Abstract
Recurrent connections are known to play a role in the coding of orientation in primary visual cortex (e.g. for contrast invariance) but many questions remain regarding their precise functional role and possible byproducts. Previously, we implemented a recurrent model of orientation selectivity and tested responses to pairs of gratings presented briefly in succession (adapter and test). The model generated repulsive shifts in tuning curves that matched those found experimentally in V1. Unlike the shifts produced with long adaptation periods (thought to underlie the tilt after-effect [TAE]), which are well accounted for by plasticity, the repulsive shifts we observed were the result of slow population dynamics only. These effects on tuning curves gave rise to attractive shifts in the population response towards the adapter, suggesting that perception of the test orientation should -under such conditions- be biased towards the adapted orientation. Methods: To test this prediction, we asked human subjects to determine which of two Gabors (test and reference) presented simultaneously for a short period of time (Ͱ4;100 ms) at either side of the fixation dot was tilted more clockwise. Crucially, the test was preceded by another oriented Gabor (adapter) in the same location (Ͱ4;100 ms), while the reference was preceded by a non-oriented pattern matched for contrast and spatial frequency. Both test and reference were succeeded by non-oriented patterns to reduce afterimages. Results: The psychometric curves exhibited significant shifts consistent with an attraction of perceived orientation of the test towards the adapter. The observed effect matches the predicted effect of population dynamics, and is opposite in direction to the more commonly studied TAE. Conclusion: Population dynamics affect the coding of information and behavioral responses at fast timescales and can produce adaptation-like effects even without any form of neural plasticity.
Meeting abstract presented at VSS 2013