Grant and Roberts found that houselight presented throughout the delay period on a delayed matching-to-sample task caused pigeons to demonstrate a much lower level of accuracy than was found when the delay was spent in darkness. A series of experiments was carried out to examine possible mechanisms responsible for this light-induced retroactive inhibition (RI). Experiments 1 and 2 were parametric investigations, which varied the length and point of exposure to light within the delay interval. It was found that brief exposure to light produced significant RI, degree of RI increased as the length of exposure increased, and placement of interpolated light at the end of the delay led to greater RI than placement at the beginning (beginning-end effect). Experiments 3 through 5 tested specific hypotheses designed to account for light-induced RI and the beginning-end effect. The perceptual attenuation, perceptual alteration, and nonmemory- pecking hypotheses all offered explanations which held that illumination affects processes active at the time of choice between comparison stimuli but does not modify memory of the sample stimulus. All three of these hypotheses failed to be supported by the results of these experiments. It was concluded that theoretical mechanisms which specify that houselight leads to loss of memory can best account for all of the evidence available on light-induced RI.