A Self-Organizing Model of “Color Blob” Formation

1996 ◽  
Vol 8 (7) ◽  
pp. 1427-1448 ◽  
Author(s):  
Harry G. Barrow ◽  
Alistair J. Bray ◽  
Julian M. L. Budd
Keyword(s):  
Visual Processing ◽  
Striate Cortex ◽  
Hebbian Learning ◽  
Principal Component ◽  
Self Organization ◽  
Color Information ◽  
Color Processing ◽  
Parvocellular Pathway ◽  
Optimal Linear ◽  
Contrast Information

This paper explores the possibility that the formation of color blobs in primate striate cortex can be partly explained through the process of activity-based self-organization. We present a simulation of a highly simplified model of visual processing along the parvocellular pathway, that combines precortical color processing, excitatory and inhibitory cortical interactions, and Hebbian learning. The model self-organizes in response to natural color images and develops islands of unoriented, color-selective cells within a sea of contrast-sensitive, orientation-selective cells. By way of understanding this topography, a principal component analysis of the color inputs presented to the network reveals that the optimal linear coding of these inputs keeps color information and contrast information separate.

scholarly journals A Neural Network Model for the Self-Organization of Cortical Grating Cells

2000 ◽  
Vol 55 (3-4) ◽  
pp. 282-291
Author(s):  
Christoph Bauer ◽  
Thomas Burger ◽  
Martin Stetter ◽  
Elmar W. Lang
Keyword(s):  
Neural Network ◽  
Network Model ◽  
Neural Network Model ◽  
Striate Cortex ◽  
Hebbian Learning ◽  
Self Organization ◽  
Response Behavior ◽  
Lateral Coupling ◽  
Saturation Effects ◽  
Activity Dependent

Abstract A neural network model with incremental Hebbian learning of afferent and lateral synaptic couplings is proposed,which simulates the activity-dependent self-organization of grating cells in upper layers of striate cortex. These cells, found in areas V1 and V2 of the visual cortex of monkeys, respond vigorously and exclusively to bar gratings of a preferred orientation and periodicity. Response behavior to varying contrast and to an increasing number of bars in the grating show threshold and saturation effects. Their location with respect to the underlying orientation map and their nonlinear response behavior are investigated. The number of emerging grating cells is controlled in the model by the range and strength of the lateral coupling structure.

Introducing Asymmetry into Interneuron Learning

1995 ◽  
Vol 7 (6) ◽  
pp. 1191-1205 ◽  
Author(s):  
Colin Fyfe
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Negative Feedback ◽  
Principal Components ◽  
Network Architecture ◽  
Hebbian Learning ◽  
Principal Component ◽  
Neural Network Architecture ◽  
Weight Decay ◽  
Artificial Neural Network Architecture

A review is given of a new artificial neural network architecture in which the weights converge to the principal component subspace. The weights learn by only simple Hebbian learning yet require no clipping, normalization or weight decay. The net self-organizes using negative feedback of activation from a set of "interneurons" to the input neurons. By allowing this negative feedback from the interneurons to act on other interneurons we can introduce the necessary asymmetry to cause convergence to the actual principal components. Simulations and analysis confirm such convergence.

Color Processing in Macaque Striate Cortex: Electrophysiological Properties

2002 ◽  
Vol 87 (6) ◽  
pp. 3138-3151 ◽  
Author(s):  
Carole E. Landisman ◽  
Daniel Y. Ts'O
Keyword(s):  
Optical Imaging ◽  
Cytochrome Oxidase ◽  
Striate Cortex ◽  
Color Patch ◽  
Color Processing ◽  
Color Properties ◽  
Electrophysiological Properties ◽  
Color Opponency ◽  
One To One ◽  
Highly Correlated

We have shown in the accompanying paper that optical imaging of macaque striate cortex reveals patches that are preferentially activated by equiluminant chromatic gratings compared with luminance gratings. These imaged color patches are highly correlated, although not always in one-to-one correspondence, with the cytochrome-oxidase (CO) blobs. In the present study, we have investigated the electrophysiological properties of neurons in the imaged color patches and the CO blobs. Our results indicate that individual blobs tend to contain cells of only one type of color opponency: either red/green or blue/yellow. Individual imaged color patches, however, can bridge blobs of similar opponency or differing opponency. When imaged color patches contain two blobs of differing opponency, the cells in the bridge region exhibit mixed color properties that are not opponent along the two cardinal color axes (either red/green or blue/yellow). Two blobs within a single imaged color patch receive input from the same eye or from different eyes. In the latter case, the bridge region between blobs contains binocular cells that are color selective. Because the cells recorded in imaged color patches were more color selective and unoriented than cells outside of color patches, color properties appear to be organized in a clustered and segregated fashion in primate V1.

