Sources · Tested · Episode 3
Does Weed Actually Make You Creative?
Every figure shown on screen in this episode, with the document, page and date it comes from, and a label for how strong the evidence is.
In the studies reviewed, cannabis did not reliably make ideas better, and in one placebo-controlled experiment performance was worse at the higher amount. The popular counter-claim, that it simply makes people rate their own ideas higher, holds only partly.
- Published
- 2 October 2026
- Length
- 8:13
- Figures checked
- 27 September 2026
- Documents
- 5
Verdict
“Cannabis reliably makes creative work better”
“Cannabis changes how positively ideas are judged”
On screen in this episode
29The document shows it, and you can check it yourself.
10Something that matters is missing from every source we checked.
What isn’t shown
- 1:57WHAT WAS TESTED · A specific group. A specific word task. · NEVER HELPS · GENERAL UPGRADE · WHAT SURVIVES A TOUGHER TEST? · The finding does not establish either universal claim
- 2:47NO DETECTED BENEFIT ≠ PROOF OF ZERO EFFECT
- 3:38STATISTICAL MODEL · PROPOSED ≠ PROVEN · A statistical model does not establish a biological mechanism
- 4:55USERS · NON-USERS · WHO THEY ARE ≠ WHAT HAPPENS WHEN HIGH · Illustrative people · Not participant portraits or counts
- 5:25OBSERVATIONAL ≠ CAUSAL
- 5:45A BRIEF BRICK TASK · One constrained response · A FINISHED ALBUM · Selection · revision · collaboration · A short task samples part of creativity
- 6:16GROUP RESULT · A group result does not predict every individual. · Illustrative people · No individual outcomes shown
- 6:22LIMITS GO BOTH WAYS · Not proof of a benefit for everyone. · Not proof of no benefit for anyone. · Broad claims need broad evidence.
- 7:06NO ONE COULD EVER BENEFIT? · NOT SHOWN · The reviewed studies do not establish this universal claim. · Verdict applies to this claim and these tests
- 7:22A MEANINGFUL EXPERIENCE · How it felt · DEMONSTRATED IMPROVEMENT · What the comparison shows · MAY · OVERLAP IS POSSIBLE. EQUIVALENCE IS NOT ESTABLISHED. · Conceptual illustration · Connection does not show causation
- 4:27It does not tell us what every occasional user would experience.
On screen, in order
- 1:21
SCHAFER ET AL. · 2012 · A WORD TASK · PREVIOUS CREATIVE ACHIEVEMENTS · Intoxicated day / sober day · LOWER · HIGHER · No placebo
Users were tested after smoking their own cannabis and on a drug-free day. The key comparison is the bottom and top quartiles of the Creative Achievement Questionnaire. The “earlier work” picture is our illustration.
- 1:35
LOWER PREVIOUS ACHIEVEMENT · MORE WORDS · Illustrative task examples · Examples are not participant responses
Verbal fluency: words beginning with a given letter in 60 seconds. The low-achievement group scored higher on the intoxicated day than on the drug-free day. BRIDGE, BUTTON and BICYCLE are our examples.
- 1:40
HIGHER PREVIOUS ACHIEVEMENT · NO SUCH IMPROVEMENT · In this comparison
The high-achievement group’s scores did not differ between the two days.
- [1] Schafer et al., Consciousness and Cognition 2012, p. 294, §3.1.1
- 1:44
A POSITIVE FINDING. WITH AN ADDRESS. · A PARTICULAR GROUP · A WORD TASK · NO PLACEBO · The test counted words. · Not a measure of song quality
- 1:57
WHAT WAS TESTED · A specific group. A specific word task. · NEVER HELPS · GENERAL UPGRADE · WHAT SURVIVES A TOUGHER TEST? · The finding does not establish either universal claim
A gain on one word task in one subgroup shows neither that cannabis never helps nor that it upgrades creativity in general. The category task showed no such day effect.
- Checked[1] Schafer et al., Consciousness and Cognition 2012, pp. 294–295, §3.1.1–3.1.3
- 2:12
CREATIVE USES FOR A BRICK · Illustrative answer · Study 1
Participants listed as many creative uses for a brick as they could in four minutes. The bookend answer is our illustration, not a participant’s idea.
- [2] Heng, Barnes & Yam, J. Applied Psychology 2023, Study 1 Method, Participants and Procedure
- 2:18
STUDY 1 · 191 · Creative uses for a brick
Analysis sample: 107 in the cannabis condition, 84 in the control condition.
