Keywords

Visual cortex, V2, texture

Abstract

The computations performed by neurons in area V1 are reasonably well understood, but computation in subsequent areas such as V2 have been more difficult to characterize. When stimulated with visual stimuli traditionally used to investigate V1, such as sinusoidal gratings, V2 neurons exhibit similar selectivity (but with larger receptive fields, and weaker responses) relative to V1 neurons. However, we find that V2 responses to synthetic stimuli designed to produce naturalistic patterns of joint activity in a model V1 population are more vigorous than responses to control stimuli that lacked this naturalistic structure (Freeman, et. al. 2013). Armed with this signature of V2 computation, we have been investigating how it might arise from canonical computational elements commonly used to explain V1 responses. The invariance of V1 complex cell responses to spatial phase has been previously captured by summing over multiple “subunits” (rectified responses of simple cell-like filters with the same orientation and spatial frequency selectivity, but differing in their receptive field locations). We modeled V2 responses using a similar architecture: V2 subunits were formed from the rectified responses of filters computing the derivatives of the V1 response map over frequencies, orientations, and spatial positions. A "V2 complex cell” sums the output of such subunits across frequency, orientation, and position. This model can qualitatively account for much of the behavior of our sample of recorded V2 neurons, including their V1-like spectral tuning in response to sinusoidal gratings as well as the pattern of increased sensitivity to naturalistic images.

Start Date

15-5-2015 11:05 AM

End Date

15-5-2015 11:30 AM

Session Number

05

Session Title

Early Vision

Share

COinS
 
May 15th, 11:05 AM May 15th, 11:30 AM

Characterizing receptive field selectivity in area V2

The computations performed by neurons in area V1 are reasonably well understood, but computation in subsequent areas such as V2 have been more difficult to characterize. When stimulated with visual stimuli traditionally used to investigate V1, such as sinusoidal gratings, V2 neurons exhibit similar selectivity (but with larger receptive fields, and weaker responses) relative to V1 neurons. However, we find that V2 responses to synthetic stimuli designed to produce naturalistic patterns of joint activity in a model V1 population are more vigorous than responses to control stimuli that lacked this naturalistic structure (Freeman, et. al. 2013). Armed with this signature of V2 computation, we have been investigating how it might arise from canonical computational elements commonly used to explain V1 responses. The invariance of V1 complex cell responses to spatial phase has been previously captured by summing over multiple “subunits” (rectified responses of simple cell-like filters with the same orientation and spatial frequency selectivity, but differing in their receptive field locations). We modeled V2 responses using a similar architecture: V2 subunits were formed from the rectified responses of filters computing the derivatives of the V1 response map over frequencies, orientations, and spatial positions. A "V2 complex cell” sums the output of such subunits across frequency, orientation, and position. This model can qualitatively account for much of the behavior of our sample of recorded V2 neurons, including their V1-like spectral tuning in response to sinusoidal gratings as well as the pattern of increased sensitivity to naturalistic images.