{"title": "Eye movements and the maturation of cortical orientation selectivity", "book": "Advances in Neural Information Processing Systems", "page_first": 261, "page_last": 267, "abstract": null, "full_text": "Eye movements and the maturation of cortical\n\norientation selectivity\n\nMichele Rucci  and Antonino Casile\u0001\n\n Department of Cognitive and Neural Systems, Boston University, Boston, MA 02215.\n\n\u0001 Scuola Superiore S. Anna, Pisa, Italy\n\nAbstract\n\nNeural activity appears to be a crucial component for shaping the recep-\ntive \ufb01elds of cortical simple cells into adjacent, oriented subregions alter-\nnately receiving ON- and OFF-center excitatory geniculate inputs. It is\nknown that the orientation selective responses of V1 neurons are re\ufb01ned\nby visual experience. After eye opening, the spatiotemporal structure of\nneural activity in the early stages of the visual pathway depends both on\nthe visual environment and on how the environment is scanned. We have\nused computational modeling to investigate how eye movements might\naffect the re\ufb01nement of the orientation tuning of simple cells in the pres-\nence of a Hebbian scheme of synaptic plasticity. Levels of correlation be-\ntween the activity of simulated cells were examined while natural scenes\nwere scanned so as to model sequences of saccades and \ufb01xational eye\nmovements, such as microsaccades, tremor and ocular drift. The speci\ufb01c\npatterns of activity required for a quantitatively accurate development\nof simple cell receptive \ufb01elds with segregated ON and OFF subregions\nwere observed during \ufb01xational eye movements, but not in the presence\nof saccades or with static presentation of natural visual input. These re-\nsults suggest an important role for the eye movements occurring during\nvisual \ufb01xation in the re\ufb01nement of orientation selectivity.\n\n1 Introduction\n\nCortical orientation selectivity, i.e. the preference to edges with speci\ufb01c orientations ex-\nhibited by most cells in the primary visual cortex of different mammal species, is one\nof the most investigated characteristics of neural responses. Although the essential ele-\nments of cortical orientation selectivity seem to develop before the exposure to patterned\nvisual input, visual experience appears essential both for re\ufb01ning orientation selectivity,\nand maintaining the normal response properties of cortical neurons. The precise mecha-\nnisms by which visually-induced activity contribute to the maturation of neural responses\nare not known.\n\nA number of experimental \ufb01ndings support the hypothesis that the development of orienta-\ntion selective responses relies on Hebbian/covariance mechanisms of plasticity. According\nto this hypothesis, the stabilization of synchronously \ufb01ring afferents onto common postsy-\nnaptic neurons may account for the segregation of neural inputs observed in the receptive\n\ufb01elds of simple cells, where the adjacent oriented excitatory and inhibitory subregions re-\n\n\fceive selective input from geniculate ON- and OFF-center cells in the same retinotopic\npositions. Modeling studies [10, 9] have shown the feasibility of this proposal assuming\nsuitable patterns of spontaneous activity in the LGN before eye opening.