Pupillometry tracks cognitive load and salience network activity in a working memory functional magnetic resonance imaging task

Abstract

The diameter of the human pupil tracks working memory processing and is associated with activity in the frontoparietal network. At the same time, recent neuroimaging research has linked human pupil fluctuations to activity in the salience network. In this combined functional magnetic resonance imaging (fMRI)/pupillometry study, we recorded the pupil size of healthy human participants while they performed a blockwise organized working memory task (N‐back) inside an MRI scanner in order to monitor the pupil fluctuations associated neural activity during working memory processing. We first confirmed that mean pupil size closely followed working memory load. Combining this with fMRI data, we focused on blood oxygen level dependent (BOLD) correlates of mean pupil size modeled onto the task blocks as a parametric modulation. Interrogating this modulated task regressor, we were able to retrieve the frontoparietal network. Next, to fully exploit the within‐block dynamics, we divided the blocks into 1 s time bins and filled these with corresponding pupil change values (first‐order derivative of pupil size). We found that pupil change within N‐back blocks was positively correlated with BOLD amplitudes in the areas of the salience network (namely bilateral insula, and anterior cingulate cortex). Taken together, fMRI with simultaneous measurement of pupil parameters constitutes a valuable tool to dissect working memory subprocesses related to both working memory load and salience of the presented stimuli.

Boundary updating as a source of history effect on decision uncertainty

“…2e; Supplementary Table. 1), consistent with previous studies 14,16,22 . At the time point with their highest correlation with BMBU's decision uncertainty (Supplementary Fig.…”

“…'The pre-congruence effect is straightforward but has not been established empirically. The neglect of this obvious source of the variability in decision variability is odd, given the fundamental roles of decision uncertainty in adaptive human behavior such as volatility monitoring 11 , learning from errors 12 , and executive control 13 , and its wide-ranging associations with behavioral (response time [14][15][16][17] ), physiological (pupil-size 14,15 ), and neural (salience network 11,[14][15][16]18 ) measures. Here, we probed human response time (RT), pupil size, and neural activity to ascertain the presence of the pre-congruence effect and assess its contribution to the variability in decision uncertainty.We recruited 41 human participants and asked them to perform a binary classification task.…”

“…1f). Moving on, as a first step to establish the presence of the pre-congruence effect, we probed RT (N=41), a well-known behavioral correlate of decision uncertainty [14][15][16] . As predicted, participants' RTs displayed both the pre-congruence effect (𝛽 !…”

“…Moving on, as a first step to establish the presence of the pre-congruence effect, we probed RT (N=41), a well-known behavioral correlate of decision uncertainty 14–16 . As predicted, participants’ RTs displayed both the pre-congruence effect ( β S t –1* C t = 0.11 ( P = 9.2 × 10 −9 )) and the current-congruence effect ( β S t * C t = −0.21, ( P = 2.7 × 10 −18 )) ( Fig.…”

Limited evidence for enhanced working memory performance and effort allocation in the face of loss versus gain incentives: A preregistered (non) replication

“…It scales with task difficulty and with motivation and task engagement (Hopstaken et al, 2015; Massar et al, 2016, 2019; Schneider et al, 2018). Pupillometry studies involving the N‐Back task similarly found increased pupil diameter with higher working memory load, and with reward motivation (Mies et al, 2018; Peysakhovich et al, 2017; Yeung et al, 2021), related to activation of effort‐related brain circuitry, including dorsal anterior cingulate cortex, and anterior insula (Fietz et al, 2022).…”