Spirituality, dimensional autism, and schizotypal traits: The search for meaning

The relationships of spirituality with human social cognition, as exemplified in autism spectrum and schizophrenia spectrum cognitive variation, remain largely unstudied. We quantified non-clinical levels of autism spectrum and schizotypal spectrum traits (using the Autism Quotient and the Schizotypal Personality Questionnaire-Brief Revised) and dimensions of spirituality (using the Hardt Spirituality Questionnaire) in a large sample of undergraduate students. We tested in particular the hypothesis, based on the diametrical model of autism and psychosis, that autism should be negatively associated, and positive schizotypal traits should be positively associated, with spirituality. Our primary findings were threefold. First, in support of the diametric model, total Spirituality score was significantly negatively correlated with total Autism Quotient score, and significantly positively correlated with Positive Schizotypal traits (the Schizotypal Personality Cognitive-Perceptual subscale), as predicted. Second, these associations were driven mainly by opposite patterns regarding the Search for Meaning Spirituality subscale, which was the only subscale that was significantly negatively associated with autism, and significantly positively associated with Positive Schizotypal traits. Third, Belief in God was positively correlated with Positive Schizotypal traits, but was uncorrelated with autism traits. The opposite findings for Search for Meaning can be interpreted in the contexts of well-supported cognitive models for understanding autism in terms of weak central coherence, and understanding Positive Schizotypal traits in terms of enhanced salience.


Introduction
The psychological bases and correlates of religious beliefs, participation and experience have recently been subject to an increasing depth and breadth of empirical work [1]. Most of these studies have focused on how variation in aspects of human cognition, sociality, and personality is related to variation in aspects of religiosity, among non-clinical populations. These analyses have revealed important insights into the psychological underpinnings of religiosity, including: (1) negative associations of religiosity and belief in God with physical-world beliefs, interests, low levels of analytical thinking. High levels of analytical thinking in autism may thus be expected to mediate, in part, any reduction in religiosity found among these individuals, given that an analytical cognitive style is negatively associated with religious and paranormal beliefs [24]; see also [3,25]. More broadly, the autism spectrum involves a considerable suite of traits that each might be expected to engender reduced, or different, belief in God or religion, including: (1) reduced social connectedness and theory of mind;, (2) restricted non-social, mechanistic, and analytical interests; (3) reduced empathizing and high systemizing; (4) reduced perception and inference of agency; (5) increased focus on details and parts rather than the 'big picture'; and (6) reduced moralizing, often expressed in utilitarian judgements (e. g., [26,27]). However, the relative roles of these factors remain to be discerned. Two studies to date have analyzed autistic and schizotypal traits together, in relation to aspects of religiosity. Thus, Gray, Jenkins, Heberlein, and Wegner [28] found that scoring highly on the Autism Quotient (AQ) was associated with reduced perception of agency in humans, and that scoring highly on the Schizotypal Personality Questionnaire (SPQ) Cognitive-Perceptual subscale (positive schizotypal traits) was linked to 'seeing mind where there is none' by ascribing mind to trees, the dead, and God, and to assignment of more agency than normal to robots and dogs, which can jointly be interpreted as belief that all entities exhibit forms of consciousness. These findings highlight central cognitive differences in mind and agency perceptions between autism and positive schizotypal traits, especially with regard to mentalistic thought.
Lindeman and Lipsanen [2], also using the AQ and a version of the SPQ, found that religiosity was significantly positively associated with schizotypal traits, but significantly negatively associated with autistic traits, although the proportion of variation accounted for was low. In their analysis, the most consistent difference between religious believers and non-believers was 'interests and skills needed to understand the physical world', a finding consistent with a notable body of previous work, as well as with theory of these conditions as cognitively opposite to one another [5,7].
