The Role of Personal Identity on Positive and Negative Symptoms in Psychosis: A Study Using the Repertory Grid Technique

Abstract

Background

People with psychosis experience disruptions in personal identity that affect positive and negative symptoms, but the complexity of these phenomena needs to be addressed in an in-depth manner. Using the Personal Construct Theory, we examined whether distinct dimensions of personal identity, as measured with the Repertory Grid Technique along with other cognitive factors, might influence psychotic symptomatology.

Method

Eighty-five outpatients with schizophrenia-spectrum disorders completed a repertory grid, an observed-rated interview of psychotic symptoms, and measures of cognitive insight, depressive symptoms, neurocognition, and theory of mind.

Results

Structural equation models revealed that interpersonal dichotomous thinking directly affected positive symptoms. Self-discrepancies influenced positive symptoms by mediation of depressive symptoms. Interpersonal cognitive differentiation and interpersonal cognitive richness mediated the impact of self-reflectivity and neurocognitive deficits in negative symptomatology.

Conclusions

This study is the first of its kind to examine the structure of personal identity in relation to positive and negative symptoms of psychosis. Results suggest interventions targeted to improving interpersonal dichotomous thinking, self-discrepancies, interpersonal cognitive differentiation, and interpersonal cognitive richness may be useful in improving psychotic symptoms.

Background

Disturbances of personal identity in schizophrenia disorders have been recognized since the origins of the concept of schizophrenia itself. Bleuler¹ conceived the self-disorders as fundamental symptoms of schizophrenia, and Jaspers² highlighted the importance of the role of the self, identity, and personal meanings in the expression of psychotic symptomatology. These notions have been recovered by modern person-centered approaches to psychopathology in schizophrenia.^3–5 Moreover, the development of a coherent sense of personal identity and meaning-making processes are key elements for recovery and well-being in people with psychosis, and also aspects for which these people are most concerned with.^6,7 Therefore, understanding the dimensions of personal identity and its possible connections to thought and emotion disruptions becomes relevant. Accordingly, current psychological models of psychotic symptoms may be improved by exploring the role of personal identity in them.

Personal identity is considered here as a complex concept formed by the integrative representations of self and interpersonal relationships.^8,9 Contributions from Personal Construct Theory (PCT)¹⁰ and its methodology may be helpful in situating personal identity within the context of the interpersonal world. Our approach also stresses the role of identity disruptions in mental disorders, including psychosis. PCT sees psychological activity as a subjective meaning-making process of the events people encounter in life.¹¹ Personal identity is a cognitive system formed by a complex (and hierarchical) network of personal constructs—ie, evaluative dimensions of personal meanings constructed by the person in a relational context throughout his/her life. Thus, people use their construct system to define and interpret their selves and the people who constitute their interpersonal world. Within PCT, the Repertory Grid Technique (RGT) is the most used instrument to measure personal identity. Its use in research settings for psychosis started with the work of Bannister and colleagues on formal thought disorder in the 1960s.^12,13 This research tool allows a more in-depth analysis of personal identity than classical questionnaires,¹⁴ capturing the personal meanings involved in the description of self and the person’s interpersonal world. The RGT captures the structure of personal identity from diverse angles: self-discrepancies, interpersonal dichotomous thinking, interpersonal cognitive differentiation, and richness. These dimensions (table 1) could be relevant for the conceptualization of psychological models of psychotic symptoms. The extant models usually associate several cognitive and emotional aspects to positive and negative symptoms, so it is likely that specific dimensions of personal identity would contribute distinctively to these models. Specifically, it is possible that interpersonal dichotomous thinking and self-discrepancies could influence positive symptoms, while interpersonal cognitive differentiation and richness could affect negative symptoms.

Table 1.

