Is eye-movement desensitization achieved through effects on memory or affect?

Is eye-movement desensitization achieved through effects on

memory or affect?

Ricardo Lieuw On

University of Amsterdam

Student number:

10381953

Mentors :

Hans Phaf & Alexander Krepel

Amount of words:

7803 (incl. voorblad en literatuurlijst)

2

Index

Abstract………3

Introduction………..4

Method………..9

Materials………... 11

Results……… 14

Discussion……….. 23

References………. 30

3

EMDR-therapy is increasingly and effectively being used for the treatment of PTSD. Its mechanisms however remain elusive, though eye movements (EM) were shown to be an effective component. Surprisingly, almost all relevant research has looked into effects of EM on memory, while little to no research has looked at effects on emotion. This study

investigated the affective regulation hypothesis, which proposes that eye movements work by specific down-regulation of negative affect, and increased dopamine production leading to enhancement of memory for negative stimuli and effective reconditioning of the memory with this reduced negative affect. The working memory hypothesis was also investigated, which proposes that EM leads to reduced WM capacity, resulting in reduced memory and

emotionality. Effects of eye movements on affect and implicit memory were investigated using affective priming for emotional pictures and their effect on explicit memory was investigated using a recognition task. Results showed a specific memory enhancement for explicit negative memories, but not for implicit negative memories, possibly implying a differential effect as to conscious and subconscious processing . No direct effect on affect was found. Results contradict the working memory hypothesis. The affective regulation was only partially supported, due to an absence of effects on affect, and a specific differential enhancing effect on explicit memory. To account for the findings, a new ‘coping hypothesis’ is put forth, suggesting that eye movements are efficacious by means of increased

production of dopamine, resulting in a shift from reactive/avoidant coping with negative stimuli to more proactive/approach coping.

4

Emerging from a sea of controversy, eye-movement desensitization and reprocessing (EMDR) is a relatively new form of therapy, which is increasingly and effectively being used for the treatment of post-traumatic stress disorder and anxiety based disorders (van den Hout & Engelhard, 2012). However, it is still not very clear which attributes of the treatment contribute to its effectiveness, and perhaps even more elusive is how these factors

contribute. Part of the therapy requires the patient to produce bilateral eye-movements by following the back and forth motion of the therapist’s hand (Engelhard, van Uijen & van den Hout, 2010). There is still some debate concerning the role of these eye-movements as to whether or not they are an effective component, but a recent meta-analysis suggests it is (Lee & Cuijpers, 2010). Up until now research has mainly investigated the effects of bilateral eye-movements on memory. Despite EMDR’s use as an intervention for disorders

characterized by emotionally charged episodes, surprisingly little research has attempted to shine light on the effects of these movements pertaining directly to emotion. Elucidating these effects on memory and emotion is of great clinical significance. For example, a direct effect on emotion would explain and justify its use for non-trauma related disorders, such as panic disorder, in which often there is no specific memory causing the emotionally charged attack. Contrarily, an effect solely on memory would deem this therapy more applicable to trauma-related disorders. In this study we investigate possible direct effects of

eye-movements on emotion and also related effects on memory.

Two currently prevailing hypotheses concerning possible mechanisms of action are the working memory hypothesis (Andrade, Kavanagh & Baddeley, 1997), and the affective regulation hypothesis (Dame, 2013; de Roo, 2012), which is based on the orienting

hypothesis (Sokolov, 1990; Armstrong & Vaughan, 1996). The working memory hypothesis posits that a decrease in emotionality and vividness of memories occurs due to competition between recall and a distraction task (Andrade, Kavanagh, & Baddeley, 1997). As both eye-movements and memories are believed to load the visuo-spatial sketchpad (VSSP) of the working memory, performing both during recall would result in less working memory capacity

5

to properly process memories, leading to reduced vividness of the memory. This reduced vividness would in turn lead to reduced emotionality regarding the memory and its impaired retrieval (Andrade, Kavanagh, & Baddeley, 1997; van den Hout & Engelhard, 2012). As both positive and negative memories would be disrupted by the dual-task, the WM hypothesis predicts a decrease in vividness and emotionality for both.

Several studies have indeed shown that bilateral eye-movements decrease vividness and emotionality for both positive and negative autobiographical memories (Van den Hout, Muris, Salemink & Kindt, 2001; Maxfeld et al., 2008). However, other studies found that eye movements reduce emotionality, but not vividness (Engelhard, Uijen & van den Hout, 2010; Lee & Drummond, 2008). This is peculiar since the WM hypothesis supposes reductions of emotionality to be mediated by reductions of vividness. Another finding that is rather puzzling in the context of the WM-hypothesis is enhanced retrieval of episodic memories due to bilateral eye-movements (Lyle, Logan & Roediger, 2012.). Under the WM-hypothesis

impaired retrieval of memories would be predicted, as memories that are less vivid should be less readily retrieved.