Taken together, the bulk of the evidence to date therefore suggests reduced religiosity on the autism spectrum, and increased religiosity on the schizophrenia spectrum, especially with regard to positive schizotypal traits and positive schizophrenia symptoms. These findings represent important first steps towards understanding the psychological bases of religious belief and experience broadly construed, but they exhibit some notable limitations. Of these, one of the most important is that almost all of the work has focused on aspects of religion considered from modern, Western perspectives. Recent large-scale religions, and modern beliefs in and conceptions of God, are sufficiently recent phenomena that they represent evolutionary novelties, to which humans are not predicted to be adapted in any specific genetic, neurological or cognitive sense (e. g., [26]). Indeed, current conceptions of the world as divisible into 'natural' and 'supernatural' beings and events are a recent phenomenon: hunter-gatherer and smallscale societies do not make this distinction, and live in a natural world imbued with spirituality. How might spirituality, then, be related to psychological predicates?
Lindeman, Svedholm-Häkkinen, and Lipsanen [29] found, in a large sample of typical Finnish individuals, that religious beliefs, paranormal beliefs, and belief in supernatural purpose were highly predicted not by mentalizing abilities, but by what they call 'core ontological confusions', considered as cognitive biases that involve ascribing mental states and agency to entities in the physical world. To a close approximation, such 'core ontological confusions' represent forms of spiritual animism, the 'animation' with spirits of all salient phenomena in the natural world, be they rocks, water holes, or goanna lizards. These authors also found similarly strong prediction of religious beliefs, paranormal beliefs, and belief in supernatural purpose from 'hyper-empathizing' (high empathizing combined with low systemizing), a construct also linked positively with aspects of subclinical psychosis traits by Brosnan, Ashwin, Walker, and Donaghue [30]. Most directly, Willard and Norenzayan [31] analyzed the psychological correlates of being 'spiritual but not religious', finding that such individuals scored more highly for positive schizotypy than did religious individuals and controls, as well as endorsing a higher incidence of paranormal beliefs and supernatural, mystical experiences.
These findings suggest that spirituality represents a form of human cognitive-affective experience that is both centrally important and substantially distinct from religiosity, and from the supernatural, as well as representing an ancestral, long-term, and ongoing (in many societies) state of cognition and experience. Under a spiritual world-view, then, natural and supernatural phenomena and events are not separated, one engages in searches for meaning within one's life, one's group, one's culture, and the universe most broadly, and one focuses on personal beliefs and experiences concerning existence, states of being, and one's relationship with the world and its constituents. A god or gods may be considered to exist, but not in the context of 'institutionalized' religions beyond the belief systems of one's own tribe, for small-scale societies. Spirits, 'animism', souls of ancestors and others, and origin stories, while they can certainly be viewed as 'cognitive biases' or 'ontological confusions' by scientific world-views, can also represent, under spiritual world-views, important existential features of cognitive as well as external landscapes.
Most previous studies of the psychology of religiosity in non-clinical individuals, or individuals in relation to the autism spectrum or schizotypy spectrum, have thus centered on facets of contemporary religion, which may include for example belief in a single god. In this context, and to operationalize the distinctions applied here, religiosity can be considered to comprise two primary, inter-connected components, (1) supernatural phenomena, including ideas concerning creation of the world and (2) moral precepts. These religious concepts are now seen most prominently in Christianity, Islam, and Judaism. However, the current focus on contemporary religious groups and institutions presents challenges for interpretation of the social, cultural, neurological and genetic contexts under which cognition related to religiosity evolved, given that the primary environments for the evolution of the biological and cultural underpinnings of religiosity and related phenomena must have been pre-historical, and comparable to those found in hunter-gatherer and other small-scale societies.
