Did Tai Chi Improve Inhibition or Preserve Memory in Women Recovering From Methamphetamine Use? A 72-Person Trial

Answer first: A randomized exercise comparison among 72 women in a methamphetamine rehabilitation setting found improved Go/No-Go response accuracy over twelve weeks and a difference between groups in how 3-back working-memory accuracy changed. Yet both the Tai Chi and conventional-exercise groups improved their No-Go accuracy. Overall working-memory accuracy fell, the active conventional-exercise group deteriorated more on selected measures, and the switching task did not yield a significant group-by-time improvement. Improvements over time in blood pressure, vital capacity, and body-mass index likewise were not accompanied by statistically significant treatment-by-time interactions. The findings concern laboratory cognition and physical performance after organized exercise, not blood methamphetamine concentrations or the impairment of an identified person.

This account is based on the actual original randomized trial by Shen Menglu, Liu Ruiwen, Yang Suyong, and Zhu Dong, published in Frontiers in Psychiatry in 2021, Effects of Tai Chi on the Executive Function and Physical Fitness of Female Methamphetamine Dependents: A Randomized Controlled Trial. Its most useful lesson is methodological: improvement within one exercise group should not automatically be called an effect superior to another active exercise program, and a computerized inhibition score cannot substitute for chemical proof of drug exposure.

What did the trial actually compare?

The study compared twelve weeks of supervised Tai Chi with twelve weeks of supervised conventional physical activities among women with histories of methamphetamine dependence. This was not a randomized trial of methamphetamine administration and not a comparison of exercise against doing nothing. The women in the control arm performed square dance and a structured group calisthenics program. This matters because movement, social participation, repeated testing, staff supervision, and time spent in a rehabilitation environment could affect both arms.

The investigators evaluated three distinct executive-function constructs: withholding a response to a particular signal, temporarily maintaining and updating information in a working-memory task, and switching attention between task rules. They also obtained clinical physical-fitness measurements. Those are related abilities but are not interchangeable. A favorable No-Go result cannot fill in an unsuccessful switching result; a measured blood-pressure change cannot supply a missing methamphetamine toxicology result.

How many independent participants contributed data?

Eighty women were enrolled. The authors describe four exclusions associated with eligibility and administrative circumstances before the final analysis: three involved personnel movement and one involved judicial detention. Their flow and demographic table identify 72 independent women analyzed, with 35 assigned to Tai Chi and 37 to conventional exercise. The paper also refers to 76 randomized participants. Its trial population came from the Shanghai Female Mandatory Detoxification and Rehabilitation Center, rather than from a community-based sample of drivers or patients using prescribed stimulants.

Participants were women with reported methamphetamine-use histories, generally aged within the investigators’ eligibility range of 18 to 65 years. Baseline mean ages were 39.31 years in the Tai Chi group and 39.37 in the comparator group, a difference that was not significant (reported p = .981). Baseline body weights were less balanced: 61.87 versus 66.66 kilograms (p = .031). That observed difference warrants caution when comparing physical changes between arms.

Do not multiply the number of independent people by the number of exercises, stimulus presentations, visits, physiological readings, or computerized button presses. Two test occasions per woman and numerous stimuli within a test are repeated measurements on the same biological individual. Moreover, the 3-back analysis reports degrees of freedom that do not straightforwardly correspond to every woman in the demographic table; the original does not provide a sufficiently complete task-level attrition account to resolve that discrepancy. The appropriate denominator depends on the specific statistical analysis, not a hypothetical enlargement of the sample.

What exercises did the women perform?

The detailed methods describe approximately 40 minutes per session, three times weekly for twelve weeks in both arms. The Tai Chi schedule comprised a five-minute warm-up, thirty minutes of simplified twenty-four-form Tai Chi, and five minutes of cool-down. One or more sessions were instructor-led by a Tai Chi professional, with another supervised by trained rehabilitation personnel. The active comparator involved stretching, square dance, and the ninth Guang Bo Ti Cao group calisthenics sequence within a similarly described exercise period.

The abstract gives a different session duration from the detailed methods. The measured outcomes and participant counts remain the important scientific facts; exact intervention minutes should not be asserted with greater certainty than the underlying report permits. Although the authors reference an approximate metabolic demand and target exercise heart rate for Tai Chi, they acknowledge that they did not instrumentally monitor actual activity intensity in each participant. Equal scheduled minutes do not prove equal cardiovascular workload or adherence.

The single-blind framework also has a practical limitation: people performing Tai Chi and square dancing can often recognize which activity they are doing. The authors acknowledged that familiarity with supervisors could affect blinding. Neither arm was an inert placebo. These design conditions distinguish an evaluation of a specific organized program from a demonstration that Tai Chi movements alone produced a neural mechanism.