- [2] Heng, Barnes & Yam, J. Applied Psychology 2023, Study 1 Method, Participants and Procedure
- 2:22
STUDY 2 · 140 · Revenue ideas for a band
Analysis sample of full-time employees: 62 in the cannabis condition, 78 in the control condition. “OUR EXAMPLE: SELL TICKETS” is our illustration, not a participant’s idea.
- [2] Heng, Barnes & Yam, J. Applied Psychology 2023, Study 2 Method, Participants and Procedure
- 2:27
CONDITION · Raters did not know the condition
Trained research assistants and crowdsourced raters were blind to the experimental condition in both experiments. The reviewer and the label are our illustration.
- 2:36
STUDY 1 · NO SIGNIFICANT BENEFIT DETECTED · STUDY 2 · NO SIGNIFICANT BENEFIT DETECTED · Externally rated creativity
Two sets of ratings in Study 1 and four in Study 2; none differed significantly between the cannabis and control conditions.
- 2:40
VARIED PRODUCTS · EVERYDAY SETTINGS · NO PLACEBO · IMPERFECT COMPLIANCE · Cannabis condition / abstaining control · Methods and limitations
Participants used their own cannabis in their daily lives; none was provided and the comparison was abstinence, not a placebo. In Study 1, among the 117 who completed the saliva test, 72% of the cannabis group tested positive and 64% of the control group tested negative; in Study 2, the check on the cannabis group (an image of their product or receipt) showed 79.0% compliance. The authors say they did not achieve perfect random assignment.
- [2] Heng, Barnes & Yam, J. Applied Psychology 2023, Study 1 Method, Participants and Procedure
- [2] Heng, Barnes & Yam, J. Applied Psychology 2023, Study 1 Results and Discussion (manipulation checks)
- [2] Heng, Barnes & Yam, J. Applied Psychology 2023, Study 2 Results and Discussion (compliance check)
- [2] Heng, Barnes & Yam, J. Applied Psychology 2023, General Discussion (limitations)
- 2:47
NO DETECTED BENEFIT ≠ PROOF OF ZERO EFFECT
Non-significant differences in two experiments do not establish an effect of exactly zero.
- Checked[2] Heng, Barnes & Yam, J. Applied Psychology 2023, Tables 2 and 4
- 2:50
I FELT INSPIRED. · THE STORY · Assess the work without the creator’s story
The raters scored the ideas without knowing the participants’ condition. The reviewer, her notes and the “I felt inspired” label are our illustration.
- 3:05
A RELIABLE IMPROVEMENT IN THE WORK? · DOESN’T HOLD · Not proof of no effect in every person. · Verdict applies to this claim and these tests
Our verdict on the narrow claim, from the external ratings in both experiments. The authors’ combined analysis of all six sets of ratings found d = −.04, 95% CI [−.17, .09].
- 3:13
STUDY 1 · SELF-RATINGS · HIGHER · Cannabis versus control
Mean self-rating 3.52 in the cannabis condition and 3.23 in the control condition on a 1–5 scale; F = 4.98, p < .05.
- [2] Heng, Barnes & Yam, J. Applied Psychology 2023, Study 1 Results and Discussion, Table 2
- 3:17
STUDY 2 · SELF-RATINGS · NO SIGNIFICANT OVERALL DIFFERENCE · Revised rating questions · The self-rating measures are not identical
Global self-rating 3.75 in the cannabis condition and 3.80 in the control condition on a 1–5 scale; F = 0.15. Study 2 added two usefulness items to the Study 1 scale. Idea-level self-ratings did not differ significantly either.
- 3:20
STATISTICAL MODEL · CHEERFUL MOOD - - - MORE FAVOURABLE JUDGMENTS · Modelled indirect association · Mediation model
“Cheerful mood” is the authors’ measure of state joviality. The indirect effect via joviality on self-ratings was significant in Study 1 (.06, 95% CI [.01, .13]) and Study 2 (global: .17, 95% CI [.02, .34]); in Study 2 it also reached ratings of other people’s ideas.
- 3:25
SUPPLEMENT D · SELF-EVALUATIONS · (close-up of the supplement) Main Effect of Cannabis Use on Creativity Self-Evaluations: “…significant in Study 1 but not Study 2. … d = .18, 95% CI = [-.04, .39], Z = 1.58, p = .114, for a fixed effect model, and … d = .16, 95% CI = [-2.43, 2.74], Z = .77, p = .584, for a random effects model.”
The authors’ own paragraph, shown as a close-up of the supplement. Not our calculation.
