Alan Yu of the University of Chicago will present a colloquium.
Click to join the Zoom meeting. The passcode is 635410. (The meeting ID is 647 6334 5823.)
Individual differences in categorization gradience: evidence from eye movement and neurophysiological measurements
Categorization is an important and fundamental human cognitive ability which allows a perceiver to treat objects with different physical features as functionally equivalent. A classic demonstration of this ability is the phenomenon of categorical perception, where a continuous sequence of equal physical changes in a stimulus can be perceived and grouped as two different categories of stimuli. Within the speech domain, however, recent studies documented extensive individual variations in speech categorization, even among typically-developed individuals.
In this talk, I report the results of two recent experiments. The first experiment investigated the relationship between categorization gradience and cue weighting across two sets of English contrasts, showing that secondary cue weighting, derived from eye movements in an eye-tracking experiment, correlates with how gradiently the listeners categorize speech sounds. Greater categorization gradience is associated with larger secondary cue weights. The relationship between categorization gradience and secondary cue weighting is observed across different types of cues and contrasts, suggesting that categorization gradience and cue weighting may be consistent speech processing properties at the individual level, even if the specific magnitudes might differ across contrasts. We also found evidence that individuals who integrate secondary cues during processing utilize a buffer processing strategy, suggesting that delayed reaction to an early-arriving cue until all relevant cues are available may facilitate the integration of multiple cues in the signal.
The second study investigated the underlying neuro-mechanisms behind individual variability in categorization gradience, by examining the subcortical and cortical speech-evoked responses. Specifically, we found that individuals differ in how the subcortical and cortical representations diverge from each other; higher redundancy (i.e. higher similarity between the two representations) in speech information relayed from subcortical to cortical level correlates with more gradient categorization, whereas more categorical categorization correlates with more information reduction and abstraction (i.e. decreased similarity between the two representations) along successive processing stages. These studies highlight the importance of understanding individual variability in speech processing and what such variation might mean for speech learning in general.