Approach and avoidance in fear of spiders

https://doi.org/10.1016/j.jbtep.2006.10.001Get rights and content

Abstract

We examined attitudes towards spiders by employing an Approach-Avoidance Task, in which participants respond to pictures by pulling a joystick towards themselves or by pushing it away from themselves. For spider fearfuls, this stimulus–response assignment is either compatible (push spiders away) or incompatible (pull spiders closer). Specific compatibility effects were found: compared to non-anxious controls and control pictures, highly spider fearful participants responded to spider pictures more quickly by pushing than by pulling, even when picture contents was task-irrelevant. Moreover, compatibility effects predicted fear-related behavior independently of questionnaires. Potential applications, extensions, and limitations of the findings are discussed.

Introduction

Is it possible to assess someone's attitudes without asking him or her explicitly? Are we able to measure people's fear of animals or social situations without questionnaires or interviews? During the past few years, a whole new class of tasks have been developed in order to examine these questions. These indirect measures of attitudes promise to assess attitudes and associations in a variety of domains, in the hope that they are less affected by problems associated with direct measures (questionnaires and interviews), such as social desirability. The new measures have been termed “indirect” because participants are not explicitly asked to reveal their attitudes. Instead, the attitudes are inferred from seemingly unrelated responses. In most of these tasks, reaction times (RTs) are measured as the dependent variable. The common principle of the tasks is that the participants’ response speed (and/or accuracy) in a categorization task is affected by the compatibility between the response and the valence of the stimuli (see De Houwer, 2003a). If the two are incompatible, responses are slowed down, for instance, if one has to respond positively to a threatening stimulus. This principle has been applied to various clinical populations using a number of indirect tasks (e.g., Ellwart, Becker, & Rinck, 2005; Ellwart, Rinck, & Becker, 2006; Huijding & de Jong, 2006; Teachman, Gregg, & Woody, 2001). These studies and many others indicate that indirect measures may be a helpful tool in the study of anxiety disorders, because they promise to assess the structure and state of specific threat-related associations.

A limitation of these tasks is that they mainly tap into the semantic aspects of emotional information processing. Emotional reactions, however, consist of a much more complex pattern of responses, involving behavioral and physiological responses in addition to cognitive ones (e.g., Lang, 1994). In fear and phobias, this is particularly relevant: Fear activates cognitive representations such as “I am afraid of spiders” or “spiders are dangerous”, but also physiological reactions of the sympathetic nervous system (e.g., increased heart beat) and a behavioral tendency to avoid the threatening stimulus (e.g., running away from a spider). Thus, the studies mentioned in the previous paragraph indicate that spider fearful individuals do indeed associate spiders more strongly with negative valence than non-anxious individuals do. However, they tell us little about the processes involved in behavioral responses, that is, whether and how spider fearfuls differ from non-anxious individuals in approach-avoidance reactions. To assess negative attitudes by means of these affective behavioral reactions, the Approach-Avoidance Task (AAT) was designed. In this task, single stimuli are presented to the participants on a computer screen, for instance pictures of animals. In one version of the task (employed in Experiments 1 and 2), there are two types of stimuli, namely negatively valenced ones and approximately neutral ones (e.g., spider pictures and pictures showing an empty background). The participants’ task is to respond to each stimulus by means of a joystick connected to the computer. They are instructed to pull the joystick towards themselves whenever one stimulus type is presented, and to push the joystick away from themselves whenever the other type is presented. While the joystick is moved, the size of the picture changes: pulling the joystick makes it larger, while pushing makes it smaller. This creates the visual impression that the picture is pulled closer or pushed away, respectively.

Depending on the combination of response direction and stimulus valence, the stimulus–response assignment is either compatible (pull spider-free and push spiders) or incompatible (push spider-free and pull spiders). If response times in the compatible condition are shorter than in the incompatible condition, a compatibility effect is said to occur, suggesting the existence of a negative attitude towards spiders, reflected in a behavioral avoidance tendency. Moreover, the size of the compatibility effect should be related to the strength of the negative attitudes, such that spider fearful individuals should show larger compatibility effects than non-fearful individuals. According to the taxonomy suggested by De Houwer (2003a), this compatibility effect may be characterized as a traditional stimulus–response compatibility effect. Moreover, according to the definition mentioned above, the task is an indirect one because it aims to assess attitudes without asking participants to explicitly state their attitude. We use the general term “negative” here because it is unclear which negative emotion drives the compatibility effect (e.g., fear, disgust). We will get back to this topic in the Section 5.

