Single-unit recordings in macaque V1 have identified two populations of color selective cells. One contains neurons that prefer low spatial frequencies and respond best to red-green (L-M cone contrast). Another contains neurons that prefer higher spatial frequencies; some of these cells respond best to L-M, but most respond best to "non-cardinal" directions (e.g., M cone contrast). The goal of this study was to identify perceptual mechanisms consistent with these two sets of neurons. We used a selective adaptation procedure with two types of adapting stimuli targeted to each set of neurons. Subjects adapted to Gaussian blobs and 2 cyc/deg Gabor patterns, each containing L-M or M cone contrast. After viewing a 3 deg adapting stimulus in one hemifield, subjects adjusted the contrast of a stimulus in the unadapted field to match the appearance of a test stimulus (either L-M, M, L or L+M cone contrast) presented in the adapted field. Adapting to Gabor L-M patterns reduced the apparent contrast of L-M tests more than the other tests. Similarly, adapting to a Gabor M pattern reduced the apparent contrast of M tests more than the others. Adapting to a Gaussian L-M pattern reduced the apparent contrast of the L-M test most, but, critically, adapting to a Gaussian M pattern reduced the apparent contrast of M and L-M patterns to an equal degree. Selective adaptation is most frequently interpreted as reduced responsiveness of a mechanism that prefers the most strongly affected test. Accordingly, our results indicate that adapting to Gabor stimuli can reduce the responsiveness of mechanisms that prefer non-cardinal directions. Adapting to Gaussian stimuli mainly reduces the responsiveness of mechanisms that prefer L-M, but does not appear to greatly affect non-cardinal mechanisms. Our data suggest that low and high spatial frequency patterns are encoded by two separable populations of color selective neurons, both of which contribute to color appearance.