Find the Best Available Evidence

 Quick Analysis Plan Copied

 Long Analysis Plan

Evidence by Population

Randomized controlled trials demonstrate that computerized cognitive training improves executive function measures, though the strength and specificity of effects vary by population and control condition.

Clinical populations (severe mental illness): A two-week RCT (N=25) showed EF training improved composite T-scores from 45.16 to 50.49 versus 50.26 to 50.64 in active sham control, accompanied by increased frontal theta power during high-load tasks (F(1,24)=7.11, p=0.013, partial η²=0.209) EF training in SMI: EEG and cognitive gains.

Children (8–9 years): A school-based RCT (N=168) found VR and tablet Mastermind Cognitive Training improved eye-tracking attention versus active control (RightEye overall F(2,150)=11.2, p=0.001; saccades F=8.70, p=0.004), with minimal dose effects VR and tablet cognitive training in children RCT. A broader review confirms computerized training among interventions improving EF in children aged 4–12, particularly those with initially poorer skills Science review of EF interventions in children.

Adolescents: A 14-day RCT (N=85) produced greater Raven's Matrices gains in the training group (+2.56) than no-training control (+1.11; interaction F(1,83)=13.843, p<0.001, η²=0.143), with lower baseline IQ predicting larger gains (r=−0.255, p=0.018) Short-term EF training and fluid intelligence RCT.

Older adults: Evidence is heterogeneous. In the oldest-old (mean 81.5, N=182), CCT added to strength-balance training improved Trail Making Test B, divided attention, and dual-task gait (all p<0.001) versus physical training alone Combined CCT and physical training in oldest-old adults. However, in a frailty RCT (N=347, age 50+), Frontal Assessment Battery gains were comparable between CCT+exercise groups and video-watching+exercise control (Multi-CCT: +1.4; Two-CCT: +0.9; Control: +1.3) CCT with exercise in frail older adults RCT.

Blue = computerized training; gray = control. Units differ across studies (T-scores, raw scores, FAB 0–18); the pattern is comparable but magnitudes are not. Training-specific gains are clearest in clinical and student populations.

Computational Analysis Plan

• Extract: Pre/post means, SDs, and Ns for each EF outcome per group from all RCTs.
• Quantify: Compute Hedges' g per comparison; pool via random-effects meta-analysis (I², Q statistic).
• Test moderators: Population (clinical vs healthy), control type (active vs passive), training modality (VR vs tablet), duration.
• Bias assessment: Funnel plot asymmetry, Egger's regression; subgroup by active vs passive control to avoid inflating pooled effects.
• Sensitivity: Exclude studies confounding CCT with physical exercise to isolate training-specific contributions.

Three issues qualify conclusions: active controls improve substantially, narrowing contrasts; near transfer is robust while far transfer to fluid intelligence remains modest and may partly reflect retest effects; and non-blinded assessors in some trials introduce detection bias Far-transfer and retest-effect limitations.