Abstract
Visual evoked potential (VEP) contrast response functions (CRFs) frequently manifest two linear limbs on log-contrast axes, a shallow limb at low contrasts, transitioning into a steep limb at high contrasts, implying contribution of multiple mechanisms. These could reflect magnocellular (M) and parvocellular (P) dominated responses, respectively (Nakayama & Mackeben, 1982; Bobak et al., 1984; Tyler & Apkarian, 1985; Souza et al., 2007, 2008). Preliminary principal component analyses of individual differences in sweep VEPs (sVEP) supported this hypothesis (Hamer, Souza et al. 2013). To further elucidate the underlying mechanisms, we apply factor analysis to CRFs obtained with the efficient sweep VEP (sVEP) recorded from 9 subjects viewing 0.4, 2, 10 or 16 cycles/degree achromatic sinewave gratings (161 cd/m2 mean luminance), phase-reversing at 4.29, 6, 10 or 15 Hz. In each 10-sec sVEP trial, grating contrast increased logarithmically from 1.4 to 90%. A vector average Discrete Fourier Transform amplitude (without adaptive filtering, to prevent correlations across contrast) at the second harmonic was calculated for 10 contrast sweeps per spatiotemporal condition. We estimated underlying factors in log-transformed amplitudes (to normalize distributions), examining covariances, and rotated (Varimax) the principal components to approximate "simple structure" and maximize the number of zero or non-zero loadings. Our analysis revealed two factors that accounted for the 74.7% of the data's variance: Factor 1 (FC1) scores increased monotonically with contrast, loading more onto high contrasts. FC2 decreased with contrast, loading more onto low contrasts, the two crossing between 10-20% contrast. These results were similar to prior findings (Hamer et al., 2013; Peterzell, Bubl & Bach, 2015). FC1 (high-contrast) scores were higher at low temporal frequencies (TFs) across spatial (SFs). FC2 (low-contrast) scores were highest at 2 c/deg for all TFs. The results will be discussed in relation to potential M- and P-contributions to the sVEP.
Meeting abstract presented at VSS 2016