情绪心理学研究资料Specifying Emotional Information

Specifying Emotional Information: Regulation of Emotional Intensity via Executive Processes Pierre Philippot, Ce´line Baeyens, and Ce´line Douilliez University of Louvain The degree of specificity at which emotional information is activated might determine evoked emotional intensity. However, the nature of this effect remains unclear. Four studies tested (a) whether people hold the naı¨ve theory that activating specific details of emotional information arouses acute feelings; (b) whether an emotionally distressed population (social phobics) also holds that theory; and (c) whether voluntarily focusing on specific aspects of a distressing situation reduces its emotional impact. Results indicate that control as well as emotionally distressed people hold a naı¨ve theory that specifying emotion increases its intensity. However, Studies 3 and 4 showed that voluntarily elaborating specific aspects of a distressing situation reduces distress. Results are discussed in terms of voluntary versus automatic processing of emotional information. Keywords: emotion regulation, executive processes, psychopathology The regulation of emotion is a hallmark of everyday life. How- ever, when confronted with unwanted and painful emotions, a condition common to most people suffering from emotional dis- orders, the issue of the regulation of emotion becomes particularly crucial. The development of cognitive and behavior therapy by clinical researchers and of appraisal theories by emotion research- ers has fostered the notion that emotion regulation is partly deter- mined by the way emotional information is processed (e.g., Clark, 1999; Foa & Kozak, 1991; Scherer, 1999; Williams, Watts, Mac- Leod, & Mathews, 1999). One dimension of emotional information processing that seems particularly clinically relevant is the level of specificity versus overgenerality in such processing. This notion is present in many theoretical accounts of emotion regulation in clinical populations. For instance, Beck has argued that overgeneralizing negative pre- dictions to whole clusters of events precipitate negative mood (Beck, Emery, & Greenberg, 1985). Relatedly, Borkovec (Bork- ovec & Inz, 1990; Sto¨ber & Borkovec, 2002) has observed that worries in generalized anxiety are essentially a verbal thought activity that tends to remain at an overgeneral and abstract level. They conceptualized this overgeneral processing as a coping strat- egy used to avoid somatic activation and imagery associated with negative affect. This conceptualization is congruent with Lang’s (1993) claim that thinking in specific details about emotion entails activating emotionally relevant and acute cognitions linked to the activation of emotional arousal. However, Sto¨ber and Borkovec (2002) argue that overgeneral processing can lead to maintenance of anxiety and failure in emotional processing. Consequently, they consider that processing more specifically and concretely emo- tional information is useful for improving problem solving and facilitating emotional processing. In sum, although the notions of generality and specificity of processing have been defined and operationalized differently according to the authors, this overview of the literature suggests that overgeneral processing of emotional information is related to the maintenance of emotional disorders, and that specific processing might lead to emotional change and better emotional regulation, but that it bears a cost: the activation of acute emotional feelings. This notion is particularly well illustrated by the work of Wil- liams (1996). He has evidenced that depressed individuals suffer from an overgenerality bias in retrieving personal memories of past emotional experiences (e.g. Williams, 1996; Williams, Stiles, & Shapiro, 1999) as well as in imagining possible future experi- ences (Williams, Ellis, Tyers, Macleod, & Rose, 1996). For in- stance, when asked to recall experiences of anger, depressed individuals tend to report overgeneral events (e.g., “when I am with my girlfriend”) rather than a specific event (e.g., “last Sun- day, I had an argument with my neighbor whose dog was endlessly barking”). This overgenerality bias has also been observed in other clinical populations (e.g., Wenzel, Werner, Cochran & Holt, 2004). Williams et al. (1999) have proposed that by remaining at a general—and thus more abstract—level of information, individ- uals attempt to avoid the reactivation of acute and painful emotions felt in specific personal experiences. However, several studies have demonstrated that voluntarily specifying emotional information diminishes emotional intensity while activating this information at an overgeneral level maintains emotional intensity. In these studies, specific emotion processing Pierre Philippot, Ce´line Baeyens, and Ce´line Douilliez, Department of Psychology, University of Louvain, Belgium. The writing of this paper has been facilitated by grants