Wefirst trained pigeons to peck one button (same) after the successive presentation of 16identi-cal pictures and to peck a second button (different) after the successive presentation of 16 nonidentical pictures. Later, we tested the birds with other lists of same and different items composed of completely novel pictures. Accuracy to the testing lists reliably exceeded chance levels, thus demonstrating same-different conceptualization by pigeons under conditions that, for the first time, (1) eliminated purely perceptual mechanisms of discrimination learning and transfer and (2) required memory-based processing of the experimental stimuli.The search for abstract conceptualization in nonhuman animals has been a protracted and arduous one that has spanned a full century of research in animal intelligence (Wasserman, 1993). Different methods of investigation and species of animals have been used in this search, with decidedly mixed results. With only a very few combinations of tasks and species has any positive evidence of abstract conceptual behavior been obtained; most other combinations of tasks and species have yielded negative evidence (for reviews see Delius, 1994;Herrnstein, 1990;Thompson, 1995).There is now growing awareness that most laboratory investigations of conceptual behavior in nonhuman animals have probably underestimated the subjects' cognitive capacities (see Wright, 1992, andZentall, 1993, for more on this issue). One problem has been the use of too few experimental tasks (e.g., matching to sample); another has been the use of too few and too simple stimuli. Any of these limiting factors might lead researchers to arrive at an inappropriately pessimistic view of animal cognition.Wasserman, Hugart, and Kirkpatrick-Steger (1995) recently reported an experiment that used a still-uncommon behavioral task (same-different discrimination training) and a large number of highly complex visual items (Macintosh computer icons) in the hope of obtaining more encouraging data on the conceptual abilities of a familiar laboratory animal-the pigeon. Their results provided supportive evidence that the pigeon can learn a general same-different concept that transcends the particular stimuli with which it was originally trained. Specifically, pigeons were first taught to peck one button (same) when they viewed an array of computer icons that comprised 16 copies of the same icon and to peck a second button (different) when they viewed an array that comprised 16 distinctly different icons. These same and different training displays were all created from one set of computer icons. The birds were later tested with novel same and novel different displays that were created from a second set of computer icons. Accuracy to the training stimuli averaged 83% correct, and accuracy to the testing stimuli averaged 71% correct, in each case reliably exceeding the chance score of 50%.That experiment used arrays of simultaneously visible computer icons that occupied a sizable area of the computer monitor. Therefore, it is possible tha...