Social cues can impact complex behavior unconsciously

The study examines whether a task-irrelevant but socially meaningful cue—human gaze direction—can automatically affect complex, whole-body actions without conscious awareness or intention, and whether such effects are subject to top-down control. Traditional dual-process accounts define automatic processes as lacking intention and awareness, being uncontrollable yet efficient, whereas controlled processes are intentional, conscious, controllable, and resource-demanding. Recent critiques argue for a graded view of automaticity where these features can dissociate. Using masked priming, the authors test three core features of automaticity—awareness, intentionality, and controllability—in the context of a sports-like action (defending a basketball pass).

Prior work shows subliminal stimuli can influence judgments and simple button-press responses, but evidence for effects on complex, whole-body behavior is mixed. Supraliminal gaze cues reflexively orient attention even when uninformative or detrimental, suggesting intention-free and uncontrollable processing. However, when gaze primes are subliminal, effects have typically required a context with predictive supraliminal primes, implying top-down intentional involvement. Conversely, subliminal face primes can influence processing when faces are task-relevant targets, likely due to inclusion in the task set and action-trigger conditions. The current study advances this literature by using photorealistic, action-relevant stimuli as both primes and targets and by directly measuring whole-body kinematics (center of pressure, CoP) alongside RTs to evaluate unconscious social priming under varying top-down contexts.

Design: Three experiments each comprised two tasks: a complex whole-body response (CR; Experiments 1–3a) and a simple button-press response (SR; Experiments 1–3b). All used a 2×2 within-subject design with factors pass congruency and gaze congruency, defined by whether prime pass/gaze matched the target-defined response direction. Participants: Three independent student groups participated (Exp. 1: N=21; Exp. 2: N=22; Exp. 3: N=22). Most were right-handed; none had formal basketball expertise beyond school lessons. Normal vision, neurologically healthy, informed consent obtained; ethics approved by Bielefeld University. Apparatus: Participants stood on an AMTI BP900-900 force plate facing a projection screen (244×183 cm) at 3.5 m. Two large convex response buttons were positioned left and right (centers 195 cm apart, 103.5 cm high). Force data were sampled at 1000 Hz. RTs were recorded with 1-ms precision. Stimuli: Life-sized photographs (512×768 px) of a basketball player passing rightward were used; images were mirrored for leftward passes, yielding 12 stimuli (three consistent gaze/pass; three head-fake with inconsistent gaze/pass; plus mirrored versions). Masks were created by scrambling/distortion of originals. Primes were always masked (pre-mask 100 ms; prime 17 ms; two post-masks 50 ms each). Targets differed by experiment: Exp. 1 used only consistent gaze-pass targets (no head fake); Exp. 2 occluded the face with a gray dot (no visible gaze); Exp. 3 used only head-fake targets (gaze inconsistent with pass). Factor combinations appeared in pseudo-random order; 144 trials per experiment with 24 practice trials. Procedure: Task was to respond as fast/accurately as possible to target pass direction: CR tasks required a lateral blocking movement to press the corresponding side button; SR tasks required a hand-held button press. Trial sequence: fixation 1000 ms, pre-mask 100 ms, prime 17 ms, post-masks 2×50 ms, target until response. Error feedback was given for incorrect responses. After each response, participants re-centered CoP before the next trial. Session ended with a detection task (72 trials) where participants attempted to categorize prime pass direction by button press after being informed about primes. Data processing: Ground reaction forces/torques were converted to CoP (frontal plane). For each trial, a 1000-ms segment from prime onset was extracted and baseline-corrected; signs were normalized by response direction (right positive). Trials were excluded for incorrect responses, RTs >±2 SD, or excessive CoP sway in the first 150 ms (>±2 SD RMS). For each participant and condition, CoP time series were averaged. RTs were similarly averaged. CoP onset latencies were estimated with a jackknife procedure (2.5 mm threshold) and transformed for rmANOVA. Primary CoP amplitude metric was CoP at 335 ms post-prime (CoP-335), corresponding to the peak of initial erroneous ipsilateral shift in incongruent conditions. Statistical analysis: For each experiment, 2×2 rmANOVAs tested effects of pass and gaze congruency on RT, CoP-335, and CoP onset latency, with Holm–Bonferroni-corrected post hocs as needed. Awareness was assessed via signal detection (d′) on the prime detection task. Cross-experiment 2×2×3 rmANOVAs (experiment as between-subject factor) compared effect sizes. Sensitivity analyses repeated rmANOVAs without exclusions (except incorrect trials). A priori power analysis (pilot N=15, SR RTs) indicated ≥19 participants for 0.8 power to detect a gaze congruency main effect.

