Latent inhibition refers to the observation, made in both human and nonhuman animals, that learning about the relationship between a stimulus and an outcome progresses more rapidly when the stimulus is novel compared to when the stimulus has been rendered familiar by preexposure. Three experiments with human participants show that this effect can be reversed to reveal faster learning about a familiar than a novel stimulus, by manipulating the novelty/familiarity of the experimental context. In each experiment, during Stage 1, a preexposed stimulus was rendered familiar by being repeatedly presented within a stream of distractor letters that constituted the experimental context. In a subsequent training stage, participants were required to respond to a target outcome that was preceded by the familiar stimulus on some trials and a novel stimulus on others. These trials were also presented within a stream of contextual distractor stimuli. The results showed that during the training stage, learning about the familiar stimulus proceeded more successfully than the novel stimulus when the distractor stimuli sustained novelty during training (Experiments 1-3), but that this effect could be reverted to latent inhibition when the distractor stimuli sustained familiarity during training (Experiments 2 and 3). The results are in keeping with a novelty-mismatch analysis of latent inhibition, and a novelty-mediated generalization explanation of the results is proposed. (PsycInfo Database Record (c) 2024 APA, all rights reserved).