Self-Reported Drug Use and Creativity: (Re)Establishing Layperson Myths

Humphrey and colleagues begin by situating the study within a longstanding lay belief that drug use enhances creativity. Earlier research has produced mixed and often weak evidence: some experimental and correlational studies report self-perceived links between substances (alcohol, marijuana, “harder” drugs, and psychotropics) and creative thinking, while objective measures of creative performance typically show little or no benefit and sometimes harm. Several studies suggest expectancy or placebo effects may account for apparent gains, and personality—particularly openness to experience—covaries with both drug use and creativity, complicating interpretation. This study set out to test whether self-reported drug use predicts creativity beyond openness to experience. Using a multi-method approach, the investigators combined multiple self-report creativity measures with a performance-based photo-caption task rated by trained judges, and assessed legal, illegal, and psychotropic drug use (recent and lifetime). The authors tested three hypotheses: H1 that people who report using legal/illegal drugs will also report higher creativity; H2 that drug use will show a weak relationship with behavioural performance on the photo-caption task; and H3 that psychotropic drug use will not be associated with self-reported or behavioural creativity. Openness to experience was included as a control variable to test incremental prediction by drug-use factors.

Participants were 787 undergraduate students at a public university in Southern California (672 females, 101 males, 14 missing), mean age 24.42 years (SD = 7.21), with 79.15% aged 19–24. Students completed self-report questionnaires for course credit and performed a photo-caption creativity task. Personality was measured with the Ten-Item Personality Inventory; the two openness items (“Open to new experiences, complex” and “conventional, uncreative”, reverse-scored) were used as the control measure. Drug-use assessment comprised three components. Psychotropic drugs: a checklist of 62 commonly prescribed psychotropic medications (from a NAMI list) with participants reporting use over the last 6 months and over the lifetime on a 7-point frequency scale (ranging from Never to Daily without prescription). Legal and illegal recreational drugs: items from the FIPSE Alcohol and Drug Survey Long Form asking about use in the last month and lifetime on a 7-point frequency scale (None to 16 or more). The authors note an inverted “J” distribution for psychotropic items (about 95% zeros) and describe aggregation strategies to address low base rates. Creativity was assessed with three self-report measures and one performance measure. Self-Assessment of Creativity (SAC), a 6-item scale (5-point Likert), Everyday Creative Behaviours (22 yes/no items) and Artistic Creative Behaviours (20 yes/no items) comprised the self-report/behavioural checklists. The performance task required writing a caption for an ambiguous photograph; captions were rated for creativity using the Consensual Assessment Technique by five quasi-expert raters (inter-rater ICC = .82). For statistical analysis the investigators used Mplus. They began with Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) for ordered categorical variables, estimated with Weighted Least Squares Mean and Variance-Adjusted (WLSMV) methods. Measurement models were estimated separately for psychotropic items, the FIPSE drug items, and the creativity measures. After establishing latent structures, they performed Structural Equation Modelling (SEM) to test whether the drug-use latent variables predicted the creativity measures while controlling for openness. To reduce model complexity and improve identification, the authors aggregated items into subscales for use in the SEM (e.g. psychotropic items collapsed into four summary variables, illegal-drug items summed into alcohol/marijuana, lifetime illegal use, and last-month illegal use). Reported fit indices (chi-square, df, RMSEA, CFI, SRMR) were used to evaluate model fit.

