Choice behavior under differential outcomes: Sample stimulus control versus expectancy control

It has been hypothesized that in the differential outcomes (DO) procedure, each discriminative stimulus comes to evoke a specific expectancy or representation of its unique reward and that expectancy exerts “stimulus” control over choice behavior in the same manner as a discriminative stimulus. It has also been suggested that expectancy control may reduce—or even replace—control by the discriminative stimulus under DO, and that under DO, subjects may show considerably more ability to choose a correct response on the basis of expectancy than on the basis of the discriminative stimulus. The present experiment, using DO in the delayed matching-to-sample procedure with pigeons, demonstrates that subjects under DO show more ability to match based on expectancy alone than sample alone when direct comparisons are made, and when presented with two choices, one indicated by sample and one indicated by expectancy, subjects typically made the choice indicated by expectancy. The implications of these findings of greater control of choice behavior by expectancies for traditional accounts of behavioral control in choice behavior are discussed.