Slide Slam P1 Sandbox Series
Towards an electrophysiological hallmark for agents vs. patients in working memory
Xinchi Yu1, Jialu Li2,3,4, Hao Zhu2,3,4, Xing Tian2,3,4, Ellen Lau1; 1University of Maryland, College Park, 2Division of Arts and Sciences, New York University Shanghai, 3Shanghai Key Laboratory of Brain Functional Genomics (Ministry of Education), School of Psychology and Cognitive Science, East China Normal University, 4NYU-ECNU Institute of Brain and Cognitive Science at NYU Shanghai
Events, whether perceived visually or described linguistically, appear to be encoded by humans with conceptual relations to participant roles like “agent” (the “doer” of an action) or “patient” (the “receiver” of an action). For instance, when viewing a visual scene in which a cat is chasing a dog, or comprehending a linguistic expression that “a dog is being chased by a cat”, one can conceptually represent that the cat is the agent, and the dog is the patient. Where and how the brain represents abstract conceptual features such as agent and patient and binds them to mental model representations is still elusive. In the current (ongoing) experiment, we aim at identifying the EEG/ERP signatures distinguishing the encoding of conceptual agents vs. patients. In the current experiment, we employ a novel method inspired by a “pinging” paradigm used in vision neuroscience to study the content of working memory (e.g., Wolff et al., 2017). Participants’ task was to judge the fit between a drawing of an event (presented for 600 ms, e.g., a lion hitting an elephant; Hultén et al., 2014) and a matching/non-matching linguistic expression presented 1700 ms after the offset of the drawing (e.g., “The elephant was hit by the lion”/“The lion was hit by the elephant”, in Chinese; in total 72 trials, involving 4 unique animals and 6 actions). In between the drawing and the sentence judgment (900 ms after drawing offset), we “ping” the agent or patient entity in working memory by briefly flashing (200 ms) the corresponding word (e.g., “lion” with a larger font size). In this way, we are able to “highlight” the agent-entity binding or the patient-entity binding selectively in the ad hoc conceptual structure of the event. By comparing the EEG/ERP upon the presentation of the ping, we are able to compare the EEG/ERP component corresponding to agents vs. patients. We used both active and passive sentences for the match task, in order to encourage conceptual representation of the event and not mere linguistic prediction during the delay interval. We also included a baseline condition where no event was depicted in the picture (72 trials, where e.g., the lion and elephant are simply standing beside each other). Although this search for an EEG/ERP index of event role binding is exploratory, one hypothesis is that it may share features with EEG/ERP components associated with numerical magnitude representations (e.g., Spitzer, Waschke & Summerfield, 2017; Luyckx et al., 2019), based on recent converging evidence that (conceptual) event and magnitude representation both involve human posterior parietal cortex (event: e.g., Thompson et al., 2007; Centelles et al., 2011; magnitude: e.g., Bueti & Walsh, 2009; Summerfield, Luyckx & Sheahan, 2020). Data collection for this experiment is still ongoing, and we are expecting to be able to present some preliminary data by SNL 2021.