Misattributions of Facial Affect and Symptomatology in Schizophrenia:

Misattributions of Facial Affect and Symptomatology in Schizophrenia:

Misattributions of Facial Affect and Symptomatology in Schizophrenia: Evidence for a Social Cognitive Bias in Paranoid Symptoms?
Jason Peer, Rachel Penrod, Thea L. Rothman and Will Spaulding
University of Nebraska-Lincoln

Previous research has found that facial affect recognition abilities are impaired in schizophrenia (i.e., Morrison et al. 1988).
However, relatively little research has been conducted on the types of facial affect misattributions persons with
schizophrenia are likely to make on these tasks and the relationships of these misattributions to specific symptoms of
schizophrenia. That is, when making an error on a facial affect recognition task how often would a participant respond with
a specific affect type (i.e. selecting anger instead of the correct response surprise)? Garfield et al. (1987) found that
persons with schizophrenia made significantly greater misattributions of an emotion as disgust when compared to a
control group. Interestingly, in this study, poor affect recognition was also correlated with defensive self-enhancement. The
authors suggest that the greater misattribution of disgust may be related to a diagnosis of paranoia but did not test this
hypothesis in their original study. This suggestion is consistent with studies that have shown that patients diagnosed with
paranoid schizophrenia tend to judge pictures of faces as more negative (Smari et al. 1994) and tend to express more
unfavorable feelings about the picture (Izard 1959) when compared to controls. Intuitively, these findings are also consistent
with current social cognitive models of paranoia (i.e., Bentall and Kinderman 1998; Chadwick and Trower 1997). These
models suggest that patients with paranoid symptoms demonstrate a number of social cognitive biases. Research has found
an exaggerated self-serving bias combined with a personalizing attribution bias (attributing negative events to other people
as opposed to situations) in persons with persecutory delusions (Bentall and Kinderman 1998). In addition, research has
demonstrated that persons with paranoid symptoms show biases in their evaluations of others (i.e., they evaluate others more
negatively) and biases in their beliefs about others evaluations of themselves (i.e., others hold negative evaluations of them)
(Chadwick and Trower 1997). These models propose that social cognitive biases are utilized to protect against depressive
affect by seeking external explanations for negative events and in turn result in attentional and evaluative biases.
Collectively, these findings suggest that patients with paranoid symptoms demonstrate biases in the processing of social
information and it is possible that these biases would be evident in the types of errors made on social information processing
tasks such as facial affect recognition.
The present study investigated differences in overall facial affect recognition, the specific pattern of facial affect
misattributions, and symptomatology within a sample of persons diagnosed with a schizophrenia spectrum disorder.
Exploratory analyses were conducted to evaluate different patterns of facial affect misattributions based on symptomatology.
Based on social cognitive models of paranoid symptoms, it was predicted that participants with paranoid symptoms would
make more misattributions of interpersonal threat (anger and disgust). Based on previous research (Lewis and Garver
1995), it was also hypothesized that participants with paranoid symptoms would demonstrate better facial affect recognition
abilities than the other symptom groups.


Analysis of FA Misattributions by Symptom Group: The results indicate some support for the first hypothesis of
the study. The high Paranoid symptom group made more misattributions of interpersonal threat than the low
paranoid symptom group (see Figure 1). Specifically, significant differences were noted for misattributions of:
disgust (t=-2.40; p=0.02), anger (t=-2.62; p=0.01) and threat (t=-3.28; p=0.002). Significant differences were also
noted for misattributions of sad (t=-3.03; p=0.004). The evaluation of other symptom groups indicated some
similar patterns, although results were not as robust. Some differences between the high and low
Anxiety/Depression symptom groups that approached significance were also observed. Specifically, differences
were noted for misattributions of: disgust (t=-1.67; p=0.10) and threat (t=-1.93; p=0.06). Some differences
between the high and low Disorganized symptom groups that approached significance were also observed.
Differences were noted in misattributions of: disgust (t=-1.72; p=0.09); threat (t=-1.91; p=0.06) and happy (t=2.03; p=0.048). One significant difference for misattributions of sad (t=-2.06; p=0.048) was noted between the
Agitation/Elation symptom groups. No significant differences were noted between the high and low Blunted
symptom groups or between the high and low Hallucination/Delusion symptom groups.
Analysis of Overall FA Recognition Ability: When comparing the high and low Paranoid symptom groups on
overall FA recognition ability, the high paranoid symptom group was observed to be significantly more impaired
(t=2.91; p=0.005). The analysis of high and low Anxiety/Depression symptom groups yielded a similar result with
the high anxiety/depression group demonstrating more impairment (t=2.05; p=0.05). No significant differences
were noted in the remaining symptom group analyses (see Figure 2).
Post-Hoc Analyses: Given that the Disorganized and Anxiety/Depression symptom analyses yielded similar results
as the Paranoid analyses, multiple regression equations were constructed to evaluate the contributions of these
symptoms to the paranoid group differences. These equations included the paranoid grouping variable (used in the
analyses above) and raw BPRS factor scores for the respective symptoms (disorganized and anxiety/depression) to
predict the FA misattribution scores of interest (anger, disgust and threat). Results are shown in Table 1. Analyses
indicated that anxiety/depression factor scores did not contribute to the observed differences between paranoid
symptom groups. However, analyses indicated that disorganization contributed to differences between paranoid
symptom groups for FA threat misattribution scores, but not for FA disgust or FA anger misattribution scores.
Figure 2

Figure 1
Affect Misattributions for Paranoid Symptom Groups

Data from 72 participants with a schizophrenia spectrum disorder involved in a large study of a cognitive rehabilitation
intervention (Spaulding et al. 1999) were used for the current study. All measures were collected as part of the preintervention assessment.


