Frontal spectral entropy as a candidate individual-difference correlate of within-session accuracy change in youth soccer: an exploratory fixed-order study

Abstract

Choking under evaluative pressure is often described in sport, yet its within-session neural and behavioural correlates in youth athletes remain poorly characterised. In this exploratory, fixed-order study we asked whether changes in soccer decision accuracy over the course of a single session covaried with EEG and eye-tracking measures. Thirteen male youth soccer players completed an image-based decision-making task across three condition-labelled blocks presented in a fixed sequence: a neutral-labelled baseline (b1), an outcome-labelled block (b2), and a monitoring-labelled block (b3), while EEG from 19 scalp electrodes referenced online to A2 and mobile eye tracking were recorded. At the block-summary level, accuracy showed an omnibus effect (Friedman χ²(2) = 12.12, p = 0.002, Kendall’s W = 0.47) and was lowest in b3, but no pairwise contrast survived maxT correction, and a trial-level generalised linear mixed-effects model with crossed random intercepts for participant and scene revealed no reliable block effect (b3 vs. b1 OR = 0.92, 95% CI [0.73, 1.17], p = 0.501; omnibus block LRT χ²(2) = 0.45, p = 0.800). Response time did not vary reliably across blocks. Instead, the clearest signal emerged between individuals: the size of each player’s b1-to-b3 accuracy change tracked the change in frontal spectral entropy, computed over a filter-compatible 1–35 Hz band (Spearman ρ = −0.70, 95% bootstrap CI [− 0.91, − 0.23], permutation p = 0.010, n = 13). This association held across a broad set of sensitivity analyses, including a 1–40 Hz entropy band, exclusion of Fp1 and Fp2, a 30–35 Hz muscle-proxy control, aperiodic exponent and offset controls, and outlier and winsorisation checks, although it weakened once retained clean EEG duration was taken into account. Central alpha-range relative power showed a weaker and spatially non-specific secondary association, and we do not interpret it as a sensorimotor mechanism. Eye-tracking measures changed little across blocks and were unrelated to accuracy change. CSAI-2R self-report scores indicated higher cognitive anxiety in the outcome-labelled and monitoring-labelled blocks than in the neutral block, but condition remained fully confounded with block order. We therefore frame frontal spectral entropy as a candidate individual-difference correlate of within-session accuracy change rather than as evidence of pressure-specific causation.