Treatments that attenuate latent inhibition (U) were examined using conditioned suppression in rats. In Experiment 1, retarded conditioned responding was produced by nonreinforced exposure to the CS prior to the CS-US pairings used to assess retardation (Le., conventional U). In Experiment la, retarded conditioned responding was induced by preexposure to pairings of the CS and a weak US prior to retardation-test pairings of the CS with a strong US (Le., Hall-Pearce [1979] U). Both types of U were attenuated by extensive exposure to the training context (Le., context extinction) following the CS-US pairings of the retardation test. Experiment 2 examined the specificity of the attenuated U effect observed in Experiment 1. After preexposure to two different CSs in two different contexts, each CS was paired with a US in its respective preexposure context. One of the two contexts was then extinguished. This attenuated U to a greater degree for the CS that had been trained in the extinguished context. Experiment 3 differentiated the roles in 11 of CS-context associations and context-US associations. Following preexposure to the CS in the training context, U was reduced by further exposure to the CS outside the training context. This observation was interpreted as implicating the CS-context association as a factor in U. Thus, the results of these experiments suggest that U is a performance deficit mediated by unusually strong CS-context associations. Implications for Wagner's (1981) SOP model andMatzel's (1988) comparator hypothesis are discussed.Latent inhibition (Ll), also known as the CS-preexposure effect, is a deficit in Pavlovian responding to a reinforced conditioned stimulus (CS) observed when subjects are exposed to nonreinforced presentations of that CS prior to reinforced training (e.g., Lubow, 1973). Numerous theoretical explanations ofLl have been proposed, almost all of which have presumed that Ll results from a failure by subjects to acquire the CS-unconditionedstimulus (US) association (see, e.g., Lubow, Schnur, 8? Rifkin, 1976;Mackintosh, 1975;Pearce & Hall, 1980;Wagner, 1981). Typically, these theories propose that the nonreinforced pretraining exposure to the target CS decreases attentional or memorial processing of the CS, which impairs the subject's ability to associate the CS with a US when these elements are later paired.However, theories explaining Ll as solely the result of impaired acquisition have been challenged by recent findings which indicate that responding to the CS can be increased by various treatments administered after the retardation-test CS-US pairings. These include a "reminder treatment," which consists of the US administered alone outside the training context (Kasprow, Catterson, Schachtman, & Miller, 1984), and testing that is extensively delayed after the CS-US pairings (Kraemer, Randall, & Carbary, 1991). These observations suggest that CS preexposure produces a reversible performance failure, whereby subjects are slow to respond in accord with the novel contingency encount...