Stewart Shipp

The Organization of Connections between Areas V5 and V1 in Macaque Monkey Visual Cortex

Area V5 or MT of primate extrastriate visual cortex is specialized for involvement in the analysis of motion and receives input from two layers, 4B and 6, of the striate cortex or V1. Injections of horseradish peroxidase ‐ wheatgerm agglutinin into V5 reveal a patchy distribution of labelled cells and axonal terminals in layer 4B, suggesting the presence of a segregated and functionally specialized subsystem within the layer. The patches are similar in size and frequency to the cytochrome oxidase blobs of layers 2 and 3, but bear little systematic relationship to them. V5‐efferent cells in layer 6, however, tend to avoid the cores of the blobs.

Functional demarcation of a border between areas V6 and V6A in the superior parietal gyrus of the macaque monkey

We have compared physiological data recorded from three alert macaque monkeys with separate observations of local connectivity, to locate and characterize the functional border between two related but distinct visual areas on the caudal face of the superior parietal gyrus. We refer to these areas as V6 and V6A. They occupy almost the entire extent of the anterior bank of the parieto‐occipital sulcus, V6A being the more dorsal. These two areas are strongly interconnected. Anatomically, we have defined the border as the point at which labelled axon terminals first adopt a recognizably ‘descending’ pattern in their laminar characteristics, after injections of wheatgerm agglutinin‐horseradish peroxidase into the dorsal half of the gyrus (in presumptive V6A). A similar principle was used to recognize the same border by the pattern of input from area V5, except that in this case the relevant transition in laminar characteristics is that between an ‘intermediate’ pattern (in V6) and an ‘ascending’ pattern (in V6A). V6A was found to be distinct from V6 in a number of its physiological properties. Unlike V6, it contains visually unresponsive cells as well as units with craniotopic receptive fields (‘real‐position’ cells), units tuned to very slow stimulus speeds, units with complex visual selectivities and units with activity related to attention. V6A was also found to have a larger mean receptive field size and scatter than V6. By contrast, response properties related to the basic orientation and direction of moving bar stimuli were indistinguishable between V6 and V6A, as was the influence of gaze direction on cell activity in the two areas. Two‐dimensional maps of the recording sites allowed reconstruction of the V6/V6A border. For comparison, the anatomical results were rendered on two‐dimensional maps of identical format to those used to summarize the physiological data. After normalizing for relative size, the physiological and connectional estimates of the border between V6 and V6A were found to coincide, at least within the range of individual variation between hemispheres. An architectonic map in the same format was also made from a hemisphere stained for myelin and Nissl substance. Area PO, defined by its general density of myelination was not distinct in this material, but several architectural features were traceable and one of these was also found to approximate the V6/V6A border. The particular criteria that distinguish V6 from V6A differ from a recent description of areas PO and POd in the Cebus monkey; we believe it most likely that PO and POd together may correspond to V6.

Predictions not commands: active inference in the motor system.

The descending projections from motor cortex share many features with top-down or backward connections in visual cortex; for example, corticospinal projections originate in infragranular layers, are highly divergent and (along with descending cortico-cortical projections) target cells expressing NMDA receptors. This is somewhat paradoxical because backward modulatory characteristics would not be expected of driving motor command signals. We resolve this apparent paradox using a functional characterisation of the motor system based on Helmholtz’s ideas about perception; namely, that perception is inference on the causes of visual sensations. We explain behaviour in terms of inference on the causes of proprioceptive sensations. This explanation appeals to active inference, in which higher cortical levels send descending proprioceptive predictions, rather than motor commands. This process mirrors perceptual inference in sensory cortex, where descending connections convey predictions, while ascending connections convey prediction errors. The anatomical substrate of this recurrent message passing is a hierarchical system consisting of functionally asymmetric driving (ascending) and modulatory (descending) connections: an arrangement that we show is almost exactly recapitulated in the motor system, in terms of its laminar, topographic and physiological characteristics. This perspective casts classical motor reflexes as minimising prediction errors and may provide a principled explanation for why motor cortex is agranular.