How was response inhibition measured?

The computerized Go/No-Go task displayed one orientation of a triangle as a signal to press a keyboard key and an inverted triangle as the signal to withhold responding. Participants completed practice and formal stimulus trials using E-Prime software. Accuracy when pressing for a Go signal and accuracy when refraining from a response to a No-Go signal were recorded separately, along with reaction time for Go responses.

This is a form of response inhibition under a tightly controlled visual protocol. It does not test a participant’s behavior at a traffic intersection, response to emergency lights, susceptibility to distraction during real driving, or ability to avoid taking illicit drugs. It gives a useful standardized behavioral endpoint but only within the conditions of the stimulus set, task instructions, and timing.

Did Tai Chi uniquely improve No-Go accuracy?

No unqualified claim of unique superiority is supported by the descriptive group data. Tai Chi participants’ mean No-Go accuracy rose from 93 ± 6% to 96 ± 4%, while the active control participants’ No-Go accuracy rose from 94 ± 5% to 97 ± 4%. Both improvements are three percentage points in the printed table. The paper reports a main effect of time and a time-by-stimulus-condition interaction: F(1,68) = 9.6 and F(1,68) = 20.52, respectively. Neither is automatically a treatment-group-by-time interaction demonstrating that Tai Chi outperformed conventional exercise on No-Go accuracy.

The Go accuracy results were also different in direction from the No-Go results. Tai Chi participants averaged 99 ± 1% at both assessments. Controls averaged 99 ± 2% before and 98 ± 6% afterward. Such high Go accuracy approaches a ceiling, limiting the amount of apparent improvement possible. The report’s statement that Tai Chi improved inhibitory control is a reasonable description of performance after the program, but the narrower causal question—whether it was better than another exercise intervention—requires the correct comparative effect.

What happened to Go reaction time?

Across the study, the authors reported a decline in average Go response time from 426.65 ± 6.45 to 408.23 ± 6.52 milliseconds, with a main effect of time F(1,68) = 16.86, p < .001. Faster correct responses can be valuable in a reaction-time task. However, the existence of that pooled time effect does not by itself show a unique causal Tai Chi benefit relative to the comparator. Repetition of a computerized task also introduces familiarity and practice as possible contributors.

Timing on one keyboard task is not evidence that all complex psychomotor performance improved. A complete appraisal keeps latency, accuracy, omission, and inhibition endpoints separate. Apparent acceleration can be advantageous, neutral, or accompanied by an error tradeoff, depending on the test and the demand. This paper does not validate a transformation from its milliseconds to an intoxication probability or accident risk.

Which findings are directly shown in the original cognitive data?

Percentages are task accuracy rather than probabilities of legal incapacity; millisecond values are performance latencies, not pharmacokinetic intervals. Values are presented as the original report printed them, including its distinct pooled versus arm-specific analyses. A significant overall time effect has a different interpretation from a significant difference in trajectories between two randomized treatment groups.

Did working memory improve after Tai Chi?

The 3-back result is better described as possible relative preservation than a broad improvement. Participants viewed a serial sequence of digits and judged whether the current item matched an earlier element in the sequence. That exercise taxes short-term maintenance, updating, attention to task instructions, and correct matching across multiple trials. The investigators separately analyzed congruent and incongruent conditions, accuracy, and response time.

The reported overall accuracy decreased from 45 ± 2% to 37 ± 2% over twelve weeks, F(1,61) = 15.43, p < .001. Overall response latency became slower, from 575.59 to 622.73 milliseconds, F(1,61) = 4.66, p < .05. A favorable headline that says the working-memory battery simply improved would therefore misrepresent major measured endpoints.

There was a group-by-time interaction for accuracy, F(1,61) = 4.73, p < .05. The authors describe a more pronounced adverse course in the conventional-exercise group, including a reported control accuracy decrease from 47 ± 2% to 35 ± 3% for the relevant comparison. Tai Chi participants’ profile was interpreted as relatively stable. The distinction is important: preventing worsening compared with another active activity is not the same estimand as increasing every person’s working-memory capacity above baseline.

Why does the 3-back denominator require attention?

The demographic comparison identifies 72 women, whereas several task tests use a residual degrees-of-freedom value of 61. That means one should not casually present every F statistic as having exactly 72 contributing observations. The article does not fully reconcile this task-level analytical denominator with the demographic counts. In forensic interpretation, degrees of freedom are not merely a formality: they help reveal which observations contributed to a particular statistical test.