Reduced Representation by Neural Networks with Restricted Receptive Fields

1995 ◽  
Vol 7 (3) ◽  
pp. 507-517 ◽  
Author(s):  
Marco Idiart ◽  
Barry Berk ◽  
L. F. Abbott
Keyword(s):  
Neural Networks ◽  
Biological Networks ◽  
Input Data ◽  
Hebbian Learning ◽  
Receptive Fields ◽  
Principal Component ◽  
Data Sets ◽  
Reduced Representation ◽  
Learning Rules ◽  
Reduced Representations

Model neural networks can perform dimensional reductions of input data sets using correlation-based learning rules to adjust their weights. Simple Hebbian learning rules lead to an optimal reduction at the single unit level but result in highly redundant network representations. More complex rules designed to reduce or remove this redundancy can develop optimal principal component representations, but they are not very compelling from a biological perspective. Neurons in biological networks have restricted receptive fields limiting their access to the input data space. We find that, within this restricted receptive field architecture, simple correlation-based learning rules can produce surprisingly efficient reduced representations. When noise is present, the size of the receptive fields can be optimally tuned to maximize the accuracy of reconstructions of input data from a reduced representation.

Spatial properties of neurons in the monkey striate cortex

1987 ◽  
Vol 231 (1263) ◽  
pp. 251-288 ◽  
Keyword(s):  
Receptive Field ◽  
Contrast Sensitivity ◽  
Visual Processing ◽  
Striate Cortex ◽  
Receptive Fields ◽  
Spatial Filter ◽  
Cortical Cells ◽  
Simple Cells ◽  
Gabor Function ◽  
Difference Of Gaussians

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 were 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.

Dissociating early and late visual processing via the Ebbinghaus illusion

2016 ◽  
Vol 33 ◽  
Author(s):  
FILIPP SCHMIDT ◽  
ANDREAS WEBER ◽  
ANKE HABERKAMP
Keyword(s):  
Visual Processing ◽  
High Frequency ◽  
Time Course ◽  
Temporal Dynamics ◽  
Low Frequency ◽  
Response Speed ◽  
Ebbinghaus Illusion ◽  
Priming Effects ◽  
Parvocellular Pathway ◽  
The Difference

AbstractVisual perception is not instantaneous; the perceptual representation of our environment builds up over time. This can strongly affect our responses to visual stimuli. Here, we study the temporal dynamics of visual processing by analyzing the time course of priming effects induced by the well-known Ebbinghaus illusion. In slower responses, Ebbinghaus primes produce effects in accordance with their perceptual appearance. However, in fast responses, these effects are reversed. We argue that this dissociation originates from the difference between early feedforward-mediated gist of the scene processing and later feedback-mediated more elaborate processing. Indeed, our findings are well explained by the differences between low-frequency representations mediated by the fast magnocellular pathway and high-frequency representations mediated by the slower parvocellular pathway. Our results demonstrate the potentially dramatic effect of response speed on the perception of visual illusions specifically and on our actions in response to objects in our visual environment generally.

scholarly journals Monocular Cells Without Ocular Dominance Columns

2006 ◽  
Vol 96 (5) ◽  
pp. 2253-2264 ◽  
Author(s):  
Daniel L. Adams ◽  
Jonathan C. Horton
Keyword(s):  
Squirrel Monkey ◽  
Visual Processing ◽  
Striate Cortex ◽  
Single Cells ◽  
Ocular Dominance ◽  
Squirrel Monkeys ◽  
Ocular Dominance Columns ◽  
Receptive Field Property ◽  
Field Property ◽  
Made In

In many regions of the mammalian cerebral cortex, cells that share a common receptive field property are grouped into columns. Despite intensive study, the function of the cortical column remains unknown. In the squirrel monkey, the expression of ocular dominance columns is variable, with columns present in some animals and not in others. By searching for differences between animals with and without columns, it should be possible to infer how columns contribute to visual processing. Single-cell recordings outside layer 4C were made in nine squirrel monkeys, followed by labeling of ocular dominance columns in layer 4C. In the squirrel monkey, compared with the macaque, cells outside layer 4C were more likely to respond to stimulation of either eye whether ocular dominance columns were present or not. In three animals lacking ocular dominance columns, single cells were recorded from layer 4C. Remarkably, 20% of cells in layer 4C were monocular despite the absence of columns. This observation means that ocular dominance columns are not necessary for monocular cells to occur in striate cortex. In macaques each row of cytochrome oxidase (CO) patches is aligned with an ocular dominance column and receives koniocellular input serving one eye only. In squirrel monkeys this was not true: CO patches and ocular dominance columns had no spatial correlation and the koniocellular input to CO patches was binocular. Thus even when ocular dominance columns occur in the squirrel monkey, they do not transform the functional architecture to resemble that of the macaque.

