Fig 1.
Top: A) Participants completed the modular mental arithmetic task on 8 separate biweekly sessions over the course of 16 weeks. B) During the modular mental arithmetic task participants evaluated statements by subtracting the two numbers and indicating whether the remainder was divisible by the modulo. C) Baseline, peak amplitude, and peak latency features were extracted from the pupil response.
Fig 2.
Scatter plots showing response time for every individual trial in the experiment by peak latency measurements (left) and peak task-evoked amplitude measurements (in arbitrary units: A.U.) (right). Black lines represent the least squares linear fit to the data. Peak latency, in particular, showed a striking relationship with response time, reflected in the diagonal band of data points.
Table 1.
Modeling the relationship between RT and pupil predictors at the trial-level indicate that peak latency is more strongly associated with response time suggesting that this feature maps onto decision processes that unfold over seconds.
Fig 3.
(A) Mean task-evoked pupil responses (in arbitrary units: A.U.) by task condition (easy, hard) and accuracy (correct, incorrect). (B) Bar plots illustrate how performance (mean response time) was modulated by difficulty and accuracy, demonstrating a main effect of each factor as well as an interaction effect. (C) Bar plots illustrate a similar pattern of results for peak amplitude of the pupil response in relation to task variables. (D) Peak latency was modulated by difficulty and accuracy, and showed a marginally significant interaction effect. Error bars represent standard error of the mean.
Table 2.
Modeling the relationship between pupil features and IQR indicates that baseline and peak latency are associated with the distribution of response time between sessions in this longitudinal study.