Task-irrelevant stimuli reliably boost phasic pupil-linked arousal but do not affect decision formation

Abstract

The arousal systems of the brainstem, specifically the locus coeruleus-noradrenaline system, respond "phasically" during decisions. These central arousal transients are accompanied by dilations of the pupil. Mechanistic attempts to understand the impact of phasic arousal on cognition would benefit from temporally precise experimental manipulations. Here, we evaluated a non-invasive candidate approach to manipulate arousal in humans: presenting task-irrelevant auditory stimuli at different latencies during the execution of a challenging task. Task-irrelevant auditory stimuli drive responses of brainstem nuclei involved in the control of pupil size, but it is unknown whether such sound-evoked responses mimic the central arousal transients evoked during cognitive computations. A large body of evidence has implicated central arousal transients in reducing bias during challenging perceptual decisions. We thus used challenging visual decisions as a testbed, combining them with task-irrelevant sounds of varying onset latency or duration. Across three experiments, the sounds consistently elicited well-controlled pupil responses that superimposed onto task-evoked responses. While we replicated a negative correlation between task-evoked pupil responses and bias, the task-irrelevant sounds had no behavioral effect. This dissociation suggests that cognitive task engagement and task-irrelevant sounds may recruit distinct neural systems contributing to the control of pupil size.

Fig. 1

Behavioral tasks. Schematic sequence of events during the simple forced choice task of Experiments 1 and 2. Participants reported the presence or absence of a faint grating signal superimposed onto dynamic noise. Signal contrast is high for illustration only. In Experiment 1, the task-irrelevant, auditory white-noise (70 dB) stimulus started simultaneously to baseline onset (25% of trials). In Experiment 2, brief white noise stimuli (100 ms, 70dB) were played randomly between 3 s before and 0.5 s after decision interval onset (80% of trials).

Fig. 2

Time courses of pupil responses. (A) Pupil response time course on trials with no task-irrelevant sound, binned post-hoc by size of task-evoked pupil response, time-locked to button-press (choice). Shading, S.E.M. across participants. Grey shading, interval used for quantifying task-evoked pupil responses (Methods). (B) Differential pupil size time courses between task-irrelevant sound trials and trials without task-irrelevant sound. Baselined with 0.5 s intervals from before the first possible task-irrelevant sound. For Experiment 2, data were binned with a window sliding along possible task-irrelevant sound onset asynchronies (SOAs; window size, 800ms; step size 350 ms). Shading, S.E.M. across participants. Grey shading, interval used for quantifying task-irrelevant sound-evoked pupil responses (bottom, exemplary for last SOA window centered at -0.15 s; Methods). (C) Participant-wise task-irrelevant sound-evoked pupil response magnitude. Black bars, mean across participants.

Fig. 3

Pupil responses and choice bias. (A) Absolute choice bias (criterion c) plotted against task-evoked pupil response for experiments 1 and 2, for trials without a task-irrelevant sound. Error bars, S.E.M. across participants. (B) Mean absolute choice bias (criterion c) on trials with and without task-irrelevant sound plotted against mean task-irrelevant sound-evoked pupil responses. Error bars, S.E.M. across participants. (C) Participant-wise difference in absolute bias on task-irrelevant sound versus no task-irrelevant sound trials. Black bars, mean across participants.

Fig. 4

Task design, pupil responses and choice biases in Experiment 3. (A) Schematic sequence of events during the category level averaging task of Experiment 3. Participants reported the category (cardinal vs. diagonal) of the average orientation of gratings shown during each trial. White noise stimuli (75 dB) were played at the start of the inter-trial-interval, the first grating or the fifth grating (75% of trials). (B) Differential pupil size time courses between task-irrelevant sound trials and trials without task-irrelevant sound by task-irrelevant sound timing. Shading, S.E.M. across participants. Baselined with 0.5 s intervals from before the first possible task-irrelevant sound. (C) Participant-wise task-irrelevant sound-evoked pupil response magnitude. Black bars, mean across participants. (D) Participant-wise difference in absolute bias (shift of psychometric function) on task-irrelevant sound versus no task-irrelevant sound trials. Black bars, mean across participants. (E) Regression weights of visual stimulus position as predictors of choice (psychometric kernels). Error bars, S.E.M. across participants.

Fig. 5

Effects of task-irrelevant sound as function of pre-trial baseline pupil size. (A) Task-irrelevant sound-evoked pupil response plotted against pupil baseline size (binned post-hoc by size). Shading, S.E.M. across participants. Statistical metrics, results of repeated measures ANOVA. (B) As (A) but for difference in absolute bias (criterion c, shift of psychometric function) between trials with and without task-irrelevant sound.