Abstract
Recent evidence has demonstrated that the cortex can exhibit the same patterns of oriented activity during presentation of a blank screen (spontaneous ‘ongoing’ activity) as during presentation of a grating stimulus (c.f. Tsodyks et al. Science 1999). Here, we have made paired electrode recordings between oriented cells and luxotonic cells (cells responsive to large full-field flashes) in Area 17 and 18 of cats.
We present evidence that spontaneous ‘ongoing’ activity is not a flat baseline of activity, but an inherently biased network of interactions. We find such biased interactions between oriented and luxotonic cells in Areas 17 and 18. Luxotonic cells, which have been suggested to encode luminance, illuminance, and lightness constancy, tend to fire 5 to 10 milliseconds before nearby (< 5 deg in V1) oriented cells. This bias is present during spontaneous activity as well as during surface brightness modulation (oriented cell on contrast border, luxotonic cell at center of surface; cf. Hung et al. Vision Res 2001). This suggests that local brightness information tends to be calculated prior to edge extraction, perhaps for the purpose of enhancing or building border response.
We also show evidence of biased spontaneous interactions between pairs of Area 17 and 18 luxotonic cells. The direction of these biases is dependent on their receptive field separation, being 17-to-18 for adjacent pairs (< 5 deg separation) and 18-to-17 for distant pairs (5 to 35 deg separation). These biased interactions also persist with visual stimulation. This suggests that ‘surround’ interactions occur mostly as feedback from higher areas, whereas ‘center’ interactions tend to be feedforward.
Thus we find inherent biases in neuronal interaction during spontaneous conditions that persist during stimulated conditions. We posit that these inherent biases of information flow relate to existing functional organizations in the brain and may predict visual function at low contrasts.
Support: Packard Foundation, NIH