Personal Identity Dimensions Measured With the Repertory Grid Technique

Concept	Measure	Operationalization	Interpretation
Self-discrepancies	Self-ideal discrepancy	Euclidean distance between the “self now” and “ideal self” elements, considered as a measure of self-esteem.	High scores indicate low self-esteem, meaning that the person would be construing current self as very different from ideal self, while low scores indicate high self-esteem, seeing both selves as similar.
	Self-others discrepancy	Euclidean distance between the “self now” and estimated “others” element (computed as the mean of the ratings to elements other than current self and ideal self). This is considered a measure of perceived social isolation.	High scores indicate a high degree of perceived differences between self and others, while low scores indicate few perceived differences between these elements, a perception of being similar to others.
	Ideal-others discrepancy	Euclidean distance between the “ideal self” and the estimated “others” element. This is considered a measure of perceived adequacy of significant others.	High scores indicate a high degree of perceived differences between ideal self and others, and a negative perception of others, while low scores reflect a construction of these elements as similar, signaling a satisfactory view of significant others.
Interpersonal dichotomous thinking	Polarization	Percentage of extremity ratings (“1” and “7”) of the grid data matrix. This is considered a measure of dichotomous or extreme thinking in the interpersonal domain, a form of cognitive rigidity.	High scores represent extremity, with the person having a tendency towards a dichotomous thinking style, while low scores are an indicator of flexible thinking.
Interpersonal cognitive differentiation	Percentage of Variance Accounted for the First Factor (PVAFF)	This index is calculated by correspondence analysis of the grid data matrix and it represents the first factor. It allows simultaneous consideration of variance attributable to both constructs and elements. It is considered a measure of interpersonal cognitive differentiation, representing the ability of the construct system to discriminate in a given set of elements. In other words, it is a measure of how individuals construe their experiences as being either more different or more similar.	A high percentage indicates low differentiation, which reflects a tendency of the system of constructs to see themselves and others as more similar (unidimensional or simple thinking). A low percentage means high differentiation, a high-perceived difference in the construct system to discriminate among persons (multidimensional or differentiated thinking).
Interpersonal cognitive richness	Number of elicited constructs	This number depends on the number of similarities and differences that the person can establish among the self and main others. It is the total count of dimensions expressed by the person.	Higher scores show an important capacity for naming constructs to describe self and others. This is considered a measure of interpersonal cognitive richness, possibly related to verbal abilities.

Psychological models of positive symptoms describe the development and maintenance of delusions and hallucinations. These symptoms have been linked to cognitive and emotional processes. Regarding cognition, recent literature has associated delusional proneness to poor cognitive insight, low self-reflectivity, and high self-certainty.¹⁵ Another cognitive process linked to delusional proneness is belief inflexibility.¹⁶ This process, also defined as dichotomous thinking,¹⁷ has been discussed mostly in the context of reasoning about delusions.¹⁸ It is possible that dichotomous thinking might apply not only to delusions, but may be a general thinking style that also affects the interpersonal context in which positive symptoms happen. Therefore, delusional beliefs might be the crystallization of a general thinking style. Dichotomous thinking, as captured by the polarization index of the RGT, may be a characteristic of the personal construct system that directly affects positive symptomatology.

Regarding emotional processes, the literature has addressed their importance, along with other self-concept related factors, in their relation to positive symptoms.¹⁹ High levels of depression, which have also been connected to high self-reflectivity,¹⁵ contribute to positive symptomatology.²⁰ Similarly, negative perception of self and others has been associated with positive symptoms.²¹ It remains unclear what the nature of the relationship is among these perceptions of self and others, and depressive and positive symptoms. Regarding their possible roles as mediators of positive symptoms, the results are mixed, as depression has been found to function as a mediator of self-perceptions,^22,23 but self-perceptions may also work as mediators of depression.^24,25 To advance our knowledge about this issue, self-discrepancies, as measured with the RGT, provide a personalized assessment of the person’s view of self and others, as they capture the constructs that are relevant for that person. Thus, grid measures might be preferable for research in person-centered approaches to psychosis (García-Mieres et al. 2019, under review) and as evaluative measures for case conceptualization in clinical practice.²⁶

Psychological models of negative symptoms emphasize the role of neurocognitive factors such as low premorbid intelligence and poor executive functioning,^27,28 social cognition related factors such as limited theory of mind (ToM),^29,30 and metacognitive aspects such as poor self-reflectivity or reduced ability to think about one’s mental state.^31–33 Negative symptoms are characterized by the diminishment of a set of essential human capacities, including emotional experience, and a loss of volition and impoverishment of internal experiences.^34,35 One factor connected to the self is the richness of self-narratives, as a mediator of the impact of cognition on negative symptoms.³⁶ Simple and diminished narratives of self might result in poorer experiences of affect and volition, leading to negative symptoms. Considering that these self-narratives emerge from a cognitive structure, it is possible that their lack of complexity might be derived from a cognitive structure which is too simplistic and poorly elaborated.³⁷ The complexity of the cognitive structure on which personal identity is based, according to PCT, is composed of 2 factors: interpersonal cognitive differentiation and interpersonal cognitive richness. Therefore, these 2 factors may mediate between other well-known cognitive factors and negative symptomatology.

