Pigeons' rate of learning a two-color oddity task increased as a function of the number of incorrect alternatives from 2 to 24 in Experiments 1, 2, and 3. In general, pigeons that were transferred from many-incorrect-alternative to two-incorrect-alternative oddity performed better than controls, but considerably below baseline (Experiments 2 and 3). In Experiment 4, pigeons showed no unconditioned tendency to peck the odd stimulus among 24 incorect alternatives, when pecks were nondifferentially reinforced, and in Experiment 5, when this procedure was preceded by oddity training, a progressive drop in odd-stimulus pecking was found. In Experiment 6, pigeons exposed to a nine-stimulus array in which the odd stimulus appeared (a) in the center or (b) separate from the array learned faster than when the odd stimulus was at the edge. This outcome suggests ththe figure-ground relation between the odd stimulus and the incorrect alternatives plays a role in the facilitation produced by increasing the number of incorrect alternatives but that poor performance on the standard, three-alternative oddity task appears to be due to center-odd trials which provide a difficult size or number discrimination.