Presentation

Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams

Multimodal behavioral and electrophysiological measures in agrammatic patients reveal a complex relationship between cognitive control and sentence processing

Poster B59 in Poster Session B and Reception, Thursday, October 6, 6:30 - 8:30 pm EDT, Millennium Hall

Malathi Thothathiri¹, Jeremy Kirkwood², Erica Middleton², Abhijeet Patra³; ¹George Washington University, ²Moss Rehabilitation Research Institute, ³Manchester Metropolitan University

One feature of agrammatism, a sub-profile of aphasia, is difficulty understanding sentences. Traditionally, this difficulty was thought to arise from overreliance on semantics and a failure to process sentences syntactically (“Syntactic Deficit” hypothesis). Using contrastive cases and multimodal measures, we tested the alternative hypothesis that sentence comprehension difficulty might arise from deficits in cognitive control because such deficits undermine one’s ability to choose between competing interpretations (“Cognitive Control” hypothesis. Thothathiri et al., 2018). Four patients with chronic aphasia (>6 months post-onset) following a left hemisphere stroke participated: P30 71(F), P99 57(M), P100 74(F), P69 61(M). P30, P99, and P100 met all agrammatism screening criteria – (1) above chance lexical comprehension and at chance reversible sentence comprehension in the Philadelphia sentence comprehension battery, (2) better performance on canonical than non-canonical sentences in an expanded sentence comprehension task, (3) Western Aphasia Battery fluency<5, and (4) Northwestern Assessment of Verbs and Sentences sentence production score<4. P69 was used as a comparison patient because s/he showed higher WAB fluency (=9) and NAVS sentence production (=12) and above-chance reversible Actives comprehension. Patients completed (1) a sentence-picture matching task where we manipulated conflict between syntax and semantics (Active-Implausible/Conflict: The robber handcuffed the cop; Active-Plausible/No-conflict: The cop handcuffed the robber), and (2) a sentence plausibility judgment task (Saffran et al., 1998) where the syntax did or did not conflict with verb argument constraints (Verb-Impossible/Conflict: The equation taught the presenter; Verb-Possible/No-conflict: The surgeon taught the lesson). Additionally, patients completed a battery of cognitive control (e.g., Stroop, Auditory Stroop), short-term memory (e.g., digit span), and working memory (e.g., backwards digit span) tasks. We tested two patients (P99 and P100) further using electrophysiology and eye-tracking because they showed contrastive cognitive control profiles but similar sentence comprehension impairments. Contrary to the Syntactic Deficit hypothesis, not all agrammatic patients performed poorly when semantic cues conflicted with syntax. P30 performed similarly to the comparison patient P69, showing above-chance performance. Only P99 and P100 showed impairment in the critical conflict conditions. P99 was impaired in Stroop and the processing of conflict sentences, consistent with the Cognitive Control hypothesis. However, the results unexpectedly revealed that P100 was impaired too despite not showing deficits in the Stroop tasks. Follow-up testing revealed a differentiation between reactive (P99) and proactive (P100) cognitive control deficits in the two patients (AX-CPT. Braver, 2012). ERP results also revealed contrastive patterns, with P100 showing a P600 in response to conflict (like healthy adults) and P99 not. Our multimodal investigation of contrastive agrammatic patients revealed that (1) only some agrammatic patients struggle with overriding semantics with syntax; (2) different cognitive control deficits might impact sentence processing in different ways; and (3) online processing measures like ERP can reveal differences between agrammatic patients who perform similarly on offline behavioral measures. Our full dataset and analysis, including eye-tracking data and performance on other kinds of garden-path sentences, will shed further light on the complex relationship between cognitive control, online sentence processing, and offline sentence comprehension.

Topic Areas: Control, Selection, and Executive Processes, Syntax