Measurement modelling of psychotropic items indicated very low endorsement: roughly 95% of responses were zero, leading the authors to aggregate items into four summary variables (last-month with prescription, last-month without prescription, lifetime with prescription, lifetime without prescription). CFA compared three structures and the best-fitting model grouped prescription and non-prescription psychotropic use as two correlated factors (chi2 = 9.86, df = 1, RMSEA = .10, CFI = .99, SRMR = .02); the two factors correlated r = .43. EFA on the FIPSE alcohol and drug items yielded a four-factor solution (parallel analysis cutoff eigenvalue = 1.38; four factors explained 76% of variance). Fit indices supported the four-factor solution (chi2 = 809.17, df = 227, RMSEA = .06, CFI = .95, SRMR = .08). The resulting factors were interpreted as: (1) drug use within the last month, (2) alcohol consumption regardless of time frame, (3) lifetime drug use, and (4) a pattern combining amphetamine and wine use. For the creativity measures CFA showed that a four-factor model fit best, separating rated photo captions, everyday creativity, self-reported creativity, and artistic creativity (chi2 = 2757.9, df = 1319, RMSEA = .03, CFI = .92). Following measurement model derivation, the SEM tested direct paths from drug-use latent/subscale variables to the four creativity measures while including openness to experience as a predictor. The SEM fit well (chi2 = 408.9, df = 230, RMSEA = .03, CFI = .97, SRMR = .03). Key findings were: self-reported lifetime illegal drug use was significantly, positively associated with rated caption creativity; alcohol and marijuana use were positively associated with self-reported everyday creative behaviours; illegal drug use in the last month was positively associated with self-reported artistic creative behaviours; and psychotropic drug use showed a negative association with self-reported artistic creative behaviours. Openness to experience was the strongest positive predictor across all creativity outcomes and, in particular, almost perfectly predicted the self-assessed creativity scale (the authors note very high shared variance between openness and self-reported creativity). Reported explained variance (R2) by the model was .13 for rated caption creativity, .35 for everyday creative behaviours, and .14 for artistic creative behaviours. Overall, self-reported drug-use variables explained only a small additional portion of variance in creativity beyond openness.

Humphrey and colleagues interpret the results as largely consistent with prior literature: openness to experience is the dominant predictor of both self-reported and behaviourally rated creativity. For the self-report creativity measure, openness explained so much variance that drug-use variables added little. Nonetheless, there was partial support for the hypothesised associations: alcohol and marijuana use predicted greater self-reported everyday creative behaviours, recent illegal drug use predicted higher self-reported artistic behaviours, lifetime illegal drug use predicted higher rated caption creativity, and psychotropic drug use predicted lower self-reported artistic behaviours. The authors emphasise expectancy and belief-related explanations for some findings: people who believe drugs enhance creativity may report greater creative engagement (self-reports) even when objective performance shows little change, echoing prior placebo/priming findings. The negative association between psychotropic medications and artistic behaviours is discussed in the context of cultural beliefs about the “mad genius” and the possibility that medicated individuals may self-report lower artistic activity; the authors note that evidence linking ADHD medications or other psychotropics to real decrements in creativity is weak. They also consider indirect pathways—for example, rewarding or affective effects of certain drugs could influence creative behaviour indirectly—but concede that their data do not permit testing affective or causal mechanisms. Several limitations acknowledged by the investigators include reliance on self-report measures for drug use and many creativity indices (though they argue self-report is generally reliable in non-clinical populations), the brief two-item openness measure (which may under- or over-state the openness–creativity relationship), the predominantly female sample, and the cross-sectional, correlational design that precludes causal inference. The authors recommend future studies using longer personality inventories, a broader range of objective creativity tasks, experimental designs to test expectancy effects, and attention to affect and gender differences. In closing, the study's findings suggest that while some self-reported drug-use patterns show small associations with creativity measures beyond openness, these effects are modest and may be driven by beliefs and expectancies rather than direct pharmacological enhancement. The authors call for further research to disentangle expectancy, personality, and pharmacological influences on creativity.

The investigators conclude that certain self-reported drug-use measures show small predictive relationships with creativity beyond openness to experience, but overall drug variables account for little additional variance. Beliefs and expectancies about drugs and creativity likely play an important role in self-reported creative behaviour, and may influence decisions to try illegal substances or to alter prescribed medication use. The inconsistencies between self-reported creativity measures and the objective rated caption task underscore the need to study expectancy effects and to employ multiple objective creativity measures in future research.