\n\nAfter eye opening, the spatiotemporal structure of LGN activity depends not only on the\ncharacteristics of the visual input, but also on the movements performed by the animal\nwhile exploring its environment. It may be expected that changes in the visual input in-\nduced by these movements play an important role in shaping the responses of neurons in\nthe visual system. In this paper we focus on how visual experience and eye movements\nmight jointly in\ufb02uence the re\ufb01nement of orientation selectivity under the assumption of a\nHebbian mechanism of synaptic plasticity. As illustrated in Fig. 1, a necessary requirement\nof the Hebbian hypothesis is a consistency between the correlated activity of thalamic affer-\nents and the organization of simple cell receptive \ufb01elds. Synchronous activation is required\namong geniculate cells of the same type (ON- or OFF-center) with receptive \ufb01elds located\nat distances smaller than the width of a simple cell subregion, and among cells of opposite\npolarity with receptive \ufb01elds at distances comparable to the separation between adjacent\nsubregions. We have analyzed the second order statistical structure of neural activity in a\nmodel of cat LGN when natural visual input was scanned so as to replicate the oculomotor\nbehavior of the cat. Patterns of correlated activity were compared to the structure of simple\ncell receptive \ufb01elds at different visual eccentricities.\n\n2 The model\n\nModeling the activity of LGN cells\n\nLGN cells were modeled as linear elements with quasi-separable spatial and temporal com-\nponents as proposed by [3]. This model, derived using the reverse-correlation technique,\nhas been shown to produce accurate estimates of the activity of different types of LGN\ncells. Changes in the instantaneous \ufb01ring rates with respect to the level of spontaneous\nactivity were generated by evaluating the spatiotemporal convolution of the input image \nwith the receptive \ufb01eld kernel \u0001\n(1)\nis the symbol for convolution, \u0007\u001b\u0011\u001c\u0012\u001d\u0014\u001e\u000b and \t are the spatial and temporal variables,\nwhere \u0016\nindicates recti\ufb01cation (\u000f\nif \u0011&'(%!\u0012\u001d) otherwise). For each\nand the operator \u000f \u001f\nconsisted of two additive components, representing the center (* ) and\ncell, the kernel \u0001\nthe periphery (+ ) of the receptive \ufb01eld respectively. Each of these two contributions was\nseparable in its spatial (,\n\n\u0002\u0004\u0003\u0006\u0005\b\u0007\n\t\f\u000b\u000e\r\u0010\u000f\n\n\u0007\n\u0011\u0013\u0012\f\u0014\u0015\u0012\f\t\f\u000b\u0019\u0018\u0004\u001a\n\n\u0007\n\u0011\u0013\u0012\f\u0014\u0015\u0012\f\t\f\u000b\u0017\u0016\n\n\u0011!\u0018\n\n\"\u0011$#&%\n\nThe spatial receptive \ufb01elds of both center and surround were modeled as two-dimensional\nGaussians, with a common space constant for both dimensions. Spatial parameters varied\nwith eccentricity following neurophysiological measurements. As in [3], the temporal pro-\n\ufb01le of the response was given by the difference of two gamma functions, with the temporal\nfunction for the periphery equal to that for the center and delayed by 3 ms.\n\n) and temporal (-\n\u0007\n\u0011\u001c\u0012\u001d\u0014\u0015\u0012\f\t\f\u000b.\r\n\u0007\n\u0011\u0013\u0012\f\u0014!\u000b\n\n,0/\n\n) elements:\n,02\n-1/\n\n\u0007\u001b\t\f\u000b0#\n\n\u0007\u001b\u0011\u001c\u0012\f\u0014!\u000b\n\n\u0007\n\t\f\u000b\n\n-12\n\nModeling eye movements\n\nModeled eye movements included saccades (both large-scale saccades and microsaccades),\nocular drift, and tremor.\n\nSaccades\u2014 Voluntary saccadic eye movements, the fast shifts of gaze among \ufb01xation\npoints, were modeled by assuming a generalized exponential distribution of \ufb01xation times.\nThe amplitude and direction of a saccade were randomly selected among all possible sac-\ncades that would keep the point of \ufb01xation on the image. Following data described in the\n\n\u0001\n\n\u0018\n\u001a\n\u001a\n\u0001\n\fliterature, the duration of each saccade was proportional to its amplitude. A modulation\nof geniculate activity was present in correspondence of each saccade [7]. Neural activity\naround the time of a saccade was multiplied by a gain function so that an initial suppression\nof activity with a peak of 10%, gradually reversed to a 20% facilitation with peak occurring\n100 ms after the end of the saccade.