Rats received Pavlovian positive patterning training in which a simultaneous tone + light compound was reinforced with food and the light and tone elements were separately nonreinforced. Solution of the discrimination was accompanied by the emergence of a unique response topography in the presence of the compound. Control rats that received reinforced compound presentations but no separate nonreinforced presentations of the elements acquired behaviors characteristic of light and tone stimuli separately paired with food, but did not acquire the unique response topography acquired by the rats that received positive patterning discrimination training. These data supported earlier claims that the solution of such discriminations involves the conditioning of a stimulus unique to the compound stimulus.Most theories of learning assume that the associative strength of a compound stimulus is equal to the summed associative strengths of its constituent elements. This simple assumption has permitted the successful prediction of a number of common conditioning phenomena such as summation, blocking, overshadowing, and conditioned inhibition.However, the results of a variety of experiments show that under some circumstances animals may treat a stimulus compound and its constituent elements very differently. For example, animals can be taught to respond to a consistently reinforced compound stimulus (AB) but to withhold responding to the separately nonreinforced elements (A and B) of that compound, or to respond to the separately reinforced elements A and B but to withhold responding to the nonreinforced compound AB (e.g., Forbes & Holland, 1980;Rescorla, 1972;Woodbury, 1943). Similarly, there have been several recent reports (e.g., Forbes, 1981; Gillette & Bellingham, 1982) that extended nondifferential experience with stimulus compounds may also result in progressive behavioral differentiation of the compound and its separate elements. All of these findings seem inconsistent with the simple summation notion.One approach to reconciling these types of data with a simple summation notion is to assume the existence of a compound-specific or unique stimulus in addition to the explicit stimulus elements. That is, an AB compound could be described as comprising three cues, A, B, and a cue unique to their combination. Rescorla (1972Rescorla ( , 1973 noted that such an approach, coupled with the power of recent conditioning theories, allows us to account for much of the This research was supported in part by U.S. Public Health Service Grant MH31396. Reprints are not available. The authors' mailing address is: Department of Psychology, University of Pittsburgh, Pittsburgh, Pennsylvania 15260. 297 available data that seem inconsistent with a simple summation notion. In a series of experiments in which the relative validities of the explicit stimulus elements and of the implicit unique cue were manipulated, Rescorla (1973) found considerable evidence that the properties of such implicit cues paralleled those of explicit stimuli. For ex...