In human hunter-gatherer and other small-scale groups, and as inferred from phylogenetic studies of diverse human cultures [32], the ancestral human condition for 'religiosity' thus reflects spirituality, which, from cultural and empirical perspectives, is profoundly different from contemporary religion. As noted above, under traditional spiritualistic cognition, the natural and supernatural worlds are in no way distinct, and places, objects, and living creatures including humans and selves are animated by spirits or souls, that imbue them with living properties and meanings. The primary upshot of these considerations is that the human mind evolved predominantly in the context of what we now call spirituality, rather than what we call religion [32]. As such, analyzing the psychological underpinnings of human spirituality, in relation to the cognitive variation expressed in autistic and schizotypal thinking, may provide novel and useful insights into spirituality and religion.
What, then, are the psychological correlates and underpinnings of human spirituality, and how do they relate to the autism spectrum and schizotypal spectrum? In this article we have collected self-report data, from a large, healthy, non-clinical population, on (1) a measure of spiritual cognition, and its major components [33]; and (2) positions on metrics of autism spectrum (the Autism Quotient AQ: [34]) and schizotypal spectrum (the Schizotypal Personality Quotient Brief-Revised SPQ-BR: [35]) cognition. The main goal of this study is to determine how autism spectrum and positive schizotypal spectrum traits are related to spirituality and its dimensions. Our overarching prediction is that spirituality, in whole or for key components, should be positively associated with positive schizotypal traits, and negatively associated with autism spectrum traits.

Participants and ethics approval
Data was collected from 1194 undergraduate students (731 females, and 463 males, mean ages 19.4 for both; ethnicities Caucasian 42%, Asian 35%, Others 23%) at the University of Alberta. Biological sex, rather than (culturally-defined) gender was reported, for comparison with previous relevant studies, in the context of statistical differences and adjustments. Given the nonclinical nature of the population, subjects were not questioned concerning psychiatric diagnoses. The work was approved by Human Research Ethics at University of Alberta and by the Simon Fraser University Research Ethics Board, and all subjects gave written informed consent before participating in the study.

Psychological measures
The Hardt Spirituality Questionnaire (HSQ) [33] was used to assess self-report spirituality. The HSQ is a 20-item questionnaire with responses on a 5-point Likert scale of how true the question was perceived to be in relation to the respondent. Four subscales are represented: (1) Belief in God (any God, and having a relationship with them involving trust, love, faith and friendship); (2) Search for Meaning (centered on the quest for existential meaningfulness, searching for spiritual insights, an open mind and an expanded soul); (3) Mindfulness (conscious meta-cognitive perception of others and one's environment, empathy, and attentive self-awareness); and (4) Security (safety and trust in the world, and feeling at home). Total Spirituality was the sum of the subscales. Cronbach's alpha values (calculated using the alpha() function from the R 'psych' library) for these subscales were 0.94, 0.71, 0.72 and 0.73 respectively, and the overall alpha value was 0.88.
The Autism Spectrum Quotient (AQ) [34] was used to quantify endorsement of autism spectrum traits. The AQ is a 50-item self-report questionnaire that assesses autism-related traits across five domains: (1) Social Skills; (2) Communication; (3) Attention to Detail; (4) Attention Switching; and (5) Imagination. Higher autism traits are thus quantified as lower social skills and communication, higher attention to detail, less attention switching, and reduced imagination (mainly social imagination). Responses range from 'definitely agree' to 'definitely disagree' on a 4-point Likert scale; Total Autism score is the sum of the subscales. Cronbach's alpha values (calculated using the alpha() function from the R 'psych' library) for these subscales were 0.69, 0.52, 0.55, 0.48, and 0.44 respectively.
