“
“The neural encoding of the visual scene involves both linear and nonlinear processing. Linear processing detects image features defined Cisplatin order by spatiotemporal variation in luminance, and is typified by X cells in the retina and lateral geniculate nucleus (LGN) (Enroth-Cugell and Robson, 1966, Hochstein and Shapley, 1976 and So and Shapley, 1979). Nonlinear processing is required to detect non-Fourier image features such as interference patterns, and begins subcortically with Y cells (Demb et al., 2001b and Rosenberg et al.,

2010). Although it has long been established that Y cells respond nonlinearly to visual stimulation (Hochstein and Shapley, 1976), the nonlinear transformation they implement has not been determined. In this study, we ask whether Y cells implement a nonlinear signal processing technique called “demodulation. Demodulation is a nonlinear process used to detect envelope frequencies in interference patterns. For instance, to decode an amplitude-modulated (AM) radio signal created by multiplying a high-frequency carrier by low-frequency envelopes to be

communicated. Because there are no actual signal components at the envelope frequencies, their detection requires a nonlinear transformation of the input which is implemented by a demodulating circuit in the radio receiver. Interference INCB024360 patterns are also found abundantly in natural visual scenes, defining important features such as object contours (Johnson and Baker, 2004, Schofield, 2000 and Song and Baker, 2007). Theoretical work suggests that demodulation could provide an efficient method for encoding visual interference patterns and other non-Fourier image features Bumetanide (Daugman and Downing, 1995 and Fleet and Langley, 1994), but the existence of a neural mechanism for visual demodulation has only been speculated. To determine if LGN Y cells transmit a demodulated visual signal, we examined the temporal pattern of their responses to interference patterns with different carrier temporal frequencies but the same envelope temporal

frequency (TF). Y cell responses to these stimuli were found to be demodulated, oscillating at the envelope (but not the carrier) TF and with the same phase regardless of the carrier TF. To investigate if the demodulated signal transmitted by Y cells is represented in primary visual cortex, we compared the TF tuning properties of LGN Y cells with those of neurons in cortical areas 17 and 18. Like Y cells, area 18 neurons responded to interference patterns across a wide range of carrier TFs. This property could not be accounted for by the output of area 17 which represented a narrow range of low TFs. This suggests that Y cells initiate a distinct pathway that carries a demodulated representation of the visual scene to area 18.