Abstract
By watching and responding to the way a shill answered “yes-no” questions about food items, a developmentally delayed preschool boy not only greatly improved over his poor baseline “yes-no” answers to these same items, but also to “yes-no” answers during generalization probe sessions to untrained objects that included picture cards and body parts. Correct “yes-no” answers during follow-up sessions was also high for both the trained and untrained objects. The suggested mechanism for improvement was the elaborated from of answers the shill issues, which served to highlight and prompt “yes” versus “no” answers. Generalized “yes-no” answers improved as the training components (removing the shill and prior labeling questions) approximated the form of the generalized probe sessions. Experiment II confirmed that as soon as the labels for new objects were taught, “yes-no” answers to all labeled objects was immediately perfect and remained so thereafter. To nonlabled objects, “yes-no” answers were initially at chance level, but these answers for certain nonlabeled objects subsequently improved and their names were learned without explicit training, apparently through prior correct “yes-no” responding.