AsPredicted pre-registration: cueing mechanisms in Schmitz et al. (2024) Experiment 2

Abstract

This is a registered secondary analysis of public data from Experiment 2 of Schmitz et al. (2024, PLOS ONE), in which 39 participants identified a peripheral target letter after a central cue. The cue was an arrow, a schematic face, or a colour, and was either neutral or 100% counter-predictive (arrows and faces) or informative (colour) about the target’s location. Response (letter identity, left/right button) was orthogonal to cue direction. The purpose is to identify where in the decision process spatial cues act. I will fit drift-diffusion models, with trial-to-trial variability in drift and start point, in which the cue changes drift rate, start point, boundary separation, or non-decision time. I will compare them against a no-effect model using summed BIC and posterior-predictive checks of the cueing effect across RT quantiles. Expected outcomes: cueing speeds responses (H1) through non-decision time or boundary separation rather than drift or start point (H2); start-point effects are treated as mimicry unless a recovery simulation shows they are separable from drift (H3); and the arrow cueing effect exceeds the face effect (H4). The registration specifies a decision table with pre-set triggers for every fallback if fits fail: hierarchical refitting, dropping variability parameters, and added lapse or fast-guess/slow-error mixtures. All branches are reported. The data already exist; I disclose what I have seen in the attached document.