- [3] Heng et al., Supplement D (OSF), Main Effect of Cannabis Use on Creativity Self-Evaluations (first results paragraph)
- 3:30
SELF-RATINGS · BOTH STUDIES · d = 0.18 · 95% CI: −0.04 to 0.39 · Fixed effect model · p = .114 · NO DIFFERENCE · The interval includes no difference.
The authors’ internal meta-analysis of the two experiments, drawn as a chart; not our calculation. Their random-effects model gave d = .16, 95% CI [−2.43, 2.74], p = .584.
- [3] Heng et al., Supplement D (OSF), Main Effect of Cannabis Use on Creativity Self-Evaluations (first results paragraph)
- 3:32
The mood-related finding remains.
The supplement concludes that the effect on self-ratings is primarily indirect, via joviality.
- 3:38
STATISTICAL MODEL · PROPOSED ≠ PROVEN · A statistical model does not establish a biological mechanism
The indirect effect comes from the authors’ statistical model. Mood was measured, not assigned, so the model does not show a mechanism in the brain.
- Checked[2] Heng, Barnes & Yam, J. Applied Psychology 2023, Study 1 and Study 2 Results and Discussion
- 3:45
CAN IT CHANGE HOW IDEAS ARE JUDGED? · PARTLY · A universal self-rating boost is not established. · A DEBUNK NEEDS EVIDENCE, TOO. · Verdict applies to this claim and these tests
Our verdict. The indirect effect via mood was significant in both experiments; the overall self-rating difference was significant only in Study 1 and not in the combined analysis.
- 3:58
A MORE CONTROLLED COMPARISON · PLACEBO · 18 participants · LOWER AMOUNT · 18 participants · HIGHER AMOUNT · 18 participants · RANDOM ASSIGNMENT · 54 regular users
Randomized, double-blind, placebo-controlled; the placebo was cannabis plant material with very little THC. Of 59 people who took part, 54 were analysed for the alternative-uses task.
- 4:11
HIGHER AMOUNT vs PLACEBO · IDEA FLUENCY · WORSE · VARIETY · WORSE · ORIGINALITY · WORSE · Divergent-thinking measures
Fluency is the number of uses, flexibility (shown as variety) the number of different categories, originality how rare the uses were. Higher amount compared with placebo: p = 0.01, 0.011 and 0.045. The fourth score, elaboration, showed no significant effect.
- [4] Kowal et al., Psychopharmacology 2015, Results: Creativity tasks, Divergent thinking; Table 2
- 4:19
Lower amount: no benefit over placebo.
The lower-amount group showed no significant increase over placebo on any of the four scores.
- [4] Kowal et al., Psychopharmacology 2015, Results: Creativity tasks, Divergent thinking; Table 2
- 4:22
54 · 18 PER CONDITION · Regular users · Mostly men · THIS SAMPLE · Sample limitations
Analysed sample: 48 men and 6 women. Regular use meant at least four times a week for at least two years. The groups differed in their share of women, which the authors list as a limitation. The connecting-word task had 53 participants.
- 4:30
MORE ≠ BETTER · No benefit at the lower amount. Worse performance at the higher amount. · Compared with placebo
- [4] Kowal et al., Psychopharmacology 2015, Results: Creativity tasks, Divergent thinking; Table 2
- 4:40
WHICH COMPARISON? · BETTER THAN THE HIGHER-AMOUNT GROUP · DOES NOT MEAN BETTER THAN PLACEBO
In this study the lower-amount group scored significantly higher than the higher-amount group, but not higher than placebo.
- [4] Kowal et al., Psychopharmacology 2015, Results: Creativity tasks, Divergent thinking; Table 2
- 4:55
USERS · NON-USERS · WHO THEY ARE ≠ WHAT HAPPENS WHEN HIGH · Illustrative people · Not participant portraits or counts
The 2017 study tested users while sober. It compared users with non-users; it did not measure what happens during intoxication. The people are illustrations.
- Checked[5] LaFrance & Cuttler, Consciousness and Cognition 2017, Abstract
- 5:01
LAFRANCE & CUTTLER · 2017 · BOTH GROUPS SOBER · Users compared with non-users · SELF-RATED CREATIVITY · CONNECTING ANSWERS · Observational comparison
412 sober users and 309 non-users. The abstract describes the task as “a measure of convergent thinking”.
- [5] LaFrance & Cuttler, Consciousness and Cognition 2017, Abstract
- 5:12
GROUP DIFFERENCES · Self-ratings / connecting answers · OPENNESS · Differences no longer significant after statistical adjustment. · Openness-adjusted analysis
Controlling for the users’ higher openness to experience removed both differences.
