Social cognition is a crucial domain for interacting with other people and participation in society. It has been defined as the psychological processes that underlie social interactions, including perception, experience sharing, mentalizing, and experiencing and regulating emotion and information about other people and about ourselves (1, 2). Among this heterogeneous category of processes, the most important domains of social cognition include the perception of social cues like emotion perception (1–4). Individuals affected by schizophrenia spectrum disorders (SSD) show significant impairments in their ability to recognize and express emotions, as already described by Kraepelin (2, 5, 6). Previously, studies of emotion deficits in schizophrenia focused on three main domains: expression, experience, and recognition of emotions. A review of 110 studies revealed marked impairment in all three domains in individuals with SSD (7). It is essential to differentiate between emotion recognition – which refers to the ability to recognize an expression of emotion such as disgust - and finding a face disgusting, which is called subjective appraisal (7). Emotion recognition has been linked to emotional intelligence and social perception, whereas appraisal has been linked to emotion processing (7). Gur et al. (8) reported an association of greater impairment for individuals with SSD in emotion recognition via facial stimuli with the severity of negative symptoms. Social cognitive impairments and negative symptoms are core features of SSD closely associated with high burden due to impairments in social functioning and quality of life already apparent prior to illness onset (2, 5, 6, 9–12). For example attenuated impairment in emotion recognition manifests in unaffected first-degree relatives, ultra-high-risk individuals (13, 14), and young individuals with first-episode schizophrenia (5, 15, 16). A study found that young individuals with schizophrenia were impaired in their ability to recognize faces and gestures from movie scenes during the first two to 3 years of their illness after symptom onset (15, 17). According to Comparelli, Corigliano (13), facial emotion recognition impairment already present in ultra-high-risk individuals remained stable across the course of illness. Deficits in emotion recognition may be one of the most consistent and severe aspects of interpersonal problems (15) and impairment of social skills (18) and are related to poorer social outcomes in schizophrenia (19). Some researchers argue emotion recognition deficits are long-term stable features of schizophrenia (20). In one model, negative symptoms, emotion recognition latency, and processing speed were significant predictors of social functioning in individuals at ultra-high risk for psychosis (21, 22). Another model for schizophrenia onset included face emotion processing and negative symptoms as predictors of transition in individuals at high clinical risk to schizophrenia with an accuracy of 96% (11). Accordingly, this represents a particularly relevant field of research since both negative symptoms and impairments in social cognition could be apparent prior to illness onset with psychotic symptoms (10, 11).

Impaired emotional functioning has been linked to negative symptoms and can be conceptualized as a fundamental clinical feature in SSD (7, 23, 24). Newer discussions on deficits in emotion recognition raise the question of whether those belong to the negative symptom spectrum or form a separate domain in SSD (1, 25, 26). According to recent models, negative symptoms are also described as deficient social cognitions (27). The existing data non-etheless suggest that negative symptoms and social cognition are related, yet distinct constructs. Over time, deficits in social cognition could manifest behaviourally as negative symptoms (1). There are different models for a possible connection between social cognition and negative symptoms. On the one hand it is assumed that specific impairments in social cognition are linked to one or some negative symptoms but not others, which means there is a mechanistic heterogeneity. In support of this model, associations between only specific social cognition domains and negative symptoms have been reported (1). On the other hand, findings support broad impairments in basic sensory and cognitive processes (e.g., visual perception, motor output, processing speed, and implicit attention), which affect downstream processes. These are directly involved in social cognitive operations, which, in turn, result in the development of a range of negative symptoms and functional impairment (1). In studies looking at various factor models no correlations were found between the social cognitive factors and negative symptoms suggesting that social cognition and negative symptoms are largely separate constructs. They can be seen as relatively independent causes of dysfunction and disability and can be used to meaningfully classify non-acute patients (12, 28, 29). Furthermore, the entry level of negative symptoms was significantly associated with poor social functioning. Important predictive links have been found in the early course of schizophrenia, mostly indicating that higher negative symptom severity is associated with poor daily functioning and worse long-term outcomes (30–33). Given this, there is increasing interest in understanding to what amount negative symptoms reflect the expression of deficits in social cognition and the relationship between the two symptom domains (1). However, little is known about the dependence of these dimensions of illness and, whether individuals with schizophrenia can be meaningfully classified based on these dimensions and potentially differentially treated. Ruocco et al. (34), who compared emotion recognition deficits in individuals with schizophrenia, schizoaffective disorders, and bipolar disorders, discovered that compared to healthy controls, emotion recognition deficits among individuals increased progressively from bipolar disorder to schizoaffective disorder to schizophrenia. The researchers concluded that emotion recognition deficits are apparent at the first psychotic episode and prominent but different across psychotic disorders and relatively independent of mood state and antipsychotic treatment dosages (34–36). Conclusively illness-related characteristics like negative symptoms need further investigation (34–36). A priority of negative over positive symptoms in determining deficits in social cognition and functioning in chronic schizophrenia was concluded by many researchers (5, 37). To develop intervention approaches in the early phase of SSD, it is crucial to understand the longitudinal course of negative symptoms, especially in relationship to functioning (30).

