The Effects of Individual Differences and Self-Consciousness on Nonverbal Decoding Accuracy

Nonverbal decoding refers to the act of recognizing and interpreting the meaning of other people’s nonverbal cues. Decoding skills vary depending on many factors such as personality and environment (Knapp & Hall, 2009). The present study focused on six individual difference measures (the EPI, IOS, PSI, SCS-R, 20-item Shyness Scale, and TSIS) and an experimentally manipulated variable of self-consciousness, to determine their relationship with nonverbal decoding accuracy on two tasks: the METT and the VNDT. The results indicated that four individual difference measures—extraversion, sociability, shyness, and moving toward others— interacted at significant levels with the self-consciousness variable. These predictor variables were found to have a greater impact on performance on the METT than on the VNDT. It is suggested that future research utilizes real interactions as the basis of their decoding task. INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 4 The Effects of Individual Differences and Self-Consciousness on Nonverbal Decoding Accuracy Nonverbal behavior is considered the "primary medium for the communication of affect" (Feldman, Philippot, & Custrini, 1991 p. 332; also see Argyle, 1969; Buck, 1984; Cacioppo, Martzbe, Petty, & Tassinary, 1988) and has been researched quite extensively and comprehensively. Successful nonverbal communication requires social skills, which include social competence (Feldman & Rimé, 1991). Highly socially skilled individuals can naturally express their emotion and state of mind through their body language and interpret others’ nonverbal messages accurately. The act of sending out nonverbal messages is called encoding; the act of noticing and interpreting the nonverbal cues of others is called decoding. Examples of nonverbal cues include but are not limited to: facial expressions, posture, gestures, touch, personal space, and tone of voice (Knapp & Hall, 2009). Accurately decoding other people’s nonverbal cues is crucial not only in social life and personal relationships but also in academic performance (Feeney, Noller, & Callan, 1994; Halberstadt & Hall, 1980; Izar et al., 2001; Nowicki & Duke, 1992, 1994). Often times, good encoders are also good decoders; however, the mechanism of encoding differs from the mechanism of decoding. The main concern of the present study is the structure of individuals’ decoding skills. The Effect of Personality on Decoding Accuracy Decades of research have indicated that the ability to decode nonverbal cues significantly varies among individuals. Such variability is influenced by many personal factors such as gender, age, upbringing, cultural background, personality, cognitive ability, and knowledge of nonverbal cues (Knapp & Hall, 2009). In particular, personality has been closely studied as a major cause of such variability. INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 5 Introversion and extraversion. Introversion is as a good example of how personality affects decoding ability. Compared to extroverts, introverts receive lower scores on decoding tasks (Akert & Panter, 1988), identify fewer types of nonverbal cues (Knapp & Hall, 2009), and pay less attention to details in the social domain (Akert & Panter, 1988; Eysenck & Eysenck, 1968). Researchers reason that, similar to shy individuals, introverts tend to avoid face-to-face social interactions, which leads to fewer opportunities to improve their skill at nonverbal decoding (Akert & Panter, 1988). Extraversion is found to positively correlate with visual-based decoding ability (Rosenthal, Hall, DiMatteo, Rogers, & Archer, 1979), audio-based decoding ability (Mill, 1984), and the combination of both visualand audio-based decoding ability (Akert & Panter, 1988). However, the results of research in this area are often contradictory. For instance, while extroversion has been found to positively correlate with decoding skills due to extroverts’ rich social experiences (Akert & Panter, 1988; Argyle & Lu, 1990a, 1990b), Cunningham (1977) suggests that extroversion is only related to encoding skills—and not to decoding skills. He reports that decoding skills are in fact positively correlated with neuroticism: Neurotic people are excessively anxious in social situations, which in turn motivates them to acquire a high level of nonverbal perceptivity (Cunningham, 1977). In addition, a number of studies have failed to find a significant correlation between extraversion and decoding ability (Ambady, Hallahan, & Rosenthal, 1995; Riggio & Friedman, 1982; Rosenthal et al., 1979). Shyness. Contradictory results have also been found for the relationship between shyness and decoding accuracy. Some research has found that shy people perform worse than not-shy people on nonverbal decoding tasks. It has been suggested that shy individuals tend to avoid engaging in social interaction and thus get less practice in nonverbal communication, with the INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 6 result that their abilities to encode and decode nonverbal cues are poorer than not-shy individuals (Schroeder, 1995; Schroeder & Ketrow, 1997). Moreover, shyness is also related to cognitive interference when processing social information (Cheek & Melchior, 1990; Schroeder, 1995). While many