فرآیندهای رمزگذاری در اضطراب اجتماعی

According to current theories, memory processes play an important role in the maintenance of social fears. However, the empirical evidence regarding memory processes in social anxiety is controversial, and little is known about specific memory processes, such as encoding. The present study employs a release from proactive interference (RPI) technique to explore encoding-related processes in social anxiety. Eighty-four high and low socially anxious college students participated in the RPI task. The main hypothesis was that RPI effects that involve socially threatening words are greater in high than in low socially anxious subjects. Contrary to this hypothesis, however, greater RPI effects were found in low rather than in high socially anxious subjects if a social threat dimension was encoded. This suggests that low socially anxious individuals show more specific encoding strategies of threatening information than high socially anxious individuals.

Several cognitive models assume that anxious individuals process information in a biased manner (e.g., Beck & Clark, 1997). Two reviews specifically focusing on social anxiety and social phobia (Amir & Foa, 2001; Heinrichs & Hofmann, 2001) concluded that there is some (but mixed) support for the idea that high socially anxious students show an explicit memory bias for socially threatening words, while there is little persuasive evidence that such memory biases favoring recall of lexical socially threatening cues occur in individuals with social phobia. One reason for this conflicting evidence may be how participants are selected for participation in these studies; often individuals with high social anxiety are defined as students yielding high scores on social anxiety measures. However, in order to clarify differences and similarities among social anxiety and its clinical expression, social phobia, it may be necessary to separate moderate-to-high socially anxious individuals from those who also meet criteria for social phobia. For memory biases found in high socially anxious students, this differentiation has rarely been made, in contrast to studies with social phobia. In addition, a variety of other questions remain regarding high socially anxious samples. For example, it is still unclear whether the bias is related to attentional, encoding, or retrieval processes. The present study focuses on encoding processes. We distinguish between attention and encoding for several reasons (Heinrichs & Hofmann, 2001): Attention determines the quality and quantity of encoding; however, encoding is also a “top-down process” in which already existing memory organization impacts upon encoding processes. Therefore, encoding processes mediate between attentional and memory processes. Looking specifically at encoding may add to our knowledge about the presence (or absence) of a memory bias among individuals with a high level of social anxiety who are not at a clinical stage of social phobia. In addition, a well-known cognitive model of social anxiety (Clark & Wells, 1995) provides another rationale for focusing on encoding processes. The authors proposed that information about social threat was particularly well encoded in an individual's memory due to repeated processing of anxious feelings and negative self-perception prior to, during, and after social situations (Clark & Wells, 1995). As such, information about social threat should also be particularly well retrieved. However, not all theories predict such a bias. Russo, Fox, Bellinger, and Nguyen-Van-Tan (2001) pointed out that while Beck et al.'s model of anxiety (Beck, Emery, & Greenberg, 1985) states that an activated anxiety-related schema should result in better encoding of threat-related information, other models of information processing (Williams, Watts, MacLeod, & Mathews, 1997) predict the absence of such a mood-congruent explicit memory bias due to less elaborative encoding of threat in anxiety. Coles and Heimberg (2002) reviewed the data and concluded that evidence for an explicit memory bias was rare in social phobia. Despite these empirical observations, Russo et al. (2001) outline in detail that from a theoretical point of view, an explicit memory bias should be observable in anxiety. They argue that in the majority of studies in which high-trait anxiety individuals or generalized anxiety disorder patients were tested, tasks were used that promote some form of deep semantic encoding (e.g., self-referred tasks or liking orienting task). Tasks that promote deep semantic encoding, however, leave little room for emotional factors to impact recall because efficient encoding is facilitated under such conditions. Russo et al. consequently suggest that a detection of an explicit memory bias may be more likely if tasks were set up in a way that efficient encoding of targets would not be promoted, such as in a lexical-orienting task. The authors conducted two studies with high-trait anxious individuals. Participants were told to count the number of syllables in each target word. At the end of this incidental learning period, participants were asked to recall the words. The authors hypothesized that highly anxious individuals, in contrast to individuals low on anxiety, should be diverted by the threat-related meaning of the word and while dwelling on the meaning, deeper processing should occur. They found the predicted encoding bias for high-trait anxious individuals in the first and second experiment. Moreover, they also showed in the second experiment that this bias was absent when deeper semantic processing of the targets was allowed. This study suggests that the distinction of explicit/implicit memory bias may be less important than the depth of encoding. Some