Mindfulness training preserves sustained attention and resting state anticorrelation between default-mode network and dorsolateral prefrontal cortex: A randomized controlled trial.

The researchers compared two interventions delivered during regular school hours over 8 weeks:

• Mindfulness training: A structured program teaching attention regulation, present-moment awareness, and distraction management through meditation exercises, body scans, and mindful movement.
• Active control: Coding training: Students learned computer programming concepts and built simple projects. This controlled for the effects of learning a new skill, social interaction, and time away from academic instruction.

Primary outcome: Sustained attention measured by the Sustained Attention to Response Task (SART), specifically the rate of attention lapses (errors of commission on Go trials).

Secondary outcome: Resting-state functional connectivity (rsFC) between the default mode network (DMN) and the dorsolateral prefrontal cortex (DLPFC), measured via functional magnetic resonance imaging (fMRI).

Tertiary/exploratory outcomes: Response inhibition (No-Go accuracy on SART), reaction time variability, and correlations between changes in behavior and brain connectivity.

• Sample size: 99 sixth graders enrolled in the full RCT; 40 completed the neuroimaging protocol (34.3% of total).
• Age: Mean age 11.76 years (range not specified, but all sixth graders).
• Setting: Boston Collegiate Charter School, Dorchester, MA, USA. A public charter school serving a diverse, low-to-middle-income population.
• Demographics (imaging subsample): 70% female; 47.5% eligible for free/reduced price lunch; 10% Hispanic, 32.5% African American, 52.5% White, 5% other/multiple races.
• Exclusions: None explicitly stated beyond not volunteering for imaging. All sixth graders participated in the interventions as part of their school schedule.
• IQ: Measured via Wechsler Abbreviated Scales of Intelligence (WASI); groups were matched at baseline.

• Sustained Attention to Response Task (SART): A 15-minute computer task where participants press a button for every digit (0–9) except the number "3" (presented on only 5% of trials). Key metric: commission errors on Go trials (pressing for "3" when they should not) — these reflect lapses of attention. Also measured: omission errors (failing to press for non-3 digits), reaction time, and reaction time variability.
• Resting-state functional MRI (rs-fMRI): Participants lay in the scanner for ~6–8 minutes with eyes open, fixating on a crosshair. No task was performed. Researchers measured the temporal correlation (functional connectivity) between the DMN (seed regions: medial prefrontal cortex, posterior cingulate cortex) and the right DLPFC. A negative correlation (anticorrelation) indicates healthy segregation between these networks.
• Wechsler Abbreviated Scales of Intelligence (WASI): Used to match groups at baseline.
• Edinburgh Handedness Inventory: Used for stratification during randomization.

Design: Randomized controlled trial (RCT) with two parallel groups: mindfulness training vs. coding training (active control).

Randomization: Students were randomly assigned to groups, stratified by whether they participated in the imaging protocol and their handedness. The researchers ran 1,000 randomizations and selected the one that minimized baseline differences (Mahalanobis distance) on key variables. This is a robust method to ensure group equivalence.

Blinding: The paper does not explicitly state that participants or teachers were blinded to condition — given the nature of the interventions (mindfulness vs. coding), full blinding is impossible. However, the outcome assessors (those scoring the SART and analyzing fMRI data) were likely blinded to group assignment. The lack of participant blinding is a potential source of expectancy effects, though the active control condition partially mitigates this.

Duration: 8 weeks of intervention, delivered during the last class period of the school day (typically reserved for miscellaneous activities). Pre- and post-intervention assessments were conducted within 1–2 weeks before and after the 8-week period.

Statistical approach:

• Primary analysis: Mixed-effects models (ANCOVA-style) comparing pre-to-post changes between groups.
• Correlation analyses: Pearson correlations between SART performance and DMN–DLPFC connectivity at baseline, and between change scores (pre-to-post) within each group.
• Correction for multiple comparisons: Applied false discovery rate (FDR) correction for brain imaging analyses.

What this design can and cannot prove:

• Can prove: Causal effects of mindfulness training (vs. coding training) on sustained attention and brain connectivity, because of random assignment and an active control group. The RCT design eliminates many confounds (e.g., maturation, regression to the mean, selection bias).
• Cannot prove: That mindfulness is superior to all other interventions (only to coding training). Cannot prove mechanisms definitively — the correlation between brain and behavior changes is suggestive but not causal. Cannot generalize beyond sixth graders in a charter school setting. Cannot determine the optimal dose or duration of mindfulness training.

