March 26, 2019
Myhal Centre, Room 370
Engineering in reverse: How AI and humanoid robotics help reveal how real humans process real language in real time
Spoken language interfaces are an increasingly common element in advanced technologies designed for use in education, companionship, and elder care. To what degree do these interfaces align with our current understanding of the perceptual and cognitive processes underlying real-time language processing?
Although the processing of language by humans might at first glance be thought to involve a series of highly automatized mental operations, the past two decades of psycholinguistic research have highlighted important ways in which core aspects of processing are influenced by social cognition. As speech unfolds at the millisecond level, listeners draw on various sources of information to generate and dynamically update implicit assumptions about a conversational partner’s intentions, and use this knowledge to guide their analysis of the acoustic input.
In this seminar, I will describe the bidirectional relationship between this “basic science” work and engineering efforts in artificial intelligence. Although experimental studies of situated communication can provide guidance for the design of automated systems, engineering quandaries and solutions often shine a spotlight on issues that have been largely overlooked in psychology, in turn stimulating new lines of empirical research. I will illustrate these interconnections in a series of studies exploring the processing of task-related instructions in contexts involving children, younger and older adults, and social robots.
Craig Chambers is a faculty member in Psychology and directs the Perception, Action, and Language laboratory at UTM. He holds an MA in Linguistics from the University of Toronto and a PhD in Brain and Cognitive Sciences from the University of Rochester.
Please RSVP before noon Monday 25 March 2019 (or until full) using the below link: