Purchase this article with an account.
Jeffrey Tsai, Alexander Wade, Anthony Norcia; Dynamics and neural computations underlying visual masking. Journal of Vision 2011;11(15):14. doi: https://doi.org/10.1167/11.15.14.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
We study visual masking using source-imaged electroencephalography (EEG) and frequency-domain analysis in humans, examining a wide range of relative stimulus strengths and spectral components driven by individual stimuli (self terms) and those due to interaction between stimuli (intermodulation [IM] terms). Consistent with previous reports, in early visual cortex, masking manifests in the self-terms as an effective reduction of input contrast. We identify a novel signature of masking – the magnitude of the second-order IM term peaks when the input contrasts are equal and reaches a minimum when they are widely different. To account for our data, the standard divisive gain control model, parametric in response dynamics, was fitted to the self- and IM-terms simultaneously. Previous instantiations of similar models with either very short or very long temporal integration in the formulation of the gain pool response performed worse than a model with an integration time of approximately 30 ms. Finally, the magnitude of the spectral components depends only on the ratio of the input contrasts. This “contrast-contrast” invariance suggests that neurons in visual cortex operate on a representation of relative rather than absolute contrast. Together, these results provide a more complete description of masking within the framework of gain control.
This PDF is available to Subscribers Only