Overall, factors such as interpersonal dichotomous thinking, self-discrepancies, interpersonal cognitive differentiation, and interpersonal cognitive richness may influence psychotic symptomatology. Their identification within current psychological models of psychotic symptoms may contribute to detect novel foci for cognitive interventions, thereby facilitating recovery from psychosis.

Aims and Hypotheses

The aim of this study was to explore the role of distinct dimensions of personal identity within other well-known factors postulated in psychological models of psychotic symptoms. We proposed that these dimensions would play different and independent roles in the expression of positive and negative symptoms.

For the model of positive symptoms, we tested the hypothesis that interpersonal dichotomous thinking is directly associated with positive symptomatology. We also investigated whether self-discrepancies and depressive symptoms work as direct (or indirect) predictors of positive symptoms and explored their associations with cognitive insight and depression.

For negative symptoms, we tested the hypothesis that poor structural qualities of the construct system act as a mediator between the above-mentioned cognitive factors (neurocognition, ToM, and self-reflectivity) and negative symptomatology. We also explored their relationship with self-reflectivity, ToM, and neurocognition.

Methods

Participants

Participants were 85 outpatients with a confirmed diagnosis of a schizophrenia spectrum or related disorder recruited from 4 participating mental health centers in the province of Barcelona (Spain): Salut Mental Parc Taulí, Hospital de Santa Creu i Sant Pau, Centro de Higiene Mental Les Corts, and Parc Sanitari Sant Joan de Déu (coordinating center). Inclusion criteria were (1) diagnosis of schizophrenia, psychotic disorder not otherwise specified, delusional disorder, schizoaffective disorder, brief psychotic disorder, or schizophreniform disorder (according to DSM-5); (2) age between 18 and 60 years; and (3) being clinically stable enough to do the interviews. Exclusion criteria were (1) traumatic brain injury, dementia, or intellectual disability (premorbid IQ < 70); (2) current substance dependence; and (3) being hospitalized.

Measures

Personal Identity.

Structure of Personal identity was measured with the RGT, a semi-structured interview derived from PCT, which can take various flexible formats depending on the aim of the study. In our case, we adopted an idiographic and interpersonal design which assesses the structure of the construction of self and others, and the personal meanings involved in personal identity, operationalized in terms of personal constructs. RGT is idiographic because personal constructs are elicited from the participant rather than provided by the researcher, and it is interpersonal because these constructs are applied to a set of elements that represent relevant others (parents, siblings, relatives, partners, and friends) evaluated along with “self now,” “ideal self,” and a “non-grata person” (someone who you do not like). Constructs were elicited with the dyadic method, by comparing pairs of the mentioned elements and asking for similarities and differences (eg, “nervous-calm”). After the elicitation procedure, elements were rated by the participant on a 7-point Likert-type scale according to each construct. An example of a repertory grid from one of our participants appears in a published case study.³⁸ For the current study, we considered 6 measures (table 1).

Psychotic Symptoms.

The Positive and Negative Syndrome Scale (PANSS) was used to assess psychotic symptoms.^39,40 We used the positive and negative factors of the 5-factor scoring of Wallwork.⁴¹ The positive factor included 4 items: delusions, hallucinations, grandiosity, and unusual thought content. The negative factor involved 6: blunted affect, emotional withdrawal, poor rapport, passive-apathetic social withdrawal, lack of spontaneity, and motor retardation.

Depressive Symptoms.

The Beck Depression Inventory^42,43 (BDI-II) was used as a self-report measure of the severity of depressive symptoms.

Cognitive Insight.

