The Effect of Emotional Expression and Degree of Rotation on Recognition of Facial Stimuli - Assignment Example

? The Effect of Emotional Expression and Degree of Rotation on Recognition of Facial Stimuli The study attempted to verify the effectsof Rotation of image and Emotional Expression on Response Time taken by individuals when attempting to recognize face stimuli. It was hypothesized that negative emotions like Anger would be recognized faster than Happy or Neutral emotions; and that images that were less rotated from initial presentation would be recognized faster. Although the statistical analysis yielded significant differences; only the expectation about the association of degree of rotation with response time was supported. These results underline the value of exploring the contribution of other attentional factors in face recognition in general and the role of emotional expression in particular Title The Effect of Emotional Expression and Degree of Rotation on Recognition of Facial Stimuli Introduction Humans are social animals; meaning that we live and flourish in a society that consists of other human beings and have interdependent roles. Recognition of faces and expressions is thus essential to human interaction so that it becomes possible to detect the nature of information held by the other individual(Ohman, Lundqvist, & Esteves, 2001). Recognizing emotion in others plays many roles in human survival. It allows the individual to gauge the possibility of risk and benefit by identifying the emotion observed on the face of the other (Maratos, Mogg, Bradley, 2008). It also allows the individual to provide the appropriate response and thus allows for the development of an interpersonal empathetic bond that enables and enhances social interaction. Drawing from this hypothesis, research has been conducted to identify the differences in the way humans’ detect different emotions. Research shows that the human brain is able to detect subtle differences in expression; and the ability to read emotion efficiently and rapidly is associated with the extent to which the individual is familiar with the presented face (Valentine, Bruce & Baddeley, 1987). People are likely to detect an emotion quicker and more accurately when they view a face they know as compared to a face they are unfamiliar with. This shows that human face recognition processes store not only facial feature information about familiar persons; but also information about how a particular face depicts emotion (Valentine, Bruce & Baddeley, 1987). But we do not always interact only with individuals who are known to us; and thus, it becomes important to understand the way in which emotion is read in unknown faces as well as faces that are familiar. When the face presented in an unknown one, people are more likely to detect a negative emotion faster and more accurately (Calvo, Avero, & Lundqvist, 2006) as compared to a positive one (Ohman, Lundqvist, & Esteves, 2001). The threat associated with negative emotions like fear and anger has been proposed as a reason for this phenomenon. This hypothesis has been verified by a number of studies in the recognition of facial expressions. There exists a bias towards negative facial expressions whether they are presented alone (Ohman, Lundqvist, & Esteves, 2001) or in a group. People are more likely to identify an angry or fearful face in a crowd as compared to one depicting a non-threatening emotion like happiness or sadness or a neutral expression. People also tend to pay more attention to faces that depict an emotion associated with threat (Maratos, Mogg, Bradley, 2008)as compared to faces that do not(Ohman, Lundqvist, & Esteves, 2001). Thus, they pay more attention to a group of faces that depict anger or fear as compared to a group that depicts happiness or sadness. They also take longer to verify the presence or absence of a face that differs from the group when the predominant emotion depicted is associated with a threat condition (Fox, Lester, Russo, Bowles, Pichler, & Dutton, 2000). Research has also found that people are more accurate in detecting anger and fear as compared to other emotions in a face. Given the spectrum of situations that show this bias for emotions associated with threat, it is evident that this is an important ability for human survival. Research that attempted to verify if all humans show the same ability shows that the ability to detect facial expressions associated with fear and anger factor and more accurately than other exists across gender, race and age. Children are as adept at indentifying fear and anger and adults; and do so faster and more accurately than they do happy sad or neutral faces (LoBue, 2008). This ability is seen in child as young as a few months; and five year old children are able to identify threat related emotions as accurately and rapidly as adults when compared to positive or neutral expressions (LoBue, 2008). The amount of information that an individual is able to extract from a facial expression is affected by a number of factors. A face that fully in view is likely to provide more information and more accurate information about emotions. On the other hand, a partial view of a face is likely to take more time to interpret and is sometimes likely to give inaccurate information since the entire face is not visible. A number of studies have tried to understand if the manner in which a stimulus is presented affects the accuracy and speed of recognition (e.g. Valentine, Bruce & Baddeley, 