\n\nFixational eye movements\u2014 Small eye movements included \ufb01xational saccades, ocular\ndrift and tremor. Microsaccades were modeled in a similar way to voluntary saccades, with\namplitude randomly selected from a uniform distribution between 1 and 10 minutes of arc.\nNo modulation of LGN activity was present in the case of microsaccades.\n\nOcular drift and tremor were modeled together by approximating their power spectrum\nby means of a Poisson process \ufb01ltered by a second order eye plant transfer function over\n\u0001 . This term represents the\nirregular discharge rate of motor units for frequency less than 40 Hz. Parameters were\n\nthe frequency range 0-40 Hz where the power declines as \u0002\u0001\u0004\u0003\nadjusted so as to give a mean amplitude of \n3 Results\n\nwhich are the values measured in the cat [11].\n\n\u001f\u0006\u0005\n\n\b\u0007 and a mean velocity equal to \n\t\n\n\u0007 /s,\n\nWe simulated the activity of geniculate cells with receptive \ufb01elds in different positions\nof the visual \ufb01eld, while receiving visual input in the presence of different types of eye\nmovements. The relative level of correlation between units of the same and different types\n\n'\u0017\u0016\n\n, where the two terms are the correlation coef\ufb01cients evaluated between the\n\nis positive when the activity of units of the same type\ncovary more strongly than that of units of different types, and is negative when the opposite\noccurs. The average relative levels of correlation between units with receptive \ufb01elds at\n\nin the LGN was measured by means of the correlation difference, \u000e\nrespectively. \u000e D\u000f\u0011\u0010\n\nD\u000f\u0006\u0010\nat positions \f and\r\n\u000e ONOFF\n\u000e ONON\n\u000f\u0006\u0010\n\u000f\u0011\u0010\ntwo ON units at positions \f and \r , and between the ON unit at position \f and the OFF\nunit at position \n\n\u0015\u0014\ndifferent distances in the visual \ufb01eld were examined by means of the function \u000e\n\u001b\u001d\u001c , which evaluates the average correlation difference \u000e D\u000f\u0011\u0010 among all pairs of\n\u000e D\u000f\u0006\u0010\n\u000f\u0019\u0018\u001a\u0010\ncells at positions \f and\r at distance \u0012\nfrom each other. For simplicity, in the following we\nrefer to \u000e D\u0007\u001e\u0012\n\u000b as the correlation difference, implicitly assuming that a spatial averaging has\ntaken place. The correlation difference is a useful tool for predicting the emerging patterns\nof connectivity in the presence of a Hebbian mechanism of synaptic plasticity. The average\nseparation at which \u000e D\u0007\u0013\u0012\b\u000b changes sign is a key element in determining the spatial extent\nFig. 1 (\u001f ) provides an example of application of the correlation difference function to quan-\n\ntify the correlated activity of LGN cells. In this example we have measured the level of\ncorrelation between pairs of cells with receptive \ufb01elds at different separations when a spot\nof light was presented as input. An important element in the resulting level of correlation is\nthe polarity of the two cells (i.e. whether they are ON- or OFF-center). As shown in Fig. 1\n\nof the different sub\ufb01elds within the receptive \ufb01elds of simple cells.