Schizotypal traits were quantified using the Schizotypal Personality Questionnaire-Brief Revised (SPQ-BR) [35]. This questionnaire includes 32 items in a 5-point Likert-scale format, that includes seven subscales: (1) Constricted Affect; (2) Social Anxiety; (3) Magical Thinking; (4) Unusual Perceptions; (5) Ideas of Reference; and (6) Eccentric Behavior; and (7) Odd Speech. Total Schizotypy is the total of all seven subscales. Cronbach's alpha values (calculated using the alpha() function from the R 'psych' library) for these seven subscales were 0.81, 0.84, 0.76, 0.52, 0.72, 0.85 and 0.57 respectively. There are also three higher-level subscales: 'Interpersonal' (made up of 1+2), 'Cognitive-Perceptual' (3+4+5), which represents Positive Schizotypal Traits, and 'Disorganized' (6+7). We focus here on Positive Schizotypal Traits, based on both theoretical considerations and previous relevant empirical work showing that only this dimension of schizotypy is related to aspects of religiosity. Individuals with missing data, or scores that were outliers for the required range, for any of the questionnaires were excluded from analysis.

Analyses
Analyses were conducted in R (v3.4.3). Sex effects in Table 1 were tested with Welch's t-tests, and effect sizes were measured using Cohen's d with 95% confidence intervals calculated using 1000 bootstrap resamplings [36].
Independent analyses of covariance between spirituality (and its subscales) and autism and positive schizotypy (and their subscales) were conducted including sex as a cofactor using type-II sums of squares to test for significant interactions of sex with the other variables using the Anova() function from the "car" library [37]. In only two of these 55 analyses did sex show a significant interaction effect, demonstrating that there were generally no sex differences in the relationships of sex with other factors. We therefore present the main effects for sex in these analyses. The two significant cases will be discussed specifically, to the limited extent that the effects are relevant.
Partial correlation coefficients, in multiple regressions of spirituality measures on sex, autism and positive schizotypy scales, were calculated using the pcor() function in the "ppcor" library [38].

Sex differences
We found no sex differences in Total Autism scores, while Positive Schizotypal traits and Total Spirituality all showed significant female biases (Table 1), as found in previous studies (e. g., [39][40][41][42]). The three subscales of Positive Schizotypal traits showed differing sex biases, with women scoring significantly higher on Ideas of Reference and Magical Ideation, and significantly lower than men on Unusual Perceptions. Only one of the four subscales of autism, Imagination, showed a strong male bias. The sex difference in spirituality was evident in three of the four subscales, with women scoring higher for Belief in God, Search for Meaning and Mindfulness, but not Security ( Table 1). Most of these subscale sex differences have small effect Spirituality, autism and schizotypal traits sizes, but two subscales of religiosity, Belief in God and Mindfulness, and one subscale each from the autism and Positive Schizotypal traits assays, Imagination and Magical Ideation respectively, are considerably stronger (d' stronger than 0.2, Table 1). Sex was therefore included as a covariate as we examined the relationships between spirituality and these spectrum traits.

Bivariate associations
We examined the bivariate associations between Total Spirituality scores, and the predictor variables of Total Autism, and Positive Schizotypal traits including sex as a covariate (Table 2). Total Autism was negatively associated with Total Spirituality, while Positive Schizotypal traits showed a positive association ( Table 2). The p = 0.09 sex interaction effect between Positive Schizotypal traits and Total Spirituality was not due to opposing effects in the two sexes, but to the relatively stronger relationship in males. The positive correlation between these two traits was thus significant in both sexes, but relatively stronger in males. In contrast to the positive relationship of Total Spirituality with Positive Schizotypal traits, Total, Interpersonal, Disorganized Schizotypal traits were each negatively related to Total Spirituality (r = -0.07, p = 0.024; r = -0.23, p <0.00000001; r = -0.08, p = 0.0052 respectively). These results support the hypothesis that spirituality is positively related exclusively to the positive dimension of schizotypal traits, as noted above in previous studies. We further analyzed the bivariate associations between the subscales of Total Spirituality and each of the subscales of the spectrum trait (Tables 3-6). In only two of these 55 analyses did sex show a significant interaction effect: (1) between Magical Thinking and Total Spirituality, where significant positive correlations between these traits in were stronger in males than females, leading to a significant interaction term for sex, and (2) between Magical Thinking and Mindfulness, where males showed a significant positive effect and females showed no effect. These results demonstrate that while there may be sex differences in these psychological traits, there were generally no sex differences in the relationships between them.
