Abstract
Purpose: A number of neuroimaging tools are used to measure signals in the human visual system, including fMRI, MEG, EEG, and ECoG. Each measurement modality likely emphasizes different components of the neural circuitry. A better understanding of visual cortical function will require linking measurements across modalities. Here we studied visual signals with electrocorticography (ECoG) and compared previously collected fMRI responses to two different aspects of the neurophysiological response: narrow-bandwidth oscillation in the lower frequencies (centered between 25 and 70 Hz) and a broadband high-frequency signal (spanning at least 80–175 Hz). Methods: ECoG signals in visual cortex (V1-V4) were measured in response to a wide range of carefully designed stimuli (86 stimuli, 15 repetitions, 500-ms duration, random order). These stimuli varied systematically along dimensions such as spatial location, contrast, orientation, spatial frequency, and second-order contrast. Power spectra of the ECoG responses were computed and separated into broadband (80-175 Hz) and gamma responses (25-50Hz). Results: Broadband and gamma responses in V1/V2 exhibited strikingly different behaviors. Gamma responses were substantially stronger for stimuli with narrow orientation and spatial frequency content (i.e. sinusoidal gratings) compared to other stimuli, whereas broadband responses were substantially stronger for stimuli that contained broad orientation and spatial frequency content. The selectivity exhibited by the broadband responses, but not the gamma responses, better matched the selectivity exhibited by fMRI responses obtained from healthy control subjects. Conclusion: Our results show that the strongest gamma responses were elicited by high-contrast sinusoidal gratings; this was not true of the broadband response. We speculate that gamma oscillations reflect a resonant state elicited only by particular patterns of neuronal activity within a local region of cortex. The fMRI BOLD signal correlates better with the broadband response than with gamma oscillations, and likely reflects the overall neural response level within a cortical region.
Meeting abstract presented at VSS 2014