Major methodological weaknesses:

1. Small imaging sample (n=40): Only 40 of 99 students completed fMRI. This reduces statistical power and may introduce selection bias (families who opted in may differ from those who did not).
2. No passive control group: Without a "business-as-usual" group, we cannot know if coding training actively harmed attention or if mindfulness merely prevented a natural decline. The authors interpret the control group's decline as "natural," but this is an assumption.
3. No long-term follow-up: Assessments were only immediately post-intervention. We don't know if effects lasted days, weeks, or months.
4. Single school, single grade: Limits generalizability to other ages, schools, or socioeconomic contexts.
5. No blinding of participants: Expectancy effects could influence results, though the active control reduces this concern somewhat.

Baseline correlation (pre-intervention, all participants):

• Better sustained attention (fewer commission errors on SART) was significantly correlated with stronger anticorrelation between the DMN and right DLPFC at rest (r = -0.40, p = .01). In plain English: kids whose brains showed more segregation between "mind-wandering" and "executive control" networks made fewer attention errors.

Primary outcome — Sustained attention (SART commission errors):

• Mindfulness group: No significant change from pre to post (mean commission errors: pre ~11.5, post ~11.0; p > .05). Attention was preserved.
• Coding group: Significant increase in commission errors from pre to post (mean: pre ~11.0, post ~14.5; p < .05). Attention declined.
• Group × time interaction: Significant (F(1,97) = 4.82, p = .03, partial η² = 0.05). This is a small-to-medium effect size.

Secondary outcome — DMN–DLPFC anticorrelation:

• Mindfulness group: No significant change in anticorrelation strength (p > .05). Connectivity pattern was preserved.
• Coding group: Significant weakening of anticorrelation (i.e., DMN and DLPFC became less negatively correlated; p < .05). Connectivity pattern declined.
• Group × time interaction: Significant for the DMN seed in medial prefrontal cortex (MPFC) with right DLPFC (cluster-level FDR-corrected p < .05). Effect size not reported in standard units, but the cluster size was 27 voxels.

Correlation between brain and behavior changes:

• Within the mindfulness group only, the change in SART commission errors was significantly correlated with change in DMN–DLPFC anticorrelation (r = 0.52, p = .02). Kids who preserved their attention the most also preserved their brain connectivity the most.
• No such correlation was found in the coding group (r = -0.08, p = .75).

Secondary behavioral outcomes:

• No significant group differences in SART No-Go accuracy (response inhibition) or reaction time variability.
• No significant group differences in self-reported mindfulness (measured by the Child and Adolescent Mindfulness Measure; data not shown in detail).

• Attention preservation: The mindfulness group maintained ~11–12 commission errors per session, while the coding group increased to ~14–15 errors. This is a difference of about 3–4 more lapses in the control group — roughly a 30% increase in attention failures over 8 weeks.
• Brain connectivity: The anticorrelation between DMN and DLPFC weakened in the coding group by an amount equivalent to what is seen in some clinical populations (e.g., ADHD) or in normal aging over several years. The mindfulness group showed no such weakening.
• Effect size: The group × time interaction for SART errors had a partial η² of 0.05, which is considered a small-to-medium effect (Cohen's guidelines: 0.01 = small, 0.06 = medium, 0.14 = large). This means mindfulness training explained about 5% of the variance in attention changes beyond what coding training explained.
• Practical translation: If you imagine a classroom of 30 students, about 1–2 additional students in the coding group would show clinically meaningful attention decline compared to the mindfulness group.

Acknowledged by authors:

• Small imaging sample (n=40) limits statistical power for brain analyses.
• No long-term follow-up to assess durability of effects.
• Single school, single grade limits generalizability.
• The active control (coding) may have different cognitive demands than mindfulness, making it unclear which specific ingredients drove the effect.