studies indicate that not-shy individuals outperform shy individuals on decoding tasks and other social information processing tasks (e.g., McClure & Nowicki, 2001), other studies have found no significant correlation between shyness and decoding accuracy (Akert, Cheek, Rutenberg, Ahern-Seronde, & Dautoff, 2010; Akert, Dautoff, Ahern-Seronde, & Cheek, 2009; Young & Brunet, 2011). One possible reason for these mixed results may be a misunderstanding of shyness; not all shy individuals have the same social motivations. Caspi, Elder, and Bem (1988) examined shyness through the perspective of two interpersonal styles, which have distinctive affiliation needs, moving toward others and moving away from others (Horney, 1945). Moving toward others is an attachment style characterized by overly dependent behavior, motivated by a high need for understating, emotional support, and positive exchanges with others. Moving away from others is an attachment style characterized by excessive concerns over independence and selfsufficiency, which results in compulsive detachment from others. Caspi, Elder, and Bem (1988) suggest that these interpersonal styles affect one’s personal relationships more strongly than does shyness. Moreover, the distinction between shyness and sociability is key to understanding the relationship between shyness and decoding accuracy. Cheek and Buss (1981) found that shyness and sociability correlate at only -.30, which suggests that low sociability does not mean shyness: assuming shy individuals desire less affiliation is misleading. In addition, Young and Brunet (2011) found a significant relationship between sociability—not shyness— and decoding INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 7 accuracy. The authors concluded that this discrepancy in performance between those of high and low sociability individuals was not caused by anxiety, as is the case with shyness; poorer decoding accuracy in low sociable individuals was linked to less experience in social interactions (Young & Brunet, 2011). Private and public self-consciousness. Self-consciousness and social anxiety, which are positively and highly correlated with shyness, are also found to affect decoding accuracy (Schroeder, 1995). Socially anxious individuals tend to misjudge emotions conveyed in others’ nonverbal behaviors (McClure & Nowicki, 2001), and a high level of self-consciousness is linked to inadequate and ineffective social skills (Christensen, 1982). Self-consciousness was well studied by Fenigstein (1975), who created the most widely used scale for selfconsciousness, the Self-Consciousness Scale (SCS; Fenigstein, Scheier, & Buss, 1975). It consists of 23 items measuring three dimensions of self-consciousness: private selfconsciousness, public self-consciousness, and social anxiety. Private self-consciousness refers to a propensity to pay attention to the details of one’s feelings, internal processes, covert parts of the self, privately held beliefs and values, and knowledge of self-aspects (Franzoi, Davis, & Young, 1985; Harrington & Loffredo, 2007; Scheier & Carver, 1985). This subscale moderately correlates with the public self-consciousness (r ≥ .23, p < .01) subscale but not with the social anxiety subscale (Fenigstein et al., 1975). Public self-consciousness refers to a tendency to worry about one’s public image and to self-analyze it (Franzoi, Davis, & Young, 1985; Harrington & Loffredo, 2007; Scheier & Carver, 1985). This subscale moderately correlates with the social anxiety subscale (r ≥ .20, p < .01; Fenigstein et al., 1975). Social anxiety is defined as the experience of discomfort when others are present (Fenigstein et al., 1975). INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 8 While the public self-consciousness subscale positively correlates with shyness, the private self-consciousness subscale does not (Tabata, 2009). Harrington and Loffredo (2007) found that private self-cosnciousness—and not public self-consciousness—is linked to selfreflectiveness (SR), internal self-awareness (ISR), and psychological well-being (PWB). They found that increased self-consciousness is linked to an elevation of SR and ISA, but its link to PWB depends on the level of SR and ISA. When SR levels are high, PWB levels go down; however, SR levels alone does not predict PWB. The authors found a negative correlation between SR levels and PWB levels only when ISA levels are high. When looking at the subscales of PWB, this complex relationship becomes clear. The positive relations with others (PRO) subscale in PWB significantly intereacts with both SR and ISR: raised self-consciousness negatively correlates with PRO levels, which means highly self-conscious individuals would have difficulty relating to others effectively. Increased private self-consciousness causes people to be self-occupied, leading them to regard others less, which in turn causes problems in social settings. Developing Measures for Nonverbal Decoding Accuracy Hall, Andrzejewski, and Yopchick (2009) conducted a meta-analysis on the relationship between interpersonal sensitivity (IS) and decoding accuracy. Interpersonal sensitivity is: the act of accurately judging or recalling others’ behaviors and engaging in interpersonally appropriate behaviors, and psychosocial characteristics of the perceiver, which include personality traits, social and