\n\nD\u0007\u0013\u0012\b\u000b\n\nreceptive \ufb01elds overlap, the correlation between pairs of cells of the same type decreases\nwhen the separation between their receptive \ufb01elds is increased, while pairs of cells of op-\nposite types tend to become more correlated. As a consequence, the correlation difference\n\n( ), since geniculate cells tend to be coactive when the ON and OFF subregions of their\nfunction, \u000e D\u0007\u0013\u0012\b\u000b , is positive at small separations, and negative at large ones.\n\nFig. 2 shows the measured correlated activity for LGN cells located around 17 deg. of\nvisual eccentricity in the presence of two types of visual input: retinal spontaneous ac-\ntivity and natural visual stimulation. Spontaneous activity was simulated on the basis of\nMatronarde\u2019s data on the correlated \ufb01ring of ganglion cells in the cat retina [8]. As il-\nD and\nthe response pro\ufb01le of an average cortical simple cell at this eccentricity, indicating that a\n\nlustrated by the graph, a close correspondence is present between the measured \u000e\n\n\u001f\n\u000b\n\n#\n\u0016\n\f\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\u0003\u0001\u0003\u0001\u0003\u0001\u0003\n\u0004\u0001\u0004\u0001\u0004\u0001\u0004\n\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\u0001\u0001\u0001\u0001\n\u0002\u0001\u0002\u0001\u0002\u0001\u0002\u0001\u0002\n\nV1 RF\n\nLGN\nON\n\n\u0007\u0001\u0007\n\u0007\u0001\u0007\n\u0007\u0001\u0007\n\n(a)\n\n\u0012\u0010\u0012\u0010\u0012\u0010\u0012\u0010\u0012\u0010\u0012\n\u0013\u0010\u0013\u0010\u0013\u0010\u0013\u0010\u0013\u0010\u0013\n\u0012\u0010\u0012\u0010\u0012\u0010\u0012\u0010\u0012\u0010\u0012\n\u0013\u0010\u0013\u0010\u0013\u0010\u0013\u0010\u0013\u0010\u0013\n\u0012\u0010\u0012\u0010\u0012\u0010\u0012\u0010\u0012\u0010\u0012\n\u0018\u0010\u0018\u0010\u0018\n\u0013\u0010\u0013\u0010\u0013\u0010\u0013\u0010\u0013\u0010\u0013\n\u0019\u0010\u0019\n\u0012\u0010\u0012\u0010\u0012\u0010\u0012\u0010\u0012\u0010\u0012\n\u0018\u0010\u0018\u0010\u0018\n\u0013\u0010\u0013\u0010\u0013\u0010\u0013\u0010\u0013\u0010\u0013\n\u0019\u0010\u0019\n\u0012\u0010\u0012\u0010\u0012\u0010\u0012\u0010\u0012\u0010\u0012\n\u0013\u0010\u0013\u0010\u0013\u0010\u0013\u0010\u0013\u0010\u0013\n\u0012\u0010\u0012\u0010\u0012\u0010\u0012\u0010\u0012\u0010\u0012\n\u0013\u0010\u0013\u0010\u0013\u0010\u0013\u0010\u0013\u0010\u0013\n\n\u000f\u0010\u000f\u0010\u000f\u0010\u000f\u0010\u000f\u0010\u000f\n\u0011\u0010\u0011\u0010\u0011\u0010\u0011\u0010\u0011\u0010\u0011\n\u000f\u0010\u000f\u0010\u000f\u0010\u000f\u0010\u000f\u0010\u000f\n\u0011\u0010\u0011\u0010\u0011\u0010\u0011\u0010\u0011\u0010\u0011\n\u000f\u0010\u000f\u0010\u000f\u0010\u000f\u0010\u000f\u0010\u000f\n\u0016\u0010\u0016\u0010\u0016\n\u0011\u0010\u0011\u0010\u0011\u0010\u0011\u0010\u0011\u0010\u0011\n\u0017\u0010\u0017\n\u000f\u0010\u000f\u0010\u000f\u0010\u000f\u0010\u000f\u0010\u000f\n\u0016\u0010\u0016\u0010\u0016\n\u0011\u0010\u0011\u0010\u0011\u0010\u0011\u0010\u0011\u0010\u0011\n\u0017\u0010\u0017\n\u000f\u0010\u000f\u0010\u000f\u0010\u000f\u0010\u000f\u0010\u000f\n\u0011\u0010\u0011\u0010\u0011\u0010\u0011\u0010\u0011\u0010\u0011\n\u000f\u0010\u000f\u0010\u000f\u0010\u000f\u0010\u000f\u0010\u000f\n\u0011\u0010\u0011\u0010\u0011\u0010\u0011\u0010\u0011\u0010\u0011\n\n\u0014\u0010\u0014\u0010\u0014\u0010\u0014\u0010\u0014\u0010\u0014\n\u0015\u0010\u0015\u0010\u0015\u0010\u0015\u0010\u0015\u0010\u0015\n\u0014\u0010\u0014\u0010\u0014\u0010\u0014\u0010\u0014\u0010\u0014\n\u0015\u0010\u0015\u0010\u0015\u0010\u0015\u0010\u0015\u0010\u0015\n\u0014\u0010\u0014\u0010\u0014\u0010\u0014\u0010\u0014\u0010\u0014\n\u0015\u0010\u0015\u0010\u0015\u0010\u0015\u0010\u0015\u0010\u0015\n\u0014\u0010\u0014\u0010\u0014\u0010\u0014\u0010\u0014\u0010\u0014\n\u0015\u0010\u0015\u0010\u0015\u0010\u0015\u0010\u0015\u0010\u0015\n\u0014\u0010\u0014\u0010\u0014\u0010\u0014\u0010\u0014\u0010\u0014\n\u0015\u0010\u0015\u0010\u0015\u0010\u0015\u0010\u0015\u0010\u0015\n\u0014\u0010\u0014\u0010\u0014\u0010\u0014\u0010\u0014\u0010\u0014\n\u0015\u0010\u0015\u0010\u0015\u0010\u0015\u0010\u0015\u0010\u0015\n\nON\n\n\u001a\u0010\u001a\u0010\u001a\n\u001b\u0010\u001b\u0010\u001b\n\u001a\u0010\u001a\u0010\u001a\n\u001b\u0010\u001b\u0010\u001b\n\nOFF\n\nON\u2212ON, OFF\u2212OFF\nON\u2212OFF, OFF\u2212ON\ndifference\n\nLGN\nOFF\n\n...