Abstract
This paper summarizes a decade of human evaluative flavor-taste conditioning research. After a neutral fruit-flavor A (CS+) has been presented repeatedly in compound with bad-tasting Tween20 (US), participants typically develop a dislike for flavor A relative to a flavor B (CS-) that has been presented equally often in plain water. Likewise, differential pairings of neutral flavors with liked sugar (US) have been found to result in an enhanced liking for the CS+ flavor. Whereas the flavor-sugar effect is rather fragile, the flavor-Tween effect is very robust, allowing for substantial parametric variation in acquisition and lest conditions. Evaluative flavor-taste learning does not require contingency awareness, is resistant to extinction, is not affected by CS-US contingency manipulations, and is not sensitive to modulation, suggesting that it is an instance of merely referential learning. Finally, the finding that evaluative flavor-taste learning is not sensitive to post-conditioning US inflation indicates that only the affective characteristics of the US are encoded, not its sensory-descriptive properties.