Bilateral transfer of motor performance as a function of motor imagery training: a systematic review and meta-analysis

Wan X. Yao, Sha Ge, John Q. Zhang, Parisa Hemmat, Bo Y. Jiang, Xiao J. Liu, Xing Han Lù, Zayd Yaghi, Guang H. YueFrontiers in Psychology 2023 · DOI

TL;DR

This systematic review and meta-analysis demonstrates that "mental practice"—specifically motor imagery training (MIT)—is a highly effective tool for inducing bilateral transfer, where training one limb improves the performance of the opposite, untrained limb. The study finds that internal (first-person) imagery and "mixed-task" protocols are the most potent strategies, offering results comparable to traditional physical training.

What Problem Does This Paper Address?

Bilateral Transfer (BT), also known as cross-education, is the phenomenon where learning a motor skill with one limb leads to performance gains in the contralateral, untrained limb. While the benefits of physical training (PT) on BT are well-documented, the efficacy of Motor Imagery Training (MIT)—the internal rehearsal of a motor action without overt physical movement—has been less clear.

Previous studies on MIT and bilateral transfer often suffered from small sample sizes and inconsistent results. This paper addresses this gap by synthesizing existing research to determine if "mental practice" can reliably trigger these cross-limb benefits and which specific types of imagery are most effective. This has profound implications for neurorehabilitation, particularly for stroke survivors or patients with unilateral injuries who may be unable to perform physical exercises with an affected limb.

Methods

The authors conducted a systematic review and meta-analysis following PRISMA guidelines. They searched six major databases (including PubMed and Web of Science) for randomized-controlled trials published through late 2022.

Key methodological features included:

  • Quality Assessment: Studies were evaluated using the PEDro scale, with the included studies averaging a "high quality" score of 7.00.
  • Moderator Analysis: The researchers didn't just look at overall efficacy; they categorized data by transfer direction (Dominant $\rightarrow$ Non-dominant vs. vice versa), transfer type (Mirror, Normal, or Mixed tasks), and MIT type (Internal vs. External imagery).
  • Statistical Framework: Meta-analyses were performed using R-Studio, utilizing Standardized Mean Differences (SMD) to compare MIT against both no-exercise controls (CTR) and physical training (PT).

Datasets

As a meta-analysis, the "dataset" consists of previously published experimental results. From an initial pool of 728 identified studies, the authors narrowed the selection down to 9 peer-reviewed articles. These articles provided:

  • 14 studies for the comparison between MIT and no-exercise controls.
  • 15 studies for the comparison between MIT and physical training.
  • A total of 137 participants specifically within the MIT intervention groups.

Key Insights and Ideas

  • The "Equality" Surprise: Perhaps the most striking finding is that for inducing bilateral transfer, MIT is statistically as effective as physical training (ES = -0.02 difference between them). While PT is usually superior for the trained limb, the "leakage" of motor learning to the untrained side appears equally robust via mental rehearsal.
  • Internal vs. External Perspective: The study highlights a clear hierarchy in imagery. Internal MIT (IMIT)—imagining the sensation of movement from within one's own body—is significantly more effective (ES = 2.17) than External MIT (EMIT), where one visualizes themselves from a third-person perspective (ES = 0.95).
  • The Power of "Mixed-Tasks": The authors categorize tasks based on coordinate systems. "Normal" tasks use extrinsic (spatial) coordinates, while "Mirror" tasks rely on intrinsic (body-relative) coordinates. Mixed-tasks, which allow the brain to access both coordinate systems simultaneously, showed the strongest transfer effects (ES = 1.68).
  • Symmetry in Transfer: Unlike some previous theories suggesting transfer is easier from the dominant to the non-dominant limb, this meta-analysis found no significant difference in direction, supporting a more symmetrical model of mental motor transfer.

Key Results

  • MIT vs. Control: MIT showed a significant moderate-to-large benefit in inducing BT compared to no exercise (ES = 0.78, 95% CI = 0.57–0.98).
  • MIT vs. Physical Training: There was no significant difference in the magnitude of BT induced by mental vs. physical practice (ES = -0.02), suggesting they are comparable for this specific purpose.
  • Task Type Superiority: Mixed-tasks (ES = 1.68) significantly outperformed both mirror-tasks (ES = 0.46) and normal-tasks (ES = 0.56).
  • Imagery Type Superiority: Internal imagery (ES = 2.03) was found to be vastly superior to external imagery (ES = 0.49) in the MIT vs. CTR comparison.

Limitations and Future Work

The authors acknowledge several hurdles:

  • Publication Bias: A funnel plot analysis suggested a "substantial publication bias," meaning studies with non-significant results may be underrepresented in the literature.
  • Small Sample Size: Only 9 articles met the strict inclusion criteria, limiting the power of some subgroup analyses.
  • Duration of Training: Most included studies involved only a single session of MIT. The long-term "dose-response" relationship of mental practice on bilateral transfer remains unknown.
  • Healthy vs. Clinical Populations: The data was drawn from healthy subjects. Future research must confirm if these effects hold true for clinical populations, such as stroke patients, where the neurological environment is significantly different.
  • Combination Therapy: The paper suggests that while MIT is a great alternative, the combination of MIT and PT might be the "gold standard," but there wasn't enough data to meta-analyze this specific interaction for bilateral transfer.

Summary updated

Abstract

This review concludes that MIT can serve as a valuable alternative or supplement to PT in facilitating BT effects. Notably, IMIT is preferable to EMIT, and interventions incorporating tasks that have access to both intrinsic and extrinsic coordinates (mixed-task) are preferred over those that involve only one of the two coordinates (mirror-task or normal-task). These findings have implications for rehabilitation of patients such as stroke survivors.
BibTeX
@inproceedings{yao2023bilateral,
  author = {Wan X. Yao and Sha Ge and John Q. Zhang and Parisa Hemmat and Bo Y. Jiang and Xiao J. Liu and Xing Han Lù and Zayd Yaghi and Guang H. Yue},
  title = {{Bilateral transfer of motor performance as a function of motor imagery training: a systematic review and meta-analysis}},
  booktitle = {Frontiers in Psychology},
  year = {2023},
  doi = {10.3389/fpsyg.2023.1187175},
}
Figure