\n\n...\n\nV1\n\n2\n\n1\n\n0\n\nn\no\n\ni\nt\n\nl\n\na\ne\nr\nr\no\nc\n \nd\ne\nz\n\ni\nl\n\na\nm\nr\no\nn\n\n\u22121\n\n0\n\n20\n\n40\n60\ndistance (min.)\n\n80\n\n100\n\n(b)\n\nFigure 1: ( ) Patterns of correlated activity required by a Hebbian mechanism of synaptic\nplasticity to produce a segregation of geniculate afferents. On average ON- and OFF-center\nLGN cells overlapping excitatory and inhibitory subregions in the receptive \ufb01eld of a sim-\nple cell must be simultaneously active. (\u001f ) Example of application of the correlation dif-\nference function, \u000e D\u0007\u001e\u0012\n\u000b . The icons on the top of each graph represent the positions of the\nreceptive \ufb01elds of the two cells at the corresponding separations along the \u0011 axis. The bright\ndot marks the center of the spot of light. The three curves represent the correlation coef-\n\u0007\u001e\u0012\n\u000b (continuous thin line), units of opposite\n\ufb01cients for pairs of units of the same type \u000e\n\u0007\u001e\u0012\ntypes \u000e\u0017\u0007\n(bold line). Positive (negative) values of \u000e D\u0007\u0013\u0012\b\u000b\nindicate that the activity of LGN cells of\nthe same (opposite) type covary more closely than the activity of cells of opposite (same)\ntypes.\n\n\u000b (dashed line), and the correlation difference function \u000e\n\nD\u0007\u0013\u0012\b\u000b\u000e\n\n\u0007\u0013\u0012\b\u000b\u001e#\n\n\u0007\u001e\u0012\n\nHebbian mechanism of synaptic plasticity can well account for the structure of simple cell\nreceptive \ufb01elds before eye opening.\n\n\u0007\u001e\u001d!\u000b\n\nWhat happens in the presence of natural visual input? We evaluated the correlation dif-\nference function on a database of 30 images of natural scenes. The mean power spectrum\n%'& , which is consistent with the\nof our database was best approximated by \u001c\nresults of several studies investigating the power spectrum of natural images. The mean\ncorrelation difference function measured when the input images were analyzed statically\nis marked by dark triangles in the left panel of Fig. 2. Due to the wide spatial correlations\nof natural visual input, the estimated correlation difference did not change sign within the\nreceptive \ufb01eld of a typical simple cell. That is, LGN cells of the same type were found\nto covary more closely than cells of opposite types at all separations within the receptive\n\ufb01eld of a simple cell. This result is not consistent with the putative role of a direct Heb-\nbian/covariance model in the re\ufb01nement of orientation selectivity after eye opening.\n\n \u001f\"!\n\n\u0001$#\n\nA second series of simulations was dedicated to analyze the effects of eye movements on\nthe structure of correlated activity. In these simulations the images of natural scenes were\nscanned so as to replicate cat oculomotor behavior. As shown in right panel of Fig. 2,\nsigni\ufb01cantly different patterns of correlated neural activity were found in the LGN in the\npresence of different types of eye movements. In the presence of large saccades, levels\nof correlations among the activity of geniculate cells were similar to the case of static\npresentation