This finding is consistent with a demodulating system. Theoretical work has shown that a demodulating nonlinearity will detect a variety of non-Fourier image features including illusory contours (ICs) (Daugman and Downing, 1995 and Fleet and

Langley, 1994). By extension, our finding that Y cells demodulate interference patterns led us to hypothesize that GSK126 mw they will respond to other non-Fourier image features as well. To test this, abutting grating stimuli that produce ICs detected by some neurons in the primary visual cortex of cats and monkeys were drifted across the receptive fields of three LGN Y cells (Grosof et al., 1993 and Song and Baker, 2007; Figure S4A). Importantly, the spatial parameters of the stimuli were tailored

to the individual Ixazomib datasheet Y cells to ensure that only nonlinear responses could be elicited. Specifically, the carrier SF was selected to be above the linear passband of the neuron’s drifting grating SF tuning curve and near the nonlinear SF preference measured using contrast-reversing gratings. The ICs were also constrained to be oriented orthogonally to the carrier to ensure that spatial harmonics in the stimulus did not fall within the linear passband of the cell. Even with the small sample size, the result of this experiment was clear: the responses of all three Y cells oscillated at the frequency of ICs/sec, indicating that the ICs were detected (Figures S4B and S4C). Responses at this frequency are consistent with the output

of a demodulating system and cannot be explained by linear processing since a linear response would oscillate at half this frequency. This result suggests that by demodulating visual signals, Y cells may encode a variety of complex image features. Because the amplitude of Y cell responses to interference patterns depends on both the envelope TF (Rosenberg et al., 2010) and the carrier TF (Figure 2), we next wanted to compare the representations of envelope and carrier TF based on response amplitude. Envelope TF tuning curves were measured with a static carrier for 30 Y cells. These tuning curves were well-described by gamma functions (average oxyclozanide r = 0.94 ± 0.04 SD) which were used to estimate the tuning properties summarized in Table 1. For 24 of these Y cells, we also measured a carrier TF tuning curve that was well-described by a gamma function. The envelope and carrier TF tuning curves of a Y cell along with a population scatter plot of the peak envelope TFs and peak carrier TFs are shown in Figures S5A and S5B. Whereas the peak envelope TFs of these 24 Y cells were narrowly distributed around a low frequency (4.2 cyc/s ± 1.2 SD), the peak carrier TFs were widely distributed around a higher frequency (7.5 cyc/s ± 6.8 SD). The distributions of peak envelope TFs and peak carrier TFs were significantly different (p = 0.005, Mann-Whitney U test), and there was a moderate but nonsignificant correlation between them (r = 0.36, p = 0.08).