1987). Early research in this field used alpha-numeric stimuli; and showed that the extent to which a stimulus was inverted or rotated from its original presentation affected both the accuracy and speed of recognition (Cooper, 1975). Knowledge about the manner in which a test stimuli would be presented and time to prepare for it nullified this effect (Cooper, 1975); but showed a new effect. The time taken to prepare for the test stimuli was found to be associated with the degree of difference from the original stimulus (Cooper, 1975; Shepard, & Cooper, 1982; Shepard, & Metzler, 1971). Thus, the more the difference, the more time the person takes to either identify the stimulus, or to prepare for it. Studies conducted with facial expressions show that this effect is seen even with stimuli that represent human faces. Faces that are seen as they were previously presented are recognized fastest; while faces that are rotated or inverted take more time relative to the extent of rotation (Valentine & Bruce, 1988). This study attempts to verify if the effect of emotion and rotation of the speed of accurate recognition of faces. As seen in previous research, facial expression plays an important role in the speed of recognition, as does the extent to which the face is rotated away from its initial presentation (Valentine & Bruce, 1988). When we combine this understanding with the fact that some expressions are recognized faster than others; it leads to the question of whether the type of emotion depicted affects the speed of facial recognition as a function of the extent to which the face is rotated away from its original presentation (Calvo, Avero, & Lundqvist, 2006). Thus, it is possible to hypothesize that given the degree of rotation, threat related emotions would be associated with faster recognition than other emotions. On the other hand, given the emotion depicted; speed of face recognition is affected by the extent of rotation such that the more a face is rotated, the more time it would take to recognize it (Valentine & Bruce, 1988). Also, a face depicting a threat – related emotion is most likely to be recognized best when the rotation is the minimum. Thus, for this study, we may hypothesize that: Faces depicting Anger will be recognized faster than those depicting Happiness or Neutral Emotion. Time taken for Face recognition will increase as a function of the extent of rotation of the test stimuli. Faces depicting anger will be recognized fastest when the test stimuli are rotated least. Method Variables This study attempted to ascertain the effect of rotation and emotional expression of a target face stimulus on the time required to recognize it. The participants were required to identify whether a test stimulus was similar or different as compared to the stimulus that preceded it, the time taken for successful trails were recorded. The independent variables in this study are: 1. Rotation of face stimulus: the effect of rotation was tested across four levels ranging from no rotation to a 60o rotation such that each level differed by 20o. the levels used were a. 0o rotation b. 20o rotation c. 40o rotation d. 60o rotation 2. Emotional Expression of Face: The effect of different Emotional Expressions was tested across three levels, one each related to positive, negative and neutral emotions. the three levels were: a. Anger b. Happy c. Neutral Thus, there were a total of 12 conditions that were derived through the combinations of these factors. The dependent variable being studied was the Reaction Time taken by the participant to respond to the test stimulus. Reaction time was measured in milliseconds, and data was collected on successful trails. Participants The 151 participants chosen to be a part of the study were recruited from the student population. Of the 151 participants, 112 were male and 39 were female. They ranged in age from 19 to 55 years with a mean age of 31.13 years (S.D. = 7.9). The same participants were exposed to all 12 conditions in order to control for interpersonal factors like response speed, age and familiarity with computerised testing. Controls like handedness and Knowledge of English were used to reduce the effects of confounding. Handedness was controlled so that all participants would have the same posture and comfort on using the designated response keys. The study was conducted with right – handed persons. Native speakers were chosen to ensure that instructions were properly understood and to reduce confounding from misinterpretation. Design The study used a factorial design with two independent variables. The first I.V. had 4 levels and the second had 3. Total number of conditions tested was 12. The study chose to use a fully repeated measures design to control for inter-personal factors. Since both the main effects as well as the interaction between factors were deemed important; the data collected was analysed using a 2 – Way ANOVA. Materials The materials used were model faces presented using a computer software called E-Prime. Model faces were considered as an ideal choice since they have been shown to provide more sensitivity to emotional expressions as they may be mathematically developed (Waters, 1987). They also allow for more control over the extent of presentation of an emotion while being more ecologically valid as compared to drawings. The model faces presented were developed so that there were three sets of stimuli; one each representing anger, happy and Neutral. Within each set, there was the original