In the bivariate analyses of spirituality subscales, only Search for Meaning (Table 4) showed the same pattern of significant negative association with Total Autism, and significant positive association with Positive Schizotypal traits, as was found for Total Spirituality. Mindfulness (Table 5) and Security (Table 6) both showed strong, significant negative correlations with Total Autism but essentially no relationship with Positive Schizotypal traits; by contrast, Belief in God (Table 3) showed a significant positive correlation with Positive Schizotypal traits, but no association with Total Autism.

Multivariate associations
To partition independent effects of sex, autism traits and Positive Schizotypy traits on aspects of spirituality, we conducted multiple regressions of Total Spirituality, and each of the four spirituality subscales, on sex and each AQ and SPQ lower-level subscale ( Table 7). The coefficients were not substantially changed in relation to the bivariate results shown in Tables 2-6

Discussion
We have conducted the first analyses of both non-clinical autism spectrum traits and positiveschizotypal spectrum traits in relation to aspects of human spirituality. The main findings are threefold. First, as predicted under the hypothesis that autism and positive schizotypy represent opposite psychological dimensions, Total Spirituality score was significantly negatively correlated with Total Autism Quotient score, and significantly positively correlated with Positive Schizotypal traits (the Schizotypal Personality Cognitive-Perceptual subscale). For autism, this overall negative association was mediated predominantly by effects from the Social Skills, Communication, and Imagination subscales, as might be expected if spirituality and religiosity involve, in part, imaginative social connectedness (e. g., [19,43]). By contrast, for Positive Schizotypal traits, the effects of variation in Magical Thinking predominated. This finding implies that magical thinking concerning supernatural entities and events may most-directly mediate links of Total Spirituality with Positive Schizotypal traits, again implicating imagination as an especially relevant psychological domain (e. g., [44]: pp. 4, 161). Spiritual, religious, supernatural, and paranormal experience have also been linked with higher scores on measures of positive schizotypy in previous studies (e. g., [6,15,18,45]); as also noted above, Willard and Norenzayan [31] showed in particular that individuals who were more spiritual (but not religious) scored higher on positive schizotypal traits, using an earlier version of the SPQ [46]. The signs and magnitudes of associations of the four spirituality components with subscales of autism and positive schizotypal traits appear to make general intuitive sense, in that (1) higher Belief in God is linked with better Social Skills and higher levels of Magical Thinking, as well as female sex, which fits with the well-established female bias in spirituality and religiosity (e. g., [40]); (2) higher Search for Meaning is associated with better Social Skills and Imagination, and higher levels of Magical Thinking and Unusual Perceptions; (3) higher Mindfulness is connected with better Social Skills, Communication, and Imagination (less-autistic traits for all three), and female sex; and (4) higher Security is linked with better Social Skills, Communication, and Imagination, lower paranoia (Ideas of Reference), and higher Magical Thinking (Tables 2 and 3). Overall, higher Social Skills combined with higher Magical Thinking appear to be most predictive of the spiritual phenotypes analyzed here (applying overall, and for 3 of 4 of the subscales), which is interesting in the context of spirituality involving social connectedness with what are, from modern Western views, 'supernatural' entities, objects, or states.
Second, the opposite associations of spirituality with Total Autism versus Positive Schizotypal traits were evident only in diametric patterns for the Search for Meaning spirituality subscale, which was significantly negatively associated with social and imaginative subscales of autism (and Total Autism), and significantly positively associated with Positive Schizotypal traits. Indeed, none of the three spirituality subscales other than Search for Meaning showed any trace of this diametric pattern.