Similarly, the original task illustration and prose contain potentially confusing descriptions of the target position and trial count. Those limitations do not justify inventing a corrected experimental protocol. The strongest grounded conclusions follow the outcomes actually reported, acknowledge denominator ambiguity, and avoid assigning precision to an unverified subset.

Did the test of cognitive flexibility succeed?

The switching task required responding under changing stimulus rules involving the shape or identity of letter displays. Different rules at different moments can impose cognitive flexibility and response-selection demands. The printed mean reaction time in Tai Chi participants was 128.58 ± 131.08 milliseconds before and 129.67 ± 109.21 afterward. In active controls it was 132.47 ± 140.17 before and 141.82 ± 117.39 afterward. The authors reported no significant main effect of time and no significant time-by-group interaction for switching-task response time.

This negative result is not a claim that Tai Chi can never affect set shifting. It is a bounded result for the reported task, sample, time frame, and analysis. Conversely, the numerically smaller worsening in one group is not sufficient to call the intervention superior when the relevant statistical comparison was not significant. A defensible summary puts this genuine null alongside the more favorable other outcomes.

Were physiological measurements collected?

Yes. Participants underwent cuff blood-pressure measurements, vital-capacity assessment using spirometry, body-size measurements for body-mass index, and tests of balance, flexibility, and grip strength. These are physical measurements, but none is a chemical assay for methamphetamine. The existence of a physiological instrument in a trial does not transform it into a therapeutic drug monitoring or forensic mass-spectrometry experiment.

The investigators reported significant effects of time for systolic blood pressure, vital capacity, and body-mass index. Importantly, the group-by-time interactions for all three variables were not significant in the printed table. Thus, the general improvement over the twelve-week period should not automatically be attributed specifically to Tai Chi rather than conventional activity, common rehabilitation conditions, or other factors.

What did the physical-data table actually show?

The authors identify significant within-period time factors for three measurements but did not establish significant differences in their trajectories by assigned program. The positive direction of a mean change should not substitute for a direct comparative test. Physical-health improvements also do not prove improved inhibitory-control ability on a real-world task.

Did the researchers quantify methamphetamine in venous whole blood?

No. The exposure condition was history of methamphetamine dependence, not a known experimental drug dose administered by the investigators. They did not collect or report a calibrated parent-methamphetamine concentration from forensic venous whole blood, and no simultaneous concentration was reported for separated plasma or serum, urine, oral fluid, hair, vitreous humor, breath, or another biological drug-testing matrix. The administered study intervention was exercise, not methamphetamine.

A cuff blood-pressure reading in millimeters of mercury is a physiological observation, not a blood specimen. A spirometry result in milliliters describes ventilatory capacity, not a drug amount. A reaction time in milliseconds is a computer-task outcome, not an analyte concentration in nanograms per milliliter. These distinctions prevent an especially consequential error: treating the word “blood” in “blood pressure” as evidence that a drug was physically recovered and measured from a biological blood sample.

For background on specimen-specific interpretations, consult the site’s published guides to biological matrices, confirmatory drug testing, and therapeutic drug monitoring. Those general laboratory frameworks do not supply absent drug measurements for this exercise trial.

Could the study establish an individual’s acute intoxication or sobriety?

No. A history of dependence and enrollment in supervised rehabilitation describe a population, not a person’s contemporaneous pharmacological state. The trial’s computer scores were not calibrated against observed whole-blood methamphetamine concentrations and were not validated as a test of time since use. A slower or faster keypress cannot prove the presence or absence of a drug in that person’s biological specimen, the amount ingested, or whether the person met any forensic impairment threshold.

Moreover, there was no controlled challenge with methamphetamine at a defined dose or route. There is therefore no dose-response relationship in this study between parent-drug blood concentration and inhibition performance. Historical exposure cannot be converted into an actual exposure concentration on the date of an unrelated traffic stop or incident. A case-specific toxicology opinion would need independently authenticated evidence of specimen collection, target drug identity, method validation, analytical results, timing, and context.

How should a forensic expert distinguish an effect over time from an intervention effect?

Three questions need separate answers. First, did the outcome change on average during twelve weeks? Second, did the Tai Chi group change more than the active conventional-exercise group? Third, would any observed between-arm difference generalize to a different person, task, or event? A significant within-group improvement or time main effect answers only the first question. The group-by-time interaction is a more direct test for a differential trajectory, although it still does not independently establish the mechanisms or external generalizability.

This distinction is decisive in the source’s No-Go data. Both treatment arms improved by the same printed absolute number of percentage points. A comparable caution applies to three physical measures that changed over time without significant assignment-specific interactions. The 3-back accuracy result, by contrast, includes a significant treatment-by-time contrast but within an overall deterioration pattern and with an incompletely explained statistical denominator. Collapsing these heterogeneous results into “exercise cured stimulant cognitive dysfunction” would be unsupported.