stimulus of a face that was fully visible which was shown as the presentation stimulus; and four variations of it such that one variation each differed from the original by being rotated through 0o, 20o, 40o, and 60o. Procedure The participants were made comfortable in the laboratory, and were guided through the following experiment by giving them the following instructions: “Each trial begins with a cross in the middle of the screen Please look directly at the cross. The cross will be replaced by a face, looking straight at you. This face will disappear, and be replaced by another. Your task is to compare the EXPRESSIONS of the two faces. If the two EXPRESSIONS were the SAME, press the ‘S’ button If the two EXPRESSIONS were DIFFERENT, press the ‘D’ button Please respond using the first and second fingers of your dominant hand Please respond as QUICKLY and as ACCURATELY as possible Please Press Either Response Key to Start the Practice Trials.” Each trail started with the presentation of a cross which was presented for 1250 milliseconds followed by the target face which was displayed for 650 milliseconds. After a delay of 650 milliseconds, the reference face or test stimuli was presented for 650 milliseconds. Response time was measured from the presentation of the reference face. Each participant was given 7 practice trails. There was an interval of 1-1.5 seconds between trails; which was set to vary in this range randomly between trails. Each participant was exposed to a total of 96 trails, of which 48 had the same reference and target expressions; while 48 had different ones. Within each of these two groups, there were an equal number of trails (4) for each of the 12 conditions described. For this study, only data on the ‘Same face’ trails was used. Response time was collected for all successful trails (where the participant gave the right response) and aggregated for each condition for the individual. Results This study collected data on 151 participants, each of who was exposed to each of the 12 research conditions. Response time was collected in milliseconds; and was aggregated across successful trails for each condition for each subject. The mean response time and the associated Standard Deviations are given in Table 1. Degree of Rotation Emotional Expressions 0o 20o 40o 60o Totals Anger 506.5693 (73.128) 502.9732 (67.178) 519.4931 (67.112) 508.8762 (70.917) 509.4779 Happy 444.7976 (64.485) 493.5055 (59.866) 513.6487 (59.382) 532.4011 (59.542) 496.0882 Neutral 445.3637 (77.8118) 475.3409 (69.873) 496.2377 (76.735) 519.3162 (58.351) 484.0646 Totals 465.5769 490.6065 509.7932 520.1978 496.5436 Table 1. Means and Standard Deviations (in parenthesis) for different conditions A Two – Way ANOVA for fully Repeated Measures was calculated to ascertain the significance of the main effects and the interaction effect. All three effects were found to be significant at the 0.001 level of significance. The main effect for Rotation gave an F value of 84.64 [F (3, 1350) = 84.64; p < 0.001]; while the F value for the main effect of Emotional Expression was 10.98 [F (2,450) = 10.98; p < 0.001]. The F value for the interaction of the two variables was 17.89 [F (6, 1350) = 17.89; p < 0.001]. Since both main effects were found to be significant; a post hoc analysis was conducted to ascertain which of the levels for each factor were significantly different from each other. The Tukey’s HSD was calculated for this purpose. Post – Hoc analysis showed that for the Independent variable of Emotional Expression, all three levels differed significantly from each other at the 0.01 level of significance. Participants responded the stimuli with a Neutral expression the fastest, followed by stimuli with Happy Expression. The participants took longest to respond to stimuli with an Angry Expression. For the independent variable of Rotation, all differences were found to be significant except the difference between 40o and 60o rotations. Thus, participants took the least time to respond to 0o rotations, as compared to all other levels. Time taken to respond to 20o rotations also differed significantly from 40o and 60o rotations; but there was no significant difference between the 40o and 60o rotations. The interaction between the factors has been represented in Chart 1. Discussion Analysis of the collected data showed that all the three F values calculated for the Two – Way Repeated Measures ANOVA were statistically significant. The main effect for the factor of Rotation showed that 0o rotation conditions differed significantly from all others, and 20o rotation conditions also differed from the 40o and 60o rotation conditions while there was no difference between the 40o and 60o conditions. This data is consistent with previous research (Shepard, & Cooper, 1982); and allows the acceptance of the hypothesis that “Time taken for Face recognition will increase as a function of the extent of rotation of the test stimuli.” These results are consistent with research done by Valentine & Bruce (1988) and Valentine, Bruce & Baddeley (1987) which show that as the degree of rotation of a face increases, the time taken to recognize it successfully also increases. Ideally each of the conditions should have been different from each other; but the lack of significant difference in the 40o and 60o conditions may be due to the fact that participants start finding stimuli difficult beyond the 20o rotation; and thus pay equal attention and time to these stimuli. The main effect for the