Facial affect recognition and schizotypy
Dr Julie D. Henry, School of Psychology, University of New South Wales, Sydney, NSW 2052, Australia. Email: firstname.lastname@example.org
Aim: Deficits in facial affect recognition are well documented in schizophrenia, and have been associated with reduced social functioning and interpersonal difficulties. The aim of the present study was to test the possibility that facial affect recognition deficits represent an endophenotypic marker of schizophrenia liability by testing this capacity in individuals with the predisposition to symptoms of schizophrenia.
Methods: Eight hundred and forty-three psychologically healthy participants completed the Schizotypal Personality Questionnaire of which 28 scoring in the upper 15% (high-schizotypy group) and 28 scoring in the lower 15% (low-schizotypy group) completed measures of facial affect identification, facial affect discrimination, facial identity recognition, and a measure of negative affect.
Results: After controlling for group differences in negative affect and facial identity recognition, negative (but not positive or disorganized) aspects of schizotypy were found to be significantly associated with reduced facial affect discrimination and facial affect recognition accuracy, and in particular, difficulties with the identification of negative emotions.
Conclusions: These results provide limited support for the potential trait status of facial affect recognition deficits in schizophrenia and schizophrenia spectrum disorders, and suggest that these deficits may be particularly associated with the predisposition to negative symptoms of schizophrenia.
‘Schizotypy’ refers to enduring, biologically determined personality and cognitive traits that indicate a predisposition to psychosis.1 Although deficits in facial affect recognition are consistently reported in schizophrenia, the six empirical studies to date that have assessed the relationship between facial affect recognition and schizotypy have failed to yield consistent results. While Mikhailova et al.2 found that high-schizotypal individuals presented with poorer facial affect recognition relative to controls, Waldeck and Miller3 reported that this difficulty is specific to the recognition of happiness and surprise. Another study reported affect identification errors in schizotypal individuals identified using a composite score derived from scales reflecting vulnerability to positive symptoms.4 In contrast, two other studies have reported no evidence of impaired affect identification in schizotypy,5,6 and while Green et al.7 found that high delusion-prone individuals took longer than low delusion-prone individuals to identify angry facial affects, recognition errors for these expressions failed to attain significance.
Potential reasons for these inconsistencies include the use of diverse facial stimuli, control tasks and methods of group classification. Of particular importance, those studies reporting reduced facial affect recognition in association with heightened levels of schizotypy all failed to use a differential design to determine whether observed impairments were specific to facial affect recognition or instead attributable to a more generalized impairment in face processing.2–4 A further distinction between tasks assessing affect identification versus discrimination is important because difficulty on the former might be associated with the verbal, rather than perceptual, component of the task. None of the three studies in question included either a facial identity recognition or a facial affect discrimination task. Additionally, two of these studies involved participants from a clinical population,2,3 and consequently, results may have been confounded by illness-related factors associated with medication and hospitalization.
Of the two studies that failed to identify an association between facial affect recognition and schizotypy, one defined schizotypy in relation to positive schizotypal characteristics only.6 In Raine’s8 model of schizotypal personality cognitive disturbances are considered to give rise predominantly to cognitive-perceptual (i.e. positive) and disorganized features of schizotypy, while affective disturbances primarily shape interpersonal deficits such as blunted affect (i.e. negative schizotypal characteristics). In the context of this model, facial affect recognition should therefore be particularly associated with negative schizotypy. Finally, Toomey and Shuldberg also failed to identify an association between facial affect recognition and schizotypy, but the relatively weak differentiation between high- and low-schizotypal participants may have been a contributing factor to this null result.5 While the high-schizotypy group scored equal to or greater than two standard deviations (SDs) above the mean on the schizotypal measure, the low-schizotypy group consisted of participants that scored less than or equal to 0.5 SDs above the mean on this measure.
