Making decisions about modeling decision-making when analyzing speech perception data

Abstract

Speech perception researchers are typically interested in, well, perception. But perception is not directly observable. An experimental participant hears (and maybe sees) a stimulus, but the researcher only observes the listener’s response (e.g., which button is pushed, how long it takes to push the button). Crucially, a listener’s response doesn’t directly, or solely, reflect perceptual processing; listeners’ responses also reflect what we might call response selection or decision-making. While there is a sizable sub-field in psychology devoted to the study of judgment and decision making, these issues are of more peripheral concern for speech perception researchers. Nonetheless, I will argue that it is important to take response selection into account when analyzing perception data. Terry Nearey is one of a small number of speech researchers who has consistently taken decision-making seriously when analyzing speech perception data. I will discuss a number of mathematical approaches to simultaneously modeling perception and response selection and how changes in modeled perception and decision-making predict distinct observable patterns in data. The importance of decision-making in the study of perception will be discussed and illustrated via a number of Terry Nearey’s studies as well as my own.