Enhanced memory and reduction of emotionality in absence of reductions in vividness may be more adeptly explained by the affective regulation hypothesis, which is based on the orienting response hypothesis (as put forward by MacCulloch & Feldman, 1996) and posits that eye-movements specifically reduce negative affect while simultaneously enhancing memory (Dame, 2012). Eye-movements are likened to visual scanning of the environment for danger, and thus are supposed to elicit an orienting response (OR), bringing the organism into a heightened state of arousal. When eye movements are continued in absence of danger, a down-regulation of arousal takes place. The OR may also increase dopamine production through direct connections of the superior colliculus (associated with the orienting response) with the substantia nigra (responsible for dopamine production). As dopamine is associated with positive affect as well as increased working memory capacity, this hypothesis accounts for both a reduction of negative affect and increased memory

6

performance (Dame, 2012), by increased WM memory capacity to process the memory and affective reconditioning of the memory with positive affect.

In line with the orientation response hypothesis Barrowcliff et al. (2004) found a specific reduction of electrodermal arousal for negative autobiographical memories after eye-movements when compared to positive ones. Surprisingly however, they found reduced emotionality for both negative and positive memories. The authors put forth the possibility that perhaps both a working memory mechanism and a physiological de-arousal were of effect here. If there were only a general WM effect, similar levels of de-arousal would be expected for both types of memory, therefore there must be an additional specific emotional effect. The specificity of the effect towards negative memories is accounted for by proposing that the down-regulation of negative affect of the orientation response is only elicited when there is a ‘perceived threat’ as would be the case for negative memories and not for positive memories.

Barrowcliff et al. (2004) and other researches discussed up to this point however have consequently measured emotionality and vividness by subjective self-report. Perhaps participants consciously perceive a reduction of emotionality and vividness due to a general distraction effect caused by the eye-movements. This conscious perception does not

necessarily have to reflect actual emotionality, as subjective self-reports have been known to be notoriously inaccurate at times (Mauss & Robinson, 2009). A more reliable way of

emotionality assessment would be by means of affective priming, which is an implicit method of measuring emotionality. Furthermore previously mentioned studies (Barrowcliff et al., 2014; van den hout et al., 2001; Engelhard et al., 2011), have dealt with autobiographical memories, to which a numerous amount of variables are intrinsic, each of which could possibly obscure effects. These include differences in emotional intensity, in passing of time since occurrence of the remembered event and in the degree of consolidation, to name but a few factors, all of which may render the memory more or less susceptible to whatever effect eye movements may have. Therefore if one wishes to reliably find an effect, it is important that participants are presented with stimuli for which possibly obscuring variations in said

7

variables have been controlled for in as far as possible, as was done in the following two studies mentioned, as well as the current study.

Dame (2012) researched the effects of bilateral eye-movements on emotionality and memory. In line with his affective regulation hypothesis he found that eye-movements lead to increased memory specifically for negative words in a free recall task. However he did not find any effects on emotionality. Perhaps emotional words do not elicit an emotional

response strong enough to engage the orientation response. Visual images may be a more effective means to provide the emotional intensity needed for the fight/flight system to respond. Moreover, because in imaginary reliving of traumatic memories as with PTSD visual representations are most common (Kemps & Tiggeman, 2007), it seems relevant to use more complex visual stimuli, for instance emotional pictures, to gain greater ecological validity.

While researching the effects of bilateral eye-movements on memory and emotion, de Roo (2014) used pictures instead of words and affective priming instead of subjective self-report. The results showed a specific down-regulation of negative affect compared to positive affect as a consequence of eye-movements. Though the study tentatively points to the direction of the AR-hypothesis, it was not fully able to dissect effects on emotion and memory, due to a conflation of repetition priming and affective priming in the data.

Interestingly eye-fixation lead to increased negative affect for the negative pictures. De Roo (2014) proposed that this could perhaps be explained by a repetition priming effect. During fixation participants were at liberty to rehearse the emotional pictures. This rehearsal could have led to increased memory resulting in shorter reaction times. Also there was a ceiling effect with regard to the recognition task, possibly caused by the fact that stimuli during the affective priming task remained visible until a response was given, thus giving participants control of encoding time, which may have further obscured possible discriminatory results. Due to this de Roo (2012) was not able to rule out the working memory hypothesis.

This study attempts to improve on that of de Roo (2012) and attends to the following research question: Is the desensitization effect of bilateral eye-movements best accounted

8

for by the affective regulation or by working memory load? Affective regulation will be

measured by calculating affect index scores based on the difference between reaction times for negatively and positively valenced emotional pictures during an affective priming task, and by comparing these scores before and after the experimental manipulation

(eye-movement/eye-fixation). To separate repetition priming effects during analysis, a new set of emotional pictures will be presented after the manipulation, together with a previously seen set, so that they can be compared. Both implicit and explicit memory will be measured. Implicit memory will be measured by means of an affective priming task; reaction times and error proportions for familiar stimuli will be compared to those for unfamiliar stimuli. Explicit memory will be measured by means of a recognition task, as in tandem they can provide additional evidential weight. During the recognition task yet another new set will be presented with the other previously seen set. To prevent possible ceiling-effects due to variations in stimulus presentation time, and thus also encoding time, the inter trial interval will be held constant at an amount of time long enough for participants to respond, but no longer. Also because Lyle et al. (2008) showed that eye movements only benefit strongly right-handed individuals, hand preference will be assessed using the van Strien

questionnaire (van Strien, 2003), and only strongly right handed individuals will be included in the study. The self-assessment manikin will be used to assess participants mood before starting the experiment and as exploratory measure to see if there is a change in affect or arousal during the experiment.

