Elsevier

Neurocomputing

Volumes 38–40, June 2001, Pages 757-762
Neurocomputing

A model of cross-orientation inhibition in cat primary visual cortex

https://doi.org/10.1016/S0925-2312(01)00427-1Get rights and content

Abstract

In cortical simple cells of cat striate cortex, the response to a visual stimulus of the preferred orientation is partially suppressed by simultaneous presentation of a stimulus at the orthogonal orientation, an effect known as “cross-orientation inhibition”. It has been argued that this is due to the presence of inhibitory connections between cells tuned for different orientations, but intracellular studies suggest that simple cells only receive inhibitory input from cells with similar orientation tuning. Here we present a model, based on local intracortical connectivity between cells of similar orientation tuning, that can account for many aspects of cross-orientation inhibition.

Introduction

In 1962 Hubel and Wiesel [10] suggested that the orientation selectivity of simple cells in V1 layer 4 derives from an oriented arrangement of LGN inputs: ON-center LGN inputs have receptive fields aligned over the simple cell's ON subregions, and OFF-center inputs are aligned on OFF subregions. Recent evidence is consistent with this model [8, and references therein]. This feedforward model of receptive fields can account for various properties of simple cells in layer 4 of V1, but is not sufficient to account for the complete behavior of these cells. One example is the contrast-invariant orientation tuning of cells in V1 [12].

Another challenge to the feedforward model is known as “cross-orientation inhibition”. A typical simple cell responds to a drifting grating shown at its preferred orientation and is silent in response to a drifting grating of the perpendicular (null) orientation. If spike responses were a linear function of the input, then a superposition of the two gratings would give a response equal to that to the preferred-orientation stimulus alone, whereas in fact the superposition evokes a smaller response [2], [5]. This has led to the suggestion that inhibition from cells preferring different orientations plays an important role in cortical orientation selectivity [3, and references therein]. However, this is not supported by evidence from intracellular recordings showing that the excitation and the inhibition received by simple cells in cat layer 4 show similar orientation tuning, with both peaked at the preferred orientation and falling to small values at the orthogonal orientation [6], [1], as well as by recent evidence that the orientation selectivity of voltage responses is neither created, nor sharpened, by intracortical circuitry [7], [4].

We recently proposed a model [13] that accounted for contrast-invariant orientation tuning using a model circuit involving intracortical connections between cells of similar orientation selectivity, as suggested by the intracellular recordings. In this model, excitatory cells tend to make connections onto cells of similar preferred orientation and similar absolute spatial phase (similar locations in visual space of ON-subregions and of OFF-subregions), while inhibitory cells tend to project to cells of similar preferred orientation and opposite absolute spatial phase. We showed that the feedforward input (LGN input plus feedforward inhibition) induced by drifting gratings splits naturally into two components: a mean (DC) component that is net inhibitory, and a temporally modulating component (first harmonic, F1) which causes spiking during the excitatory phase of the modulation.

Here we show that this circuitry, with modifications as described in [11, TZL AEK & KDM in preparation], can also account for cross-orientation inhibition. One of the more extensive experimental studies of cross-orientation inhibition was done by Bonds [2]. In our modeling studies we have followed his procedure to determine how closely our model is able to reproduce the experimental properties of cross-orientation inhibition.

Section snippets

Results

In both experiments and the model, cell response to a base grating at the preferred orientation is suppressed by the superposition of a mask grating at the orthogonal orientation, and the suppression increases with the contrast of the mask grating (Fig. 1).

Experiments indicate that aspects of this suppression are not tuned for orientation of the mask grating [2], [5]. Bonds [2] measured responses to pairs of sine gratings, a base grating at the preferred orientation of the cell, and a mask

Discussion

We have shown here that a simple correlation-based local circuit can account for many aspects of cortical inhibition. This is the first model that does not need inhibition arising from cells of multiple preferred orientations in order to account for cross-orientation suppressive effects. The model has two major assumptions [13], both suggested by experiments [9], [8]: that connectivity is phase-specific, with excitation between cells of similar preferred spatial phase and inhibition onto cells

Thomas Lauritzen received his B.S. degree in physics in 1996, and his M.Sc. in physics-biophysics in 1998 from the Niels Bohr Institute, Copenhagen, Denmark. He is currently enrolled in the Ph.D. program in biophysics at the University of California, San Francisco.

References (13)

  • S. Chung et al.

    Strength and orientation tuning of the thalamic input to simple cells revealed by electrically evoked cortical suppression

    Neuron

    (1998)
  • J.S. Anderson et al.

    Orientation tuning of input conductance, excitation, and inhibition in cat primary visual cortex

    J. Neurophysiol.

    (2000)
  • A.B. Bonds

    Role of inhibition in the specification of orientation selectivity of cells in the cat striate cortex

    Vis. Neurosci.

    (1989)
  • M. Carandini et al.

    Linearity and gain control in V1 simple cells

  • G.C. DeAngelis et al.

    Organization of suppression in receptive fields of neurons in cat visual cortex

    J. Neurophysiol.

    (1992)
  • D. Ferster

    Orientation selectivity of synaptic potentials in neurons of cat primary visual cortex

    J. Neurosci.

    (1986)
There are more references available in the full text version of this article.

Cited by (1)

Thomas Lauritzen received his B.S. degree in physics in 1996, and his M.Sc. in physics-biophysics in 1998 from the Niels Bohr Institute, Copenhagen, Denmark. He is currently enrolled in the Ph.D. program in biophysics at the University of California, San Francisco.

Anton Krukowski received his B.S. degree in mathematics and physics in 1991 from Yale University, and his Ph.D. in biophysics from the University of California, San Francisco in 2000. He is currently a post doctoral fellow at NASA Ames Research Center.

Ken Miller received his B.S. degree in biology in 1980 from Reed College, and an M.S. in physics in 1981 and Ph.D. in Neuroscience in 1989 from Stanford University. He is currently Associate Professor in the Departments of Physiology & Otolaryngology at the University of California at San Francisco.

Supported by a Danish Research Agency predoctoral fellowship (TZL), a HHMI predoctoral fellowship (AEK) and RO1-EY11001 from the NEI (KDM).

View full text