Event Date: Friday, 10 November, 2023
Location: 11:00-13:00, Sala Riunioni, Pal. Venera (II piano)
Speaker: Prof. Arthur Glenberg (Arizona State University)
Title:Â Language comprehension requires embodied processing
Abstract:Â In the first part of this talk, I discuss whether Large Language Models, such as ChatGPT, understand language the way humans do. The results of empirical work show that, if they understand at all, it is not the same understanding as humans. I suggest that this lack of human-like understanding arises because the models do not have a body that interacts with the world. In the second part of the talk, I expand on these ideas to demonstrate how most psychological models of language/semantics are embarrassingly inadequate to account for human language understanding.
Arthur Glenberg is professor emeritus in Psychology at Arizona State University. He has developed an embodied theory of cognition. Professor Glenberg’s recent work has demonstrated a) how actions in a situation are an essential prerequisite for new learning; b) how language comprehension takes advantage of one’s knowledge of how actions can be combined; and c) how linguistic structures coordinate with action-based knowledge to result in language comprehension. He has also begun to investigate application embodiment theories to enhance children’s reading comprehension. He and his team have developed EMBRACE, an iPad application designed to help dual language learners read for comprehension in English. He heads the Laboratory for Embodied Cognition in the Department of Psychology at Arizona State University.