Purchase this article with an account.
Velitchko Manahilov, Gael Gordon, Julie Calvert, William Simpson; A new subtractive normalization model for contrast processing of visual stimuli. Journal of Vision 2007;7(9):256. doi: 10.1167/7.9.256.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The responses of V1 simple cells are determined by different mechanisms: linear feedforward excitation from LGN neurones, rectification, and non-linear suppression proportional to the pooled activity of a large number of other neurons. The nonlinear properties of cortical cells were accounted for by normalization models (Albrecht and Geisler, Visual Neuroscience, 7, 531–546, 1991; Heeger, Visual Neuroscience, 9, 181–197, 1992) which propose a divisive suppression mechanism whose neural correlate is associated with the process of shunting inhibition. However, shunting inhibition may not be the main mechanism mediating intercellular inhibition (Berman et al., J. Physiology., 440, 697–722, 1991).
We propose that the linear responses of cortical cells are normalized by subtractive suppression which is proportional to a nonlinear (power) quantity of the pooled activity of neurones selective to a wide range of spatial frequencies and orientations. This model predicts successfully the saturation of the responses at higher contrast levels and the “divisive” behaviour of the contrast-response functions (downward shift in log-log coordinates) for gratings superimposed on a test grating, as the spatial frequency or orientation vary from the optimal values. Model predictions are also consistent with data from psychophysical studies (Foley and Boynton, SPIE Proceedings, 2054, 32–42, 1994) which measured threshold-contrast functions for the detection of a Gabor patch superimposed on a masking grating for various orientation differences between stimuli.
Subtractive suppression can be implemented by hyperpolarising inhibition which is a fundamental property of cortical neurones. The proposed subtractive-suppression model conforms to the known physiology of cortical cells and can account for the non-linear properties of early visual stages. This model provides an alternative explanation of cortical mechanisms processing visual information.
This PDF is available to Subscribers Only