Mechanisms underlying the receptive field properties of neurons in cat visual cortex

The search of "canonical" explanations for the cerebral cortex

This paper addresses a fundamental line of research in neuroscience: the identification of a putative neural processing core of the cerebral cortex, often claimed to be "canonical". This "canonical" core would be shared by the entire cortex, and would explain why it is so powerful and diversified in tasks and functions, yet so uniform in architecture. The purpose of this paper is to analyze the search for canonical explanations over the past 40 years, discussing the theoretical frameworks informing this research. It will highlight a bias that, in my opinion, has limited the success of this research project, that of overlooking the dimension of cortical development. The earliest explanation of the cerebral cortex as canonical was attempted by David Marr, deriving putative cortical circuits from general mathematical laws, loosely following a deductive-nomological account. Although Marr's theory turned out to be incorrect, one of its merits was to have put the issue of cortical circuit development at the top of his agenda. This aspect has been largely neglected in much of the research on canonical models that has followed. Models proposed in the 1980s were conceived as mechanistic. They identified a small number of components that interacted as a basic circuit, with each component defined as a function. More recent models have been presented as idealized canonical computations, distinct from mechanistic explanations, due to the lack of identifiable cortical components. Currently, the entire enterprise of coming up with a single canonical explanation has been criticized as being misguided, and the premise of the uniformity of the cortex has been strongly challenged. This debate is analyzed here. The legacy of the canonical circuit concept is reflected in both positive and negative ways in recent large-scale brain projects, such as the Human Brain Project. One positive aspect is that these projects might achieve the aim of producing detailed simulations of cortical electrical activity, a negative one regards whether they will be able to find ways of simulating how circuits actually develop.

Blur Unblurred-A Mini Tutorial

Optical blur from defocus is quite frequently considered as equivalent to low-pass filtering. Yet that belief, although not entirely wrong, is inaccurate. Here, we wish to disentangle the concepts of dioptric blur, caused by myopia or mis-accommodation, from blur due to low-pass filtering when convolving with a Gaussian kernel. Perhaps surprisingly-if well known in optometry-the representation of a blur kernel (or point-spread function) for dioptric blur is, to a good approximation and disregarding diffraction, simply a cylinder. Its projection onto the retina is classically referred to as a blur circle, the diameter of which can easily be deduced from a light-ray model. We further give the derivation of the relationship between the blur-disk's diameter and the extent of blur in diopters, as well as the diameter's relation to the near or far point, and finally its relationship to visual acuity.

The neural basis of tactile motion perception

The manipulation of objects commonly involves motion between object and skin. In this review, we discuss the neural basis of tactile motion perception and its similarities with its visual counterpart. First, much like in vision, the perception of tactile motion relies on the processing of spatiotemporal patterns of activation across populations of sensory receptors. Second, many neurons in primary somatosensory cortex are highly sensitive to motion direction, and the response properties of these neurons draw strong analogies to those of direction-selective neurons in visual cortex. Third, tactile speed may be encoded in the strength of the response of cutaneous mechanoreceptive afferents and of a subpopulation of speed-sensitive neurons in cortex. However, both afferent and cortical responses are strongly dependent on texture as well, so it is unclear how texture and speed signals are disambiguated. Fourth, motion signals from multiple fingers must often be integrated during the exploration of objects, but the way these signals are combined is complex and remains to be elucidated. Finally, visual and tactile motion perception interact powerfully, an integration process that is likely mediated by visual association cortex.