Under the AR-hypothesis an increase in affect index score is predicted for the eye-movement condition compared to the fixation condition, reflecting a decrease in negative affect. The AR-hypothesis predicts a memory enhancement effect specifically for negative stimuli, which should be reflected in lower reaction times (implicit memory) for negative stimuli in the eye movement group, when comparing familiar pictures to unfamiliar pictures, and a higher discrimination (explicit memory) for negative stimuli on the recognition task, when comparing the eye movement group to the fixation group.

9

account that affect should decrease equally for negative and positive pictures and that the affect index is the difference score between these. With regards to memory, the WM-hypothesis predicts reduced memory for both positive and negative pictures, which should be reflected in lower discrimination (explicit memory) for positive and negative stimuli during the recognition task when comparing the eye movement condition to the fixation condition.

Method

Participants

53 students with a mean age of 22.33 (SD 1.822) (of which, 25 female) participated. Most of which were acquired through means of convenience sampling (most of the participants were friends of the researchers). Exclusion criteria were having previously partaken in EMDR therapy, uncorrected visual impairment, proneness to dizziness and/or nausea. Only extremely right-handed individuals were included (those that scored +10 on the Van Strien 2012 questionnaire).

Design

The affective priming task has a 2x2x2 repeated measures design with condition

(movement/fixation) as the between-subjects independent variable, and familiar/unfamiliar and valence (positive/negative) as the within-subjects independent variables.

Rt’s are only calculated over correct responses, which implies that response

corresponded to valence of stimulus. Participants’ results are excluded once exceeding an error proportion of 0.15. Outliers (>1.5 SD of the condition mean) were removed using a reiterative procedure with program IBM SPSS Statistics 21.

To compare the RT’s a 2x2x2 Mixed ANOVA was carried out with eye movement condition (eye movement / fixation) as the between-subjects independent variable and reaction times as the dependent variable. Reaction times were compared between familiar and unfamiliar pictures to look for differential priming effects.

10

To interpret effects on affect, an affect index is calculated. This is done by subtracting positive Rt’s from negative Rt’s per condition. In a negative state, participants will respond relatively faster to negative stimuli and slower to positive stimuli. The equation (negative – positive) results in lower reaction times on negative stimuli and/or higher reaction times on positive stimuli, leading to a lower or negative affect index value, and higher reaction times on negative stimuli and/or lower reaction times to positive stimuli to result in higher or positive values on the affect index. Thus relatively lower values of the affect index are indicative of an increase in negative affect and relatively higher values are indicative of a decrease of negative affect. This results in an affect index for each of the moments of measurement (baseline, familiar, unfamiliar).

To analyze effects on affect a 2x2 mixed ANOVA will be performed, with eye

movement condition (eye movements vs. fixation) as the between-participants independent variable, familiarity (familiar/unfamiliar) as the within-subject independent variable, and the affect index as the dependent variable. Only new pictures will be compared to baseline to see what the effects on affect are without influence of repetition priming.

The recognition task will be analyzed in accordance with the two-high threshold model (Snodgrass & Corwin,1988). This model separates response bias from true recognition. True recognition is defined as the probability that a participant recognizes a stimulus as previously seen (old), when it is actually old. As a measure of true recognition, a discrimination measure (Pr) was calculated, which is the difference between hits and false alarms (FA). The response bias is defined as the probability that a participants’ response is previously seen (old), regardless of whether the stimulus is actually old or new. As a

measure of response bias, the bias measure (Br) was calculated, which is the proportion FA corrected for the proportion of items where FA were possible (FA / (1 - Hits + FA)). To analyze differences in discrimination and response bias between positive and negative pictures paired sampled t-tests will be used. To analyze the difference in discrimination and response bias between groups, independent samples t-test will be used. To analyze

11

ANOVA will be performed, with eye movement condition (movement/fixation) as between-groups independent variable and valence (positive/negative) as within group variable.

Materials

Stimuli

All emotional pictures were selected from the IAPS database. The total range of the IAPS for valence fell between 1.31 and 8.34, and for arousal it fell between 1.72 and 7.35. In the selection of pictures for this study, the range for the positive pictures’ valence fell between 5.79 and 7.24, and for arousal it fell between 3.73 and 5.85. For the negative pictures valence fell between 2.19 and 3.62 and for arousal it fell between 4.48 and 6.5. This range was chosen as images within this range were considered not too shocking, but sufficiently arousing to elicit the required emotional response. Also because these intervals had proven adequate in the study of de Roo, 2014, they were maintained. Based on these intervals a total of 200 pictures were selected. The pictures were then divided into 4 sets of 50 pictures. Each set contained 25 negative pictures and 25 positive pictures. These pictures were presented a total of 3 times during the experiment. During the first block two sets were shown (i.e. set 1 and set 2). During the second block an old set and a new set (one that had not previously been seen by the participant) were shown (i.e. set 1 and set 3). During the final block a different old set was shown, and a yet another new set (i.e. set 2 and set 4). Multiple combinations adhering to this “old-new principle” were possible and which sets were presented in which order was selected at random.

Tasks

The affective priming task consisted of the sequential presentation of 100 emotional pictures with a presentation time of 500 milliseconds. Participants judged the pictures either to be negative by pushing the left button or positive by pushing the right button. Rhythmic

response patterns were circumvented by random variation of the inter trial interval between 1000 and 2000 milliseconds. The dependent variable was reaction time.

