Measuring and modeling facial affect

In recent years, researchers in computer science and human-computer interaction have become increasingly interested in characterizing perception of facial affect. Ironically, this applied interest comes at a time when the classic findings on perception of human facial affect are being challenged in the psychological research literature, largely on methodological grounds. This paper first describes two experiments that empirically address Russell’s methodological criticisms of the classic work on measuring “basic emotions,” as well as his alternative approach toward modeling “facial affect space.” Finally, a user study on affect in a prototype model of a robot face is reported; these results are compared with the human findings from Experiment 1. This work provides new data on measuring facial affect, while also demonstrating how basic and more applied research can mutually inform one another.     

Factors affecting the accuracy of facial affect recognition.

The effect of subjects' locus of control of reinforcement and self-monitoring orientations on their ability to identify the emotional meaning of facial expressions was evaluated under competitive and noncompetitive instructions. On the basis of social learning principles, it was predicted that internally controlled, high-self-monitoring American college students would be the most accurate; the results confirmed the prediction. The impact of locus of control and self-monitoring orientations was greatest on male subjects who had received competitive instructions. In addition, female subjects were more accurate than male subjects, and positive emotions were more often correctly identified than negative ones.     

Field-dependence/independence: ability in relation to mobility-fixity

Previous authors have pointed out that, while the classic Rod-and-Frame Test (RFT) yields a measure of ability to perform field-independently, it is unsuitable for identifying stylistic preference, such as mobility-fixity within the field-dependence/independence dimension. To relate mobility-fixity and ability aspects of field-dependence/independence, we compared data obtained from two versions of the test: one with a "free" instruction, on which the subjects were invited to adjust the rod to any position(s) they preferred on each of 10 trials and a process-oriented version (RFT-P), involving 20 trials with the standard instruction but with a constant rod-and-frame configuration, making possible analysis of the process of adaptation. Mobility versus fixity was defined in terms of more or less frequent changes of rod positions and choice of a high or a low proportion of nongeometrical positions in the RFT-Free. 38 female and 23 male university students completed first the RFT-Free and thereafter the RFT-Process. Women were more field-dependent than men throughout the RFT-Process trials; there were no differences between the sexes over trials or in choice of positions in the RFT-Free. Compared with field-dependent subjects, field-independent subjects more often changed rod positions in the RFT-Free and preferred nongeometrical positions somewhat more. Among field-independent subjects, those with consistently low deviations in RFT-Process more often preferred a variety of nongeometrical positions in RFT-Free than did those with gradually increasing deviations. We conclude that field-independence is associated with mobility and field-dependence with fixity, as defined by the RFT-Free variables, and that a prerequisite for high mobility is a relative stability of field-independence as reflected on the RFT-Process. Theoretical implications of this conclusion are discussed.     

Positive facial affect looks healthy

A healthy appearance is linked to important behavioural outcomes. Here we investigated whether positive facial affect is a cue for perceived health. In study one, two groups of participants rated the perceived health or perceived happiness of a large set of faces with neutral expressions. Perceived happiness predicted perceived health, as did anthropometric measures of expression. In a second experimental study, we collected ratings of perceived health for a wide age range of target faces with either neutral or smiling expressions. Smiling faces were rated as being much healthier looking than neutral faces, confirming that facial expression plays a role in the perception of health. A third study investigating attractiveness as a possible mediator found that expression still had a significant direct effect on perceived health, after accounting for attractiveness. Together, these studies systematically show that facial affect plays a critical role in shaping our perceptions of health in others.     

Recognition of facial affect in depression

25 depressed patients recognized a sad face with more errors than 25 normal persons and labeled other expressions as sadness when affective content was not recognized. Correct recognitions for 6 affects were related to the portion of the face depicted. Comparisons of responses of 25 patients diagnosed as having anxiety neuroses showed differences in responses from the depressed patients and normal persons.     

Valence focus and the perception of facial affect

Individuals differ in the extent to which they emphasize feelings of pleasure or displeasure in their verbal reports of emotional experience, an individual difference termed valence focus (VF). This multimethod study indicates that VF is linked to heightened efficiency in perceptual processing of affective stimuli. Individuals higher in VF (i.e., who emphasized feelings of pleasure/displeasure in reports of emotional experiences) were more sensitive to changes in negative facial expressions than individuals lower in VF. The effect was not accounted for by current affective state or other personality characteristics. Implications for the validity of self-reported experienced emotion are discussed.     