- [5] LaFrance & Cuttler, Consciousness and Cognition 2017, Abstract
- 5:25
OBSERVATIONAL ≠ CAUSAL
An observational comparison of users and non-users; nothing was assigned, so it cannot show which way cause and effect run.
- Checked[5] LaFrance & Cuttler, Consciousness and Cognition 2017, Abstract
- 5:45
A BRIEF BRICK TASK · One constrained response · A FINISHED ALBUM · Selection · revision · collaboration · A short task samples part of creativity
The reviewed tasks were short, single-session tasks: listing uses for a brick or other everyday objects, revenue ideas for a band, or words in 60 seconds. None assessed finished creative works.
- 6:16
GROUP RESULT · A group result does not predict every individual. · Illustrative people · No individual outcomes shown
- 6:22
LIMITS GO BOTH WAYS · Not proof of a benefit for everyone. · Not proof of no benefit for anyone. · Broad claims need broad evidence.
None of the reviewed studies shows a benefit for everyone, and none shows that no one could benefit.
- 6:49
RELIABLY BETTER WORK? · DOESN’T HOLD · Across the tasks reviewed here. · Verdict applies to this claim and these tests
Our verdict across the reviewed tasks: no significant improvement in externally rated creativity (Heng et al.); worse divergent thinking at the higher amount and no benefit at the lower amount (Kowal et al.). The one positive result was a word-task gain in one subgroup, which was not a quality rating (Schafer et al.).
- [2] Heng, Barnes & Yam, J. Applied Psychology 2023, Tables 2 and 4
- [3] Heng et al., Supplement D (OSF), Main Effect of Cannabis Use on Creativity (second results paragraph)
- [4] Kowal et al., Psychopharmacology 2015, Results: Creativity tasks, Divergent thinking; Table 2
- [1] Schafer et al., Consciousness and Cognition 2012, p. 294, §3.1.1
- 6:55
A DIFFERENT JUDGMENT OF IDEAS? · PARTLY · Mood-related findings; inconsistent overall self-ratings. · Verdict applies to this claim and these tests
Our verdict. Indirect effects via mood in both experiments; overall self-ratings higher only in Study 1 and not significant when both experiments were combined.
- 7:06
NO ONE COULD EVER BENEFIT? · NOT SHOWN · The reviewed studies do not establish this universal claim. · Verdict applies to this claim and these tests
Checked all five documents: none tests or establishes that no one could ever benefit. Schafer et al. report a gain on one word task in one subgroup.
- 7:22
A MEANINGFUL EXPERIENCE · How it felt · DEMONSTRATED IMPROVEMENT · What the comparison shows · MAY · OVERLAP IS POSSIBLE. EQUIVALENCE IS NOT ESTABLISHED. · Conceptual illustration · Connection does not show causation
In Heng et al., self-ratings rose in one experiment while externally rated creativity did not. Kowal et al. note that the experience of intoxication may not reflect actual performance. The diagram is a conceptual illustration.
Also said in the narration
- 1:30
Its key comparison used groups with higher and lower previous creative achievements.
90 participants from the top and bottom quartiles of the Creative Achievement Questionnaire.
- [1] Schafer et al., Consciousness and Cognition 2012, p. 294, §2.3.2 and §2.4
- 1:35
In the lower-achievement group, people produced more words beginning with a given letter while intoxicated.
- 1:41
The higher-achievement group did not show that improvement.
- [1] Schafer et al., Consciousness and Cognition 2012, p. 294, §3.1.1
- 2:27
Participants were assigned to complete the study after their usual cannabis use or after abstaining.
- 3:13
The same paper found higher self-ratings in the first experiment, but not the second.
- [2] Heng, Barnes & Yam, J. Applied Psychology 2023, Tables 2 and 4
- 3:59
A Dutch experiment randomly assigned regular cannabis users to a placebo, a lower amount of cannabis, or a higher amount.
Approved by the Leiden University Medical Center ethics committee; participants were native Dutch speakers, recruited in part through Dutch coffee shops.
- 4:27
It does not tell us what every occasional user would experience.
All participants were regular users. The authors write that effects in occasional or drug-naïve users can only be speculated about.
- Checked[4] Kowal et al., Psychopharmacology 2015, Conclusion
- 5:05
Users rated themselves as more creative and performed better on a task that required finding a connecting answer.
The abstract describes the task as “a measure of convergent thinking”; the full text was not checked.