The specificity of the diametric results to Search for Meaning is intriguing in the context of influential theories for explaining the psychological bases of autism and positive schizotypal traits (and their direct extension, psychosis in schizophrenia). Thus, for autism, Happé [47] and Happé & Frith [48] describe the concept of Weak Central Coherence as a predominant causal factor in autistic cognition; by this theory, individuals with autism exhibit a reduced 'drive for meaning' and a lesser tendency to 'see the big picture' in extracting global, high-level form and meaning from both the world and their thoughts. In principle, then, a less-developed drive for meaning in everyday activities and events may extend to the spiritual domain, where drive for meaning and central coherence would be required to seek, and find, personal spiritual significance in the world; by contrast, in contemporary large-scale religions, 'meaning' is directly provided. This hypothesis could be evaluated more directly by measuring performance on task-based correlates of central coherence, such the embedded figures test [47], in relation to spiritual searches for meaning. For positive schizotypy and psychosis, a central framework for psychological and neurochemical understanding derives from the concept of salience (i. e., relevance, meaning or significance), whereby relatively high dopaminergic states lead to aberrant assignment of importance to particular experiences (leading to delusions) and internal representations (appearing as hallucinations) [20,49]. Indeed, the prominent schizophrenia researcher van Os [50] suggested that this condition be renamed as 'salience syndrome', given the centrality of dysregulated over-attribution of meanings in schizophrenia and psychosis more generally. In this context, increased spiritual search for meaning in high positive schizotypal individuals may represent a manifestation of upregulated salience, with non-clinical psychosis-related experiences (e. g., mild delusions and unusual perceptual experiences, magical thinking, and highly-developed imagination) as central forms of expression. Across studies, 20-70% of all delusions experienced by individuals with schizophrenia spectrum disorders have clear religion-related content, manifesting especially as forms of grandiosity and megalomania [16,51,52]. High schizotypal traits have also been associated with a tendency to perceive meaning where none exists, in experimental tests using animated shapes and responses to short stories [53]. A hypothesis based on increased salience in high schizotypy would benefit from psychological tests focused more specifically on aspects of salience in relation to spiritual search for meaning, as well as tests relating dopaminergic neurotransmission to this aspect of spiritual cognition and experience (e. g., [54]).
Third, we found that Belief in God was positively correlated with Positive Schizotypal traits, but was uncorrelated with autism traits. An association of Positive Schizotypal traits with Belief in God, in the context of spirituality (separate from religious tradition), was also reported by Willard and Norenzayan [31]. Some evidence supports a hypothesis of reduced belief in a personal God among individuals with autism or more autism-related traits, based on constraints imposed by a less-developed theory of mind [4]; by contrast, Jack et al. [5] describe evidence that belief in God is predicted by moral concerns and more-intuitive, less-analytical cognition, but not by theory of mind abilities per se. Persons with high-functioning autism have also been reported to exhibit lower levels of traditional religiosity, higher levels of atheism and agnosticism, and a greater tendency to construct their own belief-related systems [55]. The degree to which conceptions regarding spiritual or religious cognition, and God, differ between non-clinical individuals and individuals with autism or high levels of autism traits, remains to be determined. By some accounts (e. g., [56]), spirituality and religiosity in autism are expected to differ qualitatively from traditional norms, being more analytical, mechanistic, and physical-world based than magical and mentalistic [7].
The primary limitations of this study and its results include (1) the use of single measures for spirituality, autism spectrum, and schizotypal spectrum traits, (2) the low magnitudes of most of the correlations and regression coefficients, despite high levels of statistical significance, and (3) the use of a highly-educated young Western population. Further joint studies of spirituality in autism and schizotypal traits are warranted, which, given our results, might usefully focus on concepts of God, other aspects of spirituality, the role and nature of searches for meaning through spiritual cognition, and analysis of the psychology of spirituality in non-Western, small-scale, more-traditional populations and cultures. Of particular interest would be studies of cultures where individuals did not differentiate between the 'natural' and 'supernatural', as such groups represent the social-cultural milieu within which spirituality and religion are presumed to have originally evolved.