In comparison to controls, patients with psychosis presented problems in both ToM and emotion recognition (ER). Nevertheless, solely problems in ER were related to delusional distress, but not related to other measures of delusions. Problems in ER also had an impact on the quality of the therapeutic alliance and interactional problems viewed by the therapist. Finally, good ER ability reduced the negative impact of delusional frequency on both the quality of the therapeutic relationship and interactional problems during CBT/ST.
The present study is the first study reporting that problems in ER are associated with delusional distress, whereas there were no further associations with other measures of delusions. Our results suggest that problems in ER might not be directly related to the presence of delusions, but enhance the distress associated with them, in line with experimental designs suggesting that problems in ER are more pronounced in stressful situations in patients with psychosis (49). It is possible that in stressful situations, patients with psychosis are less able to interpret social cues correctly and perform more errors in recognizing emotions and in inferring emotional states of other persons (50, 51) and this might lead to interactional problems and reduced social functioning (23). Thus, it might be useful to further investigate using longitudinal assessments whether problems in ER - although not directly associated with the intensity of delusions - might increase delusional distress and thus, should then be included as on of the cognitive factors involved in the development and maintenance of delusional distress in theoretical models.
The fact that we did not find an association between ToM problems and any measure of delusions nor delusional frequency / distress is in line with findings of two recent meta-analyses (14, 18) and several other studies that did not find an association between ToM and delusions (8, 52, 53), whereas negative symptoms and symptoms of disorganisation are more constantly associated with ToM problems (14). Thus, our results support more recent theoretical models that excluded ToM problems as important cognitive factors involved in the formation and maintenance of delusions or positive symptoms (12, 13).
Our study is the first to report that therapists who treated patients with problems in ER perceived more pronounced interactional problems in these patients and rated the quality of the therapeutic relationship more negatively. Our findings are partly in line with the study of Jung and colleagues (29) who reported an association between patients’ ratings on the quality of the therapeutic alliance and ToM problems, but no associations between therapists’ ratings and ToM problems, but the size of our study sample enabled us to detect associations of medium and small effect size. Nevertheless, it has to be taken into account, that our results were not pre-specified, but obtained in an exploratory analysis of a randomized-controlled therapy trial, thus, careful replication of our results should be performed, especially in light of the current replication crisis in psychology (see (54) for a review)). If our results are successfully replicated and patients’ problems in ER influence the relationship with an empathetic and highly skilled therapist, it can be assumed that their problems in ER also have a negative impact on other social interactions in their daily life, as suggested by several other studies that directly addressed this question (21, 22).
In addition, we could provide evidence for the clinically important negative impact of delusional frequency on both the therapeutic relationship and interactional problems. Again, it is important to note these results were obtained in an exploratory analysis and thus are in need of careful replication. If our findings are replicated in longitudinal assessments, they suggest that delusional frequency negatively affects social interactions (the therapeutic interactions) and thus may also partly contribute to the association between delusions and lower social functioning (55, 56), negative family atmosphere (57), more pronounced loneliness (58, 59) and social exclusion (60, 61).
Further, preserved ER abilities might protect patients from the negative influence of their delusions on the quality of the therapeutic relationship and interactional problems, as a statistically significant moderation effect occurred. Our results are partly supported by a second study that addressed the impact of ToM on the association between delusions and self-rated social functioning (31). Their results also suggest that preserved ToM abilities moderated the relationship between persecutory delusions and self-rated social functioning (31). The fact that we did not find a similar moderation effect between ToM and the quality of the therapeutic relationship could be explained by different ToM assessments (the study used the Hinting task (62)) that is based on verbal descriptions of social situations, we used a picture sequencing test based on comics. To some degree, our findings extent their results, as we used therapists’ ratings of the therapeutic alliance as a direct measure of social functioning instead of self-ratings. Again, if our exploratory findings can be replicated and positive ER skills are a protecting factor against the negative influence of delusions on the therapeutic relationship, patients’ ER abilities might also influence the effectiveness of CBT for psychosis, as a positive therapeutic relationship is closely related to the effectiveness of CBT (28, 63, 64). Interestingly, one study indeed found general ToM abilities (including ER) to moderate change in positive symptoms in CBT (65). Thus, interventions that improve ER and ToM abilities might be beneficial in order to improve the therapeutic relationship and, further, the effectiveness of CBT.
Our results suggest for therapists of patients with psychosis to take patients’ potential ER problems into account in CBT for psychosis. First, it could be useful to assess patients with regard to their ER abilities before start of therapy. Second, if patients present ER problems, it is important for therapists to make a special effort to improve the therapeutic relationship with these patients. Third, it could be helpful to train ER in patients with psychosis using specialized interventions from several social cognition trainings in the framework of Cognitive Remediation (66): the Social Cognition and Interaction Training (67) and the Metacognitive Training (Moritz and Woodward (68)) aim on improving both ToM and ER, whereas the Training of Affect Recognition (41) aims more closely on ER. In general, these trainings were able to enhance both ToM and ER abilities (67, 69–71) and their general effect on social functioning is large (72, 73). It is also possible, that an integration in or a combination of these trainings with CBT in order to improve ER might be beneficial.
Strength and limitations
Strengths of the present study include the large sample of patients with psychosis and the detailed assessment of different dimensions of delusions. An additional strength is the longitudinal assessment of the quality of the therapeutic relationship over five sessions.
In interpreting our findings, it should be mentioned that all associations between ToM, ER, delusions and the therapeutic relationship were of small effect size according to Cohen (74). In addition, it should be noted that solely two of the four scientific hypotheses were pre-specified, whereas all associations between ER, ToM, delusions and the therapeutic relationship were of exploratory nature. Thus, the question of associations between ToM, ER and the therapeutic relationship and the moderation effect require an additional careful replication study. Finally, it should be mentioned that the patients in the present study were patients interested in participating in a therapy trial who might present better general cognitive and social functioning and less pronounced problems in their social cognition. Nevertheless, as the moderation effect occurred predominantly in patients with lower ER skills, a potential selection bias might not influence the generalisation of our exploratory results.