from the “Fonds National de la Recherche Scientifique de Belgique” 8.4510.99 and 8.4505.00. The data collection has been made possible by the UCL Psy- chology Department Consulting Center (CPS) and the help of Ana Isabel Burgos, Serge Gozlan, Vale´rie Ramon, and Severine Verhasselt. The authors appreciate helpful comments of Christopher Long and five anonymous reviewers on earlier drafts of this paper. Correspondence concerning this article should be addressed to Pierre Philippot, Faculte´ de Psychologie, Universite´ de Louvain, place du Cardi- nal Mercier, 10, B-1348, Louvain-la-Neuve, Belgium. E-mail: Pierre.Philippot@psp.ucl.ac.be Emotion Copyright 2006 by the American Psychological Association 2006, Vol. 6, No. 4, 560–571 1528-3542/06/$12.00 DOI: 10.1037/1528-3542.6.4.560 560 is defined as a voluntary activation of specific and unique episodic information focusing on the elaboration of the emotional event and experience in terms of their unique details with regard to their context and emotional facets (feelings, sensations, thoughts, etc. . .). General emotion processing is defined as a voluntary activation of personal emotional information centered on generic aspects of the experience: the type of mental state, impression, etc. . . typically and repetitively occurring in the type of situation considered. In both cases, the same processing can be applied to past, present, or future events (Wheeler, Stuss, & Tulving, 1997). In one study (Schaefer et al., 2003), participants imagined standard emotional scripts while receiving precise instructions to process emotional information in a specific or in an overgeneral manner. Overgeneral processing led to more intense emotional feelings, greater increases in heart rate, and more activation of the ventro- medial prefrontal cortex, a brain region generally involved in the generation of emotion (e.g., Bechara, Damasio, Damasio, & Lee, 1999), than did processing of similar information at a specific level. These results were replicated in another study without brain activity recording and with a control group who received no processing instruction (Schaefer & Philippot, 2005). In other stud- ies (Philippot, Schaefer, & Herbette, 2003), emotion was induced by film clips or by re-evoking past personal experiences. Emo- tional information processing mode was manipulated by priming either overgeneral or specific personal emotional memories (while priming no memories among the control group). Again, it was observed that overgeneral processing maintained the intensity of emotional feelings, whereas specific processing significantly re- duced it. This observation has recently been replicated by Raes, Hermans, Williams, & Eelen (2005). The empirical evidence presented in the two preceding para- graphs might seem contradictory: Williams et al. (1999) present overgenerality as a coping strategy to avoid acute emotion while we sustain that specifying emotional information fosters better emotion regulation. We believe, however, that this contradiction might only be superficial. Indeed, it is remarkable that the studies that found a positive relationship between emotional information specificity and emotional intensity have all operationalized spec- ificity in processing emotional information as an individual differ- ence. Conversely, all studies that found a negative relationship have manipulated the specificity dimension by instructing partic- ipants to voluntarily process emotional information in one way or another. This is not the first instance in the study of emotion regulation where correlational effects (individual differences) show patterns opposite to those from experimental manipulation effects. For example, Manstead (1991) has demonstrated that when facial expression is experimentally manipulated, less expression results in less emotional arousal; however, in correlational studies, individuals who are naturally facially less expressive respond with more arousal in emotional situations (Colby, Lanzetta, & Kleck, 1977). We have proposed a theoretical account for this reversal in effect (Philippot, Baeyens, Douilliez, & Francart, 2004). Our ra- tionale is that, in everyday life, most accessing of personal emo- tional information occurs on an automatic mode labeled “direct retrieval” (Conway & Pleydell-Pearce, 2000). Such access arises when a stable pattern of activation emanating from event-specific knowledge in autobiographical memory is established and linked to the working memory goals of the individual (Conway & Pleydell-Pearce, 2000, p. 275). This process is almost instanta- neous, requires few cognitive resources, and may result in an emotionally vivid memory, rich in specific details, although this memory may be partial and not cover the whole scene. The phenomenon of flashback in PTSD (Brewin, 2001) may constitute an extreme example of direct retrieval triggered by specific emo- tional information. Such characteristics of directly retrieved emo- tional material