Abstract
Field potentials (FPs) are commonly used to measure responses of local populations of neurons in cortex. Recent results questioned how local these populations are, and suggested that only responses at high frequencies originate locally (Henrie & Shapley, 2005; Liu & Newsome, 2006). These results, however, were obtained with stimuli eliciting tonic responses, which are hard to isolate with FPs, as they overlap greatly with ongoing activity. We used stimuli eliciting phasic responses in cat V1 and assessed how local FPs are by measuring their orientation selectivity. Selectivity would imply a local origin: the distance between orthogonal orientation domains is ∼300 µm. We first measured the amplitude spectrum of ongoing activity, recorded while presenting a blank screen. This amplitude decreased markedly with frequency, varying approximately as 1/frequency. We then recorded responses to random sequences of 32-ms flashed gratings having various orientations, spatial frequencies and spatial phases. Stimulus-triggered averaging of the FPs revealed a pronounced difference between responses to orthogonal orientations, evident mostly at low frequencies in the spectrum of these responses. Thus, in contrast to previous reports, we find a pronounced selectivity for stimulus orientation even in low frequency components of the FPs. Therefore, the FPs' origin must be in part local, on a scale of <300 µm.