Global motion percept mediated through integration of barber poles presented in bilateral visual hemifields

How is motion information that has been obtained through multiple viewing apertures integrated to form a global motion percept? We investigated the mechanisms of motion integration across apertures in two hemifields by presenting gratings through two rectangles (that form the dual barber poles) and recording the perceived direction of motion by human observers. To this end, we presented dual barber poles in conditions with various inter-component distances between the apertures and evaluated the degree to which the hemifield information was integrated by measuring the magnitude of the perceived barber pole illusion. Surprisingly, when the inter-component distance between the two apertures was short, the perceived direction of motion of the dual barber poles was similar to that of a single barber pole formed by the concatenation of the two component barber poles, indicating motion integration is achieved through a simple concatenation mechanism. We then presented dual barber poles in which the motion and contour properties of the two component barber poles differed to characterize the constraints underlying cross-hemifield integration. We found that integration is achieved only when phase, speed, wavelength, temporal frequency, and duty cycle are identical in the two barber poles, but can remain robust when the contrast of the two component barber poles differs substantially. We concluded that a motion stimulus presented in bilateral hemifields tends to be integrated to yield a global percept with a substantial tolerance for spatial distance and contrast difference.

End stopping in V1 is sensitive to contrast

Common situations that result in different perceptions of grouping and border ownership, such as shadows and occlusion, have distinct sign-of-contrast relationships at their edge-crossing junctions. Here we report a property of end stopping in V1 that distinguishes among different sign-of-contrast situations, thereby obviating the need for explicit junction detectors. We show that the inhibitory effect of the end zones in end-stopped cells is highly selective for the relative sign of contrast between the central activating stimulus and stimuli presented at the end zones. Conversely, the facilitatory effect of end zones in length-summing cells is not selective for the relative sign of contrast between the central activating stimulus and stimuli presented at the end zones. This finding indicates that end stopping belongs in the category of cortical computations that are selective for sign of contrast, such as direction selectivity and disparity selectivity, but length summation does not.

End-stopping in the visual cortex: excitation or inhibition?

In the visual cortex some neurons respond more strongly to short stimuli than to long ones. This is referred to as "end-stopping" and has been generally attributed to inhibition. The role of inhibition, however, has been difficult to demonstrate. Moreover, modeling has shown that end-stopping can be created solely from excitation. The roles of excitation and inhibition were investigated using intracellular recordings (Anderson et al., 2001, J. Neurosci. 21: 2104-2112). The results of that study were interpreted in favor of inhibition. The present report re-examines these results and finds that they may be in good, perhaps even better, agreement with an excitation model of end-stopping.

Emergence of orientation-selective inhibition in the primary visual cortex: a Bayes-Markov computational model

The recent consensus is that virtually all aspects of response selectivity exhibited by the primary visual cortex are either created or sharpened by cortical inhibitory interneurons. Experimental studies have shown that there are cortical inhibitory cells that are driven by geniculate cells and that, like their cortical excitatory counterparts, are orientation selective, though less sharply tuned. The main goal of this article is to demonstrate how orientation-selective inhibition might be created by the circuitry of the primary visual cortex (striate cortex, V1) from its nonoriented geniculate inputs. To fulfill this goal, first, a Bayes-Markov computational model is developed for the V1 area dedicated to foveal vision. The developed model consists of three parts: (i) a two-layered hierarchical Markov random field that is assumed to generate the activity patterns of the geniculate and cortical inhibitory cells, (ii) a Bayesian computational goal that is formulated based on the maximum a posteriori (MAP) estimation principle, and (iii) an iterative, deterministic, parallel algorithm that leads the cortical circuitry to achieve its assigned computational goal. The developed model is not fully LGN driven and it is not implementable by the neural machinery of V1. The model, then, is transformed into a fully LGN-driven and physiologically plausible form. Computer simulation is used to demonstrate the performance of the developed models.

