اجتماعی شدن حالات صورت در محیط های مجازی همه جانبه: مطالعه EMG صورت
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|37798||2015||5 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Biological Psychology, Volume 91, Issue 1, September 2012, Pages 17–21
Abstract Immersive virtual environment technology is increasingly used by psychologists as a tool for researching social influence in realistic, yet experimentally controllable, settings. The present study demonstrates the validity and reliability of facial electromyography as a marker of affect in immersive virtual environments and further shows that the mere presence of virtual humans is enough to elicit sociality effects on facial expressiveness. Participants viewed pleasant and unpleasant images in a virtual room either alone or with two virtual humans present. The patterns of smiling and frowning activity elicited by positive and negative stimuli in the virtual environment were the same as those found in laboratory settings. Moreover, when viewing positive stimuli, smiling activity was greater when two agents were present than in the alone condition. The results provide new psychophysiological evidence for the potency of social agents in immersive virtual environments.
. Introduction Immersive virtual environment (IVE) technology is increasingly used by psychologists as a tool for understanding human behavior in realistic, yet experimentally controllable, settings (see Fox et al., 2009 and McCall and Blascovich, 2009, for reviews of IVE use in behavioral research). The immersive visual experience is often conveyed to participants through a head-mounted display (HMD) that uses head movements to adjust the visual display of the environment in real-time, permitting visual and physical exploration of a three-dimensional, computer-simulated environment as though it were real. The immersive world instills a sense of presence in the user – a feeling that the user exists in the virtual environment (rather than merely watching it); the IVE is capable of portraying psychologically realistic scenarios (e.g., Kotlyar et al., 2008) and socially potent computer-controlled virtual humans (i.e., agents; Bailenson et al., 2001). The present study extends what is known about the social potency of agents by testing the effects of the apparent “company” of virtual others on facial expressiveness through electromyography (EMG). The social potency of IVEs has been demonstrated with a variety of methodologies. For example, the amount of interpersonal space an immersed participant grants to an agent can be augmented by having the agent engage in eye contact (Bailenson et al., 2003) or by altering the ethnic appearance of the agent (Dotsch and Wigboldus, 2008 and McCall et al., 2009). On the basis of such findings, Blascovich et al. (2002) proposed a threshold model of social influence with IVEs. Whether an agent will influence the participant's behavior depends on four variables: behavioral realism (the extent to which the agent behaves like its real-world counterpart), social presence (the degree to which the participant believes the agent is under the control of an actual human), self-relevance (the extent to which the interaction has value or meaning to the participant), and the target response system (the level of the behavioral response – automatic and low-level vs. more controlled and high-level). According to the model, virtual humans are more likely to have an impact on the participant's behavior as each of these variables increases. For example, when a participant believes the agent is controlled by another person, and the agent's behavior and appearance closely approximates real human behavior, the participant should interact with the agent much like he or she would in the real world. An aim of the present study was to examine the extent to which such agents in an IVE can influence a low-level behavior: spontaneous facial expressions. Facial expressions can serve as automatic markers of an affective state. Consistent with this interpretation, pleasant stimuli reliably potentiate smiling (i.e., zygomaticus major activity) and inhibit frowning (i.e., corrugator supercilii activity) expressions (see Tassinary et al., 2007, for a review). Yet, the correspondence between affective states and facial muscle activity varies considerably across individuals (Larsen et al., 2003) and social contexts. For instance, the perceived presence of other humans intensifies smiling toward pleasant stimuli (i.e., the “sociality effect;” Fridlund, 1991). This sociality effect on smiling is strongest when a social other is physically present, yet the implied presence of social others, such as knowing that a friend is viewing the same material in another room, is sufficient to elicit exaggerated expressions to positive stimuli (Fridlund, 1991 and Hess et al., 1995). It is unclear, however, whether agents can also elicit the sociality effect, particularly when social presence and self-relevance are low. Prior research has demonstrated that depictions of virtual humans can influence facial muscle movements, even when participants are not immersed in an IVE. Consistent with the Blascovich et al. (2002) threshold model of social influence, simply viewing the facial expression of a dynamic avatar1 (i.e., one with higher behavioral realism) on a computer screen elicits more mimicry than a static avatar (i.e., one with low behavioral realism) as measured by facial EMG (Weyers et al., 2006). In addition, when self-relevance is increased by nonconsciously priming social competition, facial mimicry when viewing happy and sad avatar faces while playing a live online game becomes counter-empathic (Weyers et al., 2009). Such unconscious nonverbal behaviors suggest that agents trigger automatic, reactive goals within us (Bargh and Chartrand, 1999). To date, however, no study has examined the sociality effect in a virtual environment. Building on this prior research, the present study examined the effects of agents in an IVE on two distinct automatic functions of facial expressions within a virtual environment: expressions as affective markers and expressions as social signals. A fuller understanding of such effects would help establish the pervasiveness of the sociality effect – that is, perhaps we are so used to modifying our facial expressions in the presence of other humans, we automatically do so at the slightest suggestion in a virtual environment. In addition, because facial EMG has been a useful measure of affect in other contexts, demonstrating that it is a valid marker of affect in response to variables manipulated in an IVE would mean that researchers would have another noninvasive measure to add to their methodological armamentarium. Participants wore a stereoscopic HMD while facial EMG was recorded from the brow and cheek regions. First, we examined whether facial expressiveness in an IVE is affected by affective visual stimuli in the same manner as in previous laboratory settings. We hypothesized that pleasant and unpleasant visual stimuli would elicit greater zygomaticus and corrugator activity, respectively. Second, we tested whether the mere presence of social agents in an immersive virtual environment is sufficient to elicit exaggerated facial expressions when viewing pleasant stimuli. We provided no information about the agents. That is, social presence was low, as participants did not believe the agents were controlled by actual humans, and self-relevance was also low because the agents had no ostensible connection to the participant. Given the low-level response of facial activity, however, we hypothesized that zygomaticus activity would be potentiated for positive stimuli when these nominally social agents were present in the virtual environment, similar to when a friend was present in the earlier studies of sociality effects (Fridlund, 1991 and Hess et al., 1995). In addition, we examined whether sociality effects would occur for unpleasant stimuli and whether corrugator activity was similarly affected by social cues, as such sociality effects have not been previously demonstrated for negative stimuli.
