The relative contributions of predictiveness and salience in flavor aversion learning

Abstract

Two experiments investigated the relative influence of cue predictiveness and stimulus salience on flavor aversion learning. In the first study experimental subjects experienced a nonsalient strawberry flavor and a different, more salient flavor prior to illness on each of the first three conditioning trials; the nonsalient strawberry and salient lemon were paired with illness on the fourth trial. This treatment resulted in a strong aversion to the strawberry flavor and on aversion to the lemon flavor. In contrast, animals which had experienced only a single pairing of strawberry, lemon, and illness acquired a strong aversion to lemon and no aversion to strawberry, while animals which had experienced only the pairing of different flavors and illness on each of the first three trials developed a strong aversion to both strawberry and lemon. Experiment 2 replicated the results of the experimental condition of the first study but also included a “blocking” condition in which subjects received three strawberry and illness pairings prior to a strawberry-lemon-illness pairing. Subjects in the blocking condition developed only a weak aversion to strawberry, and this aversion did not block the development of a strong lemon aversion. Apparently, in order for a nonsalient flavor cue to block the acquisition of an aversion to a more salient cue, the nonsalient flavor must not only be paired with illness but also be more predictive of illness than the more salient flavors also paired with illness. The data were discussed in terms of the R. A. Rescorla and A. R. Wagner's associative model (1972, in Classical conditioning II: Current research and theory, New York: Appleton-Century-Crofts), R. A. Rescorla's catalytic model (1982, Journal of Experimental Psychology: Animal Behavior Processes, 8, 131–141), and N. J. Mackintosh's attentional model (1975, Psychological Review, 82, 276–298) of conditioning.