Task specificity in nonspecific transfer and in extradimensional stimulus generalization in pigeons.

The theory of general attention predicts that true discrimination (TD) training will produce greater positive non- (stimulus-) specific transfer and sharper extradimensional stimulus generalization gradients than pseudodiscrimination (PD) training, because TD training establishes a set to discriminate (general attentiveness). Previous demonstrations have employed go-no-go (GN) discrimination training. In order to determine whether discrimination training per se is sufficient to produce the effects described, TD and PD groups of pigeons were trained in a successive-conditional (SC) discrimination. In Experiment 1, eight groups received TD or PD training between two line orientations in either an SC or a GN discrimination prior to transfer to either a true SC or a true GN discrimination between two colors. Discriminative performance in transfer was facilitated only when the original and transfer discriminations were of the same type. Experiment 2 replicated four of the conditions of Experiment 1 with procedural changes to make the transfer discrimination identical to that employed in previous "general attention experiments," with the same results. It is concluded that positive nonspecific transfer is primarily mediated by the transfer of task-specific learning rather than by the transfer of attentiveness. In Experiment 3, four groups were trained as in Stage 1 of Experiments 1 and 2, then given single-stimulus (SS) training with a chromatic stimulus, followed by a wavelength stimulus generalization test. Only TD-GN discrimination training produced sharp gradients, an effect leading to the conclusion that go-no-go discrimination training and stimulus generalization are closely related by task requirements rather than by attention. It is concluded that task-appropriate response tendencies probably mediate the effects observed.