title: Pupil Dilation and Cognitive Load
status: established
last_updated: 2026-05-31
sources: Kahneman Beatty 1966 Pupil Diameter Memory, Beatty 1982 Task Evoked Pupillary Responses, Van Der Wel 2018 Pupil Dilation Cognitive Control
tags: [pupillometry, cognitive-load, TEPR, working-memory, locus-coeruleus, mental-effort, methodology]

Pupil Dilation and Cognitive Load¶

Status: established
Last updated: 2026-05-31
Sources: Kahneman Beatty 1966 Pupil Diameter Memory, Beatty 1982 Task Evoked Pupillary Responses, Van Der Wel 2018 Pupil Dilation Cognitive Control
Tags: [pupillometry, cognitive-load, TEPR, working-memory, locus-coeruleus, mental-effort, methodology]

Summary¶

The pupil dilates as cognitive processing demands rise, a response known as the task-evoked pupillary response (TEPR). The effect was established by Kahneman and Beatty (1966), formalised and validated as a load measure by Beatty (1982), and later linked to the locus coeruleus–norepinephrine system by van der Wel and van Steenbergen (2018). Together the three sources establish pupillometry as a non-invasive, real-time index of mental effort, subject to specific measurement controls.

Body¶

Context¶

This article draws on three texts that together trace pupillometry from discovery to mechanism: Kahneman and Beatty's (1966) original short-term-memory experiment, Beatty's (1982) review formalising the task-evoked pupillary response, and van der Wel and van Steenbergen's (2018) review of the underlying neural system and the controls the measure requires. Each addresses a different layer of the same phenomenon — behaviour, measurement, and mechanism. Within this knowledge base it is the cognitive-load and mental-effort strand of eye tracking, complementary to the spatial-gaze strands in Fixational Eye Movements and Fixation Saccade Detection, and it supplies the workload rationale that applied work such as Eye Tracking In Surgery builds on.

Key Points¶

Kahneman and Beatty (1966) demonstrated the core relationship using a digit-span paradigm: the pupil dilated progressively as digits were presented and constricted as they were reported, with peak diameter ordered by the number of items held in memory (PDF pp. 1–2, orig. pp. 1583–1584). A transformation task produced steeper loading than plain recall under physically identical stimuli, showing the response tracks processing load rather than sensory input, which they read as an index of the items under active rehearsal (PDF pp. 2–3, orig. pp. 1584–1585).

Beatty (1982) formalised this as the task-evoked pupillary response and reviewed its validity as a load measure, recovering it by time-locked averaging around task events as a primary index of processing load (PDF p. 1, orig. p. 276). The response occurs at short latency — onset between 100 and 200 ms — and subsides once processing ends (PDF p. 1, orig. p. 276). He evaluated it against the three criteria Kahneman proposed — sensitivity to within-task load, to between-task differences, and to between-individual differences in ability (PDF p. 1, orig. p. 276) — and concluded it satisfies all three across digit span, mental arithmetic, reasoning, and language tasks (PDF pp. 2–11, orig. pp. 277–286). He distinguished the response from non-cognitive pupil changes driven by the light reflex and emotional arousal, establishing it as a specific index of processing load rather than general activation (PDF p. 12, orig. p. 287).

Van der Wel and van Steenbergen (2018) reviewed pupil dilation as an index of effort in cognitive-control tasks and set out its physiological basis: pupil size is governed by the sphincter and dilator muscles under parasympathetic and sympathetic control, and sympathetic inhibition of parasympathetic activity produces the dilation linked to locus-coeruleus activity (PDF p. 7, orig. p. 2011). They treat the response as the stimulus-induced increase relative to a pre-stimulus baseline, and judge the effort interpretation best supported by intra-individual within-condition correlations, which are least vulnerable to confounds (PDF p. 2, orig. p. 2006). They stress the controls the measure requires: matched luminance to rule out the pupillary light reflex, matched trial numbers to exclude an orienting response, and control for variables such as age and ethnicity in between-individual comparisons (PDF p. 6, orig. p. 2010).

Conclusion¶

The three sources are complementary rather than competing, and read as a single arc. Kahneman and Beatty (1966) established that the pupil indexes processing load; Beatty (1982) established how reliably it does so and against what criteria; van der Wel and van Steenbergen (2018) supplied the physiological account and the controls under which the measure holds. They agree on the central claim — that the dilation is a load-specific signal distinct from the light reflex and emotional arousal — and the main shift across them is one of emphasis: the later review tempers the measure's applied promise with confounds the early experimental work did not need to confront. Applied confirmation in operational settings such as surgery is consistent with this account (Tolvanen et al., 2022) Eye Tracking In Surgery.

References¶

Beatty, J. (1982) 'Task-evoked pupillary responses, processing load, and the structure of processing resources', Psychological Bulletin, 91(2), pp. 276–292. doi: 10.1037/0033-2909.91.2.276. beatty1982pupillary

Kahneman, D. & Beatty, J. (1966) 'Pupil diameter and load on memory', Science, 154(3756), pp. 1583–1585. doi: 10.1126/science.154.3756.1583. kahneman1966pupil

Tolvanen, O., Elomaa, A.-P., Itkonen, M., Vrzakova, H., Bednarik, R. & Huotarinen, A. (2022) 'Eye-tracking indicators of workload in surgery: A systematic review', Journal of Investigative Surgery, 35(6), pp. 1340–1349. doi: 10.1080/08941939.2021.2025282. tolvanen2022surgery

van der Wel, P. & van Steenbergen, H. (2018) 'Pupil dilation as an index of effort in cognitive control tasks: A review', Psychonomic Bulletin & Review, 25(6), pp. 2005–2015. doi: 10.3758/s13423-018-1432-y. vanderWel2018pupil

Open Questions¶

Can pupil dilation be used for real-time adaptive automation triggering in control-room settings?
What is the minimum sampling rate required to detect TEPR onset in fast-paced operational tasks?
How do individual differences in LC-NE reactivity affect TEPR calibration across operators?