Background and Summary

Social perceptual judgments facilitate the evaluation of another person, particularly when drawing upon emotionally salient information1,2. Individuals use facial, vocal, gestural, and postural cues to communicate emotions to others; whilst also reading such cues to determine anotherā€™sā€™ internal state3,4. Here, visual processing of the human face remains the most prevalent source of information5. Indeed, facial observations provide vital information pertaining to distinguishing characteristics, perceived attractiveness, intended behaviour, and emotional state, without use of language6. Accordingly, the nature of facial expressions serves as a primary source of information in social perception7,8.

Accurate judgments of facially expressed emotion are vital for effective social interactions and non-verbal social judgments. Therefore, perceptual deficits associated with psychiatric symptoms may contribute to negative psychosocial consequences for those involved9. Perceptual alterations of emotional faces are widely established in psychiatric populations at both disorder and symptom level (e.g. insomnia, anxiety, depression, bipolar disorder, anorexia nervosa, schizophrenia)10,11,12,13,14,15.

The present work sought to further examine the prevalence of perceptual alterations of emotional faces amongst individuals experiencing symptoms of insomnia, anxiety, depression, mania, psychotic experiences, and schizotypal tendencies. More specifically, we examined the categorisation accuracy (whether the displayed emotion was correctly identified), misperception (which emotion an incorrect judgment was perceived to be), intensity (the strength of the emotion signal strength) and emotional valence (the extent and direction of perceived affect in relation to intrinsic appeal or repulsion) of all six cross-culturally accepted facial expressions of emotion16.

Methods

Sample and Design

The study was approved by the Sheffield Hallam University Research Ethics Committee (Protocol number: ER28407811), and all participants provided online informed consent. Students from two UK universities were recruited through institutional course participation schemes, social media groups and faculty emails. This resulted in a sample of Nā€‰=ā€‰706 individuals who either began or clicked on a hyperlink to the survey which was delivered using the Qualtrics platform (Qualtrics, Provo, UT). Only complete cases were used in the analysis due to the ethical right to withdraw from the survey at any time. The data were also examined for duplicate responses based on matching IP addresses, where none were found. To maintain participant anonymity, IP addresses were subsequently discarded. In addition, no personally identifying information or contact details were recorded. Rather, subjects provided a memorable keyword at the start of the survey which was to be cited in any request to withdraw data. Therefore, Nā€‰=ā€‰572 respondents (mean ageā€‰=ā€‰21.71ā€‰Ā±ā€‰7.30 years, range 18ā€“65, 76% female; 89% student) providing complete data for the variables of interest. All participants reported: normal to corrected-to-normal vision, the absence of prosopagnosia, being ā‰„18 years old.

Materials

Facial stimuli

Fifty-six facial photographs of eight individuals (50% female) displaying the emotional expressions of fear, anger, disgust, happiness, sadness, surprise, and neutrality were gathered from the Karolinska Directed Emotional Faces database17. In the present study, we chose to use all six cross-culturally accepted facial expressions of emotion16. These six expressions are commonly used to examine the perception of emotional faces in the context of psychiatric disorder. In line with previous studies controlling for potentially distracting and confounding factors, we cropped the hair and neckline from each image18,19,20,21. Thus, leaving a series of oval-shaped neutral facial images (see Fig.Ā 1).

Fig. 1
figure 1

Example set of facial expressions.

Full size image

Psychometric measures

Insomnia

Insomnia symptoms were assessed using the Insomnia Severity Index (ISI)22. The ISI consists of seven items examining the severity of insomnia symptoms over the past 2 weeks including difficulty initiating and maintaining sleep and awakening too early. Items are scored on a 5-point likert-type scale, with total scores ranging from 0 to 28. Higher scores suggest greater insomnia severity. Total scores between 0 and 7 indicate no clinically significant insomnia, 8 and 14 subthreshold insomnia, 15 and 21 clinical insomnia (moderate severity), and 22 and 28 clinical insomnia (severe). Assessment of internal consistency Cronbachā€™s alpha) yielded an alpha of Ī±ā€‰=ā€‰0.88.

Anxiety

The 7-item Generalized Anxiety Disorder Scale (GAD-7) is a validated practical self-report anxiety questionnaire used in primary care23. The tool asks respondents how often, during the last 2 weeks, they have been bothered by each of the seven core symptoms of generalized anxiety disorder. Responses choice are 0ā€‰=ā€‰ā€œnot at allā€; 1ā€‰=ā€‰ā€œseveral daysā€; 2ā€‰=ā€‰ā€œmore than half the daysā€; and 3ā€‰=ā€‰ā€œnearly every dayā€. Total scores range between 0 and 21 with cut offs of ā‰„5, ā‰„10, and ā‰„15 indicating mild, moderate, and severe anxiety levels, respectively. The GAD-7 has been shown to exhibit good reliability, as well as criterion, construct, factorial, and procedural validity23. The Cronbachā€™s alpha in the current study was Ī±ā€‰=ā€‰0.92.

