basket cells exert a powerful inibition on the postsynaptic neuron by virtue of the proximitiy of the inhibitory synapses to the trigger zone, that is, the axon initial segmment. That axons belonging to basket cells project horizontally for relatively long distances... suggests that these cells exert a powerful inbition on cells in neighboring functional columns"

Vertically oriented neurons

Spiny Stellate cells

"The lack of a dominant, apical dendrite is the chief characteristic by which spiny stellate cells may be distinguished from pyramidal neurons; however, inmany other ways, the two cell types are identical. For instance, spiny stellate cells, like pyramidal neurons, have dendrites that bear large numbers of spines, and theiir axons, at least initially, descend toward the white matter and may even enter it (Firgure 2.6). Both types of neuron appear similar at the fine structural level and differ from all other neuronal types in that their somata receive only symmetrical synapses. For these reasons, Lund (1984) proposed that spiny cortical neurons form a continuum with the "typical" pyramidal and spiny stellate neurons ars extreme types."

Cross-species correlations of nonpyramidal neuronal types

p. 39-40: scaling argument for basket cells in mice (note: bayle: a quick skim of recent literature (e.g. doi 10.1093/cercor/12.4.395 Anatomical, Physiological, Molecular and Circuit Properties of Nest Basket Cells in the Developing Somatosensory Cortex) seems to show that the existence of basket cells is now accepted in rodents; this item is provided just because it makes a scaling argument) some say that basket cells are seen in primate sensory and motor regions, but not in other mammals. However, basket cells might look different in smaller brains: "The cerebral cortices of higher animals are thicker than those of lower animals. In accord with the argument that identical numbers of neurons underlie comparable areas of the pial surface in different species (Rockel et al, 1980) (bayle: carlo and stevens have recently confirmed Rockel's with updated methods), it sands to reason that neurons in higher animals display a lower packing density, that is, are farther apart form one another than neurons within the thinner cortices of lower animals. The number of pyramidal neurns contacted by individual basket cells is uncertain; however, it is clear that, to contact the same number of pyramids, the axon of a basket cell in man must necessarily travel much further than the axon must of the basket cell hypothesized for the mouse. Thus, by simply considering the distances involved in traversing the cortices of different species, it should come as no surprise that cells having long, horizontal axons (long, both in an absolute and even in a relative sense) are uncommon in lower animals. As stated previously, the presence of long horizontal axons has been and remains an important criterion for the identification of basket cells in higher animals (e.g. the review by Jones and Hendry, 1984)."

also, basket cells were so named because you see 'baskets' of their axons "around the somata of pyramidal cells". However, these baskets are formed by multiple basket cells and you may not see baskets if you just look at the axons of one basket cell. (DeFelipe? et al 1986 "A correlative electron microscopic study of basket cells and ...").

random scaling data:

"Cortices of smaller brains tend to be composed of neurons having smaller sizer. For instance, the somata of pyramidal cells in the visual cortex of the rat range in size froom 10 to 18 microns (Peters and Kara, 1985), whereas those in the visual cortex of the cat range in size from 20 to 60 microns (Sholl, 1953). Within a single species, the larger the pyramidal cell body, the greater the number of synapses it receives (e.g. Peters and Kara, 1985), and there is every reason to believe that a similar relationship obtains across species. Thus, the smaller pyramidal cell bodies in the brains of loewr animals can be expected to be postsynaptic to fewer axons terminals, and possibly to fewer axonal branches, than would the larger pyramidal cell bodies in higher species."

and relevance to not seeing baskets in basket cells in rodents: "The number of axonal branches that impinge on a pyramidal cell body in a lower animal may be sufficiently small that when viewed collectively the axonal branches do not exhibit a basket shape".

p.45 "The problem is that observations made with the light microscope only imply the existence of a synaptic contact between specific neuronal elements -- the synapse may not actually be formed. Indeed, Peters and Proskauer (1980) cite severeal instances in which predictions of synapses made on the basis of light microscopic observations were discounted subsequently on the basis of evidence obtained with the electron microscope" (bayle: i think he means 'suggest' instead of 'imply')

Chapter 3. Synaptic connections between identified elements

Synapses of extrinsic afferents with cortical neurons

p. 63: "Results to date indicate that, in every system examined, axon terminals belonging to extrinsic afferents to the cerebral cortex invariably form only asymmetrical, presumed excitatory synapses (e.g. Colonnier, 1981 The electron-microscopic analysis of the nuronal organization of the crerbral cortex; for a possible exception, see Einstein et al, 1987 Ultrastructure of synapses from the A-lamina of the lateral geniculate nucleus in layer IV of the cat striate cortex).

p. 65: "These findings, which include investigations of thalamocortical connectivity in different cortical areas in several species, indicate that every neuron having a dendrite in layer IV of the primary sensory areas of the neocortex forms some proportion ofits synapses with thalamocortical afferents (cf. White, 1987)....Thus, ample findings exist to refute the suggestion that spiny stellate cells are the sole or even the main recipients of thalamic input (cf. Eccles, 1984). This relatively new view of thalamocortical synaptic relations provides a direct and important challenge to the concpet that thalamic input to the cortex is processed by hierarchically organized chains of neurons (see Part III)."

p. 67: "Comparatively little information is available regarding synaptic connections involving other afferent systems that project to the primary sensory areas or to other reigons of the cerebral cortex, but in general, what is known is consistent with the notion that afferents to the cortex form asymmetrical, excitatory synapses with both pyramidal and non-pyramidal cells."

