Magical thinking in individuals with high polygenic risk for schizophrenia but no non-affective psychoses—a general population study

Familial liability is the strongest single risk factor for schizophrenia, but most individuals with elevated genetic susceptibility never develop psychosis. Polygenic risk scores (PRS) derived from GWAS capture common variant liability for schizophrenia, yet their clinical specificity is relatively low, implying broader impacts beyond frank psychotic disorders. Prior work on schizophrenia-susceptible genotypes without psychosis is limited and mixed, with some associations to brain function and select psychosocial traits. Magical thinking—beliefs in telepathy, miracles, mystical events, or a sixth sense—is a facet of positive schizotypy and can occur in various psychiatric contexts below the threshold of psychosis. The study asks whether, among individuals free of non-affective psychoses, higher schizophrenia PRS predicts elevated magical thinking and distinct developmental trajectories of such thinking from early to middle adulthood.

GWAS have identified numerous loci for schizophrenia (initially 108 loci/128 SNPs; later 176 loci/208 associations; most recently 270 loci across ancestries), with some population-specificity and contributions from rare coding variants. Estimates of variance explained by PRS vary widely. Evidence among schizophrenia-susceptible individuals without psychosis has focused on brain outcomes (e.g., no consistent structural changes; altered frontal activity during cognitive tasks) and mixed findings on cognition and health behaviors. Psychosocial outcomes linked to higher PRS include higher trait anxiety, creativity, and greater number of children in women, with null findings for apathy and cohabitation duration. Magical thinking is common in the general population in milder forms and is related to schizotypy, attenuated psychosis syndrome, and other psychiatric contexts. However, longitudinal links between magical thinking and psychosis proneness remain unclear, motivating this investigation.

Design and sample: Population-based Young Finns Study (YFS), a prospective cohort initiated in 1980 with repeated follow-ups (1983, 1986, 1989, 1992, 1997, 2001, 2007, 2012, 2017). From the original 3,596 participants across six birth cohorts (1962–1977), this study included 1,292 participants with data on schizophrenia PRS, lifetime psychiatric diagnoses, magical thinking (1997, 2001, or 2012), childhood family circumstances (1980/1983), and adulthood socioeconomic factors (2011). Participants with non-affective psychotic disorders were excluded (n = 74). Ethical approvals were obtained and informed consent provided. Genotyping and PRS: Genotyping of 2,556 samples was performed on Illumina Human 670k BeadChip. Quality control filters included sample call rate <0.95, excess heterozygosity, sex mismatch, cryptic relatedness (pi-hat >0.2), SNP call rate <0.95, MAF <0.01, HWE p<1e-6; post-QC: 2,443 samples and 546,677 SNPs. Imputation to 1000 Genomes (March 2012) via SHAPEIT v1 and IMPUTE2; SNPs with info >0.3 retained. PRS was constructed using 128 independent genome-wide significant schizophrenia SNPs from the Psychiatric Genomics Consortium GWAS, computing both weighted (by beta estimates) and unweighted allele counts; PRS was standardized (mean 0, SD 1). Magical thinking outcome: Assessed with the Spiritual Acceptance subscale (13 items, 5-point Likert) of Self-Transcendence in the Temperament and Character Inventory in 1997, 2001, and 2012, capturing beliefs in telepathy, miracles, mystical events, and sixth sense. Mean scores were computed for respondents with ≥50% item completion (most had complete data). Scores were standardized to the 1997 mean/SD and treated as time-varying outcomes. Psychiatric diagnoses: Lifetime hospital-care diagnoses (ICD-8/9/10 mapped to DSM-IV) from the Finnish Care Register up to 2017 (participants aged 40–55), grouped as non-affective psychoses, personality, affective, and substance-related disorders; only non-affective psychoses were used for exclusion. The register captures ~93% of schizophrenia-spectrum psychoses and 97% of psychotic disorders. Covariates: Adulthood socioeconomic factors (2011): income (13-point scale), occupational status (manual/lower non-manual/upper non-manual), education (comprehensive/occupational or high school/academic). Childhood environment (parent questionnaires in 1980; imputed from 1983 if missing): cumulative indices for stressful life events, adverse socioeconomic circumstances, and unfavorable emotional family atmosphere, based on multiple indicators (e.g., parental education/occupation/income, family disruptions, parental mental disorder/alcohol intoxication, child hospitalizations). Statistical analysis: Participants with non-affective psychoses were excluded first. Attrition was examined via t-tests and chi-square tests. Longitudinal associations between PRS and magical thinking (1997–2012) were modeled using growth-curve (multilevel) models with maximum likelihood. Fixed effects included PRS, age (centered at 20 years), age-squared, sex, childhood indices, and adulthood socioeconomic covariates. Model 2 additionally included PRS×age interaction. Random effects included intercept variance and residual variance. Bosker/Snijders pseudo R² values were reported for Level 1 and Level 2. Additional analyses: (a) logistic regression of PRS predicting non-affective psychosis to validate PRS; (b) excluding participants with affective disorders; (c) analogous models for other Self-Transcendence subscales (Self-forgetful experiences; Transpersonal identification).

