Gottschaldt's (
1926) study on embedded figures serves as the foundation for the claim that Gestalt factors limit the impact of experience on perception. The present experiments studied the effects of practice and transfer on the detection of embedded figures. Experiment 1 showed the clear effects of practice that might have been due to recognition. Moreover, it demonstrated the effects of transfer, thus reducing the potential impact of recognition. Experiment 2 replicated practice effects and extended them to other types of part and whole figures. It was conducted to evaluate the specificity of the transfer in more detail. In Experiments 1 and 2, transfer between old and new items only occurred when either the figures and their basic elements were of the same kind or the same search strategies could be used. Thus, perceptual abilities improve by training even when Gestalt factors are at work. Possible sources of these improvements are discussed with special attention to a distinction between figural and procedural aspects.