Membrane potential and conductance changes underlying length tuning of cells in cat primary visual cortex

Spike responses for many cells of cat primary visual cortex are optimized for the length of a drifting grating stimulus. Stimuli that are longer or shorter than this optimal length elicit submaximal spike responses. To investigate the mechanisms responsible for this length tuning, we have recorded intracellularly from visual cortical neurons in the cat while presenting drifting grating stimuli of varying lengths. We have found that the membrane potential responses of the cells also exhibit length tuning, but that the suppression of spike responses at lengths longer than the preferred is 30-50% stronger than the corresponding suppression of the membrane potential responses. This difference may be attributed to the effects of spike threshold. Furthermore, using steady injected currents, we have measured changes in the excitatory and inhibitory components of input conductance evoked by stimuli of different lengths. We find that, compared with optimal stimuli, long stimuli evoke both an increase in inhibitory conductance and a decrease in excitatory conductance. These two mechanisms differ in their contrast sensitivity, resulting in stronger end stopping and shorter optimal lengths for high-contrast stimuli. These patterns suggest that response suppression for long stimuli is generated by a combination of active inhibition from stimuli outside the excitatory receptive field, as well as decreased excitation from other cortical cells that are themselves end-inhibited.

Vernier acuity during image rotation and translation: visual performance limits

Our capacity to detect spatial misalignments a fraction of the distance between retinal receptors in the presence of image motion challenges our understanding of spatial vision. We find that vernier acuity, while robust to image translation, rapidly degrades during image rotation. This indicates that orientation is a critical cue utilized by the visual system in vernier acuity tasks. Moreover, vernier acuity is robust to translational motion only at high target strengths. Vernier acuity for translating 3-dot targets over midrange velocities can be predicted from vernier acuity data derived from static targets of different presentation durations. However, the degradation observed at higher velocities is greater than predicted. The high velocity degradation reveals that performance is limited by a 1 msec asynchrony sensitivity. The moving vernier stimulus appears to constitute an optimal configuration for the visual system to achieve a 1 msec asynchrony sensitivity by making use of an orientation cue.

Directional asymmetries in the length-response profiles of cells in the feline dorsal lateral geniculate nucleus

1. The visual cortex provides a major synaptic input to the dorsal lateral geniculate nucleus (dLGN). Cortical layer VI cells giving rise to this projection are strongly influenced by stimulus orientation, length and direction of motion. In the dLGN, a significant component of the strong length tuning exhibited by most cells follows from the corticofugal influence. We have now checked whether there are directional biases in geniculate cell responses, and whether such biases are influenced by stimulus length. 2. The responses of A-laminae dLGN cells were assessed by single-unit extracellular recording. Length preference was examined by plotting multihistogram length-tuning curves to moving bars of light of various length. 3. Over half of the cells tested (100/183) exhibited directional bias and in many cases, this bias was highly dependent on bar length, resulting in radically different length response profiles for the two directions of motion. These asymmetries are similar to those documented for cortical hypercomplex cells, but do not equate to any known facet of the centre-surround organization of dLGN cell receptive fields. 4. We suspected the directional biases followed from the influence of the corticofugal projection. To test this, we recorded from preparations where areas 17 and 18 of the visual cortex had been removed. Surprisingly, a similar proportion of cells exhibited directional biases after removal of the corticofugal input, suggesting that the biases are generated subcortically. 5. The widespread presence of systematic biases in the response profiles of dLGN cells further underlines the possibility that geniculate mechanisms may make a far greater contribution to visual processing than hitherto suspected.

The length-response properties of cells in the feline perigeniculate nucleus

Perigeniculate cells receive visual input from the dorsal lateral geniculate nucleus (dLGN) and from the visual cortex. In contrast to the extensive literature documenting dLGN and cortical cell responses, comparatively little quantitative data exists for perigeniculate nucleus cells, and very little is known about the role of the corticofugal input to the perigeniculate nucleus. We have previously shown that dLGN relay cells have sharply length-tuned receptive fields and that a significant component of this is dependent on the corticofugal system. In this report, we have explored the length-response properties of perigeniculate nucleus cells in the presence and absence of corticofugal feedback. The response profiles of most perigeniculate nucleus cells contrasted markedly with the sharply length-tuned fields of dLGN cells, but exhibited a notable resemblance to those exhibited by VI cells with short summation lengths, which have recently been shown to constitute a considerable proportion of the layer VI cell population. This might suggest that the responses of perigeniculate nucleus cells to long bars derive from their cortical input. However, our data failed to reveal a discernible change in their profiles after removal of the corticofugal drive. This surprising observation implies that their length-tuning profiles follow from subcortical circuitry. The ways in which this might occur are discussed.