factor of Emotional expression was also significant, with participants responding to Neutral expressions fastest and Anger expressions the slowest. All three conditions were significantly different from each other. The hypothesis being tested for this main effect was “Faces depicting Anger will be recognized faster than those depicting Happiness or Neutral Emotion.” this hypothesis was not supported by the data; and was thus, rejected. As the differences are significant, it is also not possible to accept the null hypothesis for this main effect that states that there are no differences in the three conditions. Thus, the collected data is inconsistent with previous research (Ohman, Lundqvist, & Esteves, 2001; Calvo, Avero, & Lundqvist, 2006), and it is important to examine possible reasons for this occurrence. Research has shown that people take longer to verify the lack on non-threatening emotional information in a stimulus that provides predominantly threatening information (Fox, Lester, Russo, Bowles, Pichler, & Dutton, 2000). People also pay more attention to and stare longer at faces that depict Anger and Fear (Maratos, Mogg, Bradley, 2008); possibly to verify the threat to self. It is possible that since the stimuli presented were novel in that they were model faces; the participants paid more attention to them when they depicted Anger (maximum threat) and least attention when they were Neutral (least valuable information). From an evolutionary perspective, People are less concerned about being wrong about a natural experience than one that contains potential threat (Maratos, Mogg, Bradley, 2008). Also, a situation that offers the promise of a possible reward (Happy face) demands more attention as compared to a neutral one. Thus, the Response Time of participants may be affected by the amount of attention they paid the stimuli, and not just the speed with which the stimuli were identified. The interaction effect was also significant; showing that the factors do affect each other. The hypothesis “Faces depicting anger will be recognized fastest when the test stimuli is rotated least” was not supported though; as the means show that the lowest mean was for Happy Faces that were rotated through 0o. This result is affected by the main effect for Emotional Expression; but the graph doe show that for all happy and Neutral faces, the expected trend is seen – i.e. – the least response time is for the 0o condition and the most for the 60o condition. For each rotation level, there are distinct differences in the times taken across each of the three conditions for Emotional expression. Since it is possible that the data on emotional expression has been confounded by the attention span it commands or other variables; the trend is the opposite of what is expected. Limitations and Recommendations From the data it is evident that the amount of attention that a stimulus commands must be taken into consideration. These effects may be affected by training. The stimuli were presented in a randomized order with no caveats. It is important to verify if on a chance factor, Angry faces were on a whole presented earlier for many participants and neutral ones last. This study also does not pay attention to data on mistakes. It is possible that participants were more careful – and thus took more time – for Angry faces than others. Thus, data on errors would allow us to verify if the participants did actually have lesser errors for Angry face stimuli as compared to other situations. The participants selected consist of a majority of males; and it is not possible to ascertain if gender has played a role in the trends seen in the data. Balancing for gender would allow the researcher to eliminate gender as a confounding variable. References Calvo, M. G., Avero, P., & Lundqvist, D. (2006). Facilitated detection of angry faces: Initial orienting and processing efficiency. Cognition and Emotion, 20, 785-811. Cooper, L. A. (1975). Mental rotation of random two-dimensional shapes. Cognitive Psychology, 7, 20-43. Fox, E., Lester, V., Russo, R., Bowles, R.J., Pichler, A., & Dutton, K. (2000). Facial Expressions of Emotion: Are Angry Faces Detected More Efficiently? Cognition and Emotion, 14, 61–92. Fox, E., Russo, R., & Dutton, K. (2002). Attentional bias for threat: Evidence for delayed disengagement from emotional faces. Cognition and Emotion, 16, 355–379. LoBue, V. (2008). More than just another face in the crowd: superior detection of threatening facial expressions in children and adults. Developmental Science, 12, 305–313. Maratos, F.A., Mogg, K., & Bradley, B.P. (2008). Identification of angry faces in the attentional blink. Cognition and Emotion, 22, 1340-1352. Ohman, A., Lundqvist, D., & Esteves, F (2001). The Face in the Crowd Revisited: A Threat Advantage With Schematic Stimuli. Journal of Personality and Social Psychology, 80, 381-396. Shepard, R. N., & Cooper, L. A. (1982). Mental images and their transformations. Cambridge, MA: MIT Press. Shepard, R. N., & Metzler, J. (1971). Mental rotation of three dimensional objects. Science, 171,701-703. Valentine, T. & Bruce, V. (1988). Mental rotation of faces. Memory & Cognition, 16 (6), 556-566 Valentine, T., Bruce, V. & Baddeley A. (1987). The Basis of the 3/4 View Advantage in Face Recognition. Applied Cognitive Psychology, 1, 109-120 Waters, K. (1987). A muscle model for animating three-dimensional facial expression. Computer graphics, 21, 17 – 24 Appendix Stimuli Presented Angry face Happy face Neutral face Read More