Abstract
This study aimed to determine what information is used during observational learning and how this depends on task constraints using a mixed-design (between- and within-subjects) experimental approach. Specifically, the study aimed to examine whether full-body information or limited-body information enhances observational learning, and whether the influence of information is mediated by task constraints. For this purpose, participants (N = 48, mean age = 24 ± 5.3, male) were assigned to one of three demonstration conditions, each observing a point-light display (PLD) of a throwing action with varying kinematic information: full-body motion (BODY, 17 markers across major joints), right-arm motion (ARM, 4 markers on the throwing arm), or wrist-only motion (WRIST, 1 marker on the wrist). Each condition was divided into two sub-groups: one replicating the throwing action with a ball to a target (BODY-ball, ARM-ball, WRIST-ball) and one replicating the action without a ball, focusing on movement form (BODY-no ball, ARM-no ball, WRIST-no ball). These conditions manipulated the scope of visual kinematic cues and the presence of object-related task constraints to investigate their effects on motor learning outcomes. During the acquisition phase, participants performed 20 acquisition trials where a demonstration was shown five times on the first trial and then once again before each of the remaining trials. Twenty-four hours later, a retention test (5 trials with no demonstration) was performed. After retention, participants performed 10 further trials as re-acquisition. All participants observed a full-body PLD model in this period. Movement outcome, the similarity of intra-limb coordination, and wrist peak velocity in relation to the model were calculated. Results showed that in shoulder-elbow coordination, the BODY-ball group performed less like the model than the other groups in the retention test, all P < 0.05. In wrist-elbow coordination, a significant difference was observed in the ball condition, and the BODY group performed less like the model than the ARM and WRIST groups, all P < 0.05. Additionally, the no ball groups performed more like the model than the ball groups, all P < 0.05. In wrist peak velocity, the WRIST group performed less like the model than the BODY and ARM groups, all P < 0.05, and also, the ball groups performed more like the model than the no ball groups, all P < 0.05. These findings highlight that learning coordination patterns benefits from focused visual cues and minimal task demands, whereas learning velocity-related parameters depends on object interaction and more comprehensive kinematic information. This suggests that effective observational learning requires matching the type of visual information to the specific motor control demands of the task, offering insights for designing motor skill training protocols.