A visuo-somatomotor pathway through superior parietal cortex in the macaque monkey: cortical connections of areas V6 and V6A

This report addresses the connectivity of the cortex occupying middle to dorsal levels of the anterior bank of the parieto‐occipital sulcus in the macaque monkey. We have previously referred to this territory, whose perimeter is roughly circumscribed by the distribution of interhemispheric callosal fibres, as area V6, or the ‘V6 complex’. Following injections of wheatgerm agglutinin conjugated to horseradish peroxidase (WGA‐HRP) into this region, we examined the laminar organization of labelled cells and axonal terminals to attain indications of relative hierarchical status among the network of connected areas. A notable transition in the laminar patterns of the local, intrinsic connections prompted a sub‐designation of the V6 complex itself into two separate areas, V6 and V6A, with area V6A lying dorsal, or dorsomedial to V6 proper. V6 receives ascending input from V2 and V3, ranks equal to V3A and V5, and provides an ascending input to V6A at the level above. V6A is not connected to area V2 and in general is less heavily linked to the earliest visual areas; in other respects, the two parts of the V6 complex share similar spheres of connectivity. These include regions of peripheral representation in prestriate areas V3, V3A and V5, parietal visual areas V5A/MST and 7a, other regions of visuo‐somatosensory association cortex within the intraparietal sulcus and on the medial surface of the hemisphere, and the premotor cortex. Subcortical connections include the medial and lateral pulvinar, caudate nucleus, claustrum, middle and deep layers of the superior colliculus and pontine nuclei.

Modular Connections between Areas V2 and V4 of Macaque Monkey Visual Cortex

We have studied the connections between two visual areas of macaque monkey cortex, V2 and V4, by injecting wheat‐germ agglutinin horseradish peroxidase (HRP‐WGA) into V4 and examining the distribution of labelled cells and terminals in V2, in relation to its characteristically striped cytochrome oxidase architecture. The cells projecting from V2 to V4 are arranged in bands and the number of bands per cycle of cytochrome oxidase stripes varies (one cycle consists of a thin stripe, a thick stripe and two interstripes). In the Type 1 connectivity pattern, there is just one band per cycle, centred over the thin stripes but normally spreading into the neighbouring interstripes. In the Type 2 connectivity pattern there are two bands per cycle, generally rather narrower and centred over the interstripes. Thick stripes are mostly free of labelled cells. The return projection from V4 to V2, whilst being concentrated in the vicinity of the labelled cells, is more diffusely distributed and invades the territory of all the stripes.

Structure and function of the cerebral cortex

The complexities of cortical circuitry are nothing short of fiendish, and the problem of integrating genetic, morphological and physiological details from diverse cortical areas and across diverse species is a worthy challenge to the burgeoning science of neuroinformatics. Though inconsistencies abound, the fact that some trans-areal, trans-specific generalisations are possible, and justified, is a quite remarkable observation. Following the strategy of ‘know thine enemy’, it appears that the cortical fiend has some interesting habits, which we can usefully begin to tag with some shorthand, functional labels.

Reflections on agranular architecture: predictive coding in the motor cortex

Highlights • Predictive coding explains the recursive hierarchical structure of cortical processes.• Granular layer 4, which relays ascending cortical pathways, is absent from motor cortex.• Perceptual inference results if ascending sensory data modify sensory predictions action, if spinal reflexes enact descending motor and/or proprioceptive predictions.• Motor layer 4 regresses as motor predictions inherently require less modification.

The importance of being agranular: a comparative account of visual and motor cortex

