1. This paper represents a continuation of our effort to examine the relationship between the physiology of distinct classes of primate lateral geniculate nucleus (LGN) cells and spatial vision. The present study focuses on modeling the contrast-sensitivity functions (CSFs) of separate LGN cell classes, examining differences in the CSFs of different classes of LGN cells and comparing the results with behaviorally defined CSFs. 2. CSFs to drifting sinusoidal gratings were obtained from single LGN relay cells in the nocturnal primate, Galago crassicaudatus. The CSFs of 14 X-like, 27 Y-like, and 6 W-like cells with standard center-surround organization were well fit by a difference of Gaussians (DOG) model with small residual errors (mean error per data point ± SEM = 0.008 ± 0.002). The larger residual errors shown by a few of the Y-like cells were not due to nonlinearity of spatial summation. 3. The CSFs of eight cells that appeared to have nonstandard center-surround organization (primarily, a silent, suppressive surround) were also well fit by the DOG model. 4. The DOG curves that best fitted the data differed considerably between the three groups. As a group, X-like cells had a small center mechanism (R(c) = 0.19°) with high sensitivity (K(c) = 76.53) and a small, sensitive surround (R(s) = 0.71°; K(s) = 5.50). These parameters produced CSFs with high cutoff frequencies (V(cutoff) = 2.5 c/deg) and low peak sensitivities (CS(pk) = 6.1) that occurred at 0.8 c/deg. 5. Y-like cells had a large center mechanism (R(c) = 0.46°) with low sensitivity (K(c) = 21.16) and a large, insensitive surround (R(s) = 2.38°; K(s) = 0.81). These parameters produced CSFs with lower cutoff frequencies (V(cutoff) = 1.2 c/deg) and higher peak sensitivities (CS(pk) = 12.5) that occurred at 0.2 c/deg. 6. The few W-like cells that responded to gratings well enough to determine a CSF were quite variable. As a group they had a large center mechanism (R(c) = 0.38°) with intermediate sensitivity (K(c) = 34.55) and a surround with intermediate size and sensitivity (R(s) = 1.59°; K(s) = 1.59). These produced CSFs with intermediate cutoffs (V(cutoff = 1.6 c/deg) and low peak sensitivities (CS(pk) = 5.0) occuring at 0.4 c/deg. 7. In a stepwise discriminant analysis, the parameters: peak spatial frequency (V(pk)), cutoff frequency (V(cutoff)), amount of low spatial frequency rolloff (CS(dec)) and peak sensitivity (CS(pk)), produced distinct W-, X-, and Y-like clusters when combined with other receptive-field properties shown previously to produce clear W-, X-, and Y-like groups. Thus the present CSF data support our previous conclusion that W-, X-, and Y-like LGN cells in this primate constitute distinct visual information channels each exhibiting a different profile of responses to visual stimuli. 8. Comparison between behaviorally derived CSFs and CSFs derived from individual cells show that the behavioral curve matches reasonably well the CSFs of cells from all three classes.
