Anticipation is central to perceptual–cognitive expertise in badminton, but the decision mechanisms through which experts use visual and auditory cues remain unclear. This study examined expert–novice differences in unimodal anticipation and latent decision processes estimated by a hierarchical drift diffusion model (HDDM). Sixty-six participants (33 experts and 33 novices) completed a landing-region anticipation task under visual-only and auditory-only conditions. Behavioral performance was analyzed using mixed analyses of variance (ANOVAs), and accuracy and reaction time were jointly modeled with HDDM parameters: boundary separation (a), drift rate (v), non-decision time (t), and starting-point bias (z). Experts showed higher overall accuracy than novices, F(1, 64) = 56.46, p < 0.001, ηp² = 0.469, whereas reaction time did not differ significantly by group. The auditory-only condition produced higher accuracy, F(1, 64) = 12.69, p = 0.001, ηp² = 0.165, but longer reaction times, F(1, 64) = 150.32, p < 0.001, ηp² = 0.701. HDDM results showed higher drift rates in experts under both visual-only (BF10 = 1992.845) and auditory-only (BF10 = 24.354) conditions; auditory-only anticipation also involved higher boundary separation (BF10 = 53.095). These findings suggest that badminton expertise is associated with higher anticipatory accuracy and more efficient evidence accumulation under unimodal cue conditions, while auditory-only anticipation may also involve a more cautious decision threshold.

This study examined whether repeated shot-direction decisions in adolescent badminton players were associated with the current spatial context and the immediately preceding choice. Fifty systematically trained players aged 13–15 years completed a hidden-probability phase (Task 1), in which no numerical outcome probabilities were displayed at choice, followed by an explicit-probability phase (Task 2), in which one current outcome-probability dimension was displayed for the two route options. Task 1 constituted the primary analysis, and Task 2 provided a secondary boundary-condition assessment of the same choice structure. Primary inference was based on trial-level binomial generalized linear mixed-effects models for Task 1, complemented by rolling forward validation of out-of-sample prediction. In Task 1, relative to the center position, a left opponent position was associated with lower odds of a straight choice (OR = 0.087, 95% CI [0.059, 0.130]), whereas a right position was associated with higher odds (OR = 6.297, 95% CI [4.400, 9.013]). After adjustment for current context, trial progress, and position sequence, a previous straight choice was associated with lower odds of choosing straight on the next trial (OR = 0.430, 95% CI [0.339, 0.546]). Adding previous choice improved out-of-sample predictive performance (ΔELPD = 18.081, 95% CI [5.277, 30.885]), although the absolute improvement was modest. Task 2 produced directionally similar estimates in the mixed model, while participant-clustered generalized estimating equations provided weaker support for this association; this phase was therefore interpreted descriptively. Outcome-dependent and contextual Q-learning models showed no stable incremental predictive support beyond the context-plus-previous-choice baseline. Because route choice, screen location, and keypress were coupled by design, the previous-choice term indexes a composite history signal spanning tactical, spatial, and motor-response components.

This study examined visually guided spatial updating and endpoint estimation under visual-only conditions and assessed whether performance varied by distance, cue direction, and swimming training background. Thirty-six right-handed male participants aged 18–20 years, including expert swimmers and amateur controls, completed a cue–target spatial orientation task with two distance conditions, short and long, and four cue directions, N, E, S, and W. Movement time, spatial error distance, and proportional movement time were analyzed using 2 × 2 × 4 mixed-design repeated-measures ANOVAs. Bayesian independent-samples comparisons were further conducted to quantify evidence for the null hypothesis of no robust group differences. The results indicated that task demands were the main determinants of performance. Compared with the short-distance condition, the long-distance condition yielded longer movement time, larger spatial errors, and higher proportional movement time, suggesting increased demands on spatial updating and movement execution. Cue direction also affected proportional movement time and spatial error distance, indicating direction-specific differences in visual cue integration and reference-frame transformation. By contrast, no robust overall group differences were found across the three primary behavioral measures, and the Bayesian analyses generally supported the null model. Although a significant Group × Distance interaction was observed for spatial error distance, this interaction reflected a group-related difference in the extent to which error increased from the short- to the long-distance condition, rather than evidence for a consistent expert advantage. Overall, performance in the present visual-only cue–target task was driven mainly by visuospatial updating load, attentional allocation, and task structure, whereas swimming-related expertise was not consistently expressed under the current task constraints.