Low Symptom Group

low symptom group


High Symptom Group





high symptom group



One explanation of these findings is based on current social cognitive theories of paranoid
symptoms. Specifically, the pattern of performance of the high paranoid symptom group may have
been the result of social cognitive biases. According to these theories, in order to maintain selfesteem and protect against depressive affect, this group may have demonstrated a biased perception
of social stimuli as more interpersonally threatening; thus, the finding of greater misattributions of
anger and disgust for this symptom group. The poorer affect recognition performance of the high
paranoid symptom group can also be understood in the context of social cognitive theories of
paranoia. Specifically, persons with paranoid symptoms have been found to show impairments on
Theory of Mind (ToM) tasks that require the ability to infer others intentions (Corcoran 2001) and it
has been argued that these ToM deficits may be implicated in paranoid symptoms (Bentall and
Kinderman 1998; Frith 1994). Specifically, Frith (1994) has suggested that persons with paranoid
delusions are able to infer their own intentions but are unable to infer the mental states of others, thus
leading to false inferences that others are concealing their intentions. Likewise, the inability to
decode affect may lead persons to assume negative affect and intentions in others, based on
preexisting paranoid social cognitive biases. In fact, it has been argued that this is an adaptive
response to cognitive deficits, as it is safer to assume harmful intent than benevolence (Neufield
1991 p. 252).
The former explanation however, while consistent with current social cognitive theories of paranoid
symptoms, does not fully explain the data. Specifically, both the anxiety/depression and disorganized
symptom analyses yielded a similar pattern of misattributions, although not as robust. An alternative,
and perhaps more parsimonious explanation, is that the observed patterns were a result of overall
poor performance on the affect recognition task. That is, when making errors on the FA task
participants are more likely to misattribute affect as anger, disgust etc. Research in the area of affect
recognition has identified patterns of misattributions on affect recognition tasks. Studies have found
that disgust and anger tend to be misidentified for each other (Phillips et al. 1999) as do fear
and surprise (Ekman and Friesen 1976). Furthermore, persons with schizophrenia have been found
to experience greater difficulty in identifying negative affect (Bell et al. 1997). Therefore, it is
possible that these findings reflect both a greater difficulty in recognizing negative affect (i.e., anger
and sadness) as well as a natural tendency to confuse affect types such as anger and disgust.


To date, the current study represents one of the few studies to systematically evaluate
symptomatology, facial affect recognition performance and the types of misattributions made in a
sample of persons with schizophrenia spectrum disorders. The results suggest some evidence for a
social cognitive bias in paranoid symptoms. However, some caveats about the results are warranted.
The current analyses were unable to tease apart the types of misattributions made on an item by item
basis. Thus it is unclear as to whether the pattern of performance represents a bias or simply the
expected pattern of errors. Ultimately, the test of the hypothesis that persons with paranoid
symptoms are more likely to misperceive affect as interpersonally threatening will depend on an item
* significant difference
by item analysis. Future studies should evaluate the number of misattributions of interpersonally
threatening affect for specific items on the FA measure. For example, counting the number of times a
person misperceives neutral affect items as disgust or anger may represent a more accurate
Split0.3872.19 (0.03)ANXIETY/DEPRESSION
of response bias.Raw
bias could
also beFactor
facial affect Raw FactorScoreANGER0.082.47
Table 1: Multiple Regression Analysis of Symptom Contributions to FA MisattributionsPredictorCriterionR F (p)Betat(p)PARANOID Factormeasure
attentional bias paradigms used in other areas of experimental psychopathology (i.e., Mogg and
Bradley 1998).



