In an aim to integrate both the understudied disorganized schizotypy dimension and the dearth of self-mentalization evidence particularly in subclinical schizotypy, this study will explore the relationship of the self-other polarity of mentalization with the three schizotypy dimensions. First, we aim to explore the associations of self- and other-mentalization with positive, negative, and disorganized schizotypy in a nonclinically ascertained sample. Secondly, we will examine, for the first time, the possible mediating role of self-and other-mentalization in the relationship between a wide range of childhood adversity experiences with schizotypy, PLE, and paranoia. To our knowledge, only one study to date has examined the mediating role of mentalization in the relationship between childhood maltreatment and psychosis, albeit in a clinical sample, which revealed that mentalizing only mediated the relationship between childhood maltreatment and negative symptoms (48). Of note, different forms of intentional (e.g., maltreatment) and nonintentional (e.g., parental loss) adversity experiences, as well as the distinction between paternal versus maternal abuse during childhood, will be examined.

We predicted that positive, negative, and disorganized schizotypy would be associated with impaired self-mentalization. That is, individuals with high schizotypy would notice and understand their own emotions, thoughts, and feelings more poorly. While some evidence suggests that high positive and negative schizotypy dimensions are linked to deficits in emotional awareness and regulation (70), given the lack of clear grounding on the differential contributions of self-mentalizing factors versus other-mentalizing in schizotypy, and that most mentalization research in psychosis has only focused on other-mentalization, the study is exploratory regarding self-mentalization. For other-mentalization, we expected that the associations with schizotypy dimensions in this nonclinical sample would be aligned with previous research (i.e., negatively associated with other-mentalization) (36, 44, 45, 47, 71, 72), albeit at an attenuated level. Next, we expected that self- and other-mentalization would mediate the association between childhood maltreatment and all schizotypy dimensions, PLE and paranoia. Finally, following previous results found in epidemiological studies (73), prospective cohorts (74) and the group’s previous findings using experience sampling methodology (75), we hypothesized that the relationship between childhood adversity and psychotic outcomes via mentalizing would be more pronounced for those types of adversity characterized by an ‘intention to harm’ as compared to accidental adversity such as loss of a parent.

Participants were recruited at a university using posters and an email distributed to all students and university staff inviting them to take part in a broader study about environmental sensitivity and mental health (approved by the Ethics Committee of the Universitat Autònoma de Barcelona, ref. 5426). Participants were excluded if they were under 18 years old or had grandparents of non-Spanish origin, an exclusion criterion placed for the context of the broader study for genetic analysis purposes. After removing n = 47 participants of non-Spanish origin, n = 38 participants with careless responses according to the Infrequency Scale (76), and data from n = 7 dropout participants, responses from the original sample of n = 1,246 were reduced to n = 1,156 (M_age = 23.29, SD = 6.49; range 18–62 years; 76.2% female). Of the final samle, n = 545 (47.1%) of participants had previously or were currently undergoing psychological treatment, and n = 204 (17.6%) had previously or were currently undergoing pharmacological treatment related to mental health.

After obtaining informed consent, participants were administered an online questionnaire via Qualtrics survey software that included all materials of the present study. Participants were able to re-enter the questionnaire to complete it in multiple sessions if desired with a maximum allotted time for completion of 3 days.

Descriptive statistics, internal consistencies and correlational analysis were conducted for all variables of interest (Table 1). Note that some subscales differ slightly in their number of respondents. For CAPE Positive PLE and MSS Disorganized Schizotypy, missing data is due to a technical error in the data collection software. For CECA.Q psychological abuse and antipathy, reduced responses are reflective of the number of respondents who only had a paternal or a maternal figure, but not both. Sample size for CECA.Q antipathy is in all cases smaller compared to psychological abuse because one item of this subscale was only responded by participants who have siblings.

To reduce factors for childhood trauma (to 5 predictor variables) and the number of mediation models, two a priori analyses were conducted using trauma measures. Given the low-endorsement rates of sexual abuse and that primarily modest-to-high correlations (0.26–0.65) were observed between CTQ non-sexual abuse subscales (emotional abuse, emotional neglect, physical abuse, and physical neglect), and following Sheinbaum et al. (86), principal components analysis (PCA) was conducted to produce a single emotional/physical adversity factor. This PCA yielded one component that explained 59% of the variance. Additionally, due to high correlations (r = 0.75 for father and r = 0.76 for mother, p < 0.001 for both) rather than exploring CECA.Q psychological abuse and CECA.Q antipathy separately for maternal and paternal figures, we elected to combine standardized (z) scores for available data on psychological abuse and antipathy into one measure of adversity for each parent: CECA.Q maternal abuse and CECA.Q paternal abuse.