The aim of the present study was to quantify the relationship between schizotypy and facial affect recognition accuracy using a methodologically rigorous design. This included examination of facial affect identification alongside measures of both facial identity recognition and facial affect discrimination, in order to differentiate between specific difficulties with facial affect recognition and more general difficulties with face processing or verbal labelling. Schizotypal traits were measured using the Schizotypal Personality Questionnaire (SPQ9) to provide an index of all three facets of schizotypy. The sample consisted of non-clinical participants to eliminate the potential influence of illness-related confounds. Finally, to ensure sufficient discrimination between the high- and low-schizotypy groups, only participants who scored in the top and bottom 15% of the SPQ distribution were allocated to the high- and low-schizotypy groups, respectively.
Eight hundred and forty-three first-year psychology students completed the SPQ-B, which was presented as part of a larger counterbalanced screening package. Sixty participants who scored in the top and bottom 15% of the distribution of SPQ-B scores were then invited to participate in an individual testing session. As SPQ-B scores did not fully predict the distribution of SPQ scores, four cases were excluded from the final analysis because their subsequent SPQ scores fell outside of the high and low ranges. As a consequence, the final analysis involved 28 high- and 28 low-schizotypy participants with mean scores of 48.4 (SD=9.85) and 10.3 (SD=5.82), respectively, on the SPQ. The high-schizotypy group consisted of nine male and 19 female participants, with a mean age of 18.8 years (SD=3.01 years). The low-schizotypy group consisted of five male and 23 female participants, with a mean age of 19.6 years (SD=3.50 years). Participants in the high- and low-schizotypy group did not differ significantly with regard to age (t=0.90, d.f.=54, P=0.37). The disproportionate number of female participants to male participants was purely a function of the greater number of female students who completed the initial screening package.
After providing informed consent, each participant completed the measures outlined below in a counterbalanced order:
The SPQ is a 74-item self-report questionnaire with a dichotomous yes/no response format.9 The SPQ consists of three schizotypy factors (positive, negative and disorganized). Vollema and Hoijink’s10 guidelines were used to allocate items to each of the three dimensions. As a measure of schizotypy, there is evidence that the SPQ has substantial reliability and validity.11
The Ekman 60 Faces Test (from the Facial Expressions of Emotion Stimuli and Tests; FEEST12) was used to measure facial affect identification. In this computer task the 60 items are presented in a random order for 5 s each. Participants are required to choose a label that best describes the emotion displayed by each face. The FEEST has been shown to be a reliable and valid measure of emotion recognition in various populations.13 In addition to a FEEST Total score, a FEEST Negative score was derived by pooling correct responses for Disgust, Anger, Sadness and Fear, and a FEEST Positive score by pooling correct responses for Happiness and Surprise.
Ekman and Friesen’s14 Pictures of Facial Affect were used as the facial stimuli in a 14-slide discrimination task constructed for this study. Participants were shown slides with eight faces per slide (one target face and seven comparison faces). Participants were instructed to choose a face displaying the same emotion as the target face. Overall, there were 14 target faces, two for each of the following emotions: Anger, Fear, Happiness, Sadness, Surprise, Disgust and Neutral. The Pictures of Facial Affect have been well validated for research into the recognition of facial expressions.
The Benton Test of Facial Recognition is a 13-item task designed to measure facial identity recognition. In this task participants are shown a photo of a target face, along with six other faces. In the first part (slides 1–6) the participant must identify which one of the six faces is the same person as the target face. In the second part (slides 7–13) the participant must identify which three of the six faces is the same person as the target face. Evidence for the reliability of this measure has been presented.15
The Hospital Anxiety Depression Scale (HADS16) was developed to provide a brief means of identifying and measuring severity of depression and anxiety in non-psychiatric clinical environments. This measure was included because anxiety and depression may negatively bias the recognition of facial expressions of emotion17,18 and are highly comorbid with schizophrenia and schizophrenia-spectrum disorders.19 Thus, inclusion of this measure was important to assess whether any observed relationships between schizotypy and facial affect recognition are simply attributable to shared variance with negative affect. The HADS has been shown to have good reliability and validity as an index of this construct.20
Eight hundred and forty-three first-year psychology students completed the SPQ-B, which was presented as part of a larger counterbalanced screening package. Sixty participants who scored in the top and bottom 15% of the distribution of SPQ-B scores were then invited to participate in an individual testing session that involved completing the SPQ, FEEST, Facial Affect Discrimination task, Benton Test of Facial Recognition and the HADS, in a counterbalanced order.
