Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Simultaneous representation of multiple object-states in language is supported by left temporal theta-gamma phase-amplitude coupling
Poster Session A, Friday, September 12, 11:00 am - 12:30 pm, Field House
Wesley Leong1,2, Gerry Altmann1,2; 1University of Connecticut, 2Connecticut Institute for the Brain and Cognitive Sciences
[Introduction] What do we represent in our minds when processing events in language? Consider the sentence: “The chef will chop the onion”. The same onion token exists in both its intact and chopped states within the event. Subsequently cueing the object for retrieval (e.g. “and then he will smell it”) elicits an effect of competition in the brain (Hindy et al., 2012; Solomon et al., 2015). This finding has been interpreted to reflect the difficulty of selecting among distinct states, bound to the same object token, for integration into the sentence. Prior to retrieval, however, these multiple object-states must be simultaneously maintained in the mind. What are the neural dynamics by which we do this? Here, we test the hypothesis that the brain recruits the same process it uses to hold objects in visual working memory: theta-gamma phase-amplitude coupling (henceforth PAC). PAC strength has been shown to increase with visual working memory load (e.g. Canolty et al., 2006; Axmacher et al., 2010; Heusser et al., 2016) and performance (e.g. Sauseng et al., 2009; Reinhart & Nguyen, 2019), and is thus a strong candidate mechanism for how the brain represents distinct object states. Across two EEG experiments, we show increased PAC for sentences with substantial change to the object (e.g. “the chef will chop the onion”) over those with minimal change (e.g. “the chef will weigh the onion”). We also show that this PAC mirrors that elicited by a visual working memory task (Experiment 2). [Methods] Experiment 1 (N=79) aggregated EEG data from 3 prior studies. Participants read sentences on a screen (one word every 600ms). Stimuli consisted of 2 sentences each: The object underwent either substantial or minimal change in the first, and then was cued for retrieval in the second (e.g. “The chef will chop the onion. And then, he will smell it.”). EEG signal was epoched from the onset of the first verb (“chop”) to the end of the sentence, and PAC was calculated using a Driven Auto-Regressive (DAR) model (Dupré la Tour et al., 2017) and subject to a cluster-based permutation analysis (Maris and Oostenveld, 2007). In Experiment 2 (N=40), participants also performed a visual working memory task – the PAC cluster obtained from this task was used as a localizer to improve sensitivity for the sentence comprehension task. [Results] In Experiment 1, we identified a left temporal cluster showing increased PAC in the theta-gamma range for substantial change sentences. This cluster was not statistically significant per a cluster-based permutation test. In Experiment 2, we generated an ROI based on the cluster identified in the visual working memory task. Within this ROI, substantial change sentences showed greater PAC than minimal change sentences (W = 421, p < 0.001). A simple linear regression showed that PAC increased with degree of change to the object, irrespective of substantial/minimal change labels (β = 0.15, p < 0.01). [Summary] Maintaining multiple object-states online appears to share a neural process with holding objects in visual working memory: left temporal theta-gamma phase-amplitude coupling.
Topic Areas: Control, Selection, and Executive Processes,