Moderate-strong correlations (0.47–0.57) were observed between mentalization domains, and thus the three mentalizing factors—attention to emotions, emotional clarity and other-mentalizing—were simultaneously entered into parallel multiple mediation models for all outcome variables. A visual depiction of the a-priori mediation models tested can be seen in Figure 1. Parallel multiple mediation analyses were conducted using Hayes (87) PROCESS Macro Model 4 for assessing indirect pathways. Compared to the use of several single mediation analyses, parallel mediation accounts for the variance of other mediators in the model and is well-suited to inter-correlated mediators, as it offers a more precise estimate of indirect effects. This technique has been repeatedly demonstrated as useful in psychosis research (86, 88, 89).

Mediations of the associations of trauma and psychosis outcomes via mentalization are demonstrated by significant indirect coefficients with lower-and upper-bound confidence intervals that do not include zero. Six mediation models were tested for each of the nonclinical psychosis outcomes (positive, negative and disorganized schizotypy, positive PLE, suspiciousness, and ideas of reference) and with each of the five trauma indicators (emotional/physical adversity, maternal abuse, paternal abuse, role reversal, and parental loss) as the independent variable and the three mentalizing domains (attention to emotions, emotional clarity, and other-mentalizing) entered simultaneously as mediators, resulting in a total of 30 mediation models analyzed. Analyses were limited to these a priori hypothesized models. Bootstrapping with 10,000 resamples was conducted to generate bias-corrected confidence intervals.

After entering the three mentalization factors as parallel mediators in the models, increased attention to emotions was a significant partial mediator for most models including intentional forms of adversity (i.e., emotional/physical adversity, maternal and paternal abuse, and role reversal) and decreased emotional clarity for some models. All significant mediations were partial. Note that the general direction of effects for attention to emotions and emotional clarity in the mediation models is consistent with correlational analysis; that is, higher scores for attention to emotions is associated with higher outcomes, while lower clarity is associated with higher outcomes. Other-mentalizing was only a significant mediator in one model that examined the relationship between role reversal and negative schizotypy.

For the model using the combined emotional/physical adversity component, parameter estimates of the direct, total, and indirect effects can be found in Table 3. Indirect effects for increased attention to emotions in this model were significant for positive schizotypy, negative schizotypy, disorganized schizotypy, PLE, suspiciousness and ideas of reference. There was a significant indirect effect for emotional/physical adversity via emotional clarity on positive schizotypy, PLE and suspiciousness. Indirect effects of emotional clarity on positive schizotypy, disorganized schizotypy and positive PLE were such that trauma was associated with lower emotional clarity which is, in turn, associated with higher scores on psychotic outcomes; however, the effect of emotional/physical adversity on suspiciousness via emotional clarity was the opposite, such that lower emotional clarity was related to decreased suspiciousness. Outcomes that were significant for both attention to emotions and emotional clarity in this model reflect relatively equivalent effect sizes for the two specific indirect effects, with the exception of disorganized schizotypy which was driven predominantly by emotional clarity.

In the model for paternal abuse (Table 4), there was a significant indirect effect of paternal abuse on all outcomes via higher attention to emotions. In the case of negative schizotypy, the indirect total effect of all mediators combined was nonsignificant, indicating that there was no parallel mediation, but that attention to emotions remained a significant mediator of the association between paternal abuse and negative schizotypy after controlling for the other mediators (emotional clarity and other-mentalizing). Unlike other outcomes, the indirect effect of increased attention to emotions for this model was related to lower scores of negative schizotypy. No significant indirect effects were found for emotional clarity or other-mentalizing in the relationships between paternal abuse and psychosis outcomes. Mediating effects of mentalization on all outcomes had relatively small effect sizes but were most pronounced for positive PLE and disorganized schizotypy, which are roughly double those of other outcomes.

The model evaluating the multiple parallel mediation model between maternal abuse and psychosis outcomes with mentalization factors as mediators is presented in Table 5. There was a significant indirect effect of maternal abuse on psychosis outcomes via greater attention to emotions for positive schizotypy, disorganized schizotypy, PLE, suspiciousness and ideas of reference. In the case of negative schizotypy, maternal abuse was associated with higher attention, but decreased negative schizotypy. Contrary to results for paternal abuse, significant indirect effects of decreased emotional clarity were also found for maternal abuse on most outcomes: positive schizotypy, disorganized schizotypy, suspiciousness and ideas of reference. All outcomes had a significant indirect total effect (Table 5) indicating multiple parallel mediation, except for negative schizotypy, which indicates that its only significant mediator, attention to emotions, has an indirect effect on negative schizotypy even after controlling for effects of other mediators. Although emotional clarity did not reach significance as a mediator for paternal abuse, in general, the effect sizes for specific indirect effects of attention to emotions between maternal abuse and outcomes are roughly half of those for paternal abuse.