of natural visual input, and they did not match the structure of simple cell\nreceptive \ufb01elds. The dark triangles in Fig. 2 represent the correlation difference function\nevaluated over a window of observation of 100 ms in the presence of both large saccades\nand \ufb01xation eye movements. In contrast, when our analysis was restricted to the periods of\nvisual \ufb01xation during which microscopic eye movements occurred, strong covariances were\n\n\u0005\n\u0005\n\u0005\n\u0006\n\u0006\n\u0006\n\b\n\b\n\b\n\t\n\t\n\t\n\n\u000b\n\u000b\n\u000b\n\f\n\f\n\f\n\n\u000e\n\u000e\n\u000e\n2\n\u000e\n2\n\u000e\n\u0007\n\u000b\n\u0018\n\fmeasured between cells of the same type located nearby and between cells of opposite types\nat distances compatible with the separation between different subregions in the receptive\n\ufb01elds of simple cells.\n\ncortical RF\nspontaneous activity\nnatural visual input\n\n0.7\n\nSaccade + Fixation\nCortical RF\nFixation\n\n \n\nd\nC\nd\ne\nz\n\ni\nl\n\na\nm\nr\no\nn\n\n0.2\n\n0.7\n\n0.2\n\nn\no\n\ni\nt\n\nl\n\na\ne\nr\nr\no\nc\n \n\nd\ne\nz\n\ni\nl\n\na\nm\nr\no\nn\n\n\u22120.3\n\n0.0\n\n0.5\n\n1.0\n\n1.5\n\n2.0\n\n2.5\n\n3.0\n\ndistance (deg.)\n\n\u22120.3\n\n0.0\n\n0.5\n\n1.0\n\ndistance (deg.)\n\n1.5\n\n2.0\n\nFigure 2: Analysis of the correlated activity of LGN units in different experimental con-\nditions. In both graphs, the curve marked by white circles is the average receptive \ufb01eld of\na simple cell, as measured by Jones and Palmer (1987) shown here for comparison. (Left)\nStatic analysis: patterns of correlated activity in the presence of spontaneous activity and\nwhen natural visual input was analyzed statically. (Right) Effects of eye movements: cor-\nrelation difference functions measured when natural images were scanned with sequence\nor saccades or \ufb01xational eye movements.\n\nFig. 3 shows the results of a similar analysis for LGN cells at different visual eccentricities.\nThe white circles in the panels of Fig. 3 represent the width of the largest sub\ufb01eld in the\nreceptive \ufb01eld of cortical simple cells as measured by [13]. The other curves on the left\npanel represent the widths of the central lobe of the correlation difference functions (the\nspatial separation over which cells of the same type possess correlated activity, measured\nas the double of the point in which the correlation difference function intersects the zero\naxis) in the cases of spontaneous activity and static presentation of natural visual input. As\nin Fig. 2, (1) a close correspondence was present between the experimental data and the\nsubregion widths predicted by the correlation difference function in the case of spontaneous\nactivity; and (2) a signi\ufb01cant deviation between the two measurements was present in the\ncase of static examination of natural visual input. The right panel in Fig. 3 shows the\ncorrelation difference functions obtained at different visual eccentricities in the presence of\n\ufb01xational eye movements. The minimum separation between receptive \ufb01elds necessary for\nobserving strong levels of covariance between cells with opposite polarity increased with\neccentricity, as illustrated by the increase in the central lobe of the estimated correlation\nfunctions at the different visual eccentricities. As for the case of spontaneous activity, a\nclose correspondence is now present between the spatiotemporal characteristics of LGN\nactivity and the organization of simple cell receptive \ufb01elds.