Purchase this article with an account.
Kristina D. Simmons, Jason S. Prentice, Gasper Tkacik, Jan Homann, Philip C. Nelson, Vijay Balasubramanian; Retinal Adaptation to Stimulus Correlations. Invest. Ophthalmol. Vis. Sci. 2012;53(14):1938.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
It is widely believed, but not directly tested, that the retina spatially decorrelates incoming stimuli using surround suppression. To test for such adaptive decorrelation, we measured pairwise correlations, linear receptive fields, and nonlinearities from retinal ganglion cells responding to stimuli with various degrees of correlation.
We tested for adaptation by recording simultaneously from approximately 40 ganglion cells on a multi-electrode array while presenting checkerboards with varying degrees of correlation. We measured the pairwise correlations between spike trains of simultaneously recorded neurons under three conditions: white noise, exponentially correlated noise, and scale-invariant (naturalistic) correlated noise. We were also able to extract spatiotemporal receptive fields and nonlinearities for white noise and exponentially correlated noise.
In response to stimuli with spatially wider pairwise correlations, we found that (a) the retina maintains a relatively fixed amount of output correlation, (b) the linear receptive field grows wider while the local surround becomes relatively weaker, and (c) the static nonlinearity becomes shallower in proportion to the "effective contrast" of the stimulus. While the measured output decorrelation was partly explained by adaptation in the nonlinearity, rather than by relative changes in surround inhibition, the best linear-nonlinear model of single cell responses failed to fully account for the properties of pairs of cells.
These results indicate that the retina is able to adaptively decorrelate its input but that this decorrelation is not accomplished through an increase in surround strength. Our results also suggest that single cell receptive fields adapt not to decorrelate stimuli but to maintain invariance of responses to stimulus scale.
This PDF is available to Subscribers Only