Birds were trained on a higher order conditional discrimination task, one that required birds to match samples and comparisons on some trials and to mismatch on other trials. Which task component was in effect was indicated by the level of chamber illumination (houselight-on and houselight-off instructions). Acquisition of the components of a color (red and green) matchlmismatch task in the first half of the experiment was not differentially affected by the level of illumination associated with each task component, by houselight changes per se, or by the level of illumination during the intertrial interval (lTI). However, when shapes (plus and circle) were used to train the task in the second half of the experiment, performance on the houselight-eued task component exceeded performance on the dark-cued task component, and ITI illumination facilitated performance on both task components. These results suggest that attention to shape stimuli, but not to colors, may vary systematically as a function of chamber illumination level.A common finding in visual discrimination research with pigeons is that discriminations between color stimuli are easier to learn thandiscriminations between line orientations (Carter & Eckerman, 1975) or geometric forms (Cumming & Berryman, 1965;Farthing & Opuda, 1974).Faster acquisition of tasks that require color discriminations, rather than form discriminations, has been obtained both in simple discrimination tasks (Carter & Eckerman, 1975) and in complex conditional discrimination tasks, such as matching-to-sample and symbolic matching (Carter & Werner, 1978). Differences in discriminative performance that can be attributed to stimulus dimension are not confined to task acquisition. Studies of delayed matching-to-sample have consistently found that memory for color samples is better than memory for form samples (e.g., Cook, 1980; Farthing, Wagner, Gilmour, & Waxman, 1977). Other studies have alsoshown that when color and form stimuli are combined to produce stimulus compounds, pigeons may selectively attend to the color element of the compound (e.g., Born & Peterson, 1969; Seraganian, 1979).The reason for this difference in the discriminability of colors and shapes is not clear. However, we recently reported the results of a study that suggested that the discriminability of forms, but not colors, may vary systemThe authors wish to thank Pamela Jackson for her assistance in running subjects and Marcia Spetch for her helpful comments on an earlier version of the manuscript. Requests for reprints may be addressed to the first author, who is now at the Department of Psychology, University of Texas at EI Paso, El Paso, TX 79968.atically as a function of the level of chamber illumination (Edwards, Miller, & Zentall, 1985). Birds were trained on a higher order conditional discrimination task, one that was equivalent to combining a matching-to-sample task with an oddity-from-sample task. Thus, birds were required to match sample and comparison stimuli on some trials and to mismatch the sa...