In this study, two groups of Italian adults were recruited: a control group and a group of artists from the Florence Academy of Fine Arts. The control group was recruited from the local university using convenience and snowball sampling methods, while the artists were specifically recruited from the academy. The inclusion criteria for both groups were age between 18 and 40 years, being Italian speakers, and providing written informed consent. The exclusion criteria for both groups included illiteracy and the inability to provide consent or complete the survey. The same questionnaire and creativity tests were administered to both groups in person, under controlled laboratory conditions, ensuring consistent procedures and anonymity standards. The study was conducted over a period of seven months, from May 2023 to May 2024. The anonymity of participants was always ensured, and written informed consent was collected before enrollment. The study procedure was approved by the local University Ethical Committee (code 893197).

Self-reported data encompassed socio-demographic details, personal or familial self-report of psychiatric conditions. In the artists group, the main artistic interests were recorded (up to three per participant).

The primary measure was the Aberrant Salience Inventory (ASI), a validated self-report scale assessing the propensity for aberrant salience attribution, a hallmark of psychosis spectrum disorders (Cicero et al., 2010). The ASI evaluates behavioral manifestations of altered salience processing, which are relevant to understanding the impact of psychosis-related conditions on behavior. It is a 29-item self-report scale with binary responses (“Yes”/"No”), evaluated alterations in salience. The ASI has demonstrated robust internal consistency, with Cronbach’s alpha values of 0.91 in psychotic populations and 0.80 in controls. The ASI scores vary from 0 to 29, where higher scores indicate increased psychosis vulnerability; the best cutoff for psychosis has been identified to be 14 (Merola et al., 2023). Recent meta-analytical findings have redefined its structure into three subscales: sensory sharpening, unveiling experiences, and heightened interpretation and emotionality (Merola et al., 2023).

The Big Five Inventory (BFI, Ubbiali et al., 2013) is a tool used to assess the five major dimensions of personality, based on the Big Five personality model. It includes 44 items. The BFI demonstrates satisfactory internal consistency, with Cronbach’s alpha values ranging from 0.69 to 0.83 across subscales. This inventory is designed for quick self-reporting while still effectively measuring the broad spectrum of personality traits.

The Italian version of the Obsessive Beliefs Questionnaire (OBS, Obsessive Compulsive Cognitions Working Group, 2003) evaluates central cognitive domains associated with obsessive-compulsive symptoms using a 46-item scale with a 7-point Likert response format. The subscales of the OBS include Perfectionism, Responsibility for Damage, Control of Thoughts, Responsibility for Omission, and Importance of Thoughts. Internal consistency for the subscales in control cases ranges from 0.82 to 0.93, indicating good to excellent reliability across all dimensions.

The Remote Associates Test (RAT, Salvi et al., 2020) is a creativity test used to measure the ability to form associations between seemingly unrelated concepts. Participants are given three words (e.g., “cream,” “skate,” “water”) and must find a fourth word (“ice”) that connects them all. This test evaluates the participant’s creative and associative thinking abilities. The test is scored based on the number of correct associations made within a time limit. The RAT, available in multiple languages, has demonstrated excellent internal consistency, with a Cronbach’s alpha of 0.89.

In the Anagrams Test (ANAG, Salvi et al., 2020) participants are presented with a single word at a time and must rearrange its letters to form a different word (anagram). This test evaluates cognitive flexibility, lexical retrieval, and verbal fluency, providing insights into linguistic processing and creative problem-solving abilities. The test scores are based on the number of the words formed. The ANAG also shows excellent internal consistency, with a Cronbach’s alpha of 0.92.

An a priori power analysis was conducted to determine the sample size required for comparing the distributions of variables between artists and controls using the Mann–Whitney U test (McKnight and Najab, 2010). The analysis assumed two-tailed tests, a significance level (alpha) of 0.05, a desired power of 0.80, and an effect size of 0.6, which corresponds to a medium effect size according to Cohen’s benchmarks (Muller, 1989).The power analysis indicated that a sample size of 44 participants per group was required to achieve the desired power. Therefore, the total sample size needed for the study was 89 participants.

To compare the distributions of the variables between the artists and control groups, a Mann–Whitney U test was performed. This non-parametric test was chosen to account for potential violations of normality assumptions and the ordinal nature of some variables. The characteristics of the sample were summarized as descriptive statistics (mean and SD reported, Table 1).

To explore the relationships between the variables, a Spearman correlation analysis was conducted. The analysis was performed separately for the artists group and the control group. The correlation analysis was corrected for age, gender and education, and the it involved the following variables of interest: ASI, OBS, Neuroticism, Extraversion, Agreeableness, Conscientiousness, Openness, Group (binary variable indicating whether the participant belonged to the artists group or the control group), RAT, ANAG. p-values were corrected for FDR.

The regression analysis aimed to identify key predictors of ASI scores in the artists group. Two models were tested: the first examined demographic variables (age, gender, education, religiosity, and family psychiatric history), and the second focused on the five personality dimensions from the Big Five Inventory (Neuroticism, Extraversion, Agreeableness, Conscientiousness, and Openness).