\n\n4 Discussion\n\nIn this paper we have used computer modeling to study the correlated activity of LGN\ncells when images of natural scenes were scanned so as to replicate cat eye movements. In\nthe absence of eye movements, when a natural visual environment was observed statically,\nsimilar to the way it is examined by animals with their eyes paralyzed, we found that the\nsimulated responses of geniculate cells of the same type at any separation smaller than the\nreceptive \ufb01eld of a simple cell were strongly correlated. These spatial patterns of covary-\ning geniculate activity did not match the structure of simple cell receptive \ufb01elds. A similar\nresult was obtained when natural scenes were scanned through saccades. Conversely, in\n\n\f10\n\n8\n\n5\n\n2\n\n)\ng\ne\nd\n(\n \n\nh\n\nt\n\ni\n\nd\nw\n\n \nl\n\na\nr\nt\nn\ne\nc\n\n0\n\n0\n\nWilson & Sherman, 1976\nspontaneous activity\nnatural input (static)\n\n10\n\n20\n\neccentricity (deg.)\n\n30\n\n10\n\n8\n\n5\n\n2\n\n)\ng\ne\nd\n(\n \nh\nt\nd\nw\n\ni\n\n \nl\na\nr\nt\nn\ne\nc\n\n0\n\n0\n\nWilson & Sherman, 1976\nvisual fixation\n\n10\n\n20\n\n30\n\neccentricity (deg.)\n\nFigure 3: Analysis of the correlated activity of LGN units at different visual eccentricities.\nThe width of the larger sub\ufb01eld in the receptive \ufb01eld of simple cells at different eccentric-\nities as measured by Wilson and Sherman (1976) (white circles) is compared to the width\nof the central lobe of the correlation difference functions measured in different conditions\n(Left) Static analysis: results obtained in the presence of spontaneous activity and when\nnatural visual input was analyzed statically. (Right) Case of \ufb01xational eye movements and\nnatural visual input.\n\nthe case of micromovements, including both microsaccades and the combination of ocular\ndrift and tremor, strong correlations were measured among cells of the same type located\nnearby and among cells of opposite types at distances compatible with the separation be-\ntween different subregions in the receptive \ufb01elds of simple cells. These results suggest a\ndevelopmental role for the small eye movements that occur during visual \ufb01xation.\n\nAlthough the role of visual experience in the development of orientation selectivity has\nbeen extensively investigated, relatively few studies have focused on whether eye move-\nments contribute to the development of the responses of cortical cells. Yet, experiments in\nwhich kittens were raised with their eyes paralyzed have shown basic de\ufb01ciencies in the\ndevelopment of visually-guided behavior [6], as well as impairments in ocular dominance\nplasticity [4, 12]. In addition, it has been shown that eye movements are necessary for the\nreestablishment of cortical orientation selectivity in dark-reared kittens exposed to visual\nexperience within the critical period [2, 5]. This indicates that simultaneous experience of\nvisual input and eye movements (and/or eye movement proprioception) may be necessary\nfor the re\ufb01nement of orientation selectivity [1]. Our \ufb01nding that the patterns of LGN ac-\ntivity with static presentation of natural images did not match the spatial structure of the\nreceptive \ufb01elds of simple cells is in agreement with the hypothesis that exposure to pattern\nvision per se is not suf\ufb01cient to account for a normal visual development.