of the studies with socially anxious individuals are, however, still at variance with Russo et al.'s suggestion. Sanz (1996), for example, employed an incidental lexical-orienting memory task and did still not find a recall bias among high socially anxious individuals. In light of the discrepant results in memory processes in social anxiety and the relevance of encoding processes for this outcome, the primary purpose of the present study was to apply a cognitive paradigm to studying encoding processes and semantic organization in social anxiety. We chose the release of proactive interference paradigm (RPI; Wickens (1970), Wickens (1972) and Wickens (1973)) as a technique to study these processes. This method entails the presentation of a successive series of word-triads (e.g., “chair,” “table,” and “bed”) with different words from the same category (e.g., “furniture”). The participant's task is to repeat each word-triad and to recall them after having been engaged in a distractor task to suppress rehearsal processes. On the next trial, participants may receive “painting,” “desk,” “bench,” and so on. Typically, recall performance declines across several trials of word-triads stemming from the same class, and reaches an asymptote after the third or the fourth trial. This effect is attributed to proactive interference (PI) from the previous words. However, when the class of items is changed on the fourth trial (e.g., from “furniture” to “body parts,” such as “eye,” “noise,” and “ankle”), recall performance recovers. This recovery effect is called RPI. RPI can be used as an index of the encoding dimensions that are used (Wickens, 1972) because the stimulus dimensions that produce RPI when the stimulus class is changed represent the processing dimensions used by participants to organize the information. Thus, this technique provides an indirect investigation of semantic memory and its organization. Moreover, it is assumed that RPI reflects automatic rather than strategic memory processes because participants often fail to notice the similarities among the target items (Guttentag, 1985; Wickens, 1970). Wickens et al. have used this technique frequently in cognitive science investigations to explore potential encoding dimensions (cf., Wickens, Born, & Allen, 1963; Wickens & Engle, 1970). In sum, they found that a variety of attributes are used to encode semantic information, including semantic changes such as from words to numbers, or changes in taxonomic categories, such as from “furniture” to “fruits” (for a review, see Wickens, 1973). The phenomenon of release of PI has also been examined in other fields of psychology, including developmental psychology (e.g., Bjorklund, Smith, & Ornstein, 1982; Dobbs, Aubrey, & Rule, 1989) and neuropsychology (e.g., Belleville et al., 1992; Binetti et al., 1996; Rößner, Rockstroh, Cohen, Wagner, & Elbert, 1999) with various patient populations. Whereas a number of experiments that used patients with dementia failed to show PI and RPI (e.g., Belleville et al., 1992), other studies involving patients with schizophrenia did report RPI effects (Traupmann, Berzofsky, & Kesselman, 1976). These effects have previously been interpreted as evidence to suggest that individuals encode the attributes of the word category in memory, and that decreased recall performance is not the result of an encoding deficiency (Traupmann et al., 1976). The absence of RPI effects seems to imply that the encoding dimensions in question were irrelevant for semantic processing (Wickens, 1972). In studies examining individual differences, a lack of RPI is typically attributed to a semantic encoding deficit (e.g., Belleville et al., 1992). To our knowledge, the RPI paradigm has not been applied to the study of anxiety and the various manifestations of anxiety disorders, although this paradigm seems to be well suited for studying memory biases in anxiety. The aim of the present study was to investigate encoding processes in social anxiety. For social anxiety, a socially threatening word should be of greater psychological relevance than a physically threatening word because social threat reflects the core of the anxiety. We assumed that individuals with social anxiety more selectively employ the social cues as an encoding dimension than individuals with low social anxiety. This should create more PI for the social than for physical or neutral threat category in high versus low socially anxious subjects, and there should be a greater release from PI when a category shift occurs towards or away from social threat in the high social anxiety group compared to the low social anxiety group.2 We employed four different semantic meanings: social threat, physical threat, emotionally positive, and emotionally neutral meanings. We expected that RPI should be greater for high socially anxious individuals (as compared to low socially anxious people) when a shift occurs from socially threatening words to physically threatening words or vice versa, because socially threatening words are members of the same threat category, which should have a particular meaning and personal significance to the high socially anxious individuals. However, we expected this effect to be weaker compared to effects anticipated when emotionally positive or neutral words are involved in the shift because physical threat is likely to be still more relevant to social anxiety than positive or neutral meanings due to the frequent fear of certain physical symptoms in social anxiety (such as blushing and trembling in front of others). We have attempted to specifically employ words that are rather typical of physical symptoms in panic disorder than of those in social phobia to limit the overlap between social and physical threats.