The Beck Cognitive Insight Scale^44,45 (BCIS) is a self-administered scale of cognitive insight, containing self-reflectivity and self-certainty subscales, which were treated separately, as recommended in a recent review.¹⁵

Theory of Mind (ToM).

The Hinting Task was used to assess ToM.^46,47

Executive Functioning.

The Wisconsin Card Sorting Test (WCST) in its abbreviated form (WCST-64, computer version⁴⁸) was used to measure executive function, set-shifting, and behavioral flexibility. We included the total number of correct categories and perseverative error scores.

Premorbid IQ.

IQ was assessed with the Wechsler Adult Intelligence Scale (WAIS-III) Vocabulary subtest.⁴⁹

Procedure

The study was approved by local ethics committees. Participants were referred by their clinicians in the outpatient mental health units of participating centers. All assessments began with the informed consent process and were conducted by the first author in the respective centers. Once confirmation of inclusion criteria was made, a demographic questionnaire and the RGT were administered in the first and second session, while a third session was used for the remaining instruments.

Analyses

We computed personal identity measures of data obtained from participants’ repertory grids with GRIDCOR v4.0⁵⁰ and entered these results in a database along with other measures.

We conducted statistical analyses in 5 stages. First, we obtained descriptive statistics for sociodemographic and clinical measures. Second, we conducted a correlational analysis for psychotic symptoms and all personal identity variables in order to map out a preliminary exploration of our main hypothesis. Third, we performed 2 separate correlation analyses for positive and negative symptoms. The inclusion criteria of psychological factors associated with these symptoms were drawn from previous research,^{20,21,27,29,31} as described in the background. For positive symptoms, we introduced self-reflectivity, self-certainty, and depressive symptoms. For negative symptoms, we included executive functioning, premorbid IQ, ToM, and self-reflectivity. We also explored correlations between psychotic symptoms and potential covariates: gender, age, age at disorder onset, stage of the disorder, antipsychotic dose, and lifetime hospitalizations. Fourth, we constructed latent variables. For positive symptoms, self-ideal, self-others, and ideal-others discrepancies were grouped into the construct of “Self-discrepancies” (factor loadings of 0.929, 0.883, and 0.641, respectively). For negative symptoms, WCST categories, WCST perseverative errors, and premorbid IQ were combined into “Neurocognition” (Factor loadings of 0.904, 0.924, and 0.78, respectively).

Finally, we conducted structural equation modeling (SEM) to test the proposed models. For both path models, we included the significant variables present in the literature based on the correlational analyses and of our hypotheses. Confidence intervals (CI) and standard errors were derived through bootstrapping. Missing data was handled with Full Maximum Likelihood Estimation. Post hoc model modifications were subsequently formed by examining path coefficients and eliminating nonsignificant paths that did not contribute to model fit, when theoretically reasonable. This procedure helped to identify better-fitting, and more parsimonious, models. Potential mediations in line with our hypotheses were also tested, a comprehensive description of the pathways checked appears in supplementary material section A. When models were significant, we replicated them controlling for potential covariates that reached statistical significance in the correlational analyses. Squared multiple correlations were extracted to get an estimate of the variance explained by the models. The fit of the models was evaluated using established indices (eg, the chi-square test, the comparative fit index [CFI], the Tucker-Lewis index [TLI], the Root Mean Square Error of Approximation [RMSEA], and the Standardized Root Mean Square Residual [SRMR]). For mediational effects, we examined the standardized coefficients as a measure of the effect size of the indirect path.

Descriptive and correlational analyses were performed with Jamovi version 0.9.5.12.⁵¹ SEM and mediational effects were computed with the lavaan package⁵² for the R software.⁵³

Results

See table 2 for the sociodemographic and clinical characteristics. Table 3 depicts the correlations between symptoms and personal identity variables. For positive symptoms, the self-discrepancies and dichotomous thinking showed significant correlations. Regarding covariates, earlier age at onset and higher antipsychotic dosage were associated with positive symptoms. In contrast, negative symptoms were associated with PVAFF and number of elicited constructs. For covariates, only gender reached significance, with females showing less negative symptoms than males. The correlations of psychotic symptoms and potential covariates can be found in the supplementary material, table section B.

Table 2.