emotional functioning, life experiences, values, attitudes, and self-concept (Hall et al., 2009). The authors found that individuals’ self-assessed nonverbal decoding skills positively predicted their measured IS. They report that higher IS is correlated with favorable personality traits, such as extraversion and empathy, indicating that decoding skills are closely related to INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 9 adaptive psychosocial functioning. With the Profile of Nonverbal Sensitivity (PONS; Rosenthal et al., 1979) as a decoding accuracy measure, Hall et al. (2009) found that IS correlates with negative personality traits such as neuroticism (mean r = -.08, p < .01), shyness (mean r = -.18, p < .01), and depression (mean r = -.09, p < .10). Although these correlations of personality traits and IS are statistically significant, the ranges of these correlations are rather large: extraversion (r = -.29 to .47), empathy (r = -.24 to .33), neuroticism (r = -.48 to .27), shyness (r = -.28 to .11), and depression (r = -.44 to .41; Hall et al., 2009). As these wide ranges signify, it is quite difficult to capture the complex relationship between personality traits and decoding accuracy. Recent research suggests three possible variables that might increase the ability of personality variables to predict decoding accuracy. These are: the use of a more complex, ecologically valid decoding task; the presence of competing tasks, creating greater cognitive load; and the brief presentation of nonverbal cues. All three of these variables create a more complex, demanding task for the decoder. It may be that only under conditions of greater complexity will individual difference measures offset decoding performance. Multi-channel nonverbal decoding task. How to measure one’s decoding ability has been the center of discussion in the field of nonverbal communication research. Measures in the past were often single-channel tasks, in which participants looked at, for example, still pictures of facial expressions and judged which emotion was expressed (e.g., Ekman & Friesen, 1975). One of the major problems with single-channel tasks is that they lack external validity. In real life, we rarely see only the face of others with whom we interact. Many interpretive clues exist in the stream of nonverbal behaviors: the combination of tone of voice and facial expressions, as well as the combination of gestures and posture, and so on (Archer & Akert, 1980). In addition, multi-channeled presentations of nonverbal cues can be presented in real-time (e.g., through the INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 10 use of videotape) including onset, off set, and duration information about the nonverbal cues. None of this information is available in still photograph presentations. Finally, multi-channeled presentations can include context or situational information. An example of a multi-channel decoding task is the Social Interpretations Tasks (SIT; Archer & Akert, 1980), which presents visual and auditory nonverbal information about social situations. The participants’ task is to interpret the nature of the interaction. For example, the participant sees a videotaped scene of a woman talking on the phone and hears the woman's speech. The participants are then asked to judge whether the woman is talking to another woman or to a man. This measure is a multichannel task, in which the participants observe a variety of naturally-occurring verbal and nonverbal cues at the same time. The multi-channel task has a higher external validity because the participants can connect one behavioral cue with another to construct the complete picture of the interaction, just as it occurs in everyday life. Researchers have hypothesized that the greater complexity of a multi-channel nonverbal decoding task is needed to test adequately the personality-accuracy hypothesis. For example, Akert and Panter (2008) developed a new measure, the “Visual Nonverbal Decoding Task” (VNDT). In this task, participants watch a series of short videos of people interacting in dyads, without sound. They are asked to choose what the conversational topic is for each scene, from a list of four choices. When decoding, people simultaneously pay attention to many different behavioral factors, such as facial expression and posture, to examine what true emotion or state of mind the encoder is experiencing. For example, if someone was smiling but crossing his arms stiffly and speaking in a flat tone, his smile would be interpreted as fake. Combining several pieces of behavioral information together is the essential component to decoding nonverbal cues accurately. The ability to process a variety of information concurrently depends upon one’s INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 11 tolerance for sensory stimulation, which varies among individuals: Interpreting different channels—whether they are verbal or nonverbal—involves handling a great deal of sensory stimuli (Akert & Panter, 1988). Research has found that extroverts are able to handle higher levels of sensory stimulation in both auditory and visual modalities (Eysenck, 1971; Friedman & Meares, 1979), and enjoy higher levels of external stimulation than introverts (Ludvigh & Happ, 1974; Weisen, 1965). Multi-tasking and decoding accuracy. Lieberman and Rosenthal (2001) shed new