Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Pupil indices of arousal predict trial-level performance during Mandarin tone category learning
Poster Session D, Saturday, September 13, 5:00 - 6:30 pm, Field House
This poster is part of the Sandbox Series.
Lyra Joson1, Bharath Chandrasekaran2, Casey Roark1; 1University of New Hampshire, 2Northwestern University
Spontaneous fluctuations in pupil size are coupled with activity in the cingulo-opercular network including the anterior cingulate and anterior insula with stronger coupling predicting better task performance (Breeden et al., 2017). Elevated activity in the cingulo-opercular network predicts improved speech recognition performance in demanding listening contexts, suggesting involvement of this network in monitoring performance and making adaptive adjustments to optimize task behavior (Vaden et al., 2013). In this study, we examined the relationship between fluctuations in pupil size and non-native speech category learning performance at the trial level. In line with Adaptive Gain Theory (Aston-Jones & Cohen, 2005), we hypothesized that trial-level learning performance would be optimal at medium pupil sizes as smaller pupil sizes would be associated with task disengagement and larger pupil sizes would be associated with higher distraction. We trained 17 monolingual English listeners on four non-native Mandarin tonal categories. Participants completed six blocks of categorization training with two talkers with feedback (40 trials per block) and a generalization test with stimuli from two novel talkers and no feedback (40 trials total). We recorded monocular left pupil diameter using an Eyelink 1000+ Desktop Mount with chin and forehead stabilization. Participants were required to fixate for 500 ms before the stimulus was automatically triggered. We calculated the average pre-trial pupil size within the 500 ms immediately preceding the presentation of the speech sounds. For each participant, we grouped the pupil size data into small, medium, and large bins such that one third of all trials fell into each bin for each participant. We calculated bins separately for early (blocks 1-3) and late (blocks 4-6) phases to understand how pupil size measures may change during speech learning. Overall, participants successfully learned the speech categories and performance improved from the early (M = 34%) to late (M = 46%) phases. In line with our hypothesis, speech learning performance significantly differed across participant-defined pupil size bins (F(2, 32) = 3.69, p = .036, eta-squared = 0.013). However, the specific pattern did not match our hypothesis – large pupil sizes (M = 37%) were associated with significantly worse performance than both small (M = 42%) and medium (M = 41%) pupil sizes, which were not different from one another. This pattern did not differ across blocks. Overall, we found that within-individual pupil sizes relate to second language speech learning outcomes at the trial level. These results indicate that pupil indices of high arousal (relatively large pupil sizes) may reflect a disengaged neural state in which the monitoring of performance and adaptive control in the cingulo-opercular network is downregulated, resulting in impaired trial-level performance. These results present new avenues for state-dependent learning interventions, such as delivering key learning information during phases of optimal neural engagement.
Topic Areas: Language Development/Acquisition, Speech Perception