Demographic and Clinical Features of Participants (N = 85)

Variable	n	%
Diagnosis
Schizophrenia	39	45.9
Schizoaffective disorder	21	24.7
Psychosis not otherwise specified	16	18.8
Delusional disorder	2	2.4
Brief psychotic disorder	3	3.5
Schizophreniform disorder	4	4.7
Gender
Male	54	63.5
Female	31	36.5
Marital status
Single	62	72.9
Married	15	17.6
Divorced/widowed	8	9.4
Level of education
Primary	7	8.3
Secondary	55	64.7
University uncompleted	10	11.8
University completed	13	15.3
Employment status
Worker	14	16.5
Student	7	8.2
Incapacity	32	37.6
Unemployed	25	29.5
Others	7	8.2
Stage of the disorder
Recent onset	26	30.6
Chronic	59	69.4
	M (SD)	Range
Age	37.1 (9.57)	19–57
Age at onset	25.6 (7.54)	13–46
Years of disorder duration	11.4 (8.78)	0.5–39
Number of hospitalizations	3.20 (3.98)	0–22
Antipsychotic dose, mg/da	228 (324)	0-2292
PANSS Positive	7.39 (3.23)	4–16
PANSS Negative	8.15 (4.17)	5–20
PANSS Depressed	11.8 (4.31)	5–23
PANSS Excitative	5.19 (1.68)	4–11
PANSS Cognitive	4.98 (1.91)	3–10
BDI	16.7 (11.9)	0–48
Hinting Task	10.3 (1.74)	3–12
Self-reflectiveness BCIS	14.7 (4.36)	4–27
Self-certainty BCIS	8.07 (3.32)	1–18
WCST (Perseverative errors)	42.5 (8.11)	29–57
WCST (Categories completed)	3.82 (2.11)	0–6
Premorbid IQ	106 (13.1)	70–140
Self-ideal discrepancy	0.38 (0.05)	0.05–0.73
Self-others discrepancy	0.28 (0.09)	0.07–0.49
Ideal-others discrepancy	0.24 (0.07)	0.05–0.44
Polarization	31.9 (19.1)	0.99–76.60
PVAFF	43.6 (12.7)	21.40–81.30
Number of elicited constructs	21.9 (8.34)	7–42

Note: ^aAntipsychotic drug doses are expressed as chlorpromazine equivalence; Psychosis of recent onset: Five or under 5 years of evolution of the disease; PANSS, Positive and Negative Symptoms Scale; BDI, Beck Depression Inventory; BCIS, Beck Cognitive Insigt Scale; WCST, Wisconsin Card Sorting Test; PVAFF, Percentage of Variance Accounted for the First Factor.

Table 3.

Correlations Between Psychotic Symptoms and Personal Identity Variables

	Negative Symptoms	Self-ideal	Self-others	Ideal-others	Polarization	PVAFF	Constructs
Positive symptoms	.02	.34**	.31**	.26*	.25*	−.11	.09
Negative symptoms	—	−.05	.13	−.19	.04	.35**	−.42***

Note: Self-ideal, self-ideal discrepancy; self-others, self-others discrepancy, ideal-others, ideal-others discrepancy; PVAFF, Percentage of Variance Accounted for the First Factor (Interpersonal cognitive differentiation); Constructs, number of elicited constructs (interpersonal cognitive richness).

*P < .05, **P < .01, ***P < .001.

After examining the correlations between variables for both models (supplementary tables C and D), they were introduced into 2 structural equation models. Figure 1 shows the final model for positive symptoms, which had good fit statistics (χ ²₍₁₄₎ = 18.21; P = .197; CFI = 0.98; TLI = 0.96; RMSEA = 0.059; SRMR = 0.065). Neither the path between positive symptoms and age at onset (z = −1.85) nor antipsychotic dosage (z = 0.68) was significant, so age at onset and antipsychotic dosage were removed from the model. The model accounted for 22.3% of the variance in positive symptoms. Additionally, depressive symptoms mediated the relationship between self-discrepancies and positive symptoms (β = .217, P = .017, ±95% CI [.038, .395]). The direct pathway did not remain significant, suggesting full mediation. Dichotomous thinking was not a mediator of self-certainty on positive symptoms (P = .148).