light on this issue and claim that personality interacts with decoding performance only when people are multi-tasking or under cognitive load; that is, when decoding nonverbal cues is the secondary task. They conducted a study to test the relationship between working memory and nonverbal decoding accuracy. Participants completed a measure of extraversion/introversion prior to their experimental sessions and were assigned to one of three conditions: PONS-focus, N-back-focus, and PONS-only. The participants in all conditions took the audio PONS test (Rosenthal et al., 1979) in which the participants were asked to choose an accurate description of a brief audio clip that they heard. Each of the 2-second filtered audio clips contained an adult female’s speech. Then, the participants were asked to choose between two descriptions (e.g., “expressing jealousy” and “scolding a child”). For the PONS-focus and N-back focus conditions, the participants had to simultaneously engage in the N-back task (O'Reilly, Braver, & Cohen, 1999) while performing the audio PONS (Lieberman & Rosenthal, 2001). The N-back task (O'Reilly et al., 1999) is a target-detection task, which has four levels of difficulty. In each level, the participants make a target response, by indicating yes or no, to each of the successive letters presented on a computer monitor. In the 0-back level, the participants are told to respond yes only when the letter B appears on the computer screen. In the 1-back level, the participants INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 12 respond yes only when a repetition of letters that appears on the screen are not intervened, that is, the repeated letter is one back in the sequence (e.g., E-Q-L-L). In the 2-back level, the participants respond yes only when a repetition of letters appear on the screen has a single intervening letter, that is, the repeated letter is two back in the sequence (e.g., E-L-Q-L). In the 3back level, the participants respond yes only when the repeated letter is three back in the sequence (e.g., E-L-A-Q-L). The main portion of the N-back task contains 360 letters that are presented on the computer screen successively. Each letter is presented for 500 milliseconds with a 2-second interval for the participants to respond. The participants are expected to press either one key to respond YES or another key to respond NO. The speed and accuracy of each response was measured. While completing two tasks simultaneously, the participants in the PONS-focus condition were told to focus more on the audio PONS test, and the participants in the N-backfocus condition were told to focus on the N-back task. In the PONS-only condition, the participants only performed the audio PONS test and had no secondary task. Lieberman and Rosenthal (2001) found that introverts and extroverts performed equally well on the PONS when no secondary task was involved. However, inroverts performed poorer than extroverts on the PONS when they had a secondary task in the 0-back level. In addition, the decoding accuracy of the introverts in the N-back-focus condition was the worst among all the other conditions. The authors conclude that introverts have less “grace period” in their multitasking ability before the cognitive intereference began; therefore, introverts’ decoding accuracy decreases only if introverts engage in multitasking and also if the nonverbal decoding task is not the primary goal. While Lieberman and Rosenthal’s (2001) study brought a new perspective to the field of nonverbal communication research, their study’s results contain some problems when applied to INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 13 real world examples. First, their study’s nonverbal task has weak external validity; the N-back task that the participants engaged in does not occur naturally. Second, their study only examined the decoding accuracy of auditory cues and not visual cues; decoding visual cues are more widely employed in everyday life. Finally, the N-back task was time-restricted; however, the author did not discuss how the brief presentation time of nonverbal cues might interact with personality traits, which might have caused the discrepancy in the decoding accuracy. Limited exposure time to nonverbal cues. Young and Brunet’s (2011) also took a new approach to measuring nonverbal decoding accuracy; they manipulated the exposure time of the presentation of facial expressions to examine how the length of presentation time influences shy/not-shy and high/low sociability individuals’ performance on a decoding task. The authors randomly assigned participants to one of two conditions: brief presentation and unlimited presentation. The participants in the brief presentation condition viewed each of twelve facial expressions for only 250 milliseconds, and were asked to judge which emotion was conveyed in each facial expression. The participants viewed a set of twelve facial expressions portraying six emotions (sadness, anger, happiness, surprise, disgust, and fear), displayed by six faces of male and six faces of female encoders; they were allowed to view each stimulus twice (Young & Brunet, 2011). The authors found that low sociability participants performed poorly on the nonverbal decoding tasks only if the exposure time of the facial expressions was brief; under the unlimited presentation condition, participants