This may thus be considered BMS-754807 ic50 an X cell pathway model for constructing a neural representation of non-Fourier image features. Alternatively, it has been hypothesized that the sensitivity of area 18 neurons to non-Fourier image features originates with a preexisting neural representation created by retinal ganglion Y cells (Demb et al., 2001b and Rosenberg et al., 2010). Critical to this model is that cat area 18 is a primary visual area, receiving substantial input from LGN Y cells (Humphrey et al., 1985 and Stone and Dreher, 1973). This may thus be considered a Y cell pathway model for constructing a neural representation of non-Fourier image features. Here we showed that both Y cells and area 18 neurons represent interference

patterns over a wide range of carrier TFs (at least as high as 25 cyc/s). Importantly, the sensitivity of area 18 neurons to interference patterns with high carrier TFs could not be accounted for by the output of area 17 which represents a narrower range of low TFs (Figure 7). Our findings are thus most consistent with the Y cell pathway model, supporting the hypothesis that the cortical representation

of non-Fourier image features is constructed from Y cell input. The functional advantages of a demodulating nonlinearity in communication and signal processing have been revealed through a variety of engineering applications. The finding that Y cells implement a demodulating DAPT mw nonlinearity helps to draw parallels between Y cell physiology and traditional demodulating circuits and suggests that demodulation can provide the basis for a conceptual framework for understanding the role of the Y cell pathway in visual processing. In this final section, we introduce some implications of a Y cell demodulating nonlinearity. Non-Fourier image

features are defined by high-order correlations describing how different Isotretinoin sinusoidal components in an image come in and out of phase (Klein and Tyler, 1986). This statistical complexity implies a greater computational expense in representing non-Fourier image features than simpler image features defined solely by changes in luminance. It would consequently be more efficient to represent non-Fourier image features after transforming them into a neural representation with less statistical complexity. Demodulation performs this transformation, recoding complex spatiotemporal patterns composed of multiple high-frequency components into a simpler form that represents the lower spatiotemporal scale at which those components covary, the envelope frequency (Figure 3, Figure 4 and Figure 5). Importantly, this transformation preserves the salient image features (the envelope information) and encodes/transmits them more efficiently (Daugman and Downing, 1995). The present results therefore suggest that the Y cell pathway reduces the statistical complexity and improves the efficiency of neural representations of complex visual features.