Hypothesis
Five classic Hess studies from the 1960s-70s, reporting that pupil size tracks interest value, arousal and mental effort, including gender differences and mimicry of seen pupils, would replicate with modern eye-tracking and luminance control.
What failed
Across two experiments (N = 182 and a preregistered N = 147), the predicted gender differences in pupil change (for stimuli of differing interest and for a mob-escape video) did not appear, nor did pupil mimicry of schematic eye pairs. Effects of mental effort and arousing words did hold up.
Broken assumption
Hess's pupil effects reflected interest and gender rather than uncontrolled luminance and small-sample noise.
Why
The authors point to Hess's small samples and undocumented luminance control; in their data pupil change correlated strongly with the luminance of where participants looked.
From the published paper: Winter et al. International Journal of Psychophysiology, 2021. ↗CC BY