Table 1 summarizes performance by the two groups on each of the dependent measures along with inferential statistical test results. Independent sample t-tests were used to ascertain whether the low- and high-schizotypy participants differed significantly on any of the dependent measures of interest. Effect sizes of group differences expressed as Cohen’s d are also presented for each relevant comparison; Cohen21 defines effect sizes of 0.2 as small, 0.5 as medium, and 0.8 as large. It can be seen that the high-schizotypy participants experience significantly greater difficulty with FEEST positive facial affect identification. Exploratory analyses that focused on each of the specific emotions targeted by the FEEST indicated that this group difference was largely attributable to Surprise, which only just failed to attain significance. No other group comparisons for the individual FEEST emotions were significant. However, high-schizotypy participants did differ significantly on the two control measures of interest, presenting with significantly higher levels of negative affect, and better performance on the Benton Test of Facial Recognition. All other group differences failed to attain significance.
Table 1. Descriptive statistics, inferential statistics and effect sizes for measures of facial affect recognition, facial affect discrimination, facial recognition and negative affect for high- (n=28) and low- (n =28) schizotypal participants
| Total FEEST||49.2||4.52||49.9||3.69||–0.62||54|| 0.54|| 0.17|
| Negative FEEST||30.7||4.05||30.7||3.89|| 0.03||54|| 0.97||<0.01|
| Positive FEEST||18.5||1.29||19.5||2.22||–2.14||54|| 0.04|| 0.58|
|Facial affect discrimination||11.8||1.34||11.6||1.48|| 0.57||54|| 0.57||–0.15|
|Facial recognition||24.1||1.88||23.1||1.84|| 2.01||54|| 0.04||–0.55|
|Negative affect||17.4||4.47|| 8.0||3.60|| 8.67||54||<0.01|| 2.36|
Given the significant group differences on the HADS and the Benton Test of Facial Recognition, the subsequent analyses assessed the relationship between schizotypy and facial affect recognition accuracy while controlling for these differences. This was achieved by calculating partial correlations between SPQ scores and performance on the measures of Facial Affect Discrimination and the FEEST, while controlling for scores on the HADS and the Benton Test of Facial Recognition. These partial correlations are reported in Table 2.
Table 2. Correlations between scores on each of the schizotypal dimensions with measures of facial affect recognition, controlling for negative affect and facial recognition (n=56)
| Total FEEST||–0.23||–0.10||–0.35*||–0.12|
| Negative FEEST||–0.21||–0.11||–0.30*||–0.09|
| Positive FEEST||–0.09||–0.06||–0.08||–0.10|
|Facial affect discrimination||–0.31*||–0.21||–0.32*||–0.19|
It can be seen that higher levels of overall schizotypy (as indexed by SPQ Total) are associated with reduced FEEST scores, but only the correlation with Facial Affect Discrimination attains significance. With respect to the specific factors of schizotypy, while positive and disorganized schizotypy are not significantly related to scores on any of the facial affect measures, negative schizotypy is significantly associated with reduced recognition accuracy on all measures, except for FEEST Positive. Thus, the relationship between negative schizotypy and the FEEST is specific to the recognition of negative (and not positive) emotions. Exploratory analyses that focused on the specific emotions targeted by the FEEST indicated that higher levels of negative schizotypy were associated with comparable difficulty recognizing all four negative emotions (r ranged from –0.15 to –0.21). For positive and disorganized schizotypy, the magnitude and direction of the correlations with specific negative emotions were more inconsistent (r ranged from 0.06 to –0.22 for the former, and 0.03 to –0.15, for the latter).
