Results of directed-forgetting research with pigeons are difficult to interpret because of alternative nonmemorial accounts of performance decrements and important procedural differences from comparable research with humans. Prior research has noted the absence of directed forgetting when artifacts have been removed (e.g., nonreward following forget cues and differences in response patterns on remember and forget trials in training). In this article, it is argued that, in human directed-forgetting research, presentation of a forget cue allows for the reallocation of memory maintenance to items to be remembered. In the present experiment, true directed forgetting is found when nonmemorial performance decrements are eliminated and forget cues allow for the reallocation of sample memory to test-relevant cues.Evidence from human memory research indicates that humans can exert active control over the maintenance of memories. In studies involving "directed-forgetting" procedures, in which presentation of a list of items includes signals to remember some of those items but not others, humans show better memory for to-be-remembered items than for items that are signaled to be forgotten (e.g., Bjork, 1972;Epstein, 1972). A possibly related phenomenon has been reported in pigeons using a variation on a task that has been used extensively in research on animal memory, delayed matching-to-sample (DMTS; see Blough, 1959). This task typically involves training pigeons to peck a sample stimulus (e.g., a red or a green hue) and then to choose, after a delay, between two comparison stimuli (e.g., red and green). A response to the stimulus that matches the sample is reinforced. The finding that pigeons perform less accurately as the delay interval between the offset of the sample and onset of the comparisons is lengthened (e.g., Blough, 1959) suggests that pigeons forget the identity ofthe sample stimulus over time. More recently, a version of DMTS has been adapted to study directed forgetting in pigeons (e.g., Grant, 1981;Maki & Hegvik, 1980;Maki, Olson, & Rego, 1981;Santi & Savich, 1985) and other nonverbal species (Grant, 1982;Roberts, Mazmanian, & Kraemer, 1984). In this task, a conditional cue is presented during the delay between sample and comparisons indicating whether or not memory for the Preparation ofthis paper was supported by National Science Foundation Grant BNS 9019080 and National Institute of Mental Health Grant I ROI MH45979 to the third author. We thank Lou M. Sherburne for her assistance. Correspondence concerning this article should be addressed to K. L. Roper, who is now at Department of Psychology, University of Wisconsin, Eau Claire, WI 54702.-Accepted by previous editor, Vincent M. LoLordo sample will be tested on that trial. Following the remember (R) cue, the trial terminates in the standard two-choice comparison test. Following the forget (F) cue, memory for the sample is not tested (e.g., the trial ends without presentation of comparisons). After substantial experience with the R and F cues, one can ...