would ground the belief that activating emotional information at a specific level triggers vivid emotion. This belief might even be more pronounced in people who are regularly confronted by intrusive emotional information, such as individuals suffering from an emotional disorder. However, there is another mode of accessing emotional auto- biographical information: “generative retrieval” (Conway & Pleydell-Pearce, 2000), an effortful process by which individuals willfully construct past personal memories, envision future per- sonal experience, or raise self-awareness of present experience. It includes intensive executive processing, such as strategic memory search for and attentional focus on relevant information as well as comparison of activated memory material with the retrieval target (Conway & Pleydell-Pearce, 2000; Rubin, 1996; Schacter, Nor- man, & Koutstaal, 1998; Wheeler, Stuss, & Tulving, 1997). In- deed, recent studies have shown that voluntary accessing of spe- cific personal memories necessitates executive resources (Williams et al., 2005). When one voluntarily evokes an emotional experience, this slow and effortful process may be disturbed by the parallel activation of emotion and therefore eventually aborted. Conway and Pleydell-Pearce (2000) have argued that to avoid such disturbances, autobiographical memories specification implicates a process of inhibition of the link between event-specific informa- tion and general emotional content. Thus, to avoid being disrupted by emotional arousal, these processes would inhibit the structures responsible for emotion activation. In contrast, activating emo- tional information at an overgeneral level would necessitate much less strategic and cognitively demanding processes. In this case, no inhibition of the structures responsible for emotion activation would occur. It is important to note that in the case of generative retrieval the same processing can be applied to past, present, or future events (Wheeler, Stuss, & Tulving, 1997). In other words, remembering a past experience or envisioning a future one mobilizes the same cognitive processes. Thus, as the same processes are at work in the different time perspectives, one can expect that the specific or general emotion processing modes will have the same effects whether applied to the evocation of a past emotional memory, to the processing of an ongoing emotion, or to the anticipation of a future emotional experience. We have demonstrated that it was indeed the case for emotional memories re-evoked in mental imagery as well as for emotions directly induced by film excerpts (Philippot et al., 2003). The present constructs of specific and general processing, ap- plied to the generative elaboration of autobiographical emotional information, present similarities and differences with the notion of generality developed in the clinical literature. For instance, the present concept of generality clearly differs from Beck et al.’s (1979) notion of generalization. The former designates a mode in the activation of episodic emotional information, while the latter refers to the process by which a situation-specific observation or conclusion is erroneously applied to many situations. In contrast, 561SPECIFYING EMOTIONAL INFORMATION the present constructs are similar to those developed by Williams, with two restrictions. First, in the present case, only generative processing is considered for specific information, while Williams (1996) does not distinguish direct from generative retrieval in his definition. Second, the notion of generality in Williams’ work refers to both extended or repetitive events (while in the present perspective, overgenerality designates the generic prototypical fea- tures of one type of experience) that are repeated over many instances and that can be generalized to that class of experiences. In contrast, the notions of specificity are more similar: Williams refers to specificity of events that occur on a single day (episodic specificity), while the present notion focuses on the details that make the events unique, i.e., the emotional and episodic (time, location, circumstances) details characterizing the events. Finally, the present constructs are very close to those developed by Bork- ovec, with the nuance that Borkovec equates the dimensions of specificity-generality and of concreteness-abstraction, while the present conception focuses exclusively on the specificity (unique)—generality (generic) dimension. In sum, generative (i.e., voluntary and strategic) processing of emotional information should entail the inhibition of emotional arousal that could potentially disturb the process (Conway & Pleydell-Pearce, 2000). In contrast, the direct activation of emo- tional information would automatically trigger a state of emotional arousal. Likewise, processing emotional information at a general, holistic level should maintain emotional arousal, as it does not require the extensive cognitive processing