LSD and creativity: Increased novelty and symbolic thinking, decreased utility and convergent thinking

Creativity is important across society and there has been renewed interest in whether serotonergic psychedelics enhance creative cognition. Previous work has been limited by small samples, varied operational definitions of creativity, naturalistic settings, lack of placebo controls and inconsistent dosing, producing mixed findings for measures such as fluency, originality and convergent thinking. Semantically oriented studies have reported increased semantic priming and larger semantic distances under psychedelics, while behavioural reports describe altered artistic output and more vivid, sometimes bizarre, imagery. Wießner and colleagues set out to address these limitations by testing the effects of a relatively low dose of LSD (50 μg) in a randomised, double-blind, placebo-controlled, crossover design. They applied a multimodal creativity battery covering visual and verbal production, scored responses using theory-driven criteria (novelty, utility, surprise), established divergent and convergent thinking metrics, semantic-distance indices and a data-driven content analysis to capture features such as symbolic thinking and ambiguity. The study aimed to map how LSD alters multiple dimensions of creative cognition and behaviour under controlled conditions.

The study used a randomised, double-blind, placebo-controlled, crossover design with two sessions (LSD 50 μg, inactive placebo) separated by a 14-day washout; treatment order was randomised. Twenty-five healthy volunteers were recruited and one withdrew after the first session, leaving 24 participants (8 women; mean age 35 ± 11 years, range 25–61). Inclusion criteria included age ⩾22, at least one prior LSD experience and specified abstinence windows for substances; exclusion criteria included psychiatric symptoms or family history of psychosis, use of psychiatric medication, substance use disorders, relevant medical conditions and non-native Brazilian Portuguese. LSD was given orally dissolved in an alcohol solution, diluted in water, with the dose chosen to produce noticeable effects while keeping participants able to perform tasks. Procedures followed ethics approvals and safety guidelines. Baseline measures began at 07:30, drug/placebo was administered at 09:30, and a standardised snack was given at 11:00. Creativity testing took place around drug peak, from 12:00 to 13:15, and participants remained supervised until 17:30. To reduce learning effects, two parallel task versions were counterbalanced across participants and sessions. Five creativity tasks were administered: the pattern meaning task (PMT; generate interpretations for abstract line patterns), alternate uses task (AUT; uncommon uses for everyday objects), picture concept task (PCT; convergent selection of pictures and divergent alternate picture combinations), creative metaphors task (MET; compose metaphors) and a figural creativity task (FIG; drawings from simple patterns and titles). Responses were evaluated across four analytic approaches: theory-driven creativity criteria (novelty, utility, surprise rated on 0–2 scales, with relative measures computed), established divergent and convergent thinking metrics (fluency, originality, flexibility, elaboration; convergent thinking operationalised as correct combinations in PCT), semantic-structure indices (semantic spread and semantic steps computed from FastText on a Portuguese Wikipedia corpus) and a data-driven special-features content analysis (categories for content, techniques, ambiguity and symbolic thinking, with relative measures computed). Two trained raters independently scored subjective variables; interrater reliability (ICC) was reported as moderate to excellent for most measures. Statistical analysis used repeated-measures general linear models with treatment (within-subject) and treatment order (between-subject) factors, evaluating treatment, period (session 1 vs 2) and order effects. Effect sizes were reported as partial eta squared. Multiple comparisons within variable groups were corrected by the Benjamini–Hochberg false discovery rate (q = 0.05). Spearman correlations were computed for LSD–placebo differences, with significance threshold adjusted by number of tasks (α = 0.01).