There are, however, no sufficient treatment options for negative symptoms. The most often employed approaches as medication, psychotherapy, psychosocial interventions, and electroconvulsive therapy have only small effect sizes (38–40). A model of negative symptoms as the behavioral manifestation of altered social cognition has the potential to reveal areas of intact functioning, including domains of social cognition that remain unaffected by the illness, which novel treatments could rely upon to enhance the recovery of individuals with schizophrenia (1). The findings regarding social cognition in schizophrenia have treatment implications and the social processes that are aberrant in schizophrenia may each require their own specific therapeutic intervention. Training programmes that target facial emotion perception and mentalizing deficits have been validated in individuals with schizophrenia (2, 41, 42).

Some researchers, such as Schneider et al. (43), concluded that especially impairments in facial emotion perception were strongly associated with the severity of negative symptoms. Since then, several studies have investigated the connection between negative symptoms and emotion deficits, but findings on the relationship between emotion recognition and negative symptoms remain inconclusive (5, 15, 43–56). There are various explanations for these mixed results. Most of the studies did not differentiate subdomains of negative symptoms and only considered negative symptoms overall. Also, some studies did not include all six basic emotions in their experimental designs but presented the subjects with a smaller selection of emotional stimuli. Paul Ekman first defined the six basic emotions: anger, disgust, fear, happiness, sadness and surprise (15, 57). Furthermore, different operationalizations and instruments for negative symptoms could be a source of variation (58, 59). To better understand the link between specific negative symptoms and emotion recognition, this study aims to explore the relationship between recognizing the six basic emotions and seven distinct negative symptoms in individuals with SSD in an exploratory approach by using the Positive and Negative Syndrome Scale (PANSS) (60) and the Emotion Recognition Test of the Cambridge Neuropsychological Test Automated Battery (ERT-CANTAB) (61).

According to the current state of research, five key sub-domains of negative symptoms have been identified: (1) avolition, (2) anhedonia, (3) blunted affect (4) social withdrawal, and (5) alogia (9, 62–64). Negative symptoms can be assessed via various instruments [e.g., PANSS, Scale for the Assessment of Negative Symptoms (SANS) (65), Brief Negative Symptom Scale (BNSS) (66), Clinical Assessment Interview for Negative Symptoms (CAINS) (67), Negative Symptom Assessment (NSA) (68)]. So, it is necessary to consider the nature of each measurement instrument, especially with the aim of relating each symptom domain to the recognition of individual basic emotions. The domain of negative symptomatology is very complex encompassing primary and secondary symptoms as a result of medication side effects, depressive symptomatology and other causes (64). In SSD, studies confirmed moderate to large associations between negative symptoms and deficits in social cognition using the SANS (65) and/or the PANSS (1, 5, 12, 28, 30, 60, 69) as well as newer negative symptom scales like the BNSS (1, 5, 12, 66, 70). In many older studies, the negative syndrome scale of the classic PANSS was used to assess the individual domains of negative symptoms. However, according to recent factor analytic studies, it must be considered that the two symptom domains difficulty in abstract thinking (N5) and stereotyped thinking (N7) are now no longer regarded as negative symptoms but are assigned to the domain of cognitive symptoms (56) and difficulty in abstract thinking (N5) to the domain of disorganization, depending on the study design (71). This should therefore be considered when evaluating results using the PANSS. Nevertheless, in the context of an explorative approach of our study and for comparability with previous studies, especially older ones, we selected the PANSS as the primary rating tool for negative symptoms as a still frequently used third-party rating instrument in clinical studies (1, 5, 12, 28, 30, 60, 69). Depending on our results, a selection of the other measurement instruments should also be included in follow-up studies for an even better understanding of the relationships between negative symptoms and social cognition. We selected the Emotion Recognition Test of the CANTAB to measure the recognition of basic emotions. It contains all six basic emotions and is easy to use as a tablet-based measurement instrument in different settings (61). It is explicitly part of the Schizophrenia Test Battery of the CANTAB and several studies with patients with SSD have already been conducted with it. For example, Gica et al. (20) could show with the ERT-CANTAB, that emotion recognition deficits are long term stable features of schizophrenia and Kanchanatawan et al. (72) showed associations of anxiety and depressive symptoms in individuals with SSD with social cognition by using the ERT-CANTAB. Glenthøj et al. (22) could show impairments in social cognition with the ERT-CANTAB in ultra-high risk for psychosis individuals and that hereby emotion recognition latency but not accuracy relates to real life functioning in these individuals (21).