Affect Recognition: Participants completed the Ekman and Friesen (1976) measure of facial affect recognition (FA). This
measure consists of 110 slides depicting an actor displaying one of seven basic emotions (happy, sad, fear, anger, surprise,
disgust and neutral). Each slide was displayed for eight seconds. Using a multiple-choice format, the participant was then
instructed to select which emotion was depicted. Based on FA errors, misattributions for each affect type were tabulated.
This yielded seven different affect misattribution scores. An additional score was calculated based on the sum of disgust and
anger misattribution scores to represent a general interpersonal threat misattribution score (threat). To measure affect
recognition performance, the overall percentage of correct responses on the FA was used.
To evaluate psychiatric symptoms the Extended Brief Psychiatric Rating Scale (BPRS) (Lukoff et al. 1986) was used.
Data Cleaning: The FA data was systematically screened for outliers. First, FA total percent correct was evaluated. Based on
the outlier analysis using Tukeys hinges two cases were excluded based on their extremely poor performance (3 standard
deviations below the mean) resulting in a sample of 70 participants used in the analyses. Three other cases were identified as
outliers, but less extreme, and were included in the analysis after a windsorizing procedure. Second, the misattribution scores
were screened for outliers using Tukeys hinges and identified outliers were included in the analysis after a windsorizing
Statistical Analyses: BPRS scores were factor analyzed using a varimax rotation procedure.. and six factors were
identified (paranoid, anxiety/depression, hallucination/delusion, disorganized, blunted affect, and agitation/elation). To
evaluate differences in the affect recognition variables and symptomatology the sample was divided into different symptom
groupings based on a median split of each BPRS factor scores. In each analysis the sample was divided into a high and low
symptom group based on a median split of the given factor score. Differences between each of these groups were analyzed
with t-tests on the following variables: total FAR percent correct and the eight misattribution scores. Given the fact that this
was an exploratory study alpha correction procedures were not used. Post hoc regression analyses were also conducted on
the variables of interest.

Affect Recognition Performance

Results indicate some support for the initial hypothesis of the study. Specifically, as predicted, when
comparing paranoid symptom groups the high symptom group made more misattributions consistent
with interpersonal threat (disgust and anger), as well as more misattributions of sadness. These
findings remained significant even after including other symptom variables in multiple regression
analyses. Contrary to the hypothesis, the high paranoid symptom group demonstrated poorer affect
recognition in comparison to the low paranoid symptom group. It should be noted that previous
research (Lewis and Garver 1995) on paranoia and affect recognition has used the paranoid
schizophrenia subtype as a categorical variable, whereas the current study utilized a specific measure
of paranoid symptoms. Furthermore, the present analysis compared high and low paranoid symptom
groups to each other as opposed to other symptom groups.

Percentage Correct

Frequency of Misattributions













Affect Misattribution Type

significant difference

Symptom Group


Some final comments are warranted about the findings regarding performance on the affect
recognition task. Significantly poorer affect recognition recognition was identified in the high
paranoid symptom grouping analysis and the high anxiety/depression grouping analysis. It is
interesting to note that these were the only two symptom groups where increased symptoms were
related to poorer affect recognition. These findings seem to suggest an increase in paranoia or
anxiety/depression symptoms in persons with schizophrenia may impair social cognitive abilities
whereas increases in other symptoms does not seem to have an effect. This finding warrants further

Recently Viewed Presentations

  • HR Lean (US & Canada) Project Owner: Leslie

    HR Lean (US & Canada) Project Owner: Leslie

    Project Owner: Leslie Thompson Project Sponsor: Jeff Minyard Project Team: Ana Abina, Monica Ek, Martin Morales FY11 & FY12 HR Lean - Agenda Overview Lean Office Strategy (HR, IT, & Finance) Overview Project Charter Key Milestones Regional HR Lean List...
  • International House Company Training Making Meaning Clear

    International House Company Training Making Meaning Clear

    Use a monolingual dictionary to define it. Translate it, using a bilingual dictionary. Translate it, using the T as a dictionary/what it . doesn't . mean (false friend). Have them guess what category of word it is and make an...
  • Basic Logarithms

    Basic Logarithms

    Solving equations involving exponents and logarithms Let's review some terms. When we write log5 125 5 is called the base 125 is called the argument Logarithmic form of 52 = 25 is log525 = 2 Remember ln and log ln...
  • Psychological Disorders The same behavior may be deemed

    Psychological Disorders The same behavior may be deemed

    4) Maladaptive behavior - Behavior is maladaptive when it causes personal distress, is self-defeating, is associated with significant health, social, or occupational problems, or the behavior does not allow the person to fit into the situational context.
  • www.bluepig.plus.com


    Presentation to Grammar School Chairs of Governors and Headteachers 22nd June 2006 Email copies [email protected] Tonight's Agenda How are grammar schools going to deal with the 1


    WAR IS THE "ENGINE" OF SCIENCE AND TECHNOLOGY: "War is the father or all things" (Heraclites) "War is the strife for power" (Plato, Aristotle) "War is the natural state of mankind" (Thomas Hobbes) ALL MODERN AREAS OF SCIENCE AND TECHNOLOGIES...
  • Presentazione di PowerPoint

    Presentazione di PowerPoint

    Courneya KS et al. Exercise interventions during cancer treatment: biopsychosocial outcomes. ... Tudor-Locke C et al. ... US food industry ensures that consumers are not told to eat less. Elliott A et al. BMJ.
  • For CHW and Mother training - World Vision International

    For CHW and Mother training - World Vision International

    Media: training via radio or video. Monitoring and evaluation. Evaluating the effectiveness of the training and the screening capability of mothers: Random tests in homes several weeks after the initial training by a supervisor + refresher session if 25% of...