Virtual reality (VR)–based motor training provides a controlled and interactive environment from which to assign training task elements that shape movement practice and motor performance [1]. Such functionality with VR platforms is especially beneficial in designing protocols for motor rehabilitation [2]. This pilot study examined changes in force-based motor performance across two adjustable elements in VR-based training: augmented sensory feedback (ASF) and task difficulty. ASF provides additional real-time information about movement performance to support improved error correction [3]. Visual ASF is well-suited to immersive VR environments and can produce significant performance gains in VR task platforms [3]. The value of augmented feedback may increase with task demands, as greater path curvature and arc length raise spatial control requirements and reliance on external visual guidance [4]. As such, task difficulty is another essential training element, readily controllable in VR, and is central to ensuring skill acquisition in relation to user capability [5].

This pilot study presents a new version of a previously described VR training platform [6] that has participants exert muscular effort with the upper extremity to command a virtual avatar to trace target trajectories. A trajectory-based approach requires spatial accuracy while preserving user autonomy during continuous upper-limb VR training [7]. Further, this study newly explores target-tracking accuracy as a function of increasing training difficulty levels (ie, larger trajectory amplitudes) with and without augmented visual feedback. This pilot study was conducted with a small cohort of neurotypicals to demonstrate initial feasibility; however, the long-term application of this platform is VR-based motor rehabilitation of upper-extremity function after neurological injury [1].

Net percentage change in tracking accuracy was positive for all 8 conditions (Figure 2A). Under NF, the largest mean improvement occurred at the lowest difficulty level, L1, whereas improvements at higher difficulty levels were smaller. In contrast, VF produced positive improvements across all difficulty levels, ranging from approximately 5.6% to 10.5%. Across L1-L4, VF exceeded NF by approximately 4.5% on average. This descriptive difference became more pronounced when only L2-L4 were evaluated, with VF exceeding NF by approximately 5.9% (Figure 2B). An exploratory mixed effects model using participant-level VF – NF differences did not confirm a feedback-by-difficulty effect, with a positive but uncertain slope of 2.32% per difficulty level, a confidence interval spanning zero (–4.16 to 8.80), and an exploratory P value of .47. Thus, the pattern was interpreted as descriptive and hypothesis-generating. Processed participant-level mean tracking-error values for each condition are provided in Table S1 in Multimedia Appendix 1.

This pilot study examined how two essential VR training design elements (ASF and task difficulty) may interact to influence motor performance during trajectory tracking. Across conditions, tracking accuracy generally improved after training, with greater improvement when visual ASF was present, consistent with our prior VR studies [3,6]. The relatively highest increase with no ASF was observed at L1; however, reliance on lower-difficulty practice can limit progression and underrepresent the control demands required for learning more complex movements [9]. In this pilot study, the benefits of visual ASF were consistently amplified at higher difficulty levels (L2 to L4) and suggest that the benefit of ASF guidance during training increases with motor challenge. Further, our findings indicate the possibility that the difference between visual ASF to no ASF increases proportionately with difficulty level, suggesting visual ASF may either offset fatigue or facilitate greater engagement in the face of greater training challenge. Mechanistically, higher-difficulty trajectories may raise spatial control demands through greater curvature and path-following complexity, where real-time visual guidance becomes an especially useful external reference because the margin for spatial error narrows at higher difficulty. Together, these descriptive patterns offer preliminary rationale for future studies on how visual guidance cues and trajectory difficulty might be jointly tuned in trajectory-training task design [10].

Key study limitations include the small sample size, which was not powered for confirmatory inference, as well as the neurotypical participant sample, single-session design without retention testing, evaluation of only one sensory modality (visual ASF), lack of transfer to activities of daily living, and the absence of formal measures for fatigue or engagement. Thus, the results in this preliminary study are presented as descriptive and treated as hypothesis-generating. Future studies should use adequately powered designs and extend testing to individuals with neurological impairments, with longer-term work potentially exploring adaptive paradigms that adjust parameters such as visual ASF and task difficulty.