What has research determined about EMDR's eye movement component?

In 1989, Francine Shapiro (1995) noticed that the emotional distress accompanying disturbing thoughts disappeared as her eyes moved spontaneously and rapidly. She began experimenting with this effect and determined that when others moved their eyes, their distressing emotions also dissipated. She conducted a case study (1989b) and controlled study (1989a), and her hypothesis that eye movements (EMs) were related to desensitization of traumatic memories was supported. The role of eye movement had been previously documented in connection to cognitive processing mechanisms. A series of systematic experiments (Antrobus, 1973; Antrobus, Antrobus, & Singer, 1964) revealed that spontaneous EMs were associated with unpleasant emotions and cognitive changes.

There have been 20 published studies that investigated the role of EMs in EMDR. Studies have typically compared EMDR-with-EMs to a control condition in which the EM component was modified (e.g., EMDR-with-eyes-focused-and-unmoving). There have been four different types of studies: (1) case studies, (2) dismantling studies using clinical participants (3) dismantling studies using nonclinical analogue participants, and (4) component action studies in which eye movements are examined in isolation.

Case studies. Four case studies evaluated the effects of adding EMs to the treatment process, and three demonstrated an effect for EMs. Montgomery and Ayllon (1994) found eye movements to be necessary for EMDR treatment effects in five of six civilian PTSD patients. They wrote that the addition of the eye movement component “resulted in the significant decreases in self-reports of distress previously addressed. These findings are reflected by decreases in psycho-physiological arousal” (Montgomery & Ayllon, 1994, p. 228). Lohr, Tolin, and Kleinknecht (1995) reported that “the addition of the eye movement component appeared to have a distinct effect in reducing the level of [SUD] ratings” (p. 149). When Lohr, Tolin and Kleinknecht (1996) treated two claustrophobic subjects, substantial changes in disturbance ratings were achieved only after EMs were added to an imagery exposure procedure that used the brief frequent exposures of EMDR. The fourth study (Acierno, Tremont, Last, & Montgomery, 1994) did not use standard EMDR protocol for phobias, nor the standard procedures for accessing the image, formulating the negative belief, or eliciting new associations. In addition, the client was instructed to relax between sets of EMs until the SUD rating was reduced to baseline, a procedure not used in EMDR. The procedures used in this study did not eliminate the phobia and no effect was found for the EM condition.

Clinical dismantling studies with diagnosed participants. There have been four controlled dismantling studies with PTSD participants, and two studies where participants were diagnosed with other anxiety disorders. These studies have tended to show that EMDR-with-EMs was slightly better than EMDR-with-modification; however such comparisons have not usually been statistically significant, and results are equivocal. For example, Devilly et al. (1998) reported rates of reliable change of 67% for the EM condition, compared to 42% of the non-EM condition; Renfrey and Spates (1994) reported a decrease in PTSD diagnosis of 85% for EM conditions and 57% for the non-EM group. These studies unfortunately are limited by severe methodological problems, including inadequate statistical power. For example, there were seven or eight persons per condition in the Renfrey and Spates (1994) PTSD study. The participants in the other three PTSD (Boudeywns & Hyer, 1996; Devilly et al., 1998; Pitman et al., 1996) studies were combat veterans, who received only two sessions or treatment of only one traumatic memories. Such an inadequate course of treatment produced only moderate effect sizes; therefore a large sample would be required to provide adequate statistical power for the detection of any possible differences between groups. There has yet to be a single rigorous dismantling study with a sample adequate to assess treatment effects.

Clinical dismantling studies with analogue participants. The controlled studies that used analogue participants with nonclinical anxiety found no effect for EMs. There are many problems with these analogue studies, which typically used normal college student participants. The EMDR protocol was often truncated (e.g., Carrigan & Levis, 1999; Sanderson & Carpenter, 1992), resulting in poor construct validity and making interpretation of results problematic. It is also unlikely that the responses of analogue participants can be generalized to persons with chronic PTSD, a disorder that appears resistant to placebo effects (Solomon, Gerrity, & Muff, 1992; Van Etten & Taylor, 1998). Analogue participants responded well to EMDR-without-EMs, a procedure which contains a number of active components. The minimal distress of the analogue participants was relieved with minimal treatment, and the assessment of differences between the EM and nonEM conditions was limited by a floor effect. Consequently it may not have been possible to detect differences between conditions.

Component action studies. Component action studies test EMs in isolation. These studies typically provide brief sets of EMs (not EMDR) to examine their effects on memory, affect, cognition, or physiology. The purpose is to investigate the effects of moving the eyes (not EMDR), and EMs are compared to control conditions such as imaging and tapping. For example, a participant might be asked to visualize a memory image, then to move their eyes for a brief period ,and then to rate the vividness of the image. This permits a pure test of the specific effects of EMs and non-EMs without the added effects of the active ingredients of the other EMDR procedures. The studies have generally used nonclinical participants and a within-subject design, that compares the differences in each individual’s responses to the various conditions. This reduces the variance of subjective responding, and eliminates possible floor effects.

Findings from these studies suggest that EMs may have an effect on physiology, decreasing arousal (e.g., Barrowcliff et al., in press; D. Wilson et al., 1996) on attentional flexaility (Kuiken, Bears, Miall & Smitth (2001-2001) and on memory processes, enhancing semantic recall (Christman et al., in press). Four studies (Andrade, Kavanagh, & Baddeley, 1997; Kavanaugh, Freese, Andrade, & May, 2001; Sharpley et al., 1996; van den Hout, Muris, Salemink, & Kindt, 2001) have demonstrated that EMs decrease the vividness of memory images and the associated emotion. No (or minimal) effect has been found for tapping conditions. These studies suggest that EMs may make a contribution to treatment by decreasing the salience of the memory and its associated affect. (See discussion below on mechanisms of action).