Principal-component analysis of macaque lateral geniculate nucleus chromatic data.

Principal-component analysis was used to find a set of orthonormal basis vectors that could span the chromatic space of neurons in the parvocellular layers of the macaque monkey lateral geniculate nucleus (LGN). A total of 441 spectral curves measured on 147 cells from a well-known study [R. L. De Valois, I. Abramov, and G. H. Jacobs, J. Opt. Soc. Am. 56, 966 (1966)] were used as the input data. The principal-component technique, in a completely objective manner, showed that 93% of the variance in the data could be accounted for by only three eigenvectors. An orthogonal Varimax rotation of the first two eigenvectors matched the mean spectra of the previously identified GR and YB opponent cell groups. A different rotation gave a photopic luminosity curve.