Different motion sensitive units are involved in recovering the direction of moving lines

We studied direction discrimination for lines moving obliquely relative to their orientation. Manipulating contrast, length and duration of motion, we found systematic errors in direction discrimination at low contrast, long length and/or short durations. These errors can be accounted for by a competition between ambiguous velocity signals originating from contour motion processing units and signals from line terminator processing units. The dynamic of this competition can be described by a simple model involving two different classes of processing units with different contrast thresholds, different integration time constants and different levels of response saturation.

The length-response properties of cells in the feline dorsal lateral geniculate nucleus

1. In this report we have systematically examined the length-response properties of a large population of cells recorded in the cat dorsal lateral geniculate nucleus (dLGN). The responses of A laminae dLGN cells were assessed by the use of conventional single-unit extracellular recording techniques. The length preference of these cells was examined by plotting multihistogram length tuning curves to moving bars of light. Bar length was randomized in an interleaved fashion under computer control. The other stimulus parameters were standardized within the limits of those routinely used to assess the length preference of cortical cells. 2. The majority of cells (186/198), whose length-response properties are considered in detail in this report, exhibited strong centre-surround antagonism and a mean degree of length tuning equivalent to, or exceeding, that seen in most cortical hypercomplex cells (71 +/- 1.18%, S.E.M., n = 186). 3. The values for X cells (74 +/- 1.41%, S.E.M., n = 100) and Y cells (67 +/- 2.13%, S.E.M., n = 74) were very similar, as were those of the on-centre (71 +/- 1.51%, S.E.M., n = 123) and off-centre (71 +/- 1.85%, S.E.M., n = 63) subgroups. 4. A distinct subgroup of the Y cell population was identified. These comprised the remaining twelve out of the 198 cells examined and their response properties were sufficiently distinct to merit classification as a discrete subpopulation of cells which we have termed nlY cells. They were characterized by very poor levels of both centre-surround antagonism and length tuning, and were most frequently encountered close to laminar borders. Their response properties have been described in detail elsewhere. 5. We quantitatively compared the degree of length tuning seen with moving bars to the strength of centre-surround antagonism assessed with flashing spots. The degree of length tuning did not necessarily follow the level of centre-surround antagonism. 6. Examination of the effects of unilaterally extending bar length to one or other side of the receptive field did not reveal the type of asymmetry frequently seen in cortical hypercomplex cells. 7. The high degree of length tuning seen in this study underlines the potential importance of geniculate response properties to the generation of the length-response properties of cortical hypercomplex cells. The findings are discussed in relation to the synaptic mechanisms contributing to the generation of length tuning at subcortical and cortical levels.

Inhibition contributes to orientation selectivity in visual cortex of cat

Neurons in the visual cortex are selectively responsive to light or dark bars presented at particular orientations. On the basis of physiological data, this orientation selectivity is hypothesized as being due at least partially to intracortical inhibitory mechanisms. But this hypothesis has been challenged by intracellular recordings indicating that excitatory inputs themselves are orientation-selective, so inhibition may not contribute to the observed selectivity. Also, there is controversy about the presence of intracortical horizontal connections mediating inhibition for selectivity and about the theoretical validity of such inhibitory connections. Using cross-correlation analysis of the activities of two neurons recorded simultaneously, we find that inhibitory interactions exist between cells with somewhat different, but not orthogonal, orientation preferences. This suggests that intracortical horizontal inhibition operates between 'orientation columns' to sharpen the orientation tuning of cortical neurons.