The final sample comprised 24 participants. Interrater reliability for subjectively rated measures was mostly moderate to excellent. Results varied by task and variable; several effects survived correction while others did not. On theory-driven creativity criteria, LSD increased measures of novelty and surprise primarily in the PMT: novelty (p = 0.038), relative novelty (p < 0.001) and relative surprise (p = 0.016). An increase in relative novelty for AUT (p = 0.034) did not survive multiple-comparisons correction. By contrast, LSD reduced utility measures: utility in PMT (p = 0.021) and relative utility in PMT (p < 0.001) and AUT (p = 0.008). Some period effects consistent with habituation or learning appeared in PCT and other tasks, but several period/order effects did not survive correction. Findings for convergent and divergent thinking showed task-dependent effects. The extracted Results text contains an inconsistency: one Results paragraph reports that LSD increased convergent thinking in PCT (p = 0.023), whereas the Discussion interprets convergent thinking as decreased under LSD. The extracted material does not clearly resolve this discrepancy. Across tasks, elaboration tended to decrease at an uncorrected level (PMT, AUT, MET, FIG), while fluency generally did not change; title originality increased but other originality parameters often did not survive correction. Semantic-structure indices showed increases under LSD for PMT: semantic spread (p = 0.047) and semantic steps (p = 0.025), indicating larger semantic distances for interpretations of abstract patterns. No other significant semantic effects were reported. The data-driven special-features analysis indicated increased ambiguity in metaphors (MET; p = 0.028) and drawings (FIG; p = 0.044) and increased relative ambiguity in MET (p = 0.039). Symbolic thinking increased across several measures: PMT (p = 0.036), AUT (p = 0.022), FIG title symbolism (p = 0.011), FIG picture symbolism (p = 0.022) and FIG colour abstractness (p = 0.001); some symbolic-thinking effects in PCT and several content/technique effects were reported but did not survive correction for multiple testing. Tendential changes included fewer depicted objects and more sensory content in metaphors, more contrast in metaphor technique and increased frame breaking in drawings, though some of these did not survive correction. Correlation analyses reported that novelty correlated positively with surprise and both correlated negatively with utility (PMT). Utility correlated positively with elaboration (PMT) and negatively with symbolic thinking (AUT). Surprise correlated positively with symbolic thinking (AUT). These relationships supported the authors’ grouping of effects into related phenomena.

Wießner and colleagues interpret the pattern of findings as indicating three overarching LSD-induced phenomena. First, a 'pattern break' characterised by increased novelty, surprise, some originality measures and larger semantic distances, particularly for abstract stimuli. Second, decreased 'organisation', evidenced by reduced utility of responses, impairments in evaluative processes such as convergent thinking (noting the inconsistency in the extracted Results as discussed above) and a consistent, marginal reduction in elaboration across tasks. Third, an increased emphasis on 'meaning', reflected in higher symbolic thinking, greater ambiguity and changes in verbal techniques. The researchers link these outcomes to dose and stimulus modality: effects were most pronounced in the PMT near drug peak and for abstract stimuli, suggesting that LSD facilitates generation of remote or novel ideas but may impair cognitive control processes necessary for detailed elaboration and evaluation. They situate their results within prior literature showing variable effects depending on dose and context, noting that microdoses have sometimes yielded improvements while higher doses commonly impair some cognitive functions. On semantic measures, the authors argue that abstract input more readily elicits semantically dispersed thinking under LSD, whereas concrete stimuli showed less semantic spread. Behaviourally, metaphors and drawings under LSD tended to contain fewer objects, more sensory emphasis, increased contrast techniques and greater colour abstractness; these patterns may reflect expression of intensified subjective experiences. Symbolic thinking and ambiguity were highlighted as cross-modal and potentially therapeutically relevant features, since symbolisation may help express unconscious material and support psychological processing in psychedelic-assisted therapy. The investigators suggest that abstract stimuli and drawing tasks could be useful therapeutic tools to elicit symbolisation. Several limitations are acknowledged: the modest sample size coupled with assessment of many variables increases the false-positive risk despite correction for multiple comparisons; newly introduced special-feature variables require psychometric validation; some classifications had small subgroup sizes and limited interrater reliability for certain measures; cultural mismatches in PCT stimuli could have influenced performance in the Brazilian sample; and the construct of symbolic thinking was broadly defined due to limited response counts. The authors recommend replication, refinement of measures and careful disentangling of constructs such as symbolism and ambiguity in future research.

Wießner and colleagues conclude that a low dose of LSD alters creativity across multiple modalities and measurement approaches. Their findings suggest a shift of cognitive resources away from conventional, organised processing and towards novelty and meaning: increased novelty and semantic dispersion (pattern break), reduced organisation and evaluative detail (decreased utility, convergent thinking and elaboration), and enhanced symbolic thinking and ambiguity (meaning). The authors propose that these effects do not merely disrupt cognition but reallocate it toward the new, with potential implications for therapeutic applications that harness symbolisation and abstract expression.