low in sociability decoded facial expressions as accurately as participants high in sociability (Young & Brunet, 2011). Given this prior research, the present investigation employs two nonverbal decoding tasks: one that presents facial expressions for only 40 milliseconds, and another that presents short video scenes of multiINDIVIDUAL DIFFERENCES ON DECODING ACCURACY 14 channeled nonverbal communication, in which people interact naturally in real-life conversations. Self-Consciousness Manipulation Davies (2005) found that the use of a video camera interacts with some personality traits, leading to a decrease in attention when performing simple tasks. He reported that the presence of a video camera, though not a mirror, negatively affected individuals high in public selfconsciousness when they performed a simple task. Davies (2005) also found that the presence of a mirror, but not a video camera, greatly affected individuals who were high in private selfconsciousness. These findings suggest that private self-consciousness intensifies self-awareness inwardly while public self-consciousness intensifies it outwardly. In order to affect participants who may have private or public self-consciousness, participants are filmed through a web camera while their face appears on the computer screen. A web camera combines the functioning of a mirror and a video camera, thus it is an effective device for inducing a high level of selfconsciousness in participants. The present study experimentally manipulates participants’ self-consciousness while they respond to two nonverbal decoding tasks, the VNDT (Akert & Panter, 2008) and the METT (Ekman, 2002). Participants are randomly assigned to one of three conditions: the low selfconsciousness condition, in which they are told that their facial expressions would be videorecorded through a web camera for later analysis; the high self-consciousness condition, in which they are told the above, but they see the video-recording of their faces on the computer screen; and the control group, who are not filmed and do not see their faces on the computer screen. Participants also completed six individual differences measures through an online survey site, Survey Monkey: extraversion, sociability, moving towards others/moving away from others, INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 15 private/public self-consciousness, shyness, and social intelligence. The dependent variables are the accuracy scores on the two nonverbal tasks. Hypotheses In the present study, it is hypothesized that participants who rate themselves high on three individual difference variables—moving away from others, shyness, and private and public selfconsciousness—will score significantly lower on each of the two nonverbal decoding tasks than participants who rate themselves low on these variables. Furthermore, it is hypothesized that the experimental manipulation of self-consciousness will have a significantly greater effect on decoding accuracy for people who rate themselves high on these variables than people who rate themselves low. Specifically, participants who rate themselves high on these variables are expected to demonstrate significantly lower nonverbal decoding accuracy in the high selfconsciousness condition than in other two conditions, and moderately lower nonverbal decoding accuracy in the low self-consciousness condition than in the control condition. For participants who rate themselves lower on these variables, it is hypothesized that the self-consciousness manipulation will have little effect on their decoding performance. The decoding scores of people in the low self-consciousness conditions and the control condition are hypothesized to be similar; decoding accuracy could be somewhat lower in the high self-consciousness condition due to the possible distractibility of the manipulation. Finally, for the other three personality variables—extraversion, sociability, and social intelligence—the same pattern of results is hypothesized, except in the opposite direction. That is, participants who rate themselves low on these variables are hypothesized to react the most strongly to the self-consciousness manipulation, leading to decreased nonverbal decoding accuracy. The decoding scores of participants who rate themselves high on these variables are INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 16 expected to show little effect of self-consciousness manipulation. Participants who rate themselves high on these variables would be more accurate decoders than those who rate themselves low, across all three experimental conditions. Method Participants One hundred and twenty female undergraduates at Wellesley College volunteered to participate in the one-hour study in exchange for ten dollars (Mage = 20 years, SD = 3.27, age range: 17-54 years). The study took place in a computer lab in the Wellesley College Psychology Department. Procedure The research participants were randomly assigned to one of three experimental conditions: the high self-consciousness condition, the low self-consciousness condition, or the control group (no manipulation of self-consciousness). Each experimental session had three to five participants. The participants in a given session were randomly assigned to the same condition so that they would not detect the manipulation. Participants were told