نتیجه گیری انگلیسی
Results 3.1. Self-report ratings A two-way factorial Valence × Agent Presence repeated measures ANOVA revealed that Valence of the IAPS stimuli affected self-reported pleasantness of the stimuli, F(1, 47) = 554.60, p < .001, View the MathML sourceηp2=.922, such that positive IAPS images (M = 4.04, 95% CI [3.82, 4.26]) were rated as more pleasant than negative IAPS images (M = 1.49, 95% CI [1.40, 1.57]). Agent Presence did not affect ratings either as a main effect, F(1, 47) = 1.69, p = .20, View the MathML sourceηp2=.035, or as an interaction with Valence, F(1, 47) = 0.16, p = .70, View the MathML sourceηp2=.003. 3.2. Facial EMG The baseline corrected post-stimulus epochs for each trial were categorized based on the muscle region and stimulus valence. A test of multivariate outliers in the EMG data revealed two participants with Mahalanobis distances greater than 3 standard deviations from the mean of the distribution. Due to these extreme responses in both zygomaticus activity and corrugator, EMG data from these participants were excluded from further analyses. Complete EMG data remained for 42 participants (n = 21 alone condition), and partial EMG data remained for another 5 participants (1 zygomaticus alone, 1 zygomaticus co-viewing, 2 corrugator alone, 1 corrugator co-viewing). Baseline zygomaticus and corrugator activity were not affected by Agent Presence, F(1, 40) = 0.67, p = .42, View the MathML sourceηp2=.016. A three-way factorial Muscle × Valence × Agent Presence repeated measures ANOVA revealed significant main effects of Agent Presence, F(1, 40) = 8.65, p < .01, View the MathML sourceηp2=.819, on post-stimulus EMG activity. The effect of Agent Presence was moderated by a significant Muscle × Agent Presence interaction, F(1, 40) = 5.31, p = .03, View the MathML sourceηp2=.613, and a significant Muscle × Valence × Agent Presence interaction, F(1, 40) = 6.59, p = .01, View the MathML sourceηp2=.707. A Muscle × Valence interaction confirmed that each muscle was differently sensitive to positive and negative stimuli, F(1, 40) = 57.56, p < .001, View the MathML sourceηp2=1.000. Separate analyses for each muscle further examine the influence of Agent Presence and Valence on facial expressions. 3.2.1. Zygomaticus major A main effect of Valence, F(1, 42) = 49.83, p < .001, View the MathML sourceηp2=.543, confirmed that zygomaticus activity was greater for positive stimuli than negative stimuli. A main effect for Agent Presence, F(1, 42) = 7.37, p = .01, View the MathML sourceηp2=.149, revealed that post-stimulus zygomaticus activity was greater in the co-viewing condition. This was further qualified by a significant Valence × Agent Presence interaction, F(1, 42) = 4.68, p = .04, View the MathML sourceηp2=.100 (see Fig. 2). Paired comparisons confirmed that zygomaticus activity for positive stimuli was greater in the co-viewing condition than the alone condition, t(42) = 2.69, p = .01. Zygomaticus activity for negative stimuli did not differ by Agent Presence condition, t(42) = 0.74, p = .46. Mean zygomaticus EMG activity as a function of Agent Presence and Valence of ... Fig. 2. Mean zygomaticus EMG activity as a function of Agent Presence and Valence of stimuli. Error bars represent 1 SEM. Figure options 3.2.2. Corrugator supercilii A main effect of Valence, F(1, 43) = 43.06, p < .001, View the MathML sourceηp2=.500, confirmed that corrugator activity was greater for negative stimuli than positive stimuli. No main effect of Agent Presence was found, F(1, 43) = 0.02, p = .89, View the MathML sourceηp2<.001; however, there was a marginally significant Valence × Agent Presence interaction, F(1, 43) = 2.88, p = .10, View the MathML sourceηp2=.063 (see Fig. 3). Mean corrugator EMG activity as a function of Agent Presence and Valence of ... Fig. 3. Mean corrugator EMG activity as a function of Agent Presence and Valence of stimuli. Error bars represent 1 SEM.