Category bias in similarity ratings: the influence of perceptual and strategic biases in similarity judgments of faces.

Introduction: After category learning, same-category items tend to be rated as more similar than items from different categories. Whether this category bias in similarity ratings reflects true changes in perception or a strategic judgment bias to rate same-category items more similarly has been debated. The current study investigated the influence of perceptual and strategic judgment biases on perceived similarity ratings of face stimuli. Method: To explore the influence of perceptual and strategic biases, post-learning category bias was measured after learning one of two category structures. In a similarity-consistent structure, faces within a category shared physical features and category bias could reflect a combination of strategic bias and true perceptual changes. In a similarity-inconsistent structure, category membership was orthogonal to physical features and category bias could only be driven by strategic bias to rate same-label faces as more similar. Results: We found a strong category bias after learning, but only when category labels could be aligned to the similarity structure. When category label conflicted with similarity structure, the mere presence of a shared label did not create a bias. Discussion: These findings indicate that category bias in this paradigm is largely driven by a perceptual bias, consistent with proposals that category learning can stretch or shrink perceptual space by biasing attention toward category-relevant and away from category-irrelevant features. More broadly, these findings contribute to our understanding of category-driven biases and may inform bias research in other domains such as social stereotypes.