A stepwise regression approach was then used to select the best combination of five predictors from the total aforementioned 10 variables. The final model was evaluated for assumptions such as linearity, multicollinearity, and normality of residuals, with fit assessed using R² values. The computations were conducted in Python 3.1.

A total of 123 subjects were recruited, 65 within the control group and 58 in the artists group. Overall, 53% of the sample identified as female (57, 33% among controls and artists respectively), while the remaining portion identified as male. No participants in any group identified as non-binary, transgender, or chose to self-describe their gender. The average age was 27.42 ± 6.511 (controls 29.97 ± 7.7; artists 22.59 ± 4.038). Years in school averaged 15.40 ± 3.662 (controls 20.07 ± 1.888; artists 15.0 ± 2.817). In the artists group, participants’ main interests were the following: sculpture (68%), painting (54%), music (31%), illustration (18%), graphic design (12%), and others (5%).

Comparison of the two groups revealed that artists exhibited significantly higher ASI scores compared to controls (artists: 18.5 ± 4.3; controls: 10.2 ± 3.7; p < 0.001), suggesting an elevated vulnerability to psychosis-related aberrant salience. The Mann–Whitney test results are displayed in Table 1.

Partial correlation analysis with gender, age, and education as covariates (and FDR correction) revealed several significant results (see Figures 1, 2). In the artist group, ASI was significantly positively correlated with Openness (r = 0.46, p = 0.0012). OBS showed a significant positive correlation with Neuroticism (r = 0.54, p < 0.001). Neuroticism was significantly negatively correlated with Extraversion (r = −0.50, p < 0.001) and Agreeableness (r = −0.43, p = 0.00281). Extraversion and Agreeableness exhibited a significant positive correlation (r = 0.41, p = 0.00429). Additionally, RAT was significantly positively correlated with ANAG (r = 0.39, p = 0.0231). See Supplementary Table 1 for more details.

In the control group, ASI showed significant positive correlations with OBS (r = 0.34, p = 0.0358) and Openness (r = 0.39, p = 0.0177). OBS was significantly positively correlated with Neuroticism (r = 0.40, p = 0.015). Neuroticism exhibited a significant negative correlation with Agreeableness (r = −0.48, p < 0.001) and Conscientiousness (r = −0.34, p = 0.0358). Extraversion was significantly positively correlated with Agreeableness (r = 0.36, p = 0.0283) and Conscientiousness (r = 0.38, p = 0.0247). Agreeableness also showed a significant positive correlation with Conscientiousness (r = 0.36, p = 0.0283). More details can be found in Supplementary Table 2.

A partial correlation analysis was conducted on the entire sample as well, controlling for age, gender, and education (see Supplementary Figure 1). Significant associations were found between ASI and both Openness and OBS.

The regression models’ results are shown in Table 2. The first model, which focused on age, gender, education, religiosity, and family psychiatric history, explained 24.5% of the variance in ASI (R² = 0.245, adjusted R² = 0.181; F = 3.831, p = 0.005). Significant predictors included religion (β = 3.755, p = 0.005), family psychiatric history (β = 2.816, p = 0.022), years of education (β = −0.625, p = 0.018), and age (β = 0.356, p = 0.022). Gender showed a marginal effect (β = −2.210, p = 0.089).

The second model, based on the Big Five personality traits, explained 28.2% of the variance (R² = 0.282, adjusted R² = 0.221; F = 4.639, p = 0.001). Significant predictors included Openness (β = 5.211, p < 0.001), with Conscientiousness almost reaching significance (β = −1.490, p = 0.051). Other traits, including neuroticism, agreeableness, and extraversion, were non-significant.

The stepwise regression combining variables from both models improved the fit (R² = 0.340, adjusted R² = 0.284; F = 6.086, p < 0.001). Key predictors included Openness (β = 4.729, p < 0.001) and Conscientiousness (β = −1.511, p = 0.039), with religiosity showing a trend (β = 2.388, p = 0.051).

The study investigates the intersection of creative behavior, psychopathological traits, and aberrant salience—a transdiagnostic marker of psychosis vulnerability—providing insights into the behavioral and neuropsychological underpinnings of psychosis-related conditions.

The investigation examines key constructs such as aberrant salience, personality traits, and creativity using validated psychometric tools, emphasizing their relevance in the psychotic spectrum and behavioral neuroscience. By focusing on artists, a group with a heightened propensity for altered salience experiences, we extend the understanding of how these traits manifest in creative communities, which are both highly relevant to psychosis research and behavioral conditions. Our findings contribute to identifying risk factors for psychosis, highlighting the roles of personality and cultural influences.

The manuscript offers valuable implications for early detection and intervention strategies in populations at risk for psychosis, aligning with the journal’s emphasis on translational neuroscience. Additionally, our research enhances the understanding of creativity and its neurobiological links to psychopathology, contributing to the broader discourse on behavioral conditions within the journal’s scope.