Prediction in Language Processing: Some Ideas About How It’s Done
Tamara Swaab, Ph.D.
Professor, Department of Psychology and
Center for Mind and Brain
University of California, Davis
Title of Talk: Prediction in Language Processing: Some Ideas About How It’s Done.
Rooted in century old ideas of Kant, recent theories of perception, cognition, language, and neuroscience propose that active generation of top-down predictions of imminent sensory input guide our perceptual experiences. These predictions can be based on long-term knowledge representations stored in memory, experience based probabilistic constraints and immediate contextual constraints. Several prominent contemporary approaches to language interpretation assign significant impact to predictive processing during language interpretation. According to these accounts, comprehenders engage in a set of processes by which they can predict and pre-activate imminent bits of the linguistic input. Some models suggest that predictive processing during language comprehension happens reflexively and under all circumstances. I will present the results of a series of studies from our lab that suggest that comprehenders do not automatically and uniformly predict, but instead flexibly adapt predictive processing depending on its utility in a given processing environment.