Perceptual dimensions influence auditory category learning

Abstract

Human category learning appears to be supported by dual learning systems. Previous research indicates the engagement of distinct neural systems in learning categories that require selective attention to dimensions versus those that require integration across dimensions. This evidence has largely come from studies of learning across perceptually separable visual dimensions, but recent research has applied dual system models to understanding auditory and speech categorization. Since differential engagement of the dual learning systems is closely related to selective attention to input dimensions, it may be important that acoustic dimensions are quite often perceptually integral and difficult to attend to selectively. We investigated this issue across artificial auditory categories defined by center frequency and modulation frequency acoustic dimensions. Learners demonstrated a bias to integrate across the dimensions, rather than to selectively attend, and the bias specifically reflected a positive correlation between the dimensions. Further, we found that the acoustic dimensions did not equivalently contribute to categorization decisions. These results demonstrate the need to reconsider the assumption that the orthogonal input dimensions used in designing an experiment are indeed orthogonal in perceptual space as there are important implications for category learning.

Keywords: Audition; Categorization; Perceptual categorization and identification.

Figure 1.

Stimulus distributions for the four conditions in this study. The black line represents the optimal decision boundary that separates the two categories.

Figure 2.

Block-by-block average normalized accuracy, normalized according to ideal observer accuracy, for all conditions. Ribbon error bars reflect standard error of the mean. Dashed line represents chance accuracy (50%).

Figure 3.

Average generalization test accuracy, normalized based on optimal accuracy for each condition. Dashed line represents chance accuracy (50%). Error bars reflect standard error of the mean around the black dot which represents the mean. Each individual point is an individual participant’s average accuracy.

Figure 4.

Proportion of participants fit by each modeling strategy across all four training blocks and the generalization test. None of the participants were best fit by the Random Responder model, so it is not included in the graph.

Figure 5.

Individual decision boundaries for each participant in generalization test (after all training blocks). The optimal decision boundary for each category is shown as the red dotted line on each plot the x-axis represents the Center Frequency dimension, and the y-axis represents the Modulation Frequency dimension.

Figure 6.

Box plots of absolute value difference in participants’ best fit decision bound angles relative to the optimal decision boundary. The optimal decision boundary angle is listed for each condition next to its name and is represented by the dashed line at 0. Each dot represents an individual participant value.