12

The recognition task consisted of the sequential presentation of 100 emotional pictures with a presentation time of 500 milliseconds. Participants were to respond whether they had previously seen the picture, but pushing the left button, or not, by pushing the right button. Rhythmic response patterns were circumvented by random variation of the inter trial interval between 1000 and 2000 milliseconds. The dependent variables were a discrimination measure and a bias measure calculated from the hits and false alarms.

Experimental manipulation

Eye-movements were facilitated by a moving plus sign (+) on the screen with a font size of 32 points, which alternated from left to right with an in-between distance of 30 centimeters. The plus sign held position during 0.5s before switching. Alternation persisted for a period of 5 minutes. In the fixation condition a plus sign (+) in center screen alternated colors,

changing from red to green to blue proceeding once again to red and so forth, also for a duration of 5 minutes.

Environment and equipment

Research was conducted in a quiet room equipped with a sound proofed door. No daylight entered the room and the lights were dimmed low. The screen was 23 inches in size and had a resolution of 1920x1080 pixels with a 120hz refresh rate. Participants were seated in front of a computer screen with a distance of 30 cm. They were required to place their heads in a support to assure the distance and orientation of the head would remain constant.

The experiment was performed using Presentation® software (Version 0.70, www.neurobs.com). Because participants were all right-handed response buttons were placed on the right side of the screen.

Self-Assessment Manikin

Participants mood before and after the experiment was measured using a self-assessment manikin (Bradley & Lang, 1994), consisting of a dimension for valence and

13

arousal for which participants could rate their mood on a 5-point scale by selecting the humanoid figure, by which variations in valence and arousal were visually represented, that seemed most congruent with their current mood; for valence, 1 was extremely negative and 5 extremely positive, and for arousal, 1 was extremely aroused and 5 was extremely calm. This was also used to exclude those with extremely low mood ratings on first measurment, as depressed persons have been shown to show a bias towards negative associations (Wiesbrod et al.,1998), which could bias the data.

Exit interview

This included questions as to what the participants’ general impression was. What he or she thought about, during fixation/eye movement, and whether or not they had followed instructions. This was included check if participants had properly followed instructions to make exclusion possible if they hadn’t.

Handedness Questionnaire

The van Strien-questionnaire for hand preference included 10 questions pertaining to which hand participants’ use for various activities such as brushing one’s teeth and drawing. To each question the answer could be either left, right or both, to which scores were

attributed, respectively 1, +1, or 0. This resulted in a total score somewhere in the range of -10, being extremely left-handed to +-10, being extremely right handed (van Strien, 2003).

Procedure

Before experimentation commenced, participants were informed that the experiment investigated the effects of eye-movements or eye-fixation (dependent upon their assigned condition) on emotion and memory. Participants were explicitly made aware of the fact that the experiment contained strong emotional images. Inclusion criteria included a +10 score on the van Strien-questionnaire for hand preference, no prior EMDR treatment, not being prone to dizziness and/or nausea, and using their corrective glasses if relevant. After reading the information brochure they were asked to read and sign the informed consent. Hereafter

14

participants rated their current mood using the self-assessment mannequin. The entire experiment was explained again to the participant verbally.

During the first phase the participant was presented with 100 emotional pictures sequentially. They were asked to judge each picture to be negative or positive as quickly and accurately as possible. They were also asked to memorize the pictures to the best of their ability. This was followed by a global-local distraction task with a duration of one minute. Participants were told that there was no right or wrong. This was followed by either the eye-fixation or the eye-movement condition, with each a duration of 5 minutes. Participants were explicitly asked to recall the emotional images they had previously seen, during the

movements/fixation, followed by block 2 of the priming task, and then by the distraction task again. This was succeeded by the recognition task. Finally the experiment was concluded with the self-assessment mannequin and a short exit interview. The experiment had a total duration of 45 minutes more or less.

Results Participants

In total fifty-three participants completed the study. Of those, ten were excluded from the analysis. Three were excluded because they did not follow instructions. Another participant showed an extremely low mood rating on the self-assessment mannequin. As depressed people can show a bias towards negative associations (Weisbrod et al., 1999) his data were also not included. The other six were excluded due to their error proportion exceeding the predetermined 15% threshold.

Affective priming task

As expected there seems to be a repetition priming effect, as is reflected in figure 1 & table 1 by lower RT’s for familiar than for unfamiliar pictures, for both valences and eye movement conditions.

