Purchase this article with an account.
Corentin Massot, Tai Sing Lee; Sensitivity to Spatial Frequency Chirp in the Early Visual Cortex . Journal of Vision 2014;14(10):731. doi: https://doi.org/10.1167/14.10.731.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Neural correlates of 3D shape from texture have been recently found in areas V3 and CIP. However the neural mechanisms that generate these neural correlates is still unknown. In previous behavioral and computational studies, we showed that 3D slant and tilt can be reliably inferred from the gradient of spatial frequency present in the texture patterns (Massot et al., 2008, 2011). In our model, such gradient can be estimated from the output of an ensemble of V1 Gabor-like filters. This leads to the hypothesis that neurons in early visual cortical areas such as V2 and V3 might combine V1 neurons' responses to develop spatial frequency gradient sensitivity. To test this hypothesis, we recorded single unit activity in V1, V2 and V3a cortical areas of non-human primates using semi-chronic multi-electrode arrays. Stimuli are static gratings displaying a spatial frequency gradient simulating a planar surface receding in depth (chirps). Slant and tilt angles are defined by the amplitude and the direction of the spatial frequency gradient. A set of 360 stimuli was created with different tilt and slant angles. Each stimulus was presented for 250ms and presented 12 times in a block design experiment onto the receptive field of the isolated neuron. Our preliminary evidence shows that V3a neurons exhibit tuning to spatial frequency gradient in addition to classical orientation and spatial frequency tuning. Neurons were tuned to either high positive gradients, high negative gradients, high absolute gradients, or did not display any preference. The recorded V1 and V2 neurons, with more localized receptive fields, were not sensitive to any gradient of spatial frequency. Overall, the obtained data suggest that the neurons in V3a have developed the building blocks for the computation of shape from texture.
Meeting abstract presented at VSS 2014
This PDF is available to Subscribers Only