Awareness: In all groups, prime detection d′ did not differ from zero, indicating subliminal primes. Complex responses (CR; Exps. 1–3a):

• Experiment 1a (consistent gaze in targets): Significant main effects on RT of pass congruency (faster by −74 ms for congruent) and gaze congruency (faster by −38 ms for congruent); no interaction. CoP-335 showed significant main effects: incongruent pass produced a more positive (erroneous ipsilateral) shift by +63 mm; incongruent gaze by +45 mm; no interaction. CoP onset latencies exhibited a significant cross-over interaction: faster initiation when prime pass and gaze were consistent (both congruent or both incongruent) versus inconsistent (difference −85 ms).
• Experiment 2a (gaze occluded in targets): RT pass congruency remained robust (−65 ms); gaze congruency persisted but was reduced (−11 ms). CoP-335 showed similar pass effect (+63 mm) and a reduced gaze effect (+12 mm). CoP onset latency cross-over interaction persisted but was smaller (−36 ms). Cross-experiment analyses confirmed gaze effects decreased from Exp. 1a to 2a (RT: t(41)=−5.88, p<.001, d=1.79; CoP-335: t(41)=−7.37, p<.001, d=2.25).
• Experiment 3a (head fakes in targets): RT showed a significant pass main effect (−83 ms). No gaze main effect; a pass×gaze interaction appeared but was sensitive to exclusion criteria (not significant when all trials included). CoP-335 showed only a pass main effect (+81 mm). CoP onset latency still showed a significant cross-over interaction but smallest magnitude (+23 ms consistent vs inconsistent). Gaze effects decreased further from Exp. 2a to 3a (RT: t(42)=−4.24, p<.001, d=1.28; CoP-335: t(42)=−3.55, p<.001, d=1.07). Simple responses (SR; Exps. 1–3b):
• Experiment 1b: RT pass congruency −34 ms; gaze congruency −11 ms; no interaction.
• Experiments 2b and 3b: Only pass congruency effects (−39 ms and −37 ms); no gaze effects. Cross-experiment analyses showed gaze RT effects decreased from 1b to 2b (t(41)=−4.17, p<.001, d=1.27) and remained absent in 3b. No CoP effects were observed in SR tasks (flat CoP traces; CoP-335 no main effects or interactions). Overall pattern: Subliminal gaze primes influenced RTs and CoP during complex actions when gaze in targets was informative (Exp. 1a) and, to a lesser extent, when gaze was absent from targets (Exp. 2a), but effects disappeared when target gaze was counter-predictive (Exp. 3a). Pass congruency effects were robust across all experiments and tasks.

Findings show that socially relevant cues (gaze direction) can modulate complex, whole-body actions without conscious awareness, evidenced by RT and CoP priming in Exp. 1a and residual effects in Exp. 2a. The persistence of gaze priming without visible target gaze (Exp. 2a) indicates that intentional, top-down setup is not necessary for subliminal social processing in an action context. However, the graded reduction of gaze effects from informative (Exp. 1a) to absent (Exp. 2a) to counter-predictive contexts (Exp. 3a) demonstrates top-down controllability: participants likely suppressed detrimental gaze information when it conflicted with task goals. CoP onset latency cross-over interactions indicate that pass and gaze are extracted from subliminal primes and interact during action planning, consistent with separate information streams and rapid-chase/ideomotor frameworks. The absence of CoP priming in simple button-press tasks suggests that intention to act with a whole-body movement is a prerequisite for unconscious behavioral priming to manifest kinematically, aligning with ideomotor theories. These results reconcile prior mixed findings by showing that subliminal social priming is graded: it does not require awareness or intention but is modulated by strategic, top-down control depending on task context.

This work demonstrates that subliminal social cues (gaze) can influence complex, whole-body behavior, affecting both response timing and kinematics, thus extending social priming effects beyond simple button-press tasks. Automaticity is partial: effects occur without awareness and intention but are subject to top-down modulation—amplified when helpful and suppressed when detrimental. The action context enhances sensitivity to subliminal social cues relative to evaluative, non-action tasks. Future research could examine how expertise, different social cues (e.g., body orientation), or dynamic stimuli modulate these effects, and leverage more sensitive kinematic measures to detect subtle unconscious influences in real-world decision-making.

One interaction effect (RT pass×gaze in Experiment 3a) was sensitive to trial exclusion criteria and became non-significant when all trials (except errors) were included, suggesting caution in interpreting this specific interaction. Participants were non-expert students with no formal basketball experience, which may limit generalizability to skilled athletes. The controlled laboratory setting with static images, while ecologically enhanced by life-sized projections and whole-body responses, may not fully capture dynamics of real gameplay. Sample sizes were modest but guided by a priori power analysis for within-subject effects.