Anticipatory Threat of Shock Reduces the Number of Items Reported From a Brief Visual Display: Evidence for Capacity Loss Along the Scanning Order
Background: Anticipatory anxiety impairs working memory, but whether it also limits the earliest, high-capacity stage of visual storage is unclear. We asked whether the mere expectation of an aversive electrical stimulus reduces how many items people can report from a briefly flashed display, and whether any loss is spatially uniform. Methods: Eighty-eight healthy adults aged 18-25 years were randomly allocated to a threat-of-shock condition (surface electrodes on the wrist and the instruction that a painful stimulus might be delivered; none was) or a control condition (n = 44 each). Participants viewed a 4 x 3 letter matrix for 400 ms and performed untimed written whole report. Correct recall (0-12), intrusions and the position of each reported letter were scored. Groups were compared with Welch's t test, the Mann-Whitney U test, bootstrapped effect sizes, a binomial generalised linear model and a Bayes factor; spatial effects were tested with generalised estimating equations (GEE) accounting for letters nested within participants. Results: The threat group reported fewer correct letters (3.41 +/- 1.09) than controls (4.70 +/- 1.29); Welch t(83.6) = 5.11, p < .001, Hedges' g = 1.08, 95% CI [0.70, 1.53], BF10 ~ 7,549. Intrusions did not differ (p = .376). Recall declined more steeply from left to right under threat (group x column chi2(3) = 11.24, p = .011), with losses confined to the third and fourth letters of the top row (FDR-corrected q <= .003). Conclusions: Anticipatory threat reduced the amount of information transferred from a brief display into a reportable form without increasing guessing. Intact first items and depleted later items along the habitual reading order indicate a reduced number of items read out of a rapidly decaying visual store, rather than a hemifield-specific bias. The single-trial design and the absence of a manipulation check limit mechanistic inference.