Spatial properties of neurons in the monkey striate cortex

Contrast sensitivity as a function of spatial frequency was determined for 138 neurons in the foveal region of primate striate cortex. The accuracy of three models in describing these functions was assessed by the method of least squares. Models based on difference-of-Gaussians (DOG) functions where shown to be superior to those based on the Gabor function or the second differential of a Gaussian. In the most general case of the DOG models, each subregion of a simple cell's receptive field was constructed from a single DOG function. All the models are compatible with the classical observation that the receptive fields of simple cells are made up of spatially discrete 'on' and 'off' regions. Although the DOG-based models have more free parameters, they can account better for the variety of shapes of spatial contrast sensitivity functions observed in cortical cells and, unlike other models, they provide a detailed description of the organization of subregions of the receptive field that is consistent with the physiological constraints imposed by earlier stages in the visual pathway. Despite the fact that the DOG-based models have spatially discrete components, the resulting amplitude spectra in the frequency domain describe complex cells just as well as simple cells. The superiority of the DOG-based models as a primary spatial filter is discussed in relation to popular models of visual processing that use the Gabor function or the second differential of a Gaussian.

Two-dimensional modeling of visual receptive fields using Gaussian subunits

Retinal ganglion cell receptive fields have been successfully described using the difference of Gaussians model introduced by Rodieck. As the basic elements of retinal receptive fields are well described by the Gaussian function, it is natural to model receptive fields beyond this level as a convergence of Gaussian subunits. In this paper the full two-dimensional solution to the problem of calculating the response to drifting gratings of a model receptive field composed of Gaussian subunits is presented. The subunits are not required to be radially symmetric, any number is allowed, with any temporal phase delays; and responses are predicted to gratings of any spatial frequency at any orientation. This solution will greatly extend the range of receptive fields that can be modeled as a convergence of Gaussian subunits, including those with orientational and directional selectivities.

Effect of localized grating adaptation as a function of separation along the length axis between test and adaptation areas

Aftereffect following adaptation to localized gratings was measured as a function of the separation along the length axis between test and adaptation gratings. When the adaptation gratings were located on or near the retinal area occupied by the test grating, contrast sensitivity greatly decreased. When the adaptation gratings were spatially separated from the test grating, contrast sensitivity significantly increased. This property is similar to that which was observed in our previous study, in which the adaptation gratings were displaced from a test grating along the modulation axis. The facilitatory aftereffect of the grating adaptation can be accounted for by assuming that there may exist two mechanisms involved in the adaptation process; one is a center mechanism responsible for the detection of a test pattern, its adaptation producing a reduction in responsiveness; the other is a surround mechanism which tonically inhibits the center mechanism, its adaptation resulting in an increase in the sensitivity of the center mechanism by releasing the tonic inhibition. The spatial property of the adaptation effect may reflect the nature of spatial integration process of the center and surround mechanism.

Spatial summation in subregions of simple-cell receptive fields in cat striate cortex as a function of slit length

Spatial summation along the optimum stimulus orientation in subregions of simple-cell receptive fields in cat striate cortex was studied quantitatively by measuring the response to stationary light slits of variable length. Before summation analysis, the cell's discharge field was mapped by flashing a test slit on and off in a sequence of positions through the receptive field. A static activation procedure was used to determine the extension of subregions of the receptive field where light stimulation increased (enhancement) or decreased (suppression) the firing rate. An activation slit in the optimum orientation was positioned in the most responsive position of the discharge field and the effects of a parallel test slit, in a series of broadside positions, were assessed from the changes induced in the discharge elicited by the activation slit. Length-response curves for on and off responses were made by positioning a test slit in the respective subregions of the discharge field. The activation procedure was used to make length-response curves for suppression. A test slit of variable length was positioned in a suppression region defined by the activation profiles and an activation slit of fixed length was centred in the most responsive discharge field position. Length summation was found for all cells, both with respect to on and off responses, and suppression. The curves for on and off responses had a maximum value beyond which the response declined or levelled off, but some cells had a secondary, more shallow increase beyond an initial, steeply rising part. Similar properties were found for summation of suppression except that the effects were opposite in sign. Curves made for both on and off regions in the same cell often differed in shape. Such differences were also found when length-response curves made in different suppression regions of the same cell were compared. The various length-summation functions were explained by a model presuming that simple-cell receptive fields consist of partially overlapping non-concentric excitatory and inhibitory fields. This arrangement would also explain why length-response curves in various subregions often had different shapes.

