A major goal in visual neuroscience is to study neuronal receptive field (RF) properties. With clear linear response properties, simple cells in mammalian primary visual cortex (V1) have been studied extensively in the past several decades. However, the traditional stimuli (such as bars or drift gratings) and linear analysis methods (such as spike-triggered average, STA) cannot be used effectively to study the RFs of V1 complex cells, because of their nonlinear response properties. Recent studies revealed several new properties of complex cell RFs in anesthetized animals by using more complex stimuli (such as white noise and natural scene) and nonlinear analysis methods (such as spike-triggered covariance, STC). In the current study, we used binary white noise and STC to study properties of neurons in awake monkey primary visual cortex. We found up to nine excitatory components (eigenvectors) for each cell, including the first one or two excitatory components with dominant contributions to the neuronal responses, along with additional excitatory and suppressive components with weaker contributions. Compared with the dominant components, the non-dominant excitatory components prefer similar orientations and spatial frequencies but have lager spatial envelopes. They contribute to response invariance to small changes in stimulus orientation, position, and spatial frequency. In contrast, the suppressive components are tuned to orientations 45°–90° different from the excitatory components, which may underlie cross orientation suppression. The nonlinear model including both excitatory and suppressive components can effectively predict neuronal responses to white noise.
Key words: RF, primary visual cortex, white noise, STC, eigenvector