Eye movement during facial affect recognition by patients with schizophrenia, using Japanese pictures of facial affect

A possible relationship between recognition of facial affect and aberrant eye movement was examined in patients with schizophrenia. A Japanese version of standard pictures of facial affect was prepared. These pictures of basic emotions (surprise, anger, happiness, disgust, fear, sadness) were shown to 19 schizophrenic patients and 20 healthy controls who identified emotions while their eye movements were measured. The proportion of correct identifications of 'disgust' was significantly lower for schizophrenic patients, their eye fixation time was significantly longer for all pictures of facial affect, and their eye movement speed was slower for some facial affects (surprise, fear, and sadness). One index, eye fixation time for "happiness," showed a significant difference between the high- and low-dosage antipsychotic drug groups. Some expected facial affect recognition disorder was seen in schizophrenic patients responding to the Japanese version of affect pictures, but there was no correlation between facial affect recognition disorder and aberrant eye movement.     

Perceptual skill in decoding facial affect

The perceptual skill in identifying the briefly exposed photographs of facial affect was examined with 75 men and 75 women who were asked to judge six cross-culturally recognizable slides, each expressing one of the six emotions exposed at three different time intervals. Each subject was required to give six judgments and the number of correct responses was considered his score. Analysis by a 3 X 2 factorial design indicated the longer the exposure of the photograph the better the recognition. Although there was no over-all sex difference, analysis by emotions confirmed significantly better accuracy by the women judging 'sadness' than by men, and men were more accurate for 'anger' than were women.     

Processing of faces by patients with unilateral hemisphere lesions. I. Dissociation between judgments of facial affect and facial identity

In this study, right-hemisphere-damaged (RHD) subjects performed significantly worse than LHD and NHD controls across a series of seven facial identity and facial affect tasks. Even when the patient groups were statistically equated on a measure of visuoperceptual ability, the RHD group remained impaired on three emotional tasks--naming, picking, and discriminating emotional faces. These findings suggest that the defects shown by RHD patients on facial affect tasks cannot be solely attributed to defects in visuoperceptual processing and that the right-hemisphere superiority for processing facial affect exists above and beyond its superiority for processing facial identity.     

Recognition of facial affect in psychopathic offenders

The authors examined the reliability of facial affect processing deficits found in psychopathic individuals (R. Blair et al., 2004) and whether they could be modified by attentional set. One hundred eleven offenders, classified using the Psychopathy Checklist--Revised (R. Hare, 2003) and Welsh Anxiety Scale (G. Welsh, 1956), performed a facial affect recognition task under 2 conditions. On the basis of research linking psychopathy, amygdala dysfunction, and deficits in facial affect recognition, the authors predicted that psychopathic offenders would display performance deficits when required to identify the emotional expression of particular faces. In addition, given evidence linking the affective processing deficits in psychopathy to focus of attention, the authors predicted that any deficits in facial affect processing would disappear when participants could anticipate which affective cues would be relevant on a given trial. Contrary to expectation, psychopathic offenders performed as well as controls in both conditions. The authors conclude that the conditions that reveal affective deficits in psychopathic individuals require further specification.     

Field-dependence, sensation-seeking, and preference of paintings

College students (122 males and 88 females) completed the Group Embedded Figures Test measure of field-dependence, and Sensation-Seeking Scale, and performed preference ratings on 40 slides of paintings. Factor analysis revealed six hypothesized and one unhypothesized preference dimensions that underlay preference ratings of paintings. It was hypothesized that the cognitive structuring ability associated with field-dependence would result in certain relationships between field-dependence and preference for paintings. Further, it was hypothesized that preference for novel, complex, and dynamic experiences, associated with sensation-seeking, would result in certain relationships between sensation-seeking and preference for paintings. For the full sample, three of five hypothesized relationships between field-dependence and preference for paintings were obtained, while three of four hypothesized relationships were found between sensation-seeking and preference for paintings.     

Field-dependence,extraversion, Eysenck and autarky

The research literature was examined with regard to relationships between Field-dependence and Extraversion. Eysenck's contention that the two variables are significantly related to an extent justifying the use of causal hypotheses based on extraversion theory to account for field-dependent behavior was not supported.     

The effects of context on facial affect recognition

In a sample of 325 college students, we examined how context influences judgments of facial expressions of emotion, using a newly developed facial affect recognition task in which emotional faces are superimposed upon emotional and neutral contexts. This research used a larger sample size than previous studies, included more emotions, varied the intensity level of the expressed emotion to avoid potential ceiling effects from very easy recognition, did not explicitly direct attention to the context, and aimed to understand how recognition is influenced by non-facial information, both situationally-relevant and situationally-irrelevant. Both accuracy and RT varied as a function of context. For all facial expressions of emotion other than happiness, accuracy increased when the emotion of the face and context matched, and decreased when they mismatched. For all emotions, participants responded faster when the emotion of the face and image matched and slower when they mismatched. Results suggest that the judgment of the facial expression is itself influenced by the contextual information instead of both being judged independently and then combined. Additionally, the results have implications for developing models of facial affect recognition and indicate that there are factors other than the face that can influence facial affect recognition judgments.