Additional critical observations:

• No passive control: Without a "do nothing" group, we cannot determine if coding training actively harmed attention (e.g., through screen time, cognitive fatigue) or if mindfulness merely prevented a natural developmental decline. The authors assume the coding group represents "typical" development, but this is untested.
• Self-report measures: The Child and Adolescent Mindfulness Measure showed no group differences, suggesting that the behavioral and brain effects were not accompanied by changes in self-reported mindfulness. This weakens the mechanistic interpretation.
• Attrition and selection bias: Only 40% of the full sample completed imaging. Families who volunteer for brain scans may differ systematically (e.g., higher socioeconomic status, more health-conscious). The authors report no baseline differences between imaging and non-imaging participants, but this cannot rule out unmeasured confounds.
• Multiple comparisons: While the authors applied FDR correction for imaging, the primary behavioral analysis (SART) was not corrected for multiple outcomes (e.g., commission errors, omission errors, reaction time). The significant finding on commission errors could be a false positive if other SART metrics were also tested.
• Industry funding: The study was funded by the Walton Family Foundation (education philanthropy) and the Poitras Center (psychiatric research). No obvious conflict of interest, but funding sources can influence research questions and interpretation.
• Age-specific effects: Results may not generalize to adults, younger children, or different school settings. Sixth graders are in a period of rapid brain development, and mindfulness may have different effects at other ages.

For someone running their own n=1 experiment:

What to test

• Intervention: Daily mindfulness practice, specifically focused on attention regulation. The school program included: body scans (10–15 min), sitting meditation (10–15 min), mindful movement (5–10 min), and brief breathing exercises (3–5 min). Total daily dose: approximately 20–30 minutes, 5 days per week.
• Dose: 8 weeks minimum. The study used 8 weeks of daily practice. For a self-experiment, aim for at least 4–6 weeks to see measurable changes, but 8 weeks is more reliable.
• Active control: To isolate the specific effect of mindfulness, compare against another structured cognitive activity (e.g., learning a new language, playing a strategy game, doing puzzles). This controls for the "learning something new" effect.

Minimum meaningful duration

• 4 weeks for initial behavioral changes (attention lapses may improve by week 3–4).
• 8 weeks for reliable brain connectivity changes (based on this study's timeline).
• 12+ weeks to test durability and habituation effects.

What to measure (specific metrics)

• Primary metric: Sustained attention lapses. Use a computerized Go/No-Go task like the SART (free versions available online). Measure commission errors (pressing when you shouldn't) — these are the most sensitive to mindfulness effects. Aim for a 10–15 minute task.
• Secondary metrics:
  • Reaction time variability (higher variability = more lapses)
  • Omission errors (failing to respond to Go trials)
  • Self-reported mind-wandering frequency (e.g., "How often did your mind wander during the task?" on a 1–10 scale)
• Brain connectivity (if accessible): Resting-state fMRI measuring anticorrelation between medial prefrontal cortex (DMN) and right dorsolateral prefrontal cortex (DLPFC). This requires a scanner, but you can approximate by tracking subjective "mind-wandering" vs. "focus" states during rest.
• Daily tracking: Use a brief (2-minute) daily log: "How many times did I notice my mind wandering during today's practice?" and "How easy was it to bring attention back?" (1–10 scale).

Key confounds to control for

• Sleep: Poor sleep dramatically impairs sustained attention. Track sleep duration and quality daily (e.g., using a sleep diary or wearable).
• Caffeine and stimulants: Standardize caffeine intake (same amount, same time each day). Avoid alcohol and other CNS depressants.
• Screen time: Excessive screen use (especially social media, gaming) can impair attention. Keep screen time constant during the experiment.
• Stress: Major life stressors (exams, relationship issues, work deadlines) can confound results. Track daily stress on a 1–10 scale.
• Physical activity: Exercise improves attention. Keep exercise routine constant.
• Time of day: Perform the attention task at the same time each day (circadian rhythms affect performance).
• Practice effects: Repeatedly taking the same attention test can improve scores due to familiarity. Use alternate versions of the task or include a 1-week baseline period to establish a stable pre-intervention score.

What a positive result would look like

• Behavioral: After 4–8 weeks of daily mindfulness, your commission errors on the SART decrease by at least 20–30% from baseline (e.g., from 15 errors to 10–12 errors). Your reaction time variability decreases by 10–15%. Your self-reported mind-wandering during the task drops by 1–2 points on a 10-point scale.
• Brain (if measured): Your resting-state DMN–DLPFC anticorrelation becomes more negative (e.g., from r = -0.10 to r = -0.25). This indicates better functional segregation between mind-wandering and executive control networks.
• Subjective: You notice that you can sustain focus on boring tasks (e.g., reading, data entry, studying) for 10–15 minutes longer before your mind wanders. You catch yourself daydreaming less frequently.
• Caution: A "null result" (no change) does not mean mindfulness doesn't work for you. It may mean the dose was too low, the duration too short, or your baseline attention was already high (ceiling effect). Try increasing practice time to 30–40 minutes daily or extending the experiment to 12 weeks.