Presentation

Xinchi Yu¹, Ellen Lau¹; ¹University of Maryland

Language enables us to express and comprehend different types of abstract relations between/among entities, including agent-patient relations. Agent-patient relations are the abstract relations between the initiator (the agent) and the receiver (the patient) of transitive events, which are ubiquitous in real life. For example, in both “the lion hit the elephant” and “the lion kicked the elephant”, the two animals are involved in an abstract agent-patient relation, although the action itself differs. We are able to represent agent-patient relations in working memory, yet it is unclear what brain region(s) are responsible. Previous studies on events and event roles (i.e. agent/patient) have suggested two candidate regions that may host agent-patient relations in working memory. One is the superior temporal cortex (Frankland & Greene, 2015, 2020), which may host the specific transitive actions themselves (e.g., hitting, kicking) (Wurm & Caramazza, 2019). The other is the posterior parietal cortex (Wang et al., 2016; Matchin et al., 2019). Based on the critical role of the posterior parietal cortex in representing relations between visual objects (Xu & Chun, 2007; Ayzenberg & Behrmann, 2022) and working memory (Robitaille et al., 2010; Xu, 2017), we hypothesize that the posterior parietal cortex also hosts abstract agent-patient relations between conceptual entities conveyed through language, and is able to carry such relations forward through time. On the other hand, although superior temporal cortex encode specific actions, these regions may not encode abstract agent-patient relations. Most relevant studies have used fMRI, whose relatively poor temporal resolution makes it hard to separate the encoding and maintenance phases. In order to identify the brain regions representing agent-patient relations in working memory, we design an MEG experiment given its higher temporal resolution. In this study, we employ a sentence-picture delayed matching paradigm, in order to examine the retention phase in between. In each trial, subjects first view a sentence (1000 ms). After a retention period (1000 ms), subjects view a picture and are asked to respond to whether the picture matches with the sentence before. In half of the trials, the sentence is about two entities under an agent-patient relation (e.g., “John kicked Mary”). In the other half of the trials, the sentence is about two entities coordinated together (e.g., “John and Mary jumped”), which does not contain agent-patient relations yet involves the same number of entities. We hypothesize that the posterior parietal cortex will show differential activity for the agent-patient relation condition compared to the coordination condition during the retention period, as the former condition contains agent-patient relations that are absent in the latter condition. Data collection for this experiment is still ongoing.

Topic Areas: Meaning: Combinatorial Semantics, Reading