In Figure 1, a correlational matrix with an overview of the findings that emerged after data collection is presented. Positive correlations within the respective scales can also be found in the literature (60, 78–80). There was a significant negative correlation between blunted affect (N1) and the total number of correct responses in emotion selection (r = −0.291, p < 0.05), the recognition of happiness (r = −0.034, p = < 0.05), fear (r = −0.246, p < 0.05) and disgust (r = −0.253, p < 0.05). A significant negative correlation of difficulties in abstract thinking (N5) with fear recognition (r = −0.337, p < 0.05) was found. There was also a significant positive correlation of the response latency in emotion recognition with difficulties in abstract thinking (N5) (r = 0.289, p < 0.05) and stereotyped thinking (N7) (r = 0.349, p < 0.05). There were no significant correlations between the negative symptom domains emotional withdrawal (N2), poor rapport (N3), passive and apathetic social withdrawal (N4), and lack of spontaneity and flow of conversation (N6) and the assessed CANTAB variables among those with SSD. In the next step, linear regressions were run to examine a significant effect of the covariates age, education, sex and diagnoses on each CANTAB variable.

When controlling for age and education, a multiple linear regression analysis was conducted to assess whether blunted affect (N1) significantly predicted the total number of correct responses in emotion selection. Hereby blunted affect (N1) significantly predicted the total number of correct responses in emotion selection [B = −0.919, t = −2.199, p = 0.032]. Blunted affect (N1) could also be identified as a significant predictor of the accurate recognition of happiness when controlling for the covariates [B = −0.241, t = −2.113, p = 0.039]. Likewise, the negative symptom domain of difficulties in abstract thinking (N5) was identified as a significant predictor of fear recognition in the regression analyses [B = −0.037, t = −2.044, p = 0.045] when controlling for education. A linear regression analysis was conducted to assess whether difficulties in abstract thinking (N5) and stereotyped thinking (N7) predicted the response latency when controlling for age. In this model stereotyped thinking (N7) significantly predicted the response latency [B = 195.921, t = 2.400, p = 0.019], whereas the negative symptom difficulties in abstract thinking (N5) was not a significant predictor in this model [B = 117.474, t = 1.535, p = 0.130]. Linear regression analysis models showed that that blunted affect (N1) was not a significant predictor for recognizing either fear [B = −0.023, t = −1.607, p = 0.113] or disgust [B = −0.027, t = −1.699, p = 0.094], while controlling for the significant covariate education. For an overview of the above significant regression analyses, please refer to Table 3. The complete respective regression tables are provided in the Supplementary Material.

The demonstrated nominal significances did not withstand correction for 111 multiple tests. After Bonferroni correction, none of the mentioned correlation and regression analyses between PANSS and the ERT items, respectively, reached statistical significance (see Supplementary Material). With a power of 0.8, an expected small effect of f2 = 0.054 of the PANSS negative scales on emotion recognition based on preliminary studies (81), and a p-value of 0.00045 corrected for 111 multiple tests, a sample size of 446 subjects would be necessary to detect a significant effect. According to the guidelines of Cohen (82) f2 ≥ 0.02, f2 ≥ 0.15, and f2 ≥ 0.35 represent small, medium, and large effect sizes, respectively.

Conclusively in our exploratory study individuals with SSD and high scores of the negative symptoms blunted affect (N1), difficulty in abstract thinking (N5) and stereotyped thinking (N7) showed impairments in recognizing basic emotions and, here, particularly happiness. However, results must be weighed against the methodological limitations of the study and follow-up studies are necessary.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Charité's Ethics Committee. The patients/participants provided their written informed consent to participate in this study.

MZ, KB, NB, IH, AB, and JK designed and executed the study and conducted the data analyses. MZ wrote the article. TT, SR, EH, and MB collaborated with the design and editing of the manuscript. All authors contributed to the article and approved the submitted version.

MZ is a participant in the BIH-Charité Junior Clinician Scientist Program funded by the Charité – Universitätsmedizin Berlin and the Berlin Institute of Health (BIH). In addition, the study was funded through the internal performance-based funding (LoM) of our research group at Charité – Universitätsmedizin Berlin.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.