| Abstract: | The rabbit's nictitating membrane response was classically conditioned to a serial tone-light compound (CS1-CS2), in which the CS1-CS2 and CS2-US intervals were each manipulated over the values of 200, 400, 800, and 1600 msec in a between-groups, factorial design. In addition, at each level of the CS2-US interval, there was a control group that received CS2 and the US. Within the serial compound, the CS2 was found to be highly efficacious in that the CS2-US interval determined the rate of CR acquisition to the compound as a whole and to CS2 in particular. Moreover, CR probability during CS2 largely overlapped the performance of the corresponding single-stimulus controls. Outside the context of the compound, CR frequency on CS2 test trials fell below control levels at the 200- and 400-msec CS2-US intervals. The results can be interpreted as indicating that a generalization decrement or information loss occurred in transferring from compound training to CS2 testing. In addition, the theories of Rescorla and Wagner (In A. Black & W. F. Prokasy, Eds., Classical conditioning II, New York: Appleton-Century-Crofts, 1972) and Mackintosh (Psychological Review, 1975, 82, 276–298) may be extended to account for the present results. |
