Physical practice, mental practice or both: a systematic review with meta-analysis

27 studies, 1,494 participants, 42 effect sizes

27 studies, 1,494 participants aged 13-31 (NOT lifespan — young adults, mostly university students); mostly novices on the assessed skill; 69% laboratory settings. Skill level could NOT be tested as a moderator (insufficient reporting); only age, type of imagery and amount of practice were.

27 studies, 1,494 participants, 42 effect sizes, explicitly partitioned into ACQUISITION, RETENTION and TRANSFER phases and into pairwise contrasts. RoB2: 48% low risk, 37% some concern, 15% high risk. Only NINE pooled comparisons actually exist (Table 2): six at acquisition, two at retention, one at transfer — the discussion states this explicitly. Heterogeneity is severe on the head-to-head contrasts (I2=92% acquisition PP vs MP, 97% retention PP vs MP). Open-access regional journal, moderate rigor. TWO REPORTING DEFECTS FOUND ON FULL-TEXT READ: (1) the ABSTRACT reports five retention/transfer comparisons against no-practice that do not appear in Table 2 and that the results section explicitly says could not be computed ("Not enough data to compare no practice with CP, MP, and PP"); (2) the small-study-effects paragraph inverts Egger's logic, calling p=0.25/0.11/0.23 "asymmetry confirmed" and p<0.05 "no asymmetry". Use Table 2, not the abstract.

The decision-relevant comparison: physical practice beats mental practice head-to-head at ACQUISITION (ES -1.16 favouring PP) and at TRANSFER (0.50), but the two are STATISTICALLY INDISTINGUISHABLE AT RETENTION (-1.29, CI -3.12 to 0.54, I2=97%) — the durable-learning contrast is null, on 9 effect sizes. Combined practice does not beat physical practice alone at either acquisition (-0.01, CI -0.31 to 0.28) or retention (0.16, CI -0.29 to 0.63), directly undercutting the 'imagery + physical > physical' claim and contradicting Behrendt 2021 and Simonsmeier 2021 which had found CP superior. Mental practice's largest values come against a weak no-practice comparator and only at acquisition (0.50). Authors' conclusion: mental practice does not replace physical practice; it can complement it under some conditions (its edge over MP shrinks as sessions accumulate — CP beats MP at one session, not at nine).

Not an unreplicated finding — this is the fourth-plus meta-analysis on the question and the paper compares itself to its predecessors directly. REPLICATED half: mental/physical/combined practice each beat no practice at acquisition (Feltz & Landers 1983; Driskell et al. 1994; Hird et al. 1991; Simonsmeier et al. 2021), and physical > mental at acquisition (Freitas 2020; Ingram 2016; Mulder 2004; Ruffino 2021; Sharif 2015; Truong 2022). CONTRADICTED half: Behrendt et al. 2021 and Simonsmeier et al. 2021 found combined practice SUPERIOR to physical practice; this meta finds them equal. Sharif et al. 2015 found physical and mental similar at both retention AND transfer; this meta reproduces that at retention but not at transfer. Hence `mixed`.