Local shading analysis

Local analysis of image shading, in the absence of prior knowledge about the viewed scene, may be used to provide information about the scene. The following has been proved. Every image point has the same image intensity and first and second derivatives as the image of some point on a Lambertian surface with principal curvatures of equal magnitude. Further, if the principal curvatures are assumed to be equal there is a unique combination of image formation parameters (up to a mirror reversal) that will produce a particular set of image intensity and first and second derivatives. A solution for the unique combination of surface orientation, etc., is presented. This solution has been extended to natural imagery by using general position and regional constraints to obtain estimates of the following: ¿ surface orientation at each image point; ¿ the qualitative type of the surface, i.e., whether the surface is planar, cylindrical, convex, concave, or saddle; ¿ the illuminant direction within a region. Algorithms to recover illuminant direction and estimate surface orientation have been evaluated on both natural and synthesized images, and have been found to produce useful information about the scene.

Spatial summation in the receptive field of simple cells in the cat striate cortex

Spatial summation was studied quantitatively through width response curves made with an optimally oriented test slit of variable width, and by comparing the response to combined presentation of several parallel slits with the response to each slit alone. Prior to summation analysis, the cell's discharge field (DF) was mapped by presenting a test slit ON and OFF across the receptive field. Activation profiles, showing the extension of subregions where light stimulation increased (enhancement) or decreased the firing rate (suppression), were made by presenting an optimally oriented activation slit in the most responsive DF-position. Against this activity the effects of a parallel test slit were determined in a series of broadside positions. Width response curves were made over the subregions of the DF and the activation profiles. Spatial summation was found in all cells, but the width of the summation region was smaller than the width of the subregions in the respective profiles. The width of the summation region was related to the degree of activation rather than to specific locations within the receptive field. The effect produced by several slits presented together deviated from the algebraic sum of the effects produced by each slit alone. Linear summation was rarely found. Accumulated response curves obtained by integration of DF or activation profiles were compared with width response curves to test linearity of summation. Linear summation throughout the whole receptive field was never found. A satisfactory fit was found only over a narrow region showing that summation was linear within a small part of the summation region. Linearity ended near response maxima or minima in the response profiles. The results indicate that the receptive field of simple cells consists of overlapping excitatory and inhibitory fields, and that the exact location and width of enhancement and suppression zones are determined by an activity-dependent balance between excitatory and inhibitory inputs.

Visual response properties of zebrafish tectal cells

The visual properties of zebrafish tectal cells have been studied with a variety of stimulation routines. These include illumination of the whole receptive field and of the surround, use of moving edges, very small spots, bars of varying orientations, moving spots with varying direction and speeds, growing discs, and pairs of spots whose presentation varies in position and sequence. A number of properties correlate with the classification scheme set forth in the preceding paper. Type B cells, unlike other types, are insensitive to moving stimuli. Experiments involving surround stimulation show that type S cells have inhibitory surrounds while those of type I do not. Type I cells, however, exhibit several properties which are consistent with an intratectal delayed inhibitory mechanism operating within the receptive field. These properties include the response to moving edges and growing stimuli, and the dependence of response duration on the size of a flashed stimulus. Various explanations of these properties are considered, and a specific model is proposed which states that cells of type I receive inhibitory input from neighbouring tectal cells of the same physiological type. The properties involved may be of direct importance in the visual behavior of the fish.

