VSS 2014 Winners > Bonnie Cooper
Spatiotemporal properties of macaque retinal ganglion cells: an harmonic analysis and relationships to psychophysical data
Bonnie Cooper1, Barry Lee1,2; 1Department of Biological Sciences, College of Optometry, SUNY, 2Max Plank Institute for Biophysical Chemistry
Here we analyze the linearity of Magnocellular and Parvocellular Retinal Ganglion Cell (MC & PC RGCs) responses to complex waveforms (square-waves and ramps) and consider the responses in the context of psychophysical sensitivity. If linearity holds, then the responses to other waveforms should be predictable from the responses to sine-waves; Campbell & Robson (1968) tested this psychophysically. However, Campbell & Robson assumed that noise is similar across spatial frequency and did not consider temporal response properties.
Here we test these assumptions with RGC responses. First we consider linearity; responses of RGCs to sine and complex waveforms were collected at 2Hz across a range of spatial frequencies and multiple contrast values. To predict the amplitude of complex waveform harmonic components, SF spectra for sine-wave responses were fit (DoG for MC and SoG for PC RGCs) and scaled to account for spatial frequency and the coefficients of the square-wave or ramp fourier expansion. Fits to fundamental harmonics were satisfactory, however, the amplitude of higher odd harmonics was better accounted for with additional scaling for the temporal response properties of RGCs. Such effects will happen with eye movements, and we then consider cell responses in the context of a Spatiotemporal surface as proposed by Kelly (1979) for psychophysical data. These findings suggest substantial linearity and spatiotemporal separability of RGC responses to complex waveforms.
Secondly we consider variance; noise in RGC responses to sinusoidal modulation increases with temporal frequency (Sun et al., 2004). Here, we extend this analysis to higher harmonic square-wave response components. Noise increased in higher harmonic response components i.e., is not invariant. However, the increased response associated with higher temporal frequencies compensates for this. This has implications for the way cortical mechanisms process (summate) RGC responses to produce psychophysically observed effects.