To the best of our knowledge, this is the first attempt to apply an AAT to anxiety disorders. However, the predictions outlined above follow directly from previous studies of non-clinical populations. The reasoning is based on a rapidly growing body of evidence showing that humans usually show a spontaneous avoidance reaction to unpleasant, threatening stimuli, and a spontaneous approach reaction to pleasant stimuli. One way to observe these approach-avoidance reactions in overt behavior is by means of arm movements: avoidance is associated with pushing unpleasant objects away from oneself, and therefore with moving the arms away from one's body. In contrast, approach to pleasant objects is associated with pulling the objects closer, and therefore with moving the arm towards the body. Indeed, starting with Solarz (1960), it has been found several times that participants respond to aversive stimuli more quickly with a “push” movement of the arm, whereas they respond to pleasant ones more quickly with a “pull” movement (e.g., Chen & Bargh, 1999; Marsh, Ambady, & Kleck, 2005). In fact, the close relation between emotional valence of stimuli and arm movements seems to be quite strong, and it is also bi-directional: arm flexion and arm extension also affect evaluative reactions to stimuli (for a review of the latter, see Neumann, Förster, & Strack, 2003).

Our hypotheses are also compatible with results reported by De Houwer and his colleagues (De Houwer, Crombez, Baeyens, & Hermans, 2001; Mogg, Bradley, Field & De Houwer, 2003), who used a more symbolic AAT. In this task, participants pressed keys to make a manikin on the screen move toward or away from stimuli. De Houwer et al. (2001) found that an unselected group of participants needed more time to make the manikin on the screen approach negative words and run away from positive words than to do the opposite. Mogg et al. (2003) used a similar procedure to compare smokers to non-smokers. They found that smokers, when compared to non-smokers, were faster when instructed to make the manikin approach smoking-related pictures and move away from smoking-unrelated pictures. These results were found with an AAT that did not involve arm flexions and extensions. Nevertheless, the results are relevant to our questions of interest: They show that AATs may be suitable for comparisons of clinical groups to control groups. Thus, it seems safe to assume that strong, bi-directional links exist between the threat value of stimuli and the spontaneous, avoidant motor responses they evoke in fearfuls and phobics. In the experiments reported here, we took advantage of these pre-existing links: we investigated whether the observed compatibility effects differ between fearful individuals and non-anxious ones, and whether they predict their behavior in response to real threat stimuli. In doing so, our theoretical goal was to gain more insight into the representations and mechanisms involved in anxiety disorders, particularly in the attitudes of phobic patients. In addition, our practical goal was to predict variance in fear-related behavior that questionnaires do not predict.

Section snippets

Experiment 1: Approach-avoidance behavior in spider fearfuls

In Experiment 1, pictures of spiders as well as “empty” pictures showing only background, but no animals, were presented, and participants pulled or pushed depending on whether a spider was visible or not. To determine whether the effects observed here are disorder-specific, we tested a group of highly spider fearful individuals (half of whom were spider phobic), and compared their responses to those of a non-anxious control group. Moreover, we validated the observed compatibility effects by

Experiment 2: Cognitive re-interpretation of arm movements

Experiment 1 revealed differences in approach-avoidance tendencies between spider fearfuls and non-anxious controls. These results are promising, but their theoretical interpretation is equivocal. This is the case because the participants’ arm movements (and the accompanying joystick movements) may be interpreted in two fundamentally different ways. In Experiment 1, they were described to each participant with reference to his or her own body: arm flexions were labeled as movements toward the

Experiment 3: Approach and avoidance in response to an irrelevant stimulus feature

Although the previous experiments revealed avoidance tendencies in highly spider fearful individuals, several problems remain. Most importantly, the instructions referred to the stimulus dimension of interest, that is, whether a stimulus shows a spider or not. This is unfortunate because it may have directed participants’ attention to the research question addressed, introducing the possibility to use response strategies, just as in questionnaires. To solve this problem, Experiment 3 employed a

General discussion

Taken together, the three experiments reported here provide evidence for the claim that negative attitudes towards spiders in spider fearfuls are related to behavioral avoidance tendencies, which can be measured by means of a simple joystick task. In all experiments, spider fearfuls showed specific spider-related compatibility effects: compared to non-anxious controls and control pictures, they responded to spider pictures more quickly by pushing than by pulling. In addition, Experiment 1

Acknowledgments

Preparation of this paper was supported by grant RI 600/6-1 from the German Research Foundation (DFG) to Mike Rinck and Eni Becker. We would like to thank Andrea Reinecke, Kathrin Heuer, Lidia Kolanko, Susann Klötzer, Gero Lange, Joyce Lutgens, Spence van Melis, and Silvana Müller for their help in testing the participants, Frank Leonhardt for programming numerous versions of the experimental software, and the reviewers for their comments. We are also grateful to Jan De Houwer for many helpful

References (19)

There are more references available in the full text version of this article.

Cited by (520)

  • The empirical status of predictive coding and active inference

    2024, Neuroscience and Biobehavioral Reviews
View all citing articles on Scopus
View full text