15

According to the affective regulation hypothesis one would expect a memory enhancing effect specifically for negative memories, resulting in differential priming in the eye movement condition, which would be reflected in the data by a greater difference in reaction time between familiar and unfamiliar negative pictures than between familiar and unfamiliar positive pictures. No such differential priming would be expected in the fixation condition. As can be seen in figure 1 and table 1, contrary to what would be expected under the AR-hypothesis, there does not seem to be a greater decrease in reaction time for negative familiar pictures compared to negative unfamiliar pictures, than there is for positive familiar pictures compared to positive unfamiliar pictures, in the eye movement condition compared to the fixation condition. Surprisingly, it seems as if though differential priming is found in the fixation condition instead of in the eye-movement condition, as in the fixation condition the difference between for negative pictures seems greater than that for positive pictures. A mixed ANOVA was performed, with eye movement condition as between-subjects variable, and familiarity (familiar/unfamiliar) and valence (positive/negative) as within-subjects variables, to see if there was a significant differential implicit memory effect of eye movement condition. There was a significant main effect of familiarity on reaction time,

F(1, 41) = 63.93, p< .01, r= .781, which shows that there was indeed a repetition priming

effect, participants responded faster to familiar pictures regardless of eye movement condition and valence. There was no significant interaction between eye movement

condition, familiarity, and valence, F(1, 41) = 9.70, p< .33, r= .437 showing that there was no greater decrease in reaction time for negative familiar pictures compared to negative

unfamiliar pictures, than there was for positive familiar pictures compared to positive unfamiliar pictures, in the eye movement condition compared to the fixation condition. Thus contrary to what was expected under the AR-hypothesis, there was no differential effect of eye movement on implicit memory, in which there is a specific memory enhancement for negative stimuli.

16

Table 1

Mean reaction times and standard deviations (sd) for baseline, familiar and unfamiliar within-group conditions for the movement and fixation within-group.

Before eye movement/fixation After eye movement/fixation

Familiar (Old) Unfamiliar (New) Negative Positive Negative Positive Negative Positive Movement 845 (118) 894 (127) 750 (110) 786 (102) 841 (136) 860 (97) Fixation 804 (99) 872 (120) 712 (74) 771 (87) 811 (86) 826 (90)

Figure 1

Mean reaction times for familiar positive and negative pictures, and unfamiliar positive and negative pictures for the movement and fixation group.

17

When considering the error proportions, according to the affective regulation hypothesis one would expect a memory enhancing effect of eye movements, specifically for negative

memories, which would be reflected in the data by a greater difference in error proportion, (in which there are less errors for familiar pictures), between familiar and unfamiliar negative pictures than between familiar and unfamiliar positive pictures in the eye movement

condition. This difference would not be expected in the fixation condition. As one can see in table 2 and figure 2, it seems that the difference for negative pictures is not greater than the difference for positive pictures, in the eye-movement condition. A mixed ANOVA was performed to see if there were indeed no significant differences. As would be expected, a main effect of familiarity was found, F(1, 41) = 9.62, p < .01, r = .436, which shows that participants made less errors on familiar pictures compared to unfamiliar pictures. Contrary to what would be expected under the working memory hypothesis, no significant interaction between familiarity and eye movement condition was found, which means that there was no greater difference in error proportion between familiar and unfamiliar pictures, for the eye movement condition, regardless of valence, F(1, 41) = .15, p= .71, r= .06. There was a significant main effect of valence on error proportion, F(1, 41) = 34.54, p < .01 r= .676, showing that participants made less errors on negative pictures than on positive pictures, as would be expected. . There was no significant interaction between eye movement condition, familiarity and valence, F(1, 41) = 0.07, p = 0.80, r= 0.04, which means there was no greater decrease in error proportion when comparing familiar negative pictures to unfamiliar

negative pictures, for the eye movement condition compared to the fixation condition, than there was for positive familiar pictures compared to positive unfamiliar pictures. This means there was no differential implicit memory effect of eye movements as expected under the AR-hypothesis, in which eye movements specifically enhance memory for negative pictures. These results are neither supportive of the working memory hypothesis or the affective regulation hypothesis.

18

Table 2

Mean error proportions and standard deviations (sd) for familiar, unfamiliar and valence within-group independent conditions for the eye movement and fixation within-group.

Before eye movement/fixation After eye movement/fixation

Familiar (Old) Unfamiliar (New) Negative Positive Negative Positive Negative Positive Movement .058 (0.03) .096 (0.04) .034 (0.04) .091 (0.05) .069 (0.06) .120 (0.07) Fixation .063 (0.04) .127 (0.07) .029 (0.04) .094 (0.08) .053 (0.06) .119 (0.08) Figure 2

Mean error proportions for familiar, unfamiliar and valence within-group independent variables for the eye movement and fixation group.

19

Under the affective regulation hypothesis one would expect a specific down-regulation for negative affect in the eye movement condition, which would be reflected in the data by a higher affect index score. As can be seen in table 3 & figure 3, the affect index is higher for both the eye movement and fixation condition, suggesting that eye movements do not specifically down-regulate negative affect or that fixation also down-regulates negative affect. To see if this down-regulation of affect was significant, a mixed ANOVA was performed with eye movement condition as between-subject variable and measurement (baseline/unfamiliar pictures) as within-subject variable. A main effect of measurement was found, F(1, 41) = 6.967, p= .01, r = .38, suggesting that both after eye-movement and eye fixation participants were less negative. However there was no interaction between

measurement and condition, F(1, 41) = .488, p= .49, r = .11, which means that against expectations the amount of decrease of negative affect was not significantly larger in the eye movement condition than in the fixation condition. A mixed ANOVA was also performed on the SAM to see if eye movements have an effect on affect and arousal. No interaction effects between eye movement condition and affect, F(1,41) = 1.46, p = .234, r = .185, or arousal were found. Contrary to what was expected under the affective regulation hypothesis eye movements did not specifically down-regulate negative affect.

Table 3

Mean affect index scores and standard deviations (sd) for baseline, familiar and unfamiliar within-group conditions for the movement and fixation group.