- Attrition: Included participants were slightly older and more often women; they had slightly lower magical thinking in 1997 and 2012, less adverse childhood environments, and higher adulthood SES than dropouts; no attrition bias for PRS. - Main effect: In Model 1, higher PRS predicted higher magical thinking across adulthood among participants without non-affective psychoses: weighted PRS B = 0.077, p = 0.001; unweighted PRS B = 0.082, p = 0.001. Magical thinking showed a curvilinear age trajectory (age B = −0.033, p < 0.001; age² B = 0.001, p < 0.001). - Age interaction: In Model 2, PRS×age was significant (weighted PRS×age B = 0.003, p = 0.034; unweighted PRS×age B = 0.004, p = 0.011), indicating that individuals with low PRS showed a steady decrease in magical thinking from ages 20 to 50, whereas in those with high PRS the decrease attenuated in middle age, yielding higher-than-expected magical thinking levels. - These associations were adjusted for sex, childhood family environment, and adulthood socioeconomic factors. Random intercept variance was significant; pseudo R² values were similar across weighted and unweighted PRS models (e.g., Model 1 Level 1 ≈ 0.114–0.115; Level 2 ≈ 0.120–0.121). - PRS validation: Logistic regression showed PRS predicted higher odds of non-affective psychoses: weighted PRS OR = 1.585, p = 0.005; unweighted PRS OR = 1.536, p = 0.010. - Sensitivity analysis: Excluding participants with affective disorders replicated the main results. - Additional outcomes: Higher PRS predicted higher self-forgetful experiences in adulthood (weighted B = 0.057, p = 0.017; unweighted B = 0.059, p = 0.012), but did not predict transpersonal identification.

Findings indicate that among individuals who do not develop non-affective psychoses, elevated schizophrenia polygenic risk is associated with greater magical thinking and with a distinct developmental pattern: while magical thinking typically declines from early to middle adulthood, this decline attenuates in those with higher PRS, especially in midlife. These results persisted after accounting for sex, childhood adversity, and adulthood socioeconomic status, suggesting that genetic susceptibility contributes to milder forms of deviant thinking independently of social marginality or early-life stress. The pattern aligns with literature linking schizotypal traits (including magical thinking and self-absorption) to psychosis spectrum features, though most high-schizotypy individuals do not develop psychosis. The additional association with self-forgetful experiences supports a broader, subtle cognitive-perceptual deviation tied to genetic liability, whereas the lack of association with transpersonal identification suggests specificity to potentially maladaptive facets rather than general spirituality. Given that participants were past the typical peak onset ages for schizophrenia at outcome assessment, magical thinking may represent an alternative or downstream manifestation of genetic risk rather than a prodromal state in many cases.

This population-based longitudinal study shows that higher schizophrenia polygenic risk, in the absence of non-affective psychoses, is associated with elevated magical thinking and an altered age trajectory in adulthood, with attenuation of the normative decline in midlife. Results were robust to adjustments and sensitivity analyses and were accompanied by small increases in self-forgetful experiences but not in transpersonal identification. The study advances understanding of non-psychotic manifestations of schizophrenia genetic liability. Future work should examine mechanisms linking PRS to specific schizotypal traits, assess cultural generalizability, and clarify temporal dynamics and clinical relevance across diverse cohorts and later life stages.

- Cultural context: Associations involving self-transcendence/spirituality may vary by culture; Finland exhibits comparatively low normative self-transcendence with different health correlations, potentially limiting generalizability. - Attrition: Included participants were more often women with higher SES and less adverse childhood environments; however, PRS and magical thinking were similar between included and excluded groups, and missingness was likely at random. - Outcome timing: Although participants were beyond typical schizophrenia onset ages, some late-onset psychosis cases could still emerge, albeit rarely, which could affect interpretation. - Measurement scope: Magical thinking assessed via a TCI subscale; while well-studied, it captures specific aspects and may not encompass all relevant cognitive-perceptual deviations. - Registry-based diagnoses: High but not complete coverage of psychotic disorders; outpatient-only cases or misclassifications could occur.