Comments on Andrade et al., and Kavanagh et al.

In this post, I will focus on the Andrade et al. (1997) and Kavanagh et al. (2001) studies. Louise has proposed that these two studies, along with the Sharpley et al. (1996) and van den Hout (2001) studies which I will address in other posts, “consistently demonstrated that EMs [eye movements] have a significant desensitization effect that is greater than that of various control conditions, including exposure.”

In this post, I am going to argue two main points. First, the results of the Andrade et al. (1997) study are open to at least two different interpretations: Desensitization on the one hand, and distraction on the other hand. Second, the results of the Kavanagh et al. study, which replicate and extend the Andrade et al. results, favor the distraction hypothesis.

Given that I have introduced a distinction between two different mechanisms for anxiety reduction, it is incumbent on me to specify how they are different and what kind of test would distinguish one type form of anxiety reduction from the other. For the purpose of this discussion, distraction produces anxiety reduction only while the distractor is present, such that fear/anxiety in response to a stimulus (or any other emotion for that matter) will still be present when the stimulus (cue, target, trigger) is confronted in the absence of the distractor. Desensitization, on the other hand, represents a kind of anxiety/fear reduction that transfers across situations such that anxiety reduction that is achieved via desensitzation remains even when the stimulus is tested in the absence of the desensitization procedures.

To further illustrate this distinction, consider the following hypothetical experiment in which there are three groups that are subjected to a three-phase procedure. In phase 1, a group of, say, spider phobics are presented with a spider. We test them for their willingness to approach the spider and their fear when they do so . Because all subjects are treated the same at this point, and subjects have been randomly assigned to conditions, we expect no differences between groups at this point. We will call phase 1 the pre-treatment assessment. Now, subjects undergo their experimental treatment. One group will get in vivo exposure to spiders (exposure only condition). Another group will get in vivo exposure but will be instructed to divide their attention during exposure between attending to the spiders and attending to some neutral task, like playing packman (distracted exposure). Every five minutes during the exposure or distracted exposure, subjects will provide ratings of fear/anxiety. We expect that both groups will show reductions in anxiety over the course of the exposure session, but we hypothesize that different mechanisms are occurring in each group. Specifically, it is hypothesized that desensitization in the distracted exposure condition is reduced to the extent that distraction is effective. The third group will be given a filler task that does not involve spiders for the same amount of time (control). Phase 3 is the critical test. All subjects are again treated exactly as they were treated in phase 1. In other words, they are tested for their willingness to approach the test spider and their level of fear in the presence of the spider without engaging in the distracting task. We call this the post-treatment assessment. What are the predicted outcomes? First, we expect that fear will decline from pre- to post-treatment for all groups simply due to repeated testing. Moreover, we expect that fear reduction will be greater in the exposure only condition than in either distracted exposure condition or the control condition. The distracted exposure condition would be predicted to fall in between the two, depending on how effective the distractor was: The more effective the distractor at reducing fear in phase 2, the less fear reduction (or the more fear) at the post-treatment assessment.

Why do I make these distinctions and differential predictions? Based on the theory of emotional processing advanced by Foa & Kozak (1986), I advance the idea that fear behavior reflects the activation of a cognitive network that includes interconnections between representations of the feared stimuli, responses made in reaction to the stimulus, and the meaning of the feared stimuli and responses. Many of the connections in a fear network are inaccurate, in that they do not accurately reflect reality. For example, a dog phobic has beliefs (meaning representations) that dogs are dangerous and are likely to bite people. This is inaccurate in that most dogs that we come across in daily life are domesticated and are not dangerous and do not bite without provocation. Successful treatment requires that the fear structure be activated and then disconfirming evidence be made available so that the fear structure can be updated. Thus, exposure to dogs in therapy would activate the structure, and the experiences of being with dogs that done bite gets incorporated into the memory. Thus, after exposure therapy is done, the person is less afraid of dogs because exposure to dogs outside of therapy will activate the revised network which includes information about dogs being safe. Distraction procedures will reduce fear while the feared stimulus is present because it reduces attending to the fear stimulus and thereby reduces activation of the fear network. Less activation, less fear behavior. However, by virtue of limiting activation of the network, it prevents the network from changing because (1) the theory assumes that a network must be activated in order to be changed, and (2) distraction from the fear stimulus prevents people from attending to the fact that feared consequences do not occur (i.e., detract from processing the disconfirming evidence).