that the study focused on how people make judgments about others, particularly by using nonverbal cues, and also how they perceived their own personalities. First, the participants in the high or low self-consciousness conditions were shown their faces on their computer screen via a web camera, and told this video would be recorded for a later analysis. The control group was not shown their faces nor were they told that their expressions would be analyzed later. Next, all participants responded to two nonverbal decoding tasks, presented individually on their own computers. After they completed their nonverbal decoding tasks, they responded to a packet of self-report individual difference measures. These INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 17 were presented in an online questionnaire format, with accompanying fiveor seven-point Likert scales. After the participants completed the questionnaires, they were thanked and debriefed. The participants’ answers to the questionnaires and the decoding tasks were kept anonymous and confidential; their recorded videos were deleted immediately after the participants were debriefed. Self-Consciousness Manipulation In the high self-consciousness condition, the participants saw two windows on their individual computer monitors: one that showed the nonverbal task and another that showed their faces as filmed through the attached web camera. Although the size of the window for showing the face was smaller (approximately 6 in. by 5 in.) than the window for the nonverbal task (approximately 12 in. by 8 in.), the participants responded to the nonverbal tasks while their faces appeared on the screen. The participants were told that the camera was recording their facial expressions as they completed the nonverbal tasks, and that the experimenter would be coding their facial expressions later. The purpose of this manipulation was to induce a high level of self-consciousness. The participants in the low self-consciousness condition did not see their faces on the computer monitor when they completed the two nonverbal tasks. However, they were told that their faces were being filmed through the web camera as they did the nonverbal tasks. To ensure the manipulation is effective in the low self-consciousness condition, the experimenter showed these participants the window in which their faces appeared, and informed them that their faces were being filmed and recorded through the web camera. They were told that the experimenter would code their facial nonverbal expressions later. Then, the experimenter minimized the window (approximately 1 cm by 1 cm) so that the participants could not see their faces while INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 18 completing the nonverbal tasks. Finally, participants in the control group completed the nonverbal tasks without being filmed or seeing their faces on the computer screen. Prior studies have shown that use of a camera affects self-consciousness and changes people’s behavior. For example, camera-induced self-attention is found to change people’s attitudes following counter-attitudinal behavior (Insko, Worchel, Songer, & Arnold, 1973; Wicklund & Duval, 1971). The use of a camera is also found to decrease task-focused attention, which lowers performance on concept formation tasks for highly self-conscious individuals (Brockner, 1979). Although a mirror has been widely employed to induce self-consciousness, use of a mirror in a computer lab would evoke a sense of strangeness in participants’ minds. The benefit of using a web camera was that it combined the effect of mirror and camera: by using a web camera, the participants knew that they were being filmed and they also saw their faces on the computer monitor like in a mirror. This combination effect bolstered the induction of selfconsciousness in participants. Measures Nonverbal Decoding Tasks. Visual Nonverbal Decoding Task (VNDT; Akert & Panter, 2008; Akert et al., 2010). The VNDT consists of 24 brief scenes taken from the 1980’s PBS and ABC talk shows hosted by Dick Cavett. The scenes are shown without sound. In the scenes, the host, Dick Cavett, interviews a number of guests; the conversation topics vary from the guests’ childhood memories to their recent work. The participants’ task was to determine what the conversational topic was in each scene by observing the host’s and the guest’s nonverbal behavior. Before watching each scene, the participants read four possible answer choices for the conversational topic in their questionnaire booklet. The answer choices included only one right answer and three other choices; these three other choices were the conversational topics covered INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 19 by the host and the guest at other times during the interview. After watching each scene, participants circled their answer choice in the questionnaire booklet. For example, in Scene 4, the world-renowned musician, YoYo Ma, spoke about meeting a master cellist whom he had always respected. In the video, Ma displayed facial expressions and gestures to describe how happy he was to meet the cellist. The answer choices for this scene were: “YoYo remembers meeting a master cellist two times,” “YoYo recalls an unpleasant experience,” “YoYo describes his college years at Harvard,” and “YoYo is talking about his second cousin