p. 68: "The identification of unlabeled postsynaptic dendrites as of nonpyramidal origin is somewhat more prblematical, because it is based mainly on the assumption that only dendrites belonging to certain types of nonpyramidal cells do not have spines, but that dendrites of all other neuronal types do. However, it has been clear for some time that the proximal dendrites of pyramidal cells may have very few spines (e.g. Cajal 1909-1911), and it has been shown also that middle and distal portions of pyramidal cell dendrites may in some instances lack spines (Hersch and White, 1981b)."

p. 68: Also, ppl tend to assume that if an asymmetrical synapse is onto a dendrite, not a spine, then the postsynaptic cell is not pyramidal, because "dendrites of pyramidal neurons usually receive asymmetrical synapses only on their spines". However, "some parts of pyramidal cell dendritic shafts are postsynaptic at relatively high frequency to asymmetrical synapses (e.g. Hersch and White, 1981b)."

p. 68: summarizing the previous two, "it is likely that in osme instances, unlabeled, nonspiny segments of pyramidal cell dendrites have been mistakenly identified as of nonpyramidal origin".

Part II by Asaf Keller

Chapter 4. Functional properties of cortical neurons

p 126 "Nearly all pyramidal cells send their axons to other areas of the brain..."

corticotectal, corticocollicular, and corticopontile cells have complex receptive fields, but corticostriatal, corticothalamic, and transcallossal neurons may have simple or complex receptive fields.

Chapter 5. Synaptic circuitry revealed by electrophysiology

p. 142: "Intrinsic excitatory interactions may involve pyramidal neurons or nonpyramidal neurons, such as spiny stellate cells, some bipolar neurons, and certain double bouquet cells, which form asymmetric, presumably excitatory synapes"

p. 143 "One approach to determining the contribution of intracotrical interactions for shaping the receptive field properties of cortical neurons is to investigate the effets on cortical neurons of the selective inactiviation of specific intracortical inputs. This approach was followed by Malpeli (1983; Malpeli et al. 1986), who showed that in the absence of thalamic input simple cells in layers IV and VI of the visual cortex are silenced, but that complex cells in the superficial layers retain their receptive field properties. These findings imply that the activity of cells with complex receptive fields are not dependent on input from cells having simple receptive fields. Moreover, simple cells respond preferentially to slow stimulus velocities, whereas complex cells respond preferentially to significantly faster stimulus velocities (e.g. Pettigrew et al., 1968, Movshon, 1974). This indicates that simple cells are unlikely to provide direct input to cells with complex receptive field properties. In addition, activation of local axon collaterals belonging to layer VI corticogeniculate neurons, some of which have complex receptive field properties (e.g. Gilbert, 1977; Harvey, 1980), produces monotynaptic EPSPs in simple cells in layer IV (Ferster and Lindstrom, 1985), suggesting that complex cells may provide input to simple cells. This assumption is supported by studies showing that the activity profile of simple cells may be modulated by input from cells having complex receptive field properties (e.g. Burr et al., 1981; Maffei, 1985).

The hierarchical model suggests that inputs from supragranular layers to the infragranular layers are responsible for the formation of the receptive field properties of neurons in the infragranular layers. However, inactivation of cells in the supragranular layers, either by cooling or lesioning these layers, results in the disarrangement of the receptive field properties of only the "special complex" cells in layer V; the receptive field properties of other cells in layers V and VI remain intact (Schwark et al., 1987).

A direct approach to studying intracortical interactions involves using cross-correlation analyses to determine the types and strengths of interactions between cells whose receptive field properties have been defined. Using this approach, several studies have shown that intracortical monosynaptic excitatory interactions occur almost exclusively between cells with complex receptive properties, and then only when the pair of cells display similar orientation selectivities (e.g. Michalski et al., 1983; Toyama et al., 1981b). Excitatory interactions from cells with simple receptive fields to cells with complex receptive fields have not been observed in these studies.

In conclusion, the results of these studies do not support the assumption included in the hierarchical model that the receptive field properties of complex cells are the result of excitatory inputs from simple cells. Simple cells may provide input to complex cells, but the available data suggest that the influence of this input is less important for shaping the receptive field properties of complex cells that the hierarchical model would have us believe. "

p. 145 "Electrical stimulation of cortical afferents does not induce monosynaptic inhibitory postsynaptic potentials (IPSPs) in the cerebral cortex (e.g. Ferster and Linsdtrom, 1983; Toyama et al., 1974), and so inhibitory interactions are presumed to be mediated only by intracortical mechanisms. In addition, afferents to the cerebral cortex form only asymmetrical synapses, which are presumed to be excitatory;..."


151 b P 152 can GABA be excitatory? 153 153 b 154 156 156 180 183 184 184 b Could the reciprocal inputs just be conveying the connection latency? Mb combined with attention? If they are actually the cortical receivers 187 187 b. 188 t

190 t 191 t 191 b 193 t 194 196 m 197 t 197 m diffuse 197 b 197 b 3/198 t 3 roles for GABA 198m 199t 199m 200m 200b 200b2 201m 201b 202t 202m 202m2 203t 203m 204m

topics 4 me reciprocal, transient assm, routing, coding, canonical cortical circuit, engram, action selection, knowledge rep, preception and perception, belief maint and inference, prediction and imagination, sequence recall, my systems, oscillations, multiplexing, cross-modal cross-species arch,