Receptive field

About: Receptive field is a research topic. Over the lifetime, 8537 publications have been published within this topic receiving 596428 citations.

The neurophysiology of figure^ground segregation in primary visual cortex

Abstract: The activity of neurons in the primary visual cortex of the awake macaque monkey was recorded while the animals were viewing full screen arrays of either oriented line segments or moving random dots. A square patch of the screen was made to perceptually pop out as a circumscribed figure by virtue of differences between the orientation or the direction of motion of the texture elements within that patch and the surround. The animals were trained to identify the figure patches by making saccadic eye movements towards their positions. Almost every cell gave a significantly larger response to elements belonging to the figure than to similar elements belonging to the background. The figure-ground response enhancement was present along the entire extent of the patch and was absent as soon as the receptive field was outside the patch. The strength of the effect had no relation with classical receptive field properties like orientation or direction selectivity or receptive field size. The response enhancement had a latency of 30-40 msec relative to the onset of the neuronal response itself. The results show that context modulation within primary visual cortex has a highly sophisticated nature, putting the image features the cells are responding to into their fully evaluated perceptual context.

Uniformity of monkey striate cortex: a parallel relationship between field size, scatter, and magnification factor.

Abstract: This paper is concerned with the relationship between orientation columns, ocular-dominance columns, the topographic mapping of visual fields onto cortex, and receptive-field size and scatter. Although the orientation columns are an order of magnitude smaller than the ocular-dominance columns, the horizontal distance corresponding to a complete cycle of orientation columns, representing a rotation through 180°, seems to be roughly the same size as a left-plus-right ocular dominance set, with a thickness of about 0.5–1 mm, independent of eccentricity at least out to 15°. We use the term hypercolumn to refer to a complete set of either type (180°, or left-plus-right eyes). In the macaque monkey several penetrations were made at various eccentricities in various parts of the striate cortex subserving the fovea, parafovea and midperiphery. As observed many times previously, in any vertical penetration there was an apparently random scatter in receptive-field positions, which was of the same order of magnitude as the individual receptive fields in that part of the cortex; the field size and the scatter increased in parallel fashion with eccentricity. The movement through the visual field corresponding to a 1 mm horizontal movement along the cortex (the reciprocal of the magnification factor) also increased with eccentricity, in a manner that was strikingly parallel with the increase in receptive field size and scatter. In parts of the cortex representing retina, at least out to about 22° from the fovea, a movement along the cortical surface of about 1 mm was enough to displace the fields so that the new position they collectively occupied half overlapped the old. Such an overlap was thus produced by moving along the cortex a distance about equal to the thickness of a left-plus-right set of ocular-dominance columns, or a complete 180° array of orientation columns. It therefore seems that, independent of eccentricity, a 2 mm × 2 mm block of cortex contains by a comfortable margin the machinery needed to analyze a region of visual field roughly equal to the local field size plus scatter. A movement of 2–3 mm corresponds to a new visual field region and to several new sets of hypercolumns. The cortex thus seems remarkably uniform physiologically, just as it is anatomically.

Understanding the Effective Receptive Field in Deep Convolutional Neural Networks

Abstract: We study characteristics of receptive fields of units in deep convolutional networks. The receptive field size is a crucial issue in many visual tasks, as the output must respond to large enough areas in the image to capture information about large objects. We introduce the notion of an effective receptive field, and show that it both has a Gaussian distribution and only occupies a fraction of the full theoretical receptive field. We analyze the effective receptive field in several architecture designs, and the effect of nonlinear activations, dropout, sub-sampling and skip connections on it. This leads to suggestions for ways to address its tendency to be too small.

Cortical feedback improves discrimination between figure and background by V1, V2 and V3 neurons

Abstract: A single visual stimulus activates neurons in many different cortical areas. A major challenge in cortical physiology is to understand how the neural activity in these numerous active zones leads to a unified percept of the visual scene. The anatomical basis for these interactions is the dense network of connections that link the visual areas. Within this network, feedforward connections transmit signals from lower-order areas such as V1 or V2 to higher-order areas. In addition, there is a dense web of feedback connections which, despite their anatomical prominence1,2,3,4, remain functionally mysterious5,6,7,8. Here we show, using reversible inactivation of a higher-order area (monkey area V5/MT), that feedback connections serve to amplify and focus activity of neurons in lower-order areas, and that they are important in the differentiation of figure from ground, particularly in the case of stimuli of low visibility. More specifically, we show that feedback connections facilitate responses to objects moving within the classical receptive field; enhance suppression evoked by background stimuli in the surrounding region; and have the strongest effects for stimuli of low salience.

The two-dimensional spatial structure of simple receptive fields in cat striate cortex

Abstract: 1. A reverse correlation (6, 8, 25, 35) method is developed that allows quantitative determination of visual receptive-field structure in two spatial dimensions. This method is applied to simple cells in the cat striate cortex. 2. It is demonstrated that the reverse correlation method yields results with several desirable properties, including convergence and reproducibility independent of modest changes in stimulus parameters. 3. In contrast to results obtained with moving stimuli, we find that the bright and dark excitatory subregions in simple receptive fields do not overlap to any great extent. This difference in results may be attributed to confounding the independent variables space and time when using moving stimuli. 4. All simple receptive fields have subregions that vary smoothly in all directions in space. There are no sharp transitions either between excitatory subregions or between subregions and the area surrounding the receptive field. 5. Simple receptive fields vary both in the number of subregions observed, in the elongation of each subregion, and in the overall elongation of the field. In contrast with results obtained using moving stimuli, we find that subregions within a given receptive field need not be the same length. 6. The hypothesis that simple receptive fields can be modeled as either even symmetric or odd symmetric about a central axis is evaluated. This hypothesis is found to be false in general. Most simple receptive fields are neither even symmetric nor odd symmetric. 7. The hypothesis that simple receptive fields can be modeled as the product of a width response profile and an orthogonal length response profile (Cartesian separability) is evaluated. This hypothesis is found to be true for only approximately 50% of the cells in our sample.