What are the strongest design limitations?

The trial’s participants were women in a single supervised rehabilitation setting, not a balanced sample of sexes, ages, nondependent people, or drivers. Its comparator was an active exercise program, a scientific strength for distinguishing scheduled activity from particular movements, but it also means the study cannot directly estimate a benefit versus no exercise. Researchers did not instrumentally monitor individual activity intensity. Practical blindness to visibly different exercise regimens was limited. Baseline body-weight imbalance may affect analysis of physical changes.

Follow-up was limited to the intervention window. The authors did not establish objective individual abstinence, incident-specific alcohol or other coexposures, validated relapse rates after discharge, or chronic neurobiological biomarkers. They discussed potential mechanisms involving neurotrophic signaling and brain circulation, but those were hypotheses drawn from surrounding scientific work, not direct measurements of those pathways in these participants. No BDNF, dopamine, transporter, or central pharmacokinetic assay may be inferred from the discussion alone.

Statistics also require care. Several results are printed as thresholds rather than exact p values, and the 3-back degrees of freedom suggest fewer task-level observations than the full demographic cohort. One must not manufacture missing individual values, post hoc uncertainty intervals, or unreported null tests to make the article appear more quantitative. The separately measured tasks answer different questions; a court-quality summary retains that granularity.

What is the defensible scientific conclusion?

After twelve weeks of structured Tai Chi or active conventional exercise, the studied women performed better on selected Go/No-Go measurements, but the improvements occurred in both arms. The relative pattern in 3-back memory scores is compatible with some preservation of performance in Tai Chi compared with a larger deterioration in the active control, while overall accuracy and response latency worsened. Switching-task time and multiple physiological group-by-time comparisons remained nonsignificant. That mixed pattern is scientifically more informative than a blanket “executive function improved” statement.

For litigation or forensic consultation, this paper can inform the scope and limits of behavioral rehabilitation measures. It cannot determine drug concentration, drug exposure time, acute methamphetamine effects in an unidentified individual, or an incident’s cause. The original experiment illustrates why study population, actual denominator, comparator, endpoint, chemical matrix, and tested contrast are necessary to any opinion drawn from empirical research.

Frequently asked questions

1. Did the study include 72 or 76 women?

The report says 76 were randomized after enrollment and shows 72 in the final analyzed cohort, divided into 35 Tai Chi and 37 active comparators. Four earlier exclusions and task-specific analytical degrees of freedom complicate a simplistic accounting. The available data should not be portrayed as 76 complete analyzed participants on every endpoint.

2. Did only the Tai Chi group improve on No-Go accuracy?

No. Tai Chi mean accuracy increased from 93% to 96%, and conventional exercise increased from 94% to 97%. Both changed by three points. Reported time and stimulus-condition effects are not, on their own, a Tai Chi-specific difference in improvement.

3. Did working memory improve in all participants?

No. Pooled 3-back accuracy dropped from approximately 45% to 37%, and response times increased. The reported group-by-time interaction suggests a different pattern between treatments, with comparatively greater deterioration in controls, rather than universal working-memory improvement.

4. Did Tai Chi outperform conventional exercise on task switching?

Not on the reported test of switching response time. Its main time and treatment-by-time comparisons were nonsignificant. A numerically smaller change is not interchangeable with a significant treatment contrast.

5. Did blood pressure improvements prove Tai Chi was uniquely beneficial?

No. Systolic pressure, vital capacity, and BMI changed significantly over time, but the original interaction statistics did not establish significant differences in the changes between the two programs. These were physical measures, not tests of methamphetamine concentration.

6. Was methamphetamine or a metabolite chemically measured in blood or urine?

The study did not report contemporaneous quantitative parent methamphetamine or metabolites from legal venous whole blood, separated plasma/serum, urine, or another forensic specimen. Exercise, cognition, cuff pressure, and spirometry were evaluated; none supplies a chemical drug assay.

7. Can these task results prove why a person crashed, acted violently, or was impaired?

No. The study observed neither a particular incident nor an individualized drug exposure and did not validate an incident prediction or impairment cutoff. An expert must assess any specific event using its independently collected biological, behavioral, medical, temporal, and case evidence.

Original primary research

Shen M, Liu R, Yang S, Zhu D. (2021). Effects of Tai Chi on the Executive Function and Physical Fitness of Female Methamphetamine Dependents: A Randomized Controlled Trial. Frontiers in Psychiatry, 12:653229. Read the complete original research article.