A model of the early stages of the human visual system: functional and topological transformations performed in the peripheral visual field

A model of the early stages of the visual system is presented, with particular reference to the region of the visual field outside the fovea and to the class of retinal and lateral geniculate nucleus cells which are most active in the processing of patter information (X-cells). The main neuroanatomical and neurophysiological properties taken into account are: the linear increase of the receptive fields diameter with eccentricity, the constancy of the overlap factor and the topological transformation operated upon the retinal image by the retino-cortical connection. The type of filtering taking place between the retina and the visual cortex is analyzed and some simulations are presented. It is shown that such a filtering is of a bandpass space variant type, with center frequencies that decrease from the center (i.e. the fovea) toward the periphery of the visual field. This processing is "form invariant" under linear scaling of the input. Moreover, considering the properties of the retino-cortical connection, it is shown that the "cortical images" undergo simple shifts whenever the retinal images are scaled or rotated.

Inhibitory circuits accounting for development of visual cortical mappings, stimulus preferences, and psychophysical performance

A developmental rationale is proposed for the circuitry underlying the generation of fine retinotopic mappings, the quantitative range of simple-cell stimulus preferences, and the psychophysical performance of the visual system. It is assumed that the retina consists of a mosaic of partially overlapping elements, or hyperfields, which are laid down in a sunflower-seed pattern. These hyperfields project to a corresponding rectilinear mosaic of hypercolumns in the cortex, according to a pattern of chemoaffinities. Each hyperfield, in turn, consists of a sunflower-seed mosaic of nonoverlapping ganglion-cell receptive-field centres, which project to a matching rectilinear mosaic of minicolumns in the corresponding hypercolumn. Retinotopic order is produced in the hyperfield-hypercolumn mapping by radially symmetric inhibitory links, between cortical cells more than two minicolumns apart, which operate on Hebb-modifiable retinocortical excitatory afferent fibres. Under this mapping, hyperfield radii map onto parallel rows of minicolumns (orientation columns), and concentric semicircles of ganglion-cell receptive fields map onto spatial-frequency columns, crossing orientation columns at right angles. The 'scatter' in this mapping is equivalent to one local average receptive-field diameter. Orientation-related stimulus preferences ar produced by asymmetrical inhibitory links between cells more than two minicolumns apart, in the same spatial-frequency columns. A third network of inhibitory circuits, with Hebb-modifiable synapses, is assumed to operate between cells in the same or immediately adjacent minicolumns. This network enhances stimulus selectivity and sensitivity in simple and hypercomplex cells, and is responsible for adaptation aftereffects and sensory information storage.

Neuronal circuits capable of generating visual cortex simple-cell stimulus preferences

The range of simple-cell stimulus preferences to be found at each point in the striate cortex can be accounted for in terms of a model of retinocortical and intracortical circuits. It is assumed that each hypercolumn represents a conformal logarithmic mapping of its aggregate field, or hyperfield. Retinal-unit-field centres within the same aggregate field are assumed to be nonoverlapping, and overlap between retinal centres is attributed to a marked overlap between, adjacent aggregate fields. Unit-field centres are assumed to be arranged regularly along aggregate-field radii, with diameters which increase linearly with ecentricity. Retinal-unit-field centres project retinotopically to overlapping clusters consisting of nine cortical pillars, and each pillar receives from a corresponding cluster of nine retinal-unit=field centres. Under this mapping, aggregate-field radii project to orientation columns, and concentric semicircules project to spatial-frequency columns. Inhibitory basket-cell axons project at right angles to the orientation columns, in parallel with the spatial-frequency columns, to produce a range of preferences and tuning curves for orientation, symmetry, directionality, and length. The same circuits also offer explanations for adaption phenomena such as tilt aftereffects, and suggest ways in which attentional mechanisms might operate.