Please note, there is independent evidence for the basic assumptions of this theory. For example, higher levels of fear during exposure therapy, which theoretically indicates that the network is activated, are empirically correlated with treatment outcome such that higher fear is associated with BETTER outcome. Moreover, there is evidence that distraction reduces fear while the distractor is present, but reduces fear reduction during the post-treatment test when the distractor is no longer present.

Now, onto evaluating the Andrade et al. and Kavanagh et al. studies.

Andrade et al. conducted four studies utilizing non-clinical subjects to investigate the effects of eye movements and various control conditions on image vividness and emotion ratings utilizing neutral and negative stimuli. Eye movements were induced by telling subjects to track a blinking letter presented on a computer screen that alternated locations left to right. The neutral and negative stimuli in Experiments 1-3 were a series of photographs. The photographs were also used in Experiment 4, which compared the photographs to autobiographical memories for happy and fearful memories. The general procedure in the eye movement condition for the emotional photographs involved the subject looking at a photograph for 5 seconds. The photograph was then removed and subjects were instructed to form an image of the photograph while engaging in the eye movement (or control) task for 8 seconds. Subjects then provided ratings of image vividness and emotionality on scales ranging from –10 to 10. In the autobiographical memory condition in Experiment 4, the subjects were cued to “recall the time when…” and then they engaged in the eye movement or control tasks. All four studies utilized a within-group design in which all subjects were tested under all conditions and different photographs/memories were used for each condition.

Results of Experiment 1: Eye movements during imagery reduced the vividness of the image for both the neutral and negative conditions from an average of 6.58 in the control condition to 6.39 in the eye movement condition. Eye movements reduced positive affect to the neutral images from an average of 2.76 (control condition) to 2.46 (eye movement condition), but had no effect on emotion ratings in response to the negative images. Numerically, emotionality was GREATER during the eye movement condition in response to the negative stimuli (-5.31) than the control condition (-5.19), although this difference was not significant. Summary: Eye movements had a small but statisticaly significant effect on reducing image vividness and a small but statistically significant effect in reducing positive affect, but no effect on reducing negative affect.

Results of Experiment 2: Eye movements decreased image vividness from an average of 7.93 in the no eye movement condition to 7.08 in the eye movement condition. Image vividness in the eye movement condition was also less than in a verbal counting condition (avg. of 7.87). On emotion ratings, the eye movement task was significantly less intense than the verbal counting, but there was no significant difference between the eye movement and no eye movement condition on emotional response. In summary, eye movements had a small but significant effect in reducing image vividness, but had no effect, relative to the no eye movement control group, on emotional ratings.

Results of Experiment 3: This compared a simple tapping task with a complex tapping task and a no tapping task. Results found the complex tapping task reduced image vividness (7.4) compared to no tapping (7.8). Complex tapping had no effect on ratings of emotionality. The authors summarized things thus far: “We have yet to demonstrate clear effects on emotional responses to images” (p. 217).

Results of Experiment 4: This study compared eye movements, complex tapping, and no eye movements using both the photograph stimuli and autobiographical memories. The eye movement condition had lower vividness ratings (avg. = 6.96) than either tapping (avg. = 7.55) or no eye movements (8.07). Eye movements had no effect on emotion for the neutral photograph condition. Emotion ratings were significantly less in the negative photograph condition for both tapping (-5.03) and eye movements (-4.83) compared to no eye movements (-5.89). For the memory condition, eye movements reduced image vividness and emotional ratings compared to tapping and no eye movements.

