Slide Slam M13
Variation in how cognitive control modulates sentence processing
Abhijeet Patra1, Jeremy Kirkwood1, Erica L Middleton1, Malathi Thothathiri2; 1Moss Rehabilitation Research Institute, USA, 2The George Washington University, USA
Prior research suggests that cognitive control assists the comprehension of garden-path sentences that create conflict between interpretations. However, doubts remain about how cognitive control could influence a seemingly specialized and temporally tuned function like sentence processing. Our hypothesis tries to reconcile the two perspectives by proposing that cognitive control might influence (1) the online processing of sentences that require the integration of multiple conflicting cues but not other complex structures; (2) the offline decision processes related to choosing an interpretation and doing the experimental task. We also hypothesized that these effects could be modulated by individual differences in cognitive control. We found suggestive correlational evidence in six persons with aphasia: patients who showed large Stroop effects also showed slowed-down processing of sentences with multiple conflicting cues. We looked for additional causal evidence by testing healthy younger adults in a “conflict modulation” paradigm. Methods: We measured word-by-word self-paced reading times and comprehension accuracy for three types of sentences in a web-based study (N=78 healthy adults aged 18-35 years). Within each type, we had congruent (sentence matches prior language and world experience) and incongruent (sentence violates expectations) conditions. (1) Syntax-Semantics: Congruent-During the filming, the actor was directed by the creative producer; Incongruent-During the first rehearsal, the conductor was directed by the creative musician (actors are usually directed, conductors usually direct) (2) Phrase-Attachment: Working at the valet stand, Colin said put the car (that’s) in the entrance into the parking lot and drive slowly (Congruent contains “that’s”, Incongruent does not) (3) Relative Clause: In the research institute, the technician (Congruent=who contacted the lab manager; Incongruent=who the lab manager contacted) was analyzing the biological samples. Sentences were interleaved with Stroop trials. A given sentence could follow a congruent (e.g., “blue” in blue font) or incongruent (e.g., “orange” in blue font) Stroop trial. A 2x2 conflict modulation design (Stroop congruence x Sentence congruence) tested whether a previous Stroop trial affected the processing of a subsequent sentence. On a separate day, participants completed cognitive control (Stroop, AXCPT, Flanker) and working memory (reading, operation, and backwards digit spans) tasks. Results: Exploratory factor analysis confirmed the separability of individuals’ cognitive control and working memory abilities. Cognitive control predicted (1) online conflict modulation at specific disambiguation points for syntax-semantics (by: t=-2.2, p=.03; the: t=2.1, p=.04) but not other sentence types; and (2) offline comprehension accuracy for syntax-semantics (t=-1.8, p=.07) and phrase-attachment (t=-4.5, p<.001) but not relative clause sentences. In contrast, working memory predicted (3) offline comprehension accuracy for relative clause sentences (t=3.2, p=.002). Conclusions: Together, the results show that the effect of cognitive control on online and offline sentence comprehension is modulated by both the type of sentence/conflict and individual differences in cognitive control ability. We suggest that the theoretical controversy regarding the conflict modulation effect could be resolved by better understanding variability. Clinically, this perspective has ramifications for characterizing and treating different individuals with aphasia.