in China.” Each scene lasted about two minutes, and there were eight-second intervals between scenes to allow participants to choose their answer and read the next scene’s choices. Because of time limitations, only the last twelve scenes of the VNDT were used in the present study. The accuracy percentage for the twelve scenes ranged from 20.8% to 79.2%. Micro Expression Training Tool (METT; Ekman, 2002). METT was created to train people to improve their accuracy in recognizing micro-momentary nonverbal expressions. A micro-momentary expression is defined as a very sudden onset and offset of a facial expression (e.g., anger); it appears as a brief flash on a person’s face (Ekman, 2002). The first two parts of the METT involve training viewers how to read emotions on the face. It mainly emphasizes quick changes in parts of a face. Examples of changes include: wrinkles in the nose area (for detecting disgust), a flash of widely opened eyes (for detecting surprise), and tightened lips (for recognizing anger). The third part of the METT provides a self-test of decoding micromomentary expressions. In each test item, a different person appears. He or she first shows no expression. Then, suddenly, a facial expression of one of seven emotions appears very briefly, for only 40 milliseconds, immediately returning back to the expressionless face. This quick presentation is accomplished by presenting three still photographs briefly on the computer screen INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 20 (i.e., neutral, emotional expression, and then neutral). After each quickly flashed expression, the participants must choose which of the seven emotions was displayed: anger, fear, disgust, contempt, sadness, surprise or happiness. The present study used only the testing section of the METT, which consists of 14 still photographs (i.e., 14 items). The METT was included in the present study as a comparison to the VNDT. For example, the METT and VNDT differ in presented channels (single channel facial expression versus multi-channel, respectively) and in exposure time (extremely brief exposure versus real-time exposure, respectively). Thus, the METT and VNDT tap into different types of nonverbal decoding ability, and together offer an expanded test of the hypothesis. A few prior studies have used the METT as a decoding task instead of as a training tool (Marsh et al., 2010; Matsumoto et al., 2000; Warren, Schertler, & Bull, 2009). The METT comes with an automatic scoring function that presents the accuracy percentage on the screen when the test is completed. Thus, participants knew how well they did on the METT as they finished it. In order to control for this confound, the METT was the second and final nonverbal task. The accuracy percentage for the present study’s sample ranged from 10% to 90%. Individual Difference Measures. Eysenck Personality Inventory (EPI; Eysenck & Eysenck, 1968). The EPI consists of 57 items with three subscales: extraversion (24 items), neuroticism (24 items), and faking (lie scale: 9 items). The present study only used the extraversion subscale. The extraversion subscale measures the degree to which one is out-going and interacts with others. Examples of items in the extraversion subscale are, “Are you a talkative person?” and “Do you like mixing with people?” Each item is answered with “yes” or “no” (Eysenck & Eysenck, 1968). Lieberman and Rosenthal (2001) conceptualized introversion INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 21 as individuals who score low on the extraversion subscale of the EPI. The extraversion subscale’s coefficient of inter-item correlation for the present study was .78. Eysneck and Eysenck’s manual (1968) reported evidence for factorial validity, construct validity, and concurrent validity of the EPI. Interpersonal Orientation Scale (IOS; Hill 1987). The IOS consists of 26 items, rated on the Likert scale from 1 (not al all true) to 5 (completely true), measuring individuals’ trait-like preference for social interaction. The items are categorized into four subscales: emotional support (6 items), attention (6 items), positive stimulation (9 items), and social comparison (5 items). The present study only used the positive stimulation subscale as a measure for sociability. The positive stimulation subscale assesses levels of affiliation and the need to belong. An example of an item on this scale is, “I would find it very satisfying to be able to form new friendships with whomever I liked.” The positive stimulation subscale’s coefficient of inter-item correlation for the present study was .71. Personal Style Inventory (PSI; Robins et al., 1994). The PSI consists of 48 items assessing individuals’ interpersonal and achievement vulnerability (sociotropy and autonomy, respectively). Half of the items measure three subtypes of sociotropy—excessive concerns of what others think (Approval: 7 items), interpersonal dependency (Dependency: 7 items), and need to please others (Pleasing Others: 10 items). The other half of the items evaluates three subtypes of autonomy—self-critical perfectionism (Self Criticism: 4 items), need for control (Control: 8 items), and defensive separation (Defensive Separation: 12 items). For the present study, only two sociotropy subscales and one autonomy subscale were used: Dependency (e.g., “I find it difficult if I have to