- [5] LaFrance & Cuttler, Consciousness and Cognition 2017, Abstract
- 7:12
The positive word-task finding belongs in the picture. So do the null results, the poorer performance in the higher-amount group, and the limits of all those tasks.
Documents
- Page numbers are the documents’ own printed page numbers where the checked copy carries them (Schafer et al.).
- Heng, Barnes & Yam and Kowal et al. are cited by section, table and figure. The copy of Heng et al. checked is the advance online version, paginated 1–12, whose page numbers differ from the printed issue (pp. 635–646). The Kowal et al. text checked is the PubMed Central version, which has no page numbers. Supplement D has no page numbers either.
- In the video, most study cards carry a DOCUMENTED label on screen. Here DOCUMENTED means a finding the study reports for its own setup, not proof that a hypothesis is true.
- The verdicts are our reading of these studies. Each applies only to the claim and the tests named on its card.
- The d, confidence interval and p values on screen come from the authors’ own combined analysis in Supplement D. The episode makes no calculations of its own.
- Illustrations and marked examples are ours and are not evidence: the lamp, the spaghetti umbrella, the brick-as-bookend answer, BRIDGE / BUTTON / BICYCLE, “OUR EXAMPLE: SELL TICKETS”, the reviewer, the people, the camera analogy, the album and the proposed better study.
- Only the published abstract of LaFrance & Cuttler (2017) was checked.
- [1]
Elsevier, Consciousness and Cognition (open copy in UCL Discovery)
Published March 2012 · Retrieved 25 September 2026
Issue date. Available online on 9 January 2012.
160 cannabis users were tested on a day when they had smoked their own cannabis and on a drug-free day; there was no placebo. The key comparison is between the top and bottom quartiles (90 people) of a questionnaire on previous creative achievements. The abstract and the methods (§2.4) give the two subgroup sizes, 43 and 47, the other way round; the video does not use them.
Used at 1:21 · 1:30 · 1:35 · 1:40 · 1:41 · 1:44 · 1:57 · 5:45 · 6:22 · 6:49 · 7:06 · 7:12
- [2]
American Psychological Association, Journal of Applied Psychology
Published April 2023 · Retrieved 25 September 2026
Issue date. Published online on 28 July 2022; the copy checked for this episode is that advance online version, paginated 1–12.
Two randomized field experiments with adult users in a U.S. state where recreational cannabis is legal. Participants used their own cannabis in their own setting or abstained; the researchers provided no cannabis and there was no placebo. The raters who scored the ideas did not know the participants’ condition. Compliance with the assigned condition was imperfect.
Used at 2:12 · 2:18 · 2:22 · 2:27 · 2:36 · 2:40 · 2:47 · 2:50 · 3:05 · 3:13 · 3:17 · 3:20 · 3:32 · 3:38 · 3:45 · 5:45 · 6:16 · 6:22 · 6:49 · 6:55 · 7:06 · 7:12 · 7:22
- [3]
Heng, Barnes & Yam, Open Science Framework (the article’s supplementary materials)
Supplement D: Internal Meta-Analyses (Supplement to doi:10.1037/apl0000599 · OSF project 3hbd6)
Published 23 June 2022 · Retrieved 25 September 2026
Upload date of the file on OSF. The article links the OSF project as its supplementary material.
The authors’ own combined analysis of their two experiments. The supplement has no page numbers; rows cite its paragraph headings.
Used at 3:05 · 3:25 · 3:30 · 3:32 · 3:45 · 6:49 · 6:55 · 7:06
- [4]
Springer, Psychopharmacology (open access in PubMed Central)
Published March 2015 · Retrieved 25 September 2026
Issue date. Published online on 7 October 2014.
Randomized, double-blind, placebo-controlled experiment with three groups of regular users in the Netherlands. The conflict-of-interest statement says one author received a salary from Bedrocan BV, the company that supplied the study cannabis.
Licence: Creative Commons Attribution License
Used at 3:58 · 3:59 · 4:11 · 4:19 · 4:22 · 4:27 · 4:30 · 4:40 · 5:45 · 6:16 · 6:22 · 6:49 · 7:06 · 7:12 · 7:22
- [5]
Elsevier, Consciousness and Cognition (abstract in PubMed)
Published November 2017 · Retrieved 25 September 2026
Issue date. Published online on 21 October 2017.
Observational comparison of 412 cannabis users, tested while sober, and 309 non-users. Only the published abstract was checked for this episode; the full text was not.
Corrections
No corrections so far.
Found an error? Email [email protected] with the time stamp and your source, or leave a comment under the video.