Binocular visual processing in the owl’s telencephalon

1979 ◽  
Vol 204 (1157) ◽  
pp. 435-454 ◽  
Keyword(s):  
Visual Processing ◽  
Striate Cortex ◽  
Functional Organization ◽  
Receptive Fields ◽  
Field Size ◽  
Evolutionary Origins ◽  
Structural Differences ◽  
Output Neurons ◽  
Global Stereopsis ◽  
Orientation Movement

Single neurons recorded from the owl’s visual Wulst are surprisingly similar to those found in mammalian striate cortex. The receptive fields of Wulst neurons are elaborated, in an apparently hierarchical fashion,from those of their monocular, concentrically organized inputs to produce binocular interneurons with increasingly sophisticated requirements for stimulus orientation, movement and binocular disparity. Output neurons located in the superficial laminae of the Wulst are the most sophisticated of all, with absolute requirements for a combination of stimuli, which include binocular presentation at a particular horizontal binocular dis­parity, and with no response unless all of the stimulus conditions are satisfied simultaneously. Such neurons have the properties required for ‘global stereopsis,’ including a receptive field size many times larger than their optimal stimulus, which is more closely matched to the receptive fields of the simpler, disparity-selective interneurons. These marked similarities in functional organization between the avian and mammalian systems exist in spite of a number of structural differences which reflect their separate evolutionary origins. Discussion therefore includes the possibility that there may exist for nervous systems only a very small number of possible solutions, perhaps a unique one, to the problem of stereopsis.

Computational Predictions on the Receptive Fields and Organization of V2 for Shape Processing

2009 ◽  
Vol 21 (3) ◽  
pp. 762-785 ◽  
Author(s):  
Yiu Fai Sit ◽  
Risto Miikkulainen
Keyword(s):  
Receptive Field ◽  
Visual Processing ◽  
Receptive Fields ◽  
Self Organization ◽  
General Shape ◽  
Orientation Component ◽  
Shape Processing ◽  
Primary Orientation ◽  
Computational Predictions

It has been more than 40 years since the first studies of the secondary visual cortex (V2) were published. However, no concrete hypothesis on how the receptive field of V2 neurons supports general shape processing has been proposed to date. Using a computational model that follows the principle of self-organization, we advance two hypotheses in this letter: (1) typical V2 orientation-selective receptive field contains a primary orientation and a secondary orientation component, forming a corner, a junction, or a cross; and (2) V2 columns with the same primary orientation form contiguous domains, divided into subdomains that prefer different secondary orientations. The first hypothesis is consistent with existing experimental evidence, and both hypotheses can be tested with current techniques in animals. In this manner, computational modeling can be used to provide verifiable predictions that eventually allow us to understand the role of V2 in visual processing.

scholarly journals Comparison of the color selectivity of macaque V4 neurons in different color spaces

2016 ◽  
Vol 116 (5) ◽  
pp. 2163-2172 ◽  
Author(s):  
Takahisa M. Sanada ◽  
Tomoyuki Namima ◽  
Hidehiko Komatsu
Keyword(s):  
Visual Processing ◽  
Color Space ◽  
Limited Range ◽  
Color Information ◽  
Color Representation ◽  
Visual Areas ◽  
Color Selectivity ◽  
Different Color ◽  
V4 Neurons ◽  
Higher Visual Areas

Chromatic selectivity has been studied extensively in various visual areas at different stages of visual processing in the macaque brain. In these studies, color stimuli defined in the Derrington-Krauskopf-Lennie (DKL) color space with a limited range of cone contrast were typically used in early stages, whereas those defined in the Commission Internationale de l'Eclairage (CIE) color space, based on human psychophysical measurements across the gamut of the display, were often used in higher visual areas. To understand how the color information is processed along the visual pathway, it is necessary to compare color selectivity obtained in different areas on a common color space. In the present study, we tested whether the neural color selectivity obtained in DKL space can be predicted from responses obtained in CIE space and whether stimuli with limited cone contrast are sufficient to characterize neural color selectivity. We found that for most V4 neurons, there was a strong correlation between responses measured using the two chromatic coordinate systems, and the color selectivities obtained with the two stimulus sets were comparable. However, for some neurons preferring high- or low-saturation colors, stimuli defined in DKL color space did not adequately capture the neural color selectivity. This is mainly due to the use of stimuli within a limited range of cone contrast. We conclude that regardless of the choice of color space, the sampling of colors across the entire gamut is important to characterize neural color selectivity fully or to compare color selectivities in different areas so as to understand color representation in the visual system.

Sign in / Sign up

Export Citation Format

Share Document