My own take on this series of studies is that there probably is a small effect of eye movements on image vividness and a possibly a small, and less consistent, effect on emotionality. Thus, eye movements seem to degrade the image somewhat. However, this does not mean it is a desensitization effect. Recall, the concept of desensitization implies some kind of longer-term change such that if the person subsequently imagines the previously desensitized image in the absence of the eye movements (or other task, such as tapping), they will be less anxious then they used to be. Although the Andrade et al. study assessed everyone under all conditions, they don’t report the data in such a way as to be able to test the desensitization vs. distraction hypotheses. Rather, the effects observed in this study could be attributed to distraction as opposed to desensitization. Moreover, this interpretation of the effects of eye movements as producing distraction is very consistent with Andrade et al.’s theory of the visual-spatial sketch pad. Specifically, they assume that visual attention is a capacity limited process such that attentional resources deployed to the distractor cannot be deployed to process the fear stimulus. Thus, the fear stimulus has less impact while the distractor is present. However, Andrade et al. do not specifically discuss the implications of reduced processing of the fear stimulus during EMDR on long-term outcome.

The Kavanagh et al. study utilized positive and negative autobiographical memories for stimuli and used a design that better allowed for assessing the distraction vs. desensitization question. They compared eye movements, no eye movements, and a “visual noise” condition. The study had three phases. First there was a baseline condition in which subjects engaged in imagery without any eye movements or white noise. Then they had several trials of imagery under each of the three conditions. The third phase, one week later, was a post-treatment assessment in which subjects engaged in imagery without any eye movements or visual noise. Results for both image vividness and emotion intensity followed a similar pattern. There were no differences between conditions at the baseline assessment before eye movements or visual noise was introduced. Both measures declined when the eye movements were introduced but not in the control conditions. Now, this can be attributed to either distraction or desensitization. The critical test is in phase 3, when subjects engaged in imagery in the absence of eye movements or visual noise. If desensitization occurred, one would predict that emotion intensity would remain low in the eye movement condition. However, if eye movements only served to distract subjects from the image, then engaging in imagery in the absence of eye movements would be expected to result in a return of image vividness and emotional reactions. Results provided evidence for both desensitization for all groups, and a unique effect of eye movements that was restricted to trials in which eye movements occurred. Although eye movements reduced image vividness and emotion intensity while they were being administered, there were no differences between groups at the post-test. All groups were less emotional at the post-test compared to baseline, but there were no differences among groups. The decrease from baseline to post-test was significant, but small.

Now, to connect these findings to EMDR research. A very similar pattern was observed in the Boudewyns et al. (1993) dismantling study. SUDs ratings obtained during treatment decreased more for the EMDR condition than the no eye movement condition and therapists were more likely to rate subjects as “treatment responders” in the eye movement condition. However, when patients engaged in imagery to the target memory in the absence of eye movements at the post-test, there were no differences between groups. D. Wilson et al. (1996) also reported an effect of eye movements on SUSDs and skin conductance observed during treatment. However, this study did not assess imagery in the absence of eye movements after treatment and did not collect any treatment outcome data that would permit an assessment of whether or not the eye movements had any effect beyond the treatment session.

The studies of EMDR for phobias can also be explained this way. Granted that they used a truncated protocol, the three studies by the Muris, Merckelbach, et al. all found that SUDs decreased during EMDR and VoC of the positive cognition increased. Yet, when they were tested on a behavioral approach test (BAT) with a spider, subjects receiving EMDR were still very avoidant and fearful. Notice the pattern: Fear reduction in session that does not transfer to confrontation with the feared stimulus in the absence of distracting activities/stimuli. By contrast, in vivo exposure without distraction resulted in reduced fear and avoidance at the post-test

Thus, while the results from the Andrade et al. and Kavanagh et al. (and Boudewyns et al., 1993, for that matter) studies indicate that eye movements may reduce anxiety while the technique is implemented, there is no evidence from these studies that eye movements contribute to outcome. Incidentally, this is essentially the same conclusion I drew in my Journal of Anxiety Disorders review back in 1999 (Cahill, Carrigan, & Frueh, 1999).

Replies:

Re:Comments on Andrade et al., and Kavanagh et al., by client, 11/07/02 Re:Comments on Andrade et al., and Kavanagh et al., by Cahill, 11/07/02 Re:Comments on Andrade et al., and Kavanagh et al., by client, 11/07/02 Andrade, Kavanagh and working memory, by Louise Maxfield, 11/09/02

| Behavior OnLine Home Page | Disclaimer |

Copyright © 1996-2004 Behavior OnLine, Inc. All rights reserved.