Baseline Familiar (old) Unfamiliar (New) Movement -49 (68) -36 (73) -18 (115)

20

Figure 3

Mean affect index scores and standarddeviations (sd) for baseline, familiar, and unfamiliar within-group conditions for the movement and fixation group.

Recognition memory task

As a measure of true recognition a discrimination measure (Pr) was used, of which higher values reflect more accurate recognition and lower values reflect less accurate recognition. As a measure of response bias, values of Br values greater than 0.5 reflect a liberal bias (more willing to indicate as old), values lower than 0.5 reflect a more conservative bias (less willing to indicate as old). There was no significant difference in response bias for both the

eye-movement condition, t(20) = - .25, two-tailed p > 0.05, and the fixation condition, t(21), two-tailed p> 0.05.

21

Table 4

Pr = discrimination measure (sd) and Br = bias measure (sd) according to the two-high threshold model for each condition.

Pr Br Bilateral Negative 0.728 (0.12) 0.273 (0.21) Positive 0.821 (0.11) 0.332 (0.31) Fixation Negative 0.620 (0.14) 0.278 (0.18) Positive 0.804 (0.10) 0.460 (0.28) Figure 4

Mean discrimination (Pr) and standard deviations (sd), and mean bias scores (Br) and standard deviations (sd) for positive and negative within-group conditions and the movement and fixation groups.

22

Under the affective regulation hypothesis a memory enhancing effect is expected specifically for negative pictures. As one can see in table 4 and figure 4, the discrimination seems higher for negative pictures and equal for positive pictures when comparing the eye movement condition to fixation, as was expected. Independent t-tests were performed to see if these differences were significant. As expected, discrimination was significantly higher for negative pictures in the eye-movement condition, t(41) = 2.727, p < 0.01 two-tailed, d=

0.834, and there was no significant difference in discrimination between positive and

negative stimuli in the fixation condition, t(42) = -.259, p = 0.797, d = 0.168. In line with the affective regulation hypothesis these results suggest that there is a specific enhancing effect of eye-movements on recognition for negative memories.

No predictions were done as to the effect of eye movements on response bias. However since a difference is observed an explorative analysis was performed. Response bias seems to be equal between positive and negative images within the eye-movement condition, a paired sampled t-test confirmed that this difference is indeed not significant, t(19)= -0.579, p

= 0.57. For the fixation condition however, there seems to be a less conservative bias for

positive pictures compared to negative ones, suggesting perhaps that eye movements make one more conservative for positive stimuli, however this difference is marginally significant, t

(21) = -.304 p = 0.06 two-tailed, Cohen’s d= -0.448. With regard to between group

differences, independent samples t-tests show that there is no significant difference in response bias when comparing movement to fixation, for both negative, t(41) =-085, p =

23

This study investigated the effect of eye-movements on affect and emotional memories. As expected an enhancing effect of eye-movements was found on explicit memory, specifically for negative stimuli; participants who had performed eye-movements had better recognition for negative stimuli than those who hadn’t. However this emotion specific enhancement was not found for implicit memory, which suggests a dissociation in the effect of eye-movements with regard to implicit and explicit memory. Furthermore, contrary to expectation, no effect of eye-movement on affect was found. Surprisingly, participants were found to be less negative after both eye movements and fixation.

The results speak against the working memory hypothesis because an enhancing effect was found specifically for negative memories, whereas the WM-hypothesis predicts reduced memory for both positive and negative memories. The results are not supportive of eye movements’ therapeutic effects being mediated by their load on working memory capacity. On the other hand the AR-hypothesis doesn’t seem to fully explain the effect of eye-movements either.

The affective regulation hypothesis posits that bilateral eye movements should have a memory enhancing effect specific to negative stimuli, possibly through increased working memory capacity, which allows for better processing of the memory, and that they should also specifically down-regulate negative affect, via the orienting response, thus effectively reconditioning the memory with reduced negative affect. Surprisingly, current findings only partially support the hypothesis, considering that the predicted memory effects were found, but effects on affect were not. These findings are in line with Dame (2013), who also found a memory enhancement effect specific to negative stimuli, in his case emotional words instead of pictures, but did not reveal any effects on affect. To explain this discrepancy in effects on affect and memory, Dame (2013) suggested that perhaps the reprocessing effect of eye movements takes place not so much by means of affective reconditioning, but predominantly through increased motivation to detect negative stimuli and through increased working

24

memory capacity to process these stimuli, both as a consequence of increased dopamine production caused by the eye movements.

However, this does not account for the discrepancy found between implicit and explicit memory, which may be accounted for by considering a possible effect of eye movements on coping or approach/avoidance behavior, with dopamine production as a mediator. Furthermore, it may be more parsimonious to assume that they are effective through coping, for this could possibly explain current findings by addressing dopamine’s role solely in the motivational system, without additional working memory assumptions.

Coping refers to the set of behaviors and physiological adaptations displayed by individuals in order to master the aversive condition (internal or external) to which they are exposed (Giorgi, et al,2003). A difference can be made in pro-active vs. reactive coping styles, in which the former is characterized by approach behavior, the individual tries to actively change the aversive condition, whereas the latter is characterized by avoidance behavior or passive behavior, such as freezing or fleeing.