be alone all day”), Pleasing Others (e.g., “I often put other people’s needs before my own”), and Defensive Separation (e.g., “I tend to keep other people at a INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 22 distance”), respectively. All 29 items were rated on the Likert scale from 1 (strongly disagree) to 6 (strongly agree). These three subscales were selected to assess what Honey (1945) called the “moving towards others” and the “moving away from others” interpersonal styles. Cheek and Krasnoperova (1999) used Dependency and Pleasing Others subscales to measure moving toward others and used Defensive Separation subscale for measuring moving away from others. The alpha coefficient of internal consistency reliability for the PSI was .70: for moving toward others, α = .78, and for moving away from others, α = .78. Self-Consciousness Scale Revised (SCS-R; Scheier & Carver, 1985). The SCS-R was developed because of the unclear wording used in the original Self-Consciousness Scale (Fenigstein et al., 1975). The SCS-R changed the wording of the items in the original scale such as “I never scrutinize myself” to “I never take a hard look at myself,” and “Generally I am not very aware of myself” to “I know the way my mind works when I work through a problem.” Scheier and Carver (1985) report that participants would ignore the items when wording confused them; thus the revised version was used in the present study to eliminate potential confusion. The SCS-R consists of 22 items rated on the Likert scale from 0 (not at all like me) to 3 (a lot like me), with the same three subscales as the original SCS. The subscale intercorrelations with the original SCS were very high (in the low to the mid .80s), suggesting that the revised scale is highly similar to the original scale. The present study employed only the private self-consciousness subscale (9 items) and the public self-consciousness subscale (7 items). The alpha coefficient of internal consistency reliability for the total scale was .81: for the private self-consciousness subscale was .74, for the public self-consciousness subscale was .76. The 20-item Shyness Scale (Melchior & Cheek, 1985). The 20-item shyness scale is a revised version of the original 9-item Shyness Scale (Cheek & Buss, 1981). It is a self-report INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 23 measure of the degree of shyness, rated on the Likert scale from 1 (very uncharacteristic) to 5 (very characteristic). This new version of the shyness scale assesses three dimensions of shyness: somatic (e.g., “I feel tense when I’m with people I don’t know well), behavioral (e.g., “I am somewhat socially awkward.”), and cognitive (e.g., “I feel painfully self-conscious when I am around strangers”). The alpha coefficient of internal consistency reliability for the 20-item Shyness Scale was .93. Tromso Social Intelligence Scale (TSIS; Silvera, Martinussen, & Dahl, 2001). The TSIS consists of 21 items rated on the Likert scale from 1 (describes me poorly) to 7 (describes me extremely well). It is a self-report measure of social intelligence, which is people’s ability to understand social context and successfully interact with others. The measurement has three sub factors: social skills (e.g., “I fit in easily in social situations”), social awareness (e.g., “People often surprise me with things they do”), and social information processing (e.g., “I can predict other peoples’ behavior”). The alpha coefficient of internal consistency reliability for the total scale was .87. These three sub factors also showed acceptable rates: social skills (α = .87), social awareness (α = .73), and social information processing (α = .82). Past research indicated a significant correlation (r = .19, p < .05) between the TSIS and nonverbal decoding accuracy (Akert et al., 2010). Results Preliminary Analyses One-way analyses of variance (ANOVA) were conducted on the independent variable of self-consciousness conditions (high self-consciousness, low self-consciousness, and control) for each of the six personality measures: the Eysenck Personality Inventory (EPI), the Interpersonal Orientation Scale (IOS), the Personal Style Inventory (PSI), the Self-Consciousness Scale INDIVIDUAL DIFFERENCES ON DECODING ACCURACY 24 Revised (SCS-R), the 20-item Shyness Scale (Shyness), and the Tromso Social Intelligence Scale (TSIS). These were conducted to determine if random assignment had in fact created groups that were equivalent. Each of these six one-way ANOVAs indicated no significant differences in mean scores across the three conditions: EPI, F (2, 117) = 2.25, p = .11; IOS, F (2, 117) = .25, p = .78; PSI, F (2, 117) = .37, p = .70; Shyness, F (2, 117) = .91, p = .41; SCS-R, F (2, 117) = .67 p = .51; TSIS, F (2, 117) = .74, p = .48. These results indicated that the three experimental conditions were equivalent in terms of personality scores. Table 1 shows the means and standard deviations for the six personality variables used in the study. Table 1 Mean Scores of the Personality Scales Across the Three Self-Consciousness Conditions High Low Control Personality Scales M SD M SD M SD EPI 9.30 (2.60) 10.38 (2.81) 9.18 (2.93) IOS 3.11 (.71) 3.17 (.59) 3.22 (.73) PSI 3.71 (.48) 3.79 (.41) 3.74 (.44) Shyness 2.72 (.76) 2.79 (.86) 2.96 (.86) SCS-R 3.02 (.54) 3.14 (.38) 3.11 (.47) TSIS 4.83 (.70) 4.96 (.93) 4.74 (.81)