The present study revealed evidence for reduced identification accuracy of negative (but not positive) facial emotions in association with negative aspects of schizotypy, after controlling for differences in both facial identity recognition and negative affect. These results thus suggest that there may be a specific relationship between negative schizotypy and the recognition of negative facial emotions, and may therefore explain why Green et al.7 and van’t Wout et al.6 failed to identify significant differences in facial affect recognition accuracy between high- and low-schizotypy groups. As noted previously, in both these studies groups were differentiated on the basis of positive schizotypal characteristics only.
It has been suggested that because facial affect recognition deficits have been observed in a stabilized first-episode outpatient sample of participants with schizophrenia, they may represent a trait feature of the disorder, and not simply a marker of state factors, chronicity, or the influence of illness-related variables such as institutionalization, long-term medication effects or psychosocial deterioration.22 Consistent with this possibility, it has been shown that facial affect recognition deficits present in the acute phase of the illness do not improve with a concomitant improvement in symptoms.23 By demonstrating an association between negative schizotypal characteristics and facial affect recognition deficits in a sample drawn from the non-clinical population, the present findings provide further support for the potential trait status of facial affect recognition deficits in schizophrenia and schizophrenia spectrum disorders.
The present study, however, has a number of limitations. In particular, the use of still, posed photographs is a poor approximation of real-life emotion recognition processes, as everyday facial expressions are brief and dynamic,24 and their recognition is supported by contextual cues.13 Future research should therefore seek to test the relationship between facial affect recognition and schizotypy using more ecologically valid measures. Additionally, although the results involving the FEEST also suggest that negative schizotypy may be associated with deficits in the recognition of negative (but not positive) emotional expressions, positive emotions are generally easier to recognize.25 Thus, the positive items on the FEEST may have been insufficiently sensitive to detect the relatively subtle impairments associated with negative schizotypy. To determine whether negative schizotypy is associated with reduced accuracy for the recognition of positive (as well as negative) emotions, future research should seek to more clearly equate the level of difficulty for the positive and negative dimensions on the measure of facial affect recognition used.
Finally, in the present study broad summary scores were selected as the primary dependent measures of interest for analyses involving the FEEST because of the relatively small sample size, and thus the need to minimize the number of pairwise comparisons (and associated inflated Type I error rate). However, it has been suggested that people with schizophrenia may have disproportionate difficulty recognizing specific negative emotions such as fear and sadness.22 Therefore, to test the possibility that schizotypy might also be associated with differential recognition of specific emotions, and in particular differential recognition of specific negative emotions, additional exploratory analyses were conducted. These preliminary analyses indicated that negative schizotypy was consistently associated with difficulty recognizing Fear, Anger, Sadness and Disgust, and that these correlations were of a comparable magnitude. In comparison, the magnitude and direction of the associations between positive and disorganized schizotypy with specific target emotions were inconsistent. However, given the noted small sample size of the present study, as well as recent evidence identifying distinct neural substrates for recognition of specific negative emotions,26,27 future, larger research studies should seek to further delineate whether different aspects of schizotypy relate differentially to recognition of specific emotions.
In conclusion, relative to their low-schizotypy counterparts, high-schizotypy participants have been shown to exhibit increased rates of neurological soft signs as well as a range of impairments on measures of emotional, social and cognitive functioning that are generally intermediate in size between individuals with schizophrenia and healthy controls,8,28 consistent with the hypothesized continuum of psychosis-proneness. Given that there is considerable evidence that schizophrenia is associated with deficits in facial affect recognition,13 the present results provide further support for the view that schizotypy reflects the continuum of vulnerability towards schizophrenia and provides some limited support for the potential trait status of facial affect recognition deficits in schizophrenia and schizophrenia spectrum disorders.