Abstract
Gamma-band (30-100Hz) activity in the local field potential (LFP) is commonly observed in different brain structures and thought to play important roles in information processing. Recent studies showed that two distinct oscillatory components peaking in gamma band (low gamma and high gamma) exist in the LFP of primary visual cortex (V1). However, the neural mechanisms for generating the two gamma components and their distinct response properties are poorly understood.
Here, we simultaneously recorded from visual area 18 and the lateral geniculate nucleus (LGN) of anesthetized cats and collected their neural activity (both LFP and multi-unit activity, MUA) to drifting grating stimuli with different orientations. We found two gamma components (low gamma, 50-65Hz, high gamma, 70-100Hz) coexist in cat V1, and the orientation tunings of the two gamma components and spike activity (measured by MUA) are significantly different. Orientation selectivity, quantified as the circular variance (CV), of MUA is significantly higher than the selectivity of both low gamma component (average CV difference is 0.12; t-test: t=6.6, p<0.001) and high gamma components (average CV difference is 0.23; t-test: t=14.9, p<0.001); but the CV of low gamma is more similar to the CV of MUA. Interestingly we found that the ratio of low-gamma power and high-gamma power is positively correlated with circular variance of MUA (r=0.55, p<0.001), low-gamma (r=0.64, p<0.001) and high-gamma (r=0.4, p<0.001). Moreover, only high gamma component is found in the LGN, indicating that low gamma component originates within V1, possibly through recurrent neural network. As the strength of recurrent neural network increases, the orientation selectivity of V1 output gets weaker. Our work demonstrates multiple sources for gamma oscillation and this property might give us a useful tool to probe feedforward and recurrent connections to functional properties in the brain.