Depression

The 9-item patient health questionnaire (PHQ-9)24 is a brief self-report depression scale used to assess depressive symptoms in the general population. Each of the nine depressive symptoms corresponds to the depression criteria of the DSM-V25. Respondents are required to indicate how much, during the previous 2 weeks, the symptom has bothered them on a scale of: 0ā€‰=ā€‰ā€œnot at allā€, 1ā€‰=ā€‰ā€œseveral daysā€, 2ā€‰=ā€‰ā€œmore than half of the daysā€ or 3ā€‰=ā€‰ā€œnearly every dayā€. Total scores indicate depression severity and range from 0 to 27 with higher scores indicating higher levels of depression. The scale has been shown to demonstrate good criterion and construct validity23,24. Cronbachā€™s alpha in the current study was Ī±ā€‰=ā€‰0.91.

Mania

The Mood Disorder Questionnaire (MDQ) was used to examine symptoms of mania (range 0ā€“13)25. Participants provided a response of yes/no when presented with a series of prompting questions. For example, ā€œHas there ever been a time when you were not your usual self andā€¦ā€. The summation of each item yields a total MDQ score where higher scores indicate increased levels of manic symptoms. The internal consistency of the scale in the present study was Ī±ā€‰=ā€‰0.85. To screen positive for possible bipolar disorder, all three parts of the following criteria should be met based on the initial 13 items and two additional questions. More specifically, participants must answer yes to ā‰„7 of the thirteen items; indicate cooccurrence of ā‰„1 items within the same period; and report moderate to serious consequences of the possible episode (i.e., inability to work; family, monetary or legal difficulties; aggressive behaviour).

Psychotic like experiences

The Prodromal Questionnaire 16 (PQ-16) was administered to assess life-time symptoms of psychotic experiences26. It was developed as a brief version of the 92-item Prodromal Questionnaire26 to enable the detection of ultra-high-risk (UHR) patients in routine adult mental health services. Sixteen items evaluate the occurrence of positive/negative symptoms and avolition on a two-point scale (yes/no). In addition, symptom distress is measures on a four-point scale between: 0ā€‰=ā€‰no distress and 3ā€‰=ā€‰severe distress. More specifically, the PQ-16 contains nine items relating to hallucinations; five items relating to delusions; and two negative symptom items. The summation of ā€˜yesā€™ item responses yield a total score between 0 and 16, where higher scores indicate an increased number of psychotic symptoms. In adults, a score of ā‰„6 predicts diagnosis of psychosis with high sensitivity (87%) and specificity (87%)27. Similarly, distress items are summated to create a total prodromal severity scale. The Cronbachā€™s alpha in the current study was Ī±ā€‰=ā€‰0.85 for total scores and Ī±ā€‰=ā€‰0.90 for severity scores.

Schizotypal traits

The Short-Form Oxford-Liverpool Inventory of Feelings and Experiences scale examined schizotypal personality traits (O-LIFE)28. The 43-item measure is comprised of four subscales: unusual experiences (12-Items), cognitive disorganisation (11-items), introverted anhedonia (10-items), and impulsive nonconformity (10-items). The unusual experiences subscale examines positive symptoms related to psychosis (i.e., perceptual aberrations, magical thinking, hallucinations). Cognitive disorganisation explores disorganised features of psychosis (i.e., concentration difficulties, impaired decision-making, social anxiety). The introverted anhedonia subscale examines negative schizotypy, primarily characterised by a lack of physical and social pleasure (i.e., intimacy, social activities). Impulsive nonconformity measures diminished self-control (i.e., impulsive, anti-social, eccentric behaviour). All items utilise a two-point scale (yes/no). The summation of ā€˜yesā€™ item responses yield a total score for each respective subscale. The Cronbachā€™s alpha in the current study was Ī±ā€‰=ā€‰0.78 for unusual experiences, Ī±ā€‰=ā€‰0.81 for cognitive disorganisation, Ī±ā€‰=ā€‰0.70 for introverted anhedonia, and Ī±ā€‰=ā€‰0.70 for impulsive nonconformity.

Procedure

Participants completed an online questionnaire, in which they were presented with the series of fifty-six images in randomized order. For each face displayed, participants were asked to select (categorise) the corresponding emotional expression (fear, anger, disgust, happiness, sadness, surprise, and neutral). Subsequently, participants were asked to rate their perceived intensity and emotional valance on a 100-mm visual analogue scale. The intensity scale was anchored at not very intense and extremely intense. Whereas valence was anchored at extremely negative and extremely positive. Each face was displayed until a response was made. Following the completion of the face-task, participants completed the ISI, GAD-7, PHQ-9, MDQ PQ-16 and O-LIFE. The procedure lasted approximately thirty minutes. An example trial is presented in Fig.Ā 2.

