Dyad versus solo training during visuomotor adaptation

Abstract

From tango dancing to paddling a canoe, humans often coordinate their actions to achieve shared goals. While previous research has shown that dyadic motor tasks can enhance performance during joint action (Ganesh et al., 2014; Takagi et al., 2017), the effects of such collaboration on later solo performance remain debated (Beckers et al., 2020; Che et al., 2016). In particular, few studies have examined how the nature of partner interaction shapes individual learning when both contributors are aware of their shared control. Here, we investigated how performing a visuomotor adaptation task with a partner affects subsequent individual performance. Participants (N = 96) completed 50 baseline trials followed by 200 adaptation trials with a 30-degree clockwise or counterclockwise cursor rotation. This was followed by 20 counter- adaptation and 50 error-clamp trials before returning after a 5-minute break to repeat the same task individually. Half of the participants completed the initial session alone (Solos, N = 48) while the other half completed it in dyads (Dyads, N = 48) with both partners contributing simultaneously to a shared cursor trajectory. All participants adapted successfully during training. However, only those in the solo condition showed robust savings in the second session. Across dyads, three characteristic interaction patterns were observed, each with distinct implications for how individuals engaged with the perturbation. Participants who trained in dyads exhibited significantly weaker early re-adaptation, particularly when their partner had taken over the majority of the initial corrections. These findings suggest that while collaboration can support immediate task success, it does not guarantee lasting individual learning. This work encourages future research to clarify the mechanisms that underlie coordinated motor adaptation between two individuals. A greater understanding of how people interact and share control could help refine motor learning protocols in rehabilitative or human-machine interaction settings.