Fig. 1.

Final model for positive symptoms.

Figure 2 depicts the final model for negative symptoms. It had excellent fit statistics (χ ² ₍₁₂₎ = 8.28, P = .763; CFI = 1.00; TLI = 1.04; RMSEA = 0.00; SRMR = 0.037). The path between negative symptoms and gender was nonsignificant (z = −0.83), so it was removed from the model. The model accounted for 21.3% of the variance in negative symptoms. Moreover, interpersonal cognitive differentiation mediated the relationship between self-reflectivity and negative symptoms (β = −.083, P = .035, ±95% CI [−.160, −.006]). The direct pathway was not significant, suggesting full mediation. Interpersonal cognitive richness also fully mediated the relationship between neurocognition and negative symptoms (β = −.203, P = .002, ±95% CI [−.330, −.075]). Self-reflectivity was not a mediator of neurocognition on interpersonal cognitive differentiation (P = .16).

Fig. 2.

Final model for negative symptoms.

Previous tested models and comprehensive explanations of the post hoc model modifications appear in the supplementary material (sections F and G).

Discussion

In this study, we tested whether personal identity disruptions were different in positive and negative symptom expressions in a heterogeneous sample of outpatients with psychosis. Our results suggest that different dimensions of personal identity are related with positive and negative symptoms. Dichotomous thinking was directly associated with positive symptoms, while self-discrepancies worked as indirect predictors of positive symptoms via depressive symptoms. Finally, we found that interpersonal cognitive differentiation and interpersonal cognitive richness directly influenced negative symptoms.

For positive symptoms, higher interpersonal dichotomous thinking influenced symptomatology, while self-certainty affected dichotomous thinking. High self-certainty suggests that the individual is excessively convinced of the accuracy of his/her beliefs; this is conceptualized as a form of rigidity.⁴⁴ Therefore, it is possible that an underlying cognitive process of rigidity manifests itself in 2 specific ways: self-certainty and rigidity in the interpersonal thinking. Known cognitive biases that shape positive symptoms such as need for closure⁵⁴ and bias against disconfirmatory evidence⁵⁵ may share a common process of rigidity. In this sense, our results suggest that psychological models of positive symptoms might be improved if they consider dichotomous thinking, a cognitive bias connected to rigidity, in the context where it happens, the domains of personal identity and interpersonal relationships. Regarding the affective pathway of our model, it suggests that internal experience of self-discrepancy may affect positive symptoms when they take the shape of depressive mood, which has also been shown to act as a key maintenance factor for positive symptoms.¹⁷

In the case of negative symptoms, the fact of having a poor cognitive system may lead to avolition and impoverished internal experiences. From PCT, both can be understood in terms of constriction, defined as a reduction of the person’s phenomenological sphere.^10,56 Constriction has been described in depression,⁵⁷ but our results suggest it also applies to negative symptoms of psychosis. It may be that people with psychosis experience decrements in verbal and set-shifting abilities, which may naturally lead to difficulties in the ability to consider different perspectives and to evaluate alternative hypotheses.⁵⁸ This poor self-reflectivity may affect set-shifting in the interpersonal context, leading to diminished differentiation of self and others. It is also possible that deficits in neurocognition become salient when individuals are unable to generate a rich system of personal meanings that can be used to understand self and others. Taken together, with a constricted construct system, the conditions necessary to experience volition and affect become diminished, leading to negative symptoms. Negative symptoms result in the constriction and avoidance of social and emotional experiences; therefore, if people do not face more complex situations which could help them enrich their construct system, the cycle is maintained by this avoidance.⁵⁶ These findings are consistent with recent models suggesting that poorer synthetic metacognition, which implies a less complex understanding of self and others, may contribute to negative symptoms.³¹ Of note, there are alternative hypotheses; the directionality of the proposed paths could be reversed or the relationship could be interactive, meaning that negative symptoms undermine the understanding of self and others, or that negative symptoms and personal identity disruptions affect one another.