There was a significant indirect effect of role reversal on all indicators of schizotypy, positive PLE and paranoia via attention to emotions, which were most pronounced for the positive dimension (Table 6). Significant indirect effects were found via emotional clarity for suspiciousness, positive schizotypy, and ideas of reference. The only significant indirect effect of other-mentalizing was found in this model for the association between role reversal and negative schizotypy. Effect sizes in this model are attenuated compared to other models.

Parental loss was not related to any psychosis outcomes via mentalizing (Table 7). The only significant results were found for negative schizotypy and suspiciousness. There was a significant effect of parental loss on negative schizotypy after controlling for all mediators and a significant total effect of parental loss on negative schizotypy, along with a significant total effect of parental loss on suspiciousness.

To our knowledge, no previous research has evaluated whether there are differential effects of mentalization on the relationship between intentional versus general, unintentional childhood adversity and psychotic outcomes, and particularly not in a nonclinically ascertained sample. Results revealed that only intentionally harmful childhood adversity (i.e., maltreatment and neglect) impacted mentalization functioning, compared to nonintentional childhood adversity as indexed by parental loss. The measure used for parental loss in this study assesses the loss of one or both parent figures before age 18 due to death, separation, or abandonment. Such losses are certainly impactful to those who suffer them, as they almost inevitably result in a pivotal destabilization of family, extensive emotional consequences, and often essential and monumental family restructuring (97). Despite the repercussions of a central loss such as that of a parent figure, no significant effects were found on psychotic-like features via mentalizing, and parental loss did not affect most outcomes, even after controlling for mentalization levels. This could indicate that mentalization is negatively affected more by central but harmful attachment figures than potential consequences to attachment relationships following parental loss such as (1) a lack of attachment figures, or (2) more ‘distant’ attachment figures, perhaps outside of the family, that develop epistemic trust with a child after they suffer parental loss. Indeed, some literature suggests that adjustment difficulties following bereavement are not consistently related to grieving but are instead accounted for by inadequate care following parental loss (98). Such a finding emphasizes the importance of a parent’s role as a supportive, understanding, and responsible adult figure, rather than a dangerous and untrustworthy one (99).

A non-hypothesized finding that merits further study was the differential effects of maternal versus paternal abuse on self-mentalizing factors. Results suggest that increased attention in the wake of childhood adversity is more impactful in the case of paternal abuse, as coefficients for the mediating role of attention to emotions are roughly double the same coefficients for maternal abuse for all outcomes. Nonetheless, in the case of maternal abuse the additional mediating effect of impaired clarity is present for all outcomes excepting negative schizotypy. Despite gender-role stereotypes whereby maternal figures are responsible for child rearing and paternal figures provide resources for the family outside the home being challenged in recent years, the mean age of our sample suggests that most maternal figures may still be the principal caregiver. Considering that mentalization is usually developed through relationships with said primary caregiver(s) (9), mentalization could be severely impacted if the mother–child relationship is damaged. For example, a ‘good enough mother’ conceptualized by Winnicott (100) and later expanded upon through attachment relationships by Bowlby (101) is suggested to be necessary for adequate child development, particularly of socioemotional abilities such as mentalization. If, for example, the maternal figure is the primary caregiver, but instead of providing security, stability and fostering epistemic trust with the child, fails to play this role and breaks epistemic trust by engaging in abusive or neglectful behavior, the child’s mentalization skills may be more highly impaired than they would in a father-child relationship. This notion is supported in a recent study which demonstrates that maternal psychological states are more impactful on children’s adult clinical psychosis status than paternal psychological states (102). In cases where the father is the primary caregiver, perhaps this dynamic could be expected to be reversed, however, extant literature does not shed light on this question.

An overwhelming pattern of significant findings for self- but not other-mentalizing was revealed in analyses of this study. Extant literature suggests that adaptive emotional strategies are helpful in preventing psychosis, and that emotional clarity has been shown to be ‘protective’ from the development of other mental health disorders (103). This, combined with evidence of (especially self-) mentalization as a transdiagnostic protective factor (8, 60, 62, 91), suggests that maintaining good self-mentalizing in the wake of adverse events could potentially result in better outcomes. The implication of an impaired understanding of the self in psychosis is well-accepted, with results suggesting that disturbances in understanding and identifying with the self may underpin self-disorders which hyper-aggregate in psychotic spectrum disorders (104, 105). Perhaps the ability to self-mentalize, developed during the formative years, could impede significant impairments in the development of self-identity, protecting from psychotic outcomes. Although the precise role of self-mentalization is not well understood, results that partially support our findings have been found in various stages of the psychosis spectrum; for example in self-concept clarity (63), misattributions of self-referential representations (106, 107), and even after traumatic life events which importantly interact with the self to affect psychosis proneness (108). Evidence supporting good mentalizing as a buffer for the impact of persecutory delusions (positive dimension) on functioning (109) further suggests that mentalizing can be protective, but, when impaired conveys risk.