that would necessitate inhibition of emotional arousal. In everyday life, however, people rarely engage in generative processing of emotional information, especially if it refers to aversive experiences. They would thus rarely have the opportunity to disconfirm the naı¨ve theory that specifying one’s emotion increases its intensity. This belief might even be stronger in clin- ical populations that confront more acute and more painful emo- tions. As a consequence, people suffering from emotional disor- ders would avoid processing specifically emotional information, especially information potentially related to aversive experience. Because emotional information would not be processed, no emo- tional change would occur and the same type of feelings should reappear (Foa & McNally, 1996). This cognitive avoidance pro- cess might thus explain the maintenance of emotional disorders (Clark, 1999). The theoretical rationale developed above suggests a set of testable hypotheses. First, people should hold a naı¨ve theory or metacognitive belief (Wells, 1995) that specifying one’s emotion increases its intensity. Second, this naı¨ve theory might be more prevalent in individuals suffering from emotional disorders, as they are more likely to be confronted with intrusive emotional information (direct retrieval of acute emotional information). Third, and opposite to the naı¨ve theory, voluntarily specifying emotional information should lead to a decrease in emotional arousal. This latter hypothesis has already been tested for the recollection of everyday life emotions or emotion induced by film clips (Philippot et al., 2003; Schaefer et al., 2003). However, it has not yet been tested in a clinically relevant situation of on-going distressing emotional experience. The aim of the present paper is to test these three hypotheses. Study 1 The first experiment was designed to examine whether people hold a naı¨ve theory about how the specificity of a forthcoming emotional event impacts on the intensity of the emotion triggered by the evocation of that event. Participants were asked to imagine someone thinking either in a specific or in a general manner for 10 minutes about a forthcoming stressing experience. They then had to predict how the emotional intensity of that person’s feeling state would evolve minute after minute as he or she is thinking about this emotional experience. Method Participants and design. Individuals were approached in public places (e.g., waiting room in train station and libraries) and invited to take part in a questionnaire study on emotion regulation. Fifty-three people (32 women), between 17 and 68 years of age (mean 26.0) participated. Each participant completed both a specific and a general thinking condition (within-subject factor). Procedure. Participants individually received a questionnaire and were asked to fill it in immediately. The questionnaire first asked them to imagine a person that would have to give a talk in the context of an important exam or job interview. They were asked to imagine that person, thinking in a state of anxious apprehension about that talk the night before. They were then asked to indicate the emotional profile of the feeling state of that person on the Differential Emotion Scale (DES; Izard, Dougherty, Bloxom, & Kotsch, 1974; French validation, Philippot, 1993), which provides 10 emotional state items that are rated in terms of intensity on a Likert scale from 1 to 5. Based on these ratings, a manipulation check showed that participants were anticipating the hypothetical person to feel predominantly anxiety (mean � 4.33, SD � 1.09), attention/interest (mean � 3.86, SD � .90), and fear (mean � 3.66, SD � 1.08) (all other means under 2.56). The questionnaire was comprised of two similar sections addressing the evolution of emotional intensity in a general thinking condition and a specific thinking condition, respectively. The order of these two first sections was randomized. In these sections, participants were asked to report how the intensity of the emotional feeling state of the hypothetical person would evolve, once in a definite state of apprehension, if she or he was thinking about his or her worries concerning the talk for 10 minutes. For the general thinking section, participants were asked to “imagine that the person would think of the speech for 10 minutes, keeping in mind his/her worries, in a general form, without trying to analyze or to detail them.” For the specific thinking section, participants were asked to “imag- ine that the person would think of the speech for 10 minutes, imagining his/her worries with the greatest possible details, imagining in details all the scenarios and different emotions that could arise.” They were asked to report how the intensity of the emotional feeling state of the hypothetical person would evolve, minute after minute, by indicating on a chart 10 intensity ratings (one for each minute) on an 11-point scale, anchored at 0