Influence of perceptual similarity on induction
Previous studies have used varying levels of perceptual similarity between target and test items, and this could be a key factor behind their contradictory findings. For example, Sloutsky and col- leagues (2007) used an explicit manipulation of appearance similarity, such that the perceptual (dis- tractor) choice had the same overall appearance as the target. In contrast, Gelman and Markman (1986) selected real biological kind examples (e.g., Squirrel 1, Squirrel 2, rabbit). Although these were selected so that the overall appearance similarity between target and distractor was greater than that between target and category choice, the salience of the perceptual distractor will have been much less than in Sloutsky and colleagues’ study, allowing more children to successfully make the category choice. Thus, a comparison of these studies reveals a pattern consistent with Osherson and colleagues (1990) and Sloman’s (1993) work with adults. Specifically, fewer inductive category choices were made when perceptual similarity was higher (as in Sloutsky et al., 2007), and more inductive category choices were made when perceptual similarity was lower (as in Gelman & Markman, 1986). This crude comparison demonstrates how the choice of stimuli can affect the apparent inductive preference choice, and ideally we need to compare stimuli at different levels of similarity within the same study. This speculation is supported by Sloutsky and Fisher (2004), who used the natural kind stimuli from Gelman and Markman’s (1986) research to examine the similarity of test items to target items. They concluded that perceptual similarity and linguistic labels make significant contributions to children’s inductive choices, which can sometimes be interpreted as a category choice, as in Gelman and Mark- man’s study. In Experiment 1a, we specifically investigated whether different levels of perceptual sim- ilarity between the target and distractor item could encourage different inductive preferences. Our findings demonstrated a clear consistent influence of perceptual similarity on induction decisions. Specifically, when children were presented with HSDs, which matched the target on all features