\n\nA main assumption of this study is that the re\ufb01nement and maintenance of orientation\nselectivity after eye opening is mediated by a Hebbian/covariance process of synaptic plas-\nticity. The term Hebbian is used here with a generalized meaning to indicate the family of\nalgorithms in which modi\ufb01cations of synaptic ef\ufb01cacies occur on the basis of the patterns\nof input covariances. While no previous theoretical study has investigated the in\ufb02uence\nof eye movements on the development of orientation selectivity, some models have shown\nthat schemes of synaptic modi\ufb01cations based on the correlated activity of thalamic affer-\nents can account well for the segregation of ON- and OFF-center inputs before eye opening\nin the presence of suitable patterns of spontaneous activity [10, 9]. By showing that, during\n\ufb01xation, the spatiotemporal structure of visually-driven geniculate activity is compatible\nwith the structure of simple cell receptive \ufb01elds, the results of the present study extend the\nplausibility of such schemes to the period after eye opening in which exposure to pattern\n\n\fvision occurs.\n\nOcular movements are a common feature of the visual system of different species. It should\nnot come as a surprise that a trace of their existence can be found even in some of the most\nbasic properties of neurons in the early stages of the visual system, such as orientation\nselectivity. Further studies are needed to investigate whether similar traces can be found in\nother features of visual neural responses.\n\nReferences\n[1] P. Buisseret. In\ufb02uence of extraocular muscle proprioception on vision. Physiol. Rev.,\n\n75(2):323\u2013338, 1995.\n\n[2] P. Buisseret, E. Gary-Bobo, and M. Imbert. Ocular motility and recovery of orienta-\ntional properties of visual cortical neurons in dark-reared kittens. Nature, 272:816\u2013\n817, 1978.\n\n[3] D. Cai, G. C. DeAngelis, and R. D. Freeman. Spatiotemporal receptive \ufb01eld organiza-\ntion in the lateral geniculate nucleus of cats and kitten. J. Neurophysiol., 78(2):1045\u2013\n61, 1997.\n\n[4] R. D. Freeman and A. B. Bonds. Cortical plasticity in monocularly deprived immo-\n\nbilized kittens depends on eye movement. Science, 206:1093\u20131095, 1979.\n\n[5] E. Gary-Bobo, C. Milleret, and P. Buisseret. Role of eye movements in developmental\nprocess of orientation selectivity in the kitten visual cortex. Vision Res., 26(4):557\u2013\n567, 1986.\n\n[6] A. Hein, F. Vital-Durand, W. Salinger, and R. Diamond. Eye movements initiate\n\nvisual-motor development in the cat. Science, 204:1321\u20131322, 1979.\n\n[7] D. Lee and J. G. Malpeli. Effect of saccades on the activity of neurons in the cat\n\nlateral geniculate nucleus. J. Neurophysiol., 79:922\u2013936, 1998.\n\n[8] D. N. Mastronarde. Correlated \ufb01ring of cat retinal ganglion cells. I spontaneously\n\nactive inputs to X and Y cells. J. Neurophysiol., 49(2):303\u2013323, 1983.\n\n[9] K. D. Miller. A model of the development of simple cell receptive \ufb01elds and the\nordered arrangement of orientation columns through activity-dependent competition\nbetween ON- and OFF- center inputs. J. Neurosci., 14(1):409\u2013441, 1994.\n\n[10] M. Miyashita and S. Tanaka. A mathematical model for the self-organization of ori-\n\nentation columns in visual cortex. Neuroreport, 3:69\u201372, 1992.\n\n[11] E. Olivier, A. Grantyn, M. Chat, and A. Berthoz. The control of slow orienting eye\nmovements by tectoreticulospinal neurons in the cat: behavior, discharge patterns and\nunderlying connections. Exp. Brain Res., 93:435\u2013449, 1993.\n\n[12] W. Singer and J. Raushecker. Central-core control of developmental plasticity in\nthe kitten visual cortex II. Electrical activation of mesencephalic and diencephalic\nprojections. Exp. Brain Res., 47:22\u2013233, 1982.\n\n[13] J. R. Wilson and S. M. Sherman. Receptive-\ufb01eld characteristics of neurons in the cat\nstriate cortex: changes with visual \ufb01eld eccentricity. J. Neurophysiol., 39(3):512\u2013531,\n1976.\n\n\f", "award": [], "sourceid": 1994, "authors": [{"given_name": "Antonino", "family_name": "Casile", "institution": null}, {"given_name": "Michele", "family_name": "Rucci", "institution": null}]}