Fig. 2
figure 2

Example trial.

Full size image

Data Records

The dataset (demographics, psychiatric measures) has been anonymised and both individual datapoints for each psychometric measure and face rating are available in CSV and SAV formats on Figshare (https://doi.org/10.6084/m9.figshare.26397817)29 and as part of the current supplemental information (see Appendix A). For copyright reasons, the original PDF files including scoring information for each measure are not included in the appendix. However, the key scoring information is provided in the aforementioned measures section, and each original measure can be individually examined where necessary. In the current dataset, reverse coding was applied where required. Finally, the total/mean or composite score for each psychiatric symptom and perceptual judgments of emotional faces displaying each expression (accuracy, and intensity) were calculated and included to simplify the usability of the dataset. Descriptions of calculated variables are provided in TableĀ 1. Details of the data cleaning procedure are available in the following section.

Table 1 Variable codes for dataset.
Full size table

Technical Validation

For the current purpose, the SAV file was exported from the Qualtrics online survey platform (Qualtrics, Provo, UT) where the data was cleaned using SPSS v. 29.0 (IBM Corp., Armonk, NT, USA). Total ratings of accuracy were summated for each expression. For each trial, correct responses yielded a score of 1 whereas incorrect responses were scored as 0. Therefore, accuracy scores ranged between 0 to 8, with higher scores indicating greater categorisation accuracy of the assessed expression. Next, mean expression intensity ratings were calculated, and ranged between a possible 0 to 100 score, with higher scores indicating greater intensity ratings. Finally, mean emotional valence ratings were calculated, and ranged between a possible 0 to 100 score, with higher scores indicating more positively valanced ratings. Descriptive statistics for all calculated variables were inspected to ensure that results fell within an expected range. Whilst normality was assessed using histograms, it is vital to consider the inherent skew observed when examining psychiatric symptoms, where a traditional disruption would not be expected. As noted, reliability analysis was performed for each psychometric measure. Here, the internal consistency (Cronbachā€™s Ī±) of each scale remained acceptable. Examination of internal consistency for the intensity of each expression yielded Ī± values between 0.82ā€“0.92. Examination of internal consistency for the emotional valence of each expression yielded Ī± values between 0.68ā€“0.88. Examination of internal consistency for the categorisation before recoding was Ī±ā€‰=ā€‰0.46ā€“0.67 after recoding. Therefore, the facial stimuli reliably depicted their corresponding expression with a degree of confidence (see TableĀ 2 for each Ī± value). The dataset was carefully inspected for abnormal response patterns and completion times before any formal analyses was conducted. Neither were observed. To facilitate the potential usability and consideration of further analysis of the current data set, the descriptive statistics (means and standard deviations), and Pearsonā€™s bivariate correlations (two-tailed) between expression intensity, emotional valance, categorization accuracy and misperception scores with symptoms of insomnia, anxiety, depression, mania, psychotic experiences, and schizotypal traits are presented in TablesĀ 2, 3 and 4 respectively. Significance was considered at the pā€‰<ā€‰0.05 level.

Table 2 Mean ratings and standard deviations (SD), and corelations between insomnia, anxiety, depression, mania, psychotic experiences and schizotypal tendencies with ratings of categorisation accuracy and intensity ratings for each emotional expression.
Full size table
Table 3 Mean ratings and standard deviations (SD) for categorization accuracy and misperception for the whole sample.
Full size table
Table 4 Corelations between insomnia, anxiety, depression, mania, psychotic experiences and schizotypal tendencies with misperception scores for each emotional expression.
Full size table

Several strengths and limitations of the current data should be noted. The cross-sectional nature limits the ability to draw any definitive explanation when concerning causal relationships. Next, the sample was not homogeneous, including a disproportionate amount of data from young white female respondents. Finally, potential users of the current data should consider limitations associated with subjective assessments which rely on self-reported information. That said, all six cross-culturally accepted facial expressions of emotion were examined16,30, with the inclusion of data concerning the emotional valance and misperception of emotional expressions above the typical inclusion of only accuracy and intensity. Whilst similar studies typically focus on a smaller set of related symptoms, the present data provides perceptual judgments across a large domain of psychiatric symptoms. Often with large scale data collection, studies may employ scales single-item or bespoke in-house scales, which fail to capture symptom severity and specificity31. Accordingly, the current data provides a comprehensive assessment of psychiatric symptoms by using well validated scales with robust psychometric properties.

Usage Notes

Data have been deidentified and are presented in the same manner across data files, which can be imported into most compatible statistical software packages. As noted, the dataset includes all individual datapoints for each de-identifiable psychometric measure and perceptual judgment. For each validated questionnaire, the initial authors scoring instructions were followed to calculate the relevant total, composite, and subscale variables which may be of interest.