Until the call for evaluation of distinct mentalization polarities (59), mentalization was evaluated solely as a general construct, without exploring differential contributions of self- vs. other- polarities of mentalization. Nonetheless, positive symptoms are highly implicated in self-identity and self-boundary (53, 54) and thus measures of self-mentalization may more precisely capture characteristics of positive schizotypy than other-mentalization. Indeed, contemporary cognition research suggests that understanding of the ‘self’ forms the stem of understanding the ‘other’ (110). Overall, this evidence combined with the consideration that psychotic symptoms are viewed as a ‘disturbance to the self’, and that self-mentalizing gives rise to self-organization, emotional regulation, and sense of agency, might account for the fact that that psychosis spectrum impairments are substantially associated with self-mentalizing. Of note, self-mentalizing not only impacted positive psychotic-like features, but also the disorganized schizotypy dimension. Potentially, impaired self-mentalizing (i.e., increased attention but decreased clarity, in alignment with our results) after the exposure to childhood adversity negatively impacts the ability to organize and express thoughts and behavior, that is, resulting in a manifestation of disorganized schizotypal features.

Overall, our lack of significant findings for other-mentalizing could be due to higher discrimination of the mentalization construct in the current study, whereby self- and other-mentalizing are separated, revealing that self-mentalizing drives associations. In fact, only one model revealed an indirect effect via other-mentalizing, in which childhood experiences of role reversal decreased negative schizotypy through increased other-mentalizing. To date, mentalization has been operationalized primarily using various Theory of Mind tasks, which overall evaluate other-mentalizing. These studies support (other-) mentalizing as a mediator of childhood neglect and psychosis symptoms (47), of trauma/expressed emotion and schizotypal symptoms (72) and have even found that (other-) mentalizing fully mediates the association between social perception difficulties and negative symptoms (71). Although role reversal and its impact on other-mentalizing has not been examined previously, one may speculate that assuming parental responsibilities and providing emotional support to the parent from a young age might subserve the development of an increased capacity to think about and understand other’s emotions and needs (i.e., other-mentalizing) and this, in turn, may increase one’s curiosity and openness to the world (i.e., diminished negative schizotypy).

This study benefits from (1) its novelty in exploring self- and other-mentalization in a combined study, (2) exploring a wide range of childhood adversity types, including distinctions between intentional and nonintentional harm, (3) a unique examination of the role of maternal versus paternal abuse, and (4) the assessment of psychosis spectrum outcomes in an extensive sample of nonclinical young adults. While assessing impairment at the clinical level is helpful, it may not be early enough to develop interventions and understanding that can ultimately prevent severe functional impairment, particularly in the case of psychosis (110). Schizotypy offers a unifying construct for the psychosis spectrum that provides benefits for understanding the role of mechanisms such as mentalization in the development of disorders (26, 35). More so, acknowledging the multidimensionality of the construct allows to unravel the distinct etiological and developmental pathways that specifically lead to positive, negative or disorganized manifestations (27). Thus, the study of schizotypy features in a nonclinically ascertained sample may, in fact, be a chief strength of the study.

Unfortunately, the cross-sectional nature of our study design does not allow for causal associations to be evaluated, although hypotheses were made based on extant literature and theoretical grounding which guided subsequent analysis; thus, the present findings should be replicated in longitudinal studies. A small amount of missing data due an error in survey administration software resulted in few items being removed from certain measures for some participants. Albeit slight, this limitation should be noted. Small effect sizes are also found throughout the study, which are often frowned upon, however, discovering significant results aligned with theoretical hypotheses in a nonclinical sample suggests that, further along the developmental trajectory for psychotic disorders when differences are more glaring, effect sizes would be more pronounced. Nonetheless, future research could evaluate a similar model in clinical psychosis expressions at a clinical level of psychosis expression.

The present findings assessing self- and other-mentalizing separately but simultaneously offered what could be new understanding of the self-other polarities in the psychosis spectrum: self-mentalizing may be the driver behind evidence of impaired mentalization, particularly in those who have experienced intentionally harmful childhood adversity. Mentalization-based treatment has already shown to be effective in reducing psychosis symptoms (111), but these findings further illuminate awareness of which specific mentalization dimensions should be targeted. Indeed, this offers compelling implications for interventions and psychoeducation across the psychosis spectrum. Psychoeducation and interventions focused on self-mentalization should be prioritized particularly in psychosis’ earliest expressions, that is, schizotypy, as interventions that target mentalization in psychosis suggest that early intervention results in better outcomes (111) and contemporary economics demonstrates that it is more beneficial to invest resources early in development in order to capture the full potential of interventions (112, 113).