The agranular cortex is an important landmark—anatomically, as the architectural flag of mammalian motor cortex, and historically, as a spur to the development of theories of localization of function. But why, exactly, do agranularity and motor function go together? To address this question, it should be noted that not only does motor cortex lack granular layer four, it also has a relatively thinner layer three. Therefore, it is the two layers which principally constitute the ascending pathways through the sensory (granular) cortex that have regressed in motor cortex: simply stated, motor cortex does not engage in serial reprocessing of incoming sensory data. But why should a granular architecture not be demanded by the downstream relay of motor instructions through the motor cortex? The scant anatomical evidence available regarding laminar patterns suggests that the pathways from frontal and premotor areas to the primary motor cortex actually bear a greater resemblance to the descending, or feedback connections of sensory cortex that avoid the granular layer. The action of feedback connections is generally described as ‘modulatory’ at a cellular level, or ‘selective’ in terms of systems analysis. By contrast, ascending connections may be labelled ‘driving’ or ‘instructive’. Where the motor cortex uses driving inputs, they are most readily identified as sensory signals instructing the visual location of targets and the kinaesthetic state of the body. Visual signals may activate motor concepts, e.g. ‘mirror neurons’, and the motor plan must select the appropriate muscles and forces to put the plan into action, if the decision to move is taken. This, perhaps, is why ‘driving’ motor signals might be inappropriate—the optimal selection and its execution are conditional upon both kinaesthetic and motivational factors. The argument, summarized above, is constructed in honour of Korbinian Brodmanns centenary, and follows two of the fundamental principles of his school of thought: that uniformities in cortical structure, and development imply global conservation of some aspects of function, whereas regional variations in architecture can be used to chart the ‘organs’ of the cortex, and perhaps to understand their functional differences.

Visuotopic organization of the lateral suprasylvian area and of an adjacent area of the ectosylvian gyrus of cat cortex: a physiological and connectional study.

We have explored the visuotopic organization of the territory surrounding the middle suprasylvian sulcus (MSS) of cat cerebral cortex by electrophysiological mapping, and by tracing the topography of its cortical and subcortical connections using wheatgerm-agglutinin horseradish peroxidase (WGA-HRP). Observations from the two approaches were concordant, and confirmed the presence of two separate visual areas in the MSS that approximate, but do not exactly correspond, to the location and internal organization of the posterior medial and posterior lateral lateral suprasylvian (PMLS, PLLS) areas of Palmer et al. (1978). We define as part of the lateral suprasylvian (LS) area the territory on the medial bank and caudal end of the lateral bank of the MSS that receives a topographically organized projection from the region of area 17 representing the lower visual quadrant. This territory is connected with other structures that are themselves striate-recipient (cortical areas 18 and 19, and the lateral division of the lateral posterior (LPl) nucleus), and with a variety of nuclei that receive direct retinal input, such as the C-laminae of the LGd, the medial interlaminar nucleus (MIN), and the superficial layers of the superior colliculus (SC). Its connections with the LPl, LGd, MIN, and SC correspond topographically with the input from area 17. Revised maps of area LS were produced from the physiological and connectional data: its rostral border is formed by a representation of lower visual elevations with the horizontal meridian represented caudally, and its lateral border is formed by the vertical meridian; area LS shares a representation of the center of gaze with the visual area of the lateral bank at its caudal end. The adjacent lateral bank area has larger receptive fields than area LS, and very different connectivity. It receives no input from area 17 and little input from striate-recipient structures, including area LS, but instead is connected to more remote extrastriate visual areas, such as the anterior ectosylvian visual (AEV) area in insular cortex, and to zones of the thalamus in receipt of tectal input (LPm and the lateromedial-suprageniculate nuclear complex). According to both mapping approaches, the lateral bank area contains representations of both the upper and lower visual quadrants but a rather limited degree of visuotopic order. We refer to it as the posterior ectosylvian visual (PEV) area, because it appears to be functionally and connectionally dissociated from area LS, but is possibly a functional antecedent of area AEV.

Cerebral hierarchies: predictive processing, precision and the pulvinar

This paper considers neuronal architectures from a computational perspective and asks what aspects of neuroanatomy and neurophysiology can be disclosed by the nature of neuronal computations? In particular, we extend current formulations of the brain as an organ of inference—based upon hierarchical predictive coding—and consider how these inferences are orchestrated. In other words, what would the brain require to dynamically coordinate and contextualize its message passing to optimize its computational goals? The answer that emerges rests on the delicate (modulatory) gain control of neuronal populations that select and coordinate (prediction error) signals that ascend cortical hierarchies. This is important because it speaks to a hierarchical anatomy of extrinsic (between region) connections that form two distinct classes, namely a class of driving (first-order) connections that are concerned with encoding the content of neuronal representations and a class of modulatory (second-order) connections that establish context—in the form of the salience or precision ascribed to content. We explore the implications of this distinction from a formal perspective (using simulations of feature–ground segregation) and consider the neurobiological substrates of the ensuing precision-engineered dynamics, with a special focus on the pulvinar and attention.