Taken together, in both models, except for self-discrepancies, personal identity variables showed significant direct paths to symptomatology, which suggests that personal identity disruptions are more proximally associated with symptomatology for individuals with psychosis than better-recognized factors such as cognitive insight, ToM, and neurocognition. The relevance of personal identity found in this study is also congruent with emerging person-centered approaches to schizophrenia.^3,5,59 These approaches stress that people with schizophrenia are not passive recipients of difficult biopsychosocial challenges. Rather, they are active meaning-making agents of life challenges aimed at forging a meaningful life.

Some limitations need to be noted. First, the cross-sectional design of the study prevents drawing conclusions regarding causality, the relationships proposed could be reversed or even interactive, so longitudinal studies are needed. Second, the RGT assessments were lengthy and taxing for many patients. Therefore, priorities had to be set. Due to the focus on personal identity rather than cognition, we only conducted a short screening of neurocognitive, ToM, and cognitive insight impairment. These variables probably would have shown a stronger impact if we had used a larger battery. Thus, more work is needed to explore the links between personal identity factors and measures of social cognition, metacognition, and dysfunctional attitudes (eg, generalized negative expectations⁶⁰). We attempted to obtain a broad spectrum of outpatients’ profiles by including recent onset and chronic patients, which was included as a potential covariate. However, although the covariate did not affect the proposed models, the proportion of chronic patients was bigger compared to the proportion of recent onset patients, and this raises the question of whether the chronic patients might be accounting for the findings. Future studies with a focus on recent onset patients would help to resolve this issue. The sample did not include inpatients nor can it be considered representative of outpatients with severe psychotic symptoms, so the results cannot be generalized to these groups. We did not have a healthy or clinical control group to compare our results. We also examined psychotic symptomatology as a whole; for positive symptoms, our sample consisted mainly of patients with low-to-moderate symptoms of persecutory delusions and distressing voices, with very few patients experiencing grandiosity. Future studies should explore the fit of this model in people with more prevalence of grandiosity. For negative symptoms, the way in which our model may apply differently to separate dimensions, such as avolition/apathy and diminished expression, needs to be explored.

Finally, our sample was drawn from routine clinical practice with wide inclusion criteria, aimed at reflecting the natural clinical context. Therefore, this sample reflects mainly outpatients with moderate symptomatology who are willing to explore their personal views, acting as potential help-seekers for psychotherapy for psychotic symptoms.

These results may have clinical implications. First, the use of the RGT in clinical settings could help clinicians to detect the personal identity disruptions of their patients, thus enhancing the personalization of the intervention. Second, this study reinforces recent cognitive interventions for positive symptoms that stress the importance of targeting dichotomous thinking¹⁸ and the negative view of self and others.⁶¹ Metacognitive training has also addressed similar cognitive biases as well as emotional well-being, showing improvements in personal meanings ascribed to psychotic illness.⁶² Accordingly, these interventions may consider dichotomous thinking as a generalized cognitive bias that involves the construction of personal identity and of significant others. In other words, working with cognitive biases with a specific focus on their interpersonal context may be more mobilizing for patients and, thus more effective. These suggestions are also congruent with observations of clinicians, and with metacognitive therapies that focus on larger senses of identity, such as Metacognitive Reflection and Insight Therapy⁶³ or Metacognition Oriented Social Skills Training.⁶⁴ The self-discrepancies involve the personal constructs of the patients, showing that the negative views mentioned above involve not only aspects such as appraised threat from others or feelings of perceived vulnerability, dimensions usually covered by traditional questionnaires, but they also reflect the individual’s system of meanings. Consequently, interventions could be tailored to help patients with psychosis to improve their views of self and others according to individual needs. Lastly, this study also supports the need for targeting personal narratives and self-reflectivity processes in psychosis for negative symptoms.^31,36

These interventions might help patients enrich the structure of their personal identity. This enrichment could be achieved by developing the complexity of the representations of self and others, or by assisting patients to incorporate more meanings in their representations. Overall, psychological interventions which focus on dichotomous thinking and self-discrepancies could improve positive symptoms, while the focus on interpersonal cognitive differentiation and interpersonal cognitive richness might ameliorate negative symptoms on people suffering psychosis.

Funding

This study was funded by the F.P.U program (ref. FPU15/01721) of the Ministerio de Ciencia e Innovación and by the Generalitat de Catalunya (refs. 2017SGR642, 2017SGR964).