In their study with rats, Giorgi et al (2003) found that individuals that proactively cope with negative stimuli show increased dopamine output in the prefrontal cortex (PFC),

whereas those that reactively cope, fail to show this increased dopamine output in the prefrontal cortex. Perhaps eye movements increase production of dopamine, which

subsequently arrives in the PFC and consequently shifts reactive coping more towards pro-active coping.

Though the previously mentioned study was of course conducted with non-human subjects, leaving questions as to generalizability, its results seem to be in line with results found in research done by Kok (2015), which found that bilateral eye movements increase motivational intensity with regards to negative memories. He based this theoretically on the ‘seeking vs fear system’ (Panksepp, 1988; cited in Kok, 2015), which relates seeking to approach and fear to avoidance behavior. Kok (2015), proposes that perhaps eye movements activate dopaminergic neurons, which in turn activates the seeking system, which stimulates the organism to actively engage with the stimulus, be it positive or negative.

25

Thus eye movements’ valence specific enhancing effect on memory can perhaps be explained by a ‘coping hypothesis’, which suggests a shift from avoidant or reactive coping with negative stimuli, towards approach or proactive coping, and thus making memories more accessible.

This is corroborated by Hermans et al. (2005), who found that an avoidant coping style was related to reduced autobiographical memory specificity. Participants that scored higher on avoidant coping style questionnaires, would recall less details and describe their memories more generally during free recall than those who scored lower on avoidant coping style. If one assumes that higher memory specificity also allows one to better discriminate between, and thus more accurately recognize, memories, this would be congruent with the higher discrimination found on the recognition task for negative memories. Perhaps eye movements make negative memories more accessible by shifting coping towards a proactive coping style.

Possibly the dissociation found in implicit and explicit memory effects can also be explained by the same account, as the person possibly displays avoidant behavior with regard to the memory, it may be less accessible for conscious retrieval, which may be interpreted as cognitive approach behavior, while implicit memory remains unaffected.

This is corroborated by Fujiwara, Levine & Anderson (2008), who found that repressive copers, those that try to forget negative memories, had lowered free recall for negative information, while implicit memory was unaffected. Repressive coping could perhaps be seen as reactive coping specifically towards internal stimuli, and thus this dissociation between implicit and explicit memory could speak in favor of emotional regulation in reactive/repressive copers as motivated forgetting, or avoidance behavior regarding the memory.

The linking of eye movements, dopamine, and coping, would therefore lead to a ‘coping hypothesis’, which suggests that eye movements are effective by increasing dopamine production via the projections from the superior colliculus to the substantia nigra

26

pars compacta, leading to a shift from reactive to a proactive coping with regards to the memory, thus motivating the individual to more actively process the memory.

A substantial limitation to the current study is the absence of measures for dopamine, which served as the foundation for the initial affective regulation hypothesis and on the basis of which the coping hypothesis was also put forward. The verification of this crucial

assumption is an important prerequisite for either hypothesis to gain well-grounded credibility. Though Kok (2015) gained some behavioral evidence suggestive of a role for dopamine and though there is some evidence suggesting a role for the superior colliculus in dopaminergic pathways and approach/avoidance behavior (Westby et al., 1990; Comoli, et al., 2003; Hikosaka, Takikawa, & Kawagoe, 2000), future research should look into including measurements on dopamine such as eye-blink rate and PET-scans. Another possibility might be to investigate if eye movements are still effective in subjects that have been given dopamine antagonists.

Another possible limitation to the current study was the fact that it comprised a convenience sample. All participants were friends and acquaintances of the researchers. However there is no reason to assume that they deviated in such a substantial manner from the general population that eye movements would perhaps affect them differently, seeing that handedness seems to be the greatest contributor to variations in the effects of eye movements and these were all strongly right handed individuals. Moreover, it is likely that they had greater motivation to participate in the experiment seriously and to carry out the eye movements properly, resulting in a better experimental manipulation. Possible distortions could have been a consequence perhaps if there were a bias in coping style when comparing the current sample to the population at large, this however seems unlikely. Nonetheless, it is of course well-advised for future research to attempt a more random sample to rule out any possible bias.

Another interesting direction for future research would be to investigate

eye-movements’ relationship to coping. Perhaps scores on reactive coping style questionnaires could be included as covariate, when investigating the effects of eye movements on

27

memory. Or perhaps the effect of eye-movements on memory could be compared between a group of pro-active copers and a group of reactive copers. If eye movements are indeed effective by shifting coping towards a more proactive strategy, those that score high on reactive coping might improve more on recognition tasks than those that score low.

If this coping hypothesis is indeed true, this would be very clinically relevant as the consequence could be that EMDR-therapy is more effective for those patients that suffer due to maladaptive reactive coping styles. Therefore, the therapy might be deemed more

applicable to disorders characterized by such coping styles, which could include both disorders associated with trauma, such as PTSD, as well as those associated with feared future events, such as anxiety disorders. Perhaps assessments of coping styles via questionnaires or other means could help to predict how effective the therapy will be for a particular patient.

