Exemplar-based accounts of "multiple-system" phenomena in perceptual categorization

No prior work had fully resolved whether human classification relies on distinct cognitive architectures or a unified mechanism. Researchers often propose that rule-based and prototype-based behaviors stem from separate mental systems. This gap motivated extensive debate regarding the necessity of multiple-system frameworks in perceptual categorization. Prior research has shown that specific patterns of generalization often appear to support these dual-process theories. However, the assumption that these observed behaviors require independent cognitive modules remains controversial. That uncertainty drove the current investigation into alternative explanations for these complex phenomena. The authors examine if a single-system approach can account for data previously attributed to multiple systems. This study addresses the theoretical tension between modular and unified models of human cognition.

The aim of this study is to demonstrate that a single-system exemplar-similarity model can account for diverse perceptual classification phenomena. Researchers often attribute these patterns to multiple-system models, which assume distinct cognitive architectures for different tasks. This work challenges that assumption by providing an alternative interpretation of rule-described and prototype-described behaviors. The authors investigate whether context-dependent similarity can explain complex dissociations between categorization and recognition. They seek to show that similarity is not an invariant relation but one that changes systematically. The study explores how selective attention to dimensions influences these similarity relations. Furthermore, the researchers examine how sensitivity settings modulate the relationship between judged similarity and psychological distance. This effort aims to simplify the theoretical landscape of perceptual categorization by promoting a unified model.

The authors employ a computational modeling approach to re-evaluate existing experimental data. This review approach focuses on synthesizing findings from various rule-described and prototype-described classification studies. The researchers construct a unified framework based on the principle of context-dependent similarity. They systematically test whether this single-system architecture can replicate complex patterns of generalization. The investigation utilizes mathematical simulations to demonstrate how selective attention influences cognitive outcomes. By adjusting sensitivity parameters, the team explores how distance in psychological space affects categorization performance. The study compares these simulated results against established empirical observations from the literature. This analytical strategy allows for a direct assessment of the single-system model against competing multiple-system theories.

Key findings from the literature indicate that a single-system model successfully accounts for a wide variety of classification phenomena. The authors demonstrate that rule-described and prototype-described patterns of generalization emerge from this unified framework. Their simulations replicate specific dissociations between categorization and similarity judgments observed in previous research. The model also accounts for documented dissociations between categorization and old-new recognition tasks. These results suggest that similarity relations change systematically due to selective attention to dimensions. The researchers show that modulation in sensitivity settings effectively explains how individuals judge distances in psychological space. Adaptive learning principles provide a mechanism for these shifts in attention and sensitivity. The analysis confirms that these diverse behaviors do not require the assumption of independent cognitive systems.

The authors propose that a single-system exemplar model effectively accounts for diverse classification phenomena. This synthesis suggests that multiple-system frameworks may be unnecessary to explain rule-described and prototype-described patterns. The researchers argue that context-dependent similarity provides a robust alternative to complex modular architectures. Their findings imply that selective attention and sensitivity adjustments drive the observed systematic changes in similarity relations. The study highlights that adaptive learning principles offer a parsimonious explanation for these cognitive processes. By reinterpreting existing data, the authors demonstrate that unified models possess greater explanatory power than previously acknowledged. The implications suggest a shift toward simpler, more flexible cognitive theories in perceptual categorization research. This work provides a foundation for future investigations into the mechanisms of human similarity judgments.