The most important contribution of this study is the discrepancy that was found in the effect of eye movements on implicit and explicit memory, in which there was an enhancing effect for explicit memory and no effect on implicit memory, thus inviting future research to explore differential effects of eye movements as to conscious and subconscious processing, hopefully further elucidating EMDR-therapy’s elusive underlying mechanisms. The memory enhancing effect that was found argues against a working memory hypothesis, suggesting that eye movements do not desensitize memories via a dual load on working memory, but quite contrarily enhance memory. Furthermore, this study showed that eye movements do not have a direct effect on affect, and do not specifically down-regulate negative affect, meaning their therapeutic effects are most likely not due to affective reconditioning. Furthermore, it gave way to a new ‘coping hypothesis’ suggesting eye movements are efficacious by generating a shift from reactive coping with negative stimuli to more proactive coping. Though this study was not fully capable of playing a decisive role in the quagmire of literature on its underlying mechanisms, it has spawned interesting new directions for future research, which could bear substantial clinical significance.

28

Armstrong, M. S., & Vaughan, K. (1996). An orienting response model of eye movement desensitization. Journal of Behavior Therapy and Experimental

Psychiatry, 27(1), 21-32.

Barrowcliff, A. L., Gray, N. S., Freeman, T. C., & MacCulloch, M. J. (2004). Eye

movements reduce the vividness, emotional valence and electrodermal arousal associated with negative autobiographical memories. Journal of Forensic

Psychiatry & Psychology, 15(2), 325-345.

Bradley, M. M., & Lang, P. J. (1994). Measuring emotion: the self-assessment manikin and the semantic differential. Journal of behavior therapy and experimental psychiatry, 25(1), 49-59.

Comoli, E., Coizet, V., Boyes, J., Bolam, J. P., Canteras, N. S., Quirk, R. H., ... & Redgrave, P. (2003). A direct projection from superior colliculus to substantia nigra for detecting salient visual events. Nature neuroscience, 6(9), 974-980.

Dame, W. (2012). Bilaterale oogbewegingen: Geheugen of Affect? De Roo, R. (2014). Moving the eyes across emotional pictures.

Engelhard, I. M., van Uijen, S. L., & van den Hout, M. A. (2010). The impact of taxing working memory on negative and positive memories. European Journal of

Psychotraumatology, 1.

Engelhard, I. M., van den Hout, M. A., & Smeets, M. A. (2011). Taxing working memory reduces vividness and emotional intensity of images about the Queen’s Day tragedy. Journal of Behavior Therapy and Experimental

Psychiatry, 42(1), 32-37.

Fujiwara, E., Levine, B., & Anderson, A. K. (2008). Intact implicit and reduced explicit memory for negative self-related information in repressive coping. Cognitive,

29

Giorgi, O., Lecca, D., Piras, G., Driscoll, P., & Corda, M. G. (2003). Dissociation between mesocortical dopamine release and fear‐related behaviours in two

psychogenetically selected lines of rats that differ in coping strategies to aversive conditions. European Journal of Neuroscience, 17(12), 2716-2726. Hermans, D., Defranc, A., Raes, F., Williams, J. M. G., & Eelen, P. (2005). Reduced autobiographical memory specificity as an avoidant coping style. British

Journal of Clinical Psychology, 44(4), 583-589.

Hikosaka, O., Takikawa, Y., & Kawagoe, R. (2000). Role of the basal ganglia in the control of purposive saccadic eye movements. Physiological reviews, 80(3), 953-978. Kok, R.M. (2015). Bilaterale oogbewegingen: werkgeheugen of dopamine verhogend? Lee, C. W., & Cuijpers, P. (2013). A meta-analysis of the contribution of eye

movements in processing emotional memories. Journal of Behavior Therapy

and Experimental Psychiatry, 44(2), 231-239.

Lyle, K. B., Logan, J. M., & Roediger, H. L. (2008). Eye movements enhance memory for individuals who are strongly right-handed and harm it for individuals who are not. Psychonomic bulletin & review, 15(3), 515-520.

MacCulloch, M. J., & Feldman, P. (1996). Eye movement desensitisation treatment utilises the positive visceral element of the investigatory reflex to inhibit the memories of post-traumaticstress disorder: a theoretical analysis.The British

Journal of Psychiatry, 169(5), 571-579.

Mauss, I. B., & Robinson, M. D. (2009). Measures of emotion: A review. Cognition and

emotion, 23(2), 209-237.

Snodgrass, J. G., & Corwin, J. (1988). Pragmatics of measuring recognition memory: applications to dementia and amnesia. Journal of Experimental Psychology: General, 117(1), 34

Sokolov, E. N. (1990). The orienting response, and future directions of its

30

Van den Hout, M. A., & Engelhard, I. M. (2012). How does EMDR work?.Journal of

Experimental Psychopathology, 3(5), 724-738.

Van den Hout, M., Muris, P., Salemink, E., & Kindt, M. (2001). Autobiographical memories become less vivid and emotional after eye movements. British

Journal of Clinical Psychology, 40(2), 121-130.

Weisbrod, M., Trage, J., Hill, H., Sattler, H. D., Maier, S., Kiefer, M., ... & Spitzer, M. (1916). Emotional priming in depressed patients. Age (SD), 48, 13-3.

Westby, G. W. M., Keay, K. A., Redgrave, P., Dean, P., & Bannister, M. (1990). Output pathways from the rat superior colliculus mediating approach and avoidance have different sensory properties. Experimental brain research, 81(3), 626-638.