From unraveling the working mechanisms of imagery rescripting to context manipulation in a new generation of ABA-Fear Conditioning Paradigms : a series of 5 pilots

(1)

From Unraveling the Working Mechanisms of Imagery Rescripting to Context Manipulation

in a New Generation of ABA-Fear Conditioning Paradigms: A Series of 5 Pilots

Karen Fischer

University of Amsterdam

Supervisors:

Anna E. Kunze, M.Sc.

Prof. Dr. Arnoud Arntz

Prof. Dr. Merel Kindt

(2)

Abstract

Imagery rescripting (IR) is a promising psychological technique to treat traumatic fear memories. IR may even have greater potential in reducing fear associated with negative emotional memories than traditional treatment techniques such as exposure therapy. This is due to the fact that IR might work through unconditioned stimulus (UCS) devaluation, a mechanism that can assign new, less aversive meanings to the UCS and thereby diminish fear responses. Because UCS devaluation is theorized to change the meaning of the original memory directly, rather than facilitating the formation of a new memory (i.e., exposure), it is hypothesized that UCS devaluation reduces the commonly observed fear renewal after a context-switch. Thus, the aim of the current study was to compare the two treatment techniques (i.e., rescripting vs. exposure) on the return of fear due to a context-switch after successful fear acquisition and extinction. For this purpose, we used a 3-day ABA-fear conditioning paradigm, where fear acquisition (context A, day 1), intervention (context B, day 2), extinction (context B, day 3) and fear renewal (context A, day 3) took place in the respective context. The context was manipulated by changing the color of the laboratory lights. It was expected that robust fear renewal could be observed in the exposure condition on subjective distress and physiological measure (EMG), while less fear renewal will occur in the IR condition. However, contrary to the expectations, pilot testing of the control condition (exposure) resulted in no fear renewal after the context-switch. Given that fear renewal in the control condition constitutes an effective context-switch and is therefore a prerequisite to investigate the effects of IR on fear renewal, it was concluded that the paradigm was not suited to test the hypothesis. In order to improve the paradigm, a series of five pilot studies was conducted within the control condition, which aimed at adjusting the context-switch paradigm. However, none of the pilots showed successful fear renewal after the context-switch on either subjective distress or EMG. In the present report, the pilot studies will be described and important issues concerning the interpretation and validation of the results will be discussed.

(3)

From Unraveling the Working Mechanisms of Imagery Rescripting to Context Manipulation in a New Generation of ABA-Fear Conditioning Paradigms: A Series of 5 Pilots

1.1 A New Generation of Fear Conditioning Paradigms

Pavlovian fear conditioning and extinction have brought indispensable insights to the underlying processes of acquisition, consolidation and expression of traumatic fear memories (e.g. see Craske, Hermans, & Vansteenwegen, 2006; Fanselow & Poulos, 2005; LeDoux, 2000). Exposure therapy, an extinction-based therapeutic technique, has been applied since the 1950’s and still constitutes one of the principle pillars in the treatment of traumatic fear memories. More recently, in recognition of the multifaceted and subjective nature of human fear memories, fear conditioning paradigms and their respective stimuli have been adjusted from their early simplistic implications to model more complex human learning processes (e.g. Kunze, Arntz, & Kindt, 2015).

For example, a recent study by Kunze et al. (2015) adjusted the previously simple conditioned stimuli (CS; e.g. geometric shapes and figures) and unconditioned stimulus (UCS; e.g. electroshock) to human faces and film clips and embedded them in a relevant context to mimic the more complex learning processes and subjectivity involved in human fear learning. The findings validated that implementing complex and meaningful CS and UCS is effective in eliciting psychophysiological (startle) as well as subjective fear responses. Furthermore, embedding the reinforced CS (CS+) and UCS in a relevant context (i.e. a movie scene from where faces and film clip were taken) prior to fear acquisition resulted in stronger fear responses than if exposed to a neutral film clip. The study by Kunze et al. (2015) corroborates the findings of Davey (1997), which show that providing additional and meaningful information about the UCS (e.g. a film) resulted in a greater conditioned response to the CS+ and higher negative valence ratings of the UCS.

1.2 UCS Devaluation

The field of fear conditioning has been invigorated by a new generation of human differential fear conditioning paradigms, which have been successfully updated to emulate the complex and subjective nature of human traumatic fear memories (Kunze et al., 2015; Pile, Barnhofer, & Wild, 2015; Wegerer, Blechert, Kerschbaum, & Wilhelm, 2013). As one of the most crucial implications of this research, attention was brought to the importance of focusing on the meaning of the UCS, above and beyond the CS-UCS associative strength in predicting fear response strength. Davey (1989), who coined the term UCS devaluation, had early on recognized the clinical potential of devaluing the UCS to reduce fear responses: he argued that the clinical applications of exposure are limited because they focus on prediction of the context-vulnerable CS-UCS contingency strength and not altering the actual fear response itself (Davey, 1989). Whereas exposure aims to diminish the conditioned response (CR) by altering the CS-UCS associative strength, UCS devaluation encompasses the idea that directly altering the meaning assigned to the aversive memory (UCS) can directly diminish the strength of the CR (Kunze et al., 2015; Dibbets, Poort, & Arntz, 2012).

(4)

1.3 Advantages of IR over Exposure

More research is accumulating which suggests that psychological techniques based on UCS devaluation might have greater potential to reduce fear responses than exposure (e.g. Bouton, 2002; Davey, 1989; Kunze et al., 2015). One such technique, which has been suggested to work through UCS devaluation is imagery rescription (IR) (e.g. Arntz, 2012). In IR, a patient mentally manipulates the traumatic fear memory (UCS) in such a way that the aversive event acquires a less negative meaning. IR has thereby been shown to be an effective technique in reducing patients fear responses (Arntz & Weertman, 1999; Arntz, 2011) and anxiety. The advantage of this psychological technique over mere imaginal exposure (IE) is that the strength of the conditioned response (CR) is not mediated by the contingency-based CS-UCS associative strength (Dibbets et al., 2012), but rather by direct alteration of the meaning of the aversive event itself (Davey, 1997; Davey, 1989).

The advantage of UCS devaluation-based techniques such as IR over exposure-based treatments becomes most apparent in light of the following well-known scenario: A patient overcomes his spider phobia in the therapy setting through guided exposure until the fear is extinguished. When the patient then leaves the therapy setting and is confronted with the phobic stimuli again, a robust renewed fear response is commonly observed in most patients when they leave the context where exposure was conducted in (Bouton, 2004). As this context-induced return of fear is assumed to be a product of the context dependency of a learned CS-UCS association (Bouton, 2002), exposure therapy is limited in treating traumatic fear memories, which are by nature multi-context laden. Hence, if IR indeed works as a psychological technique to devalue the UCS directly, then CR strength can be diminished independently of the CS-UCS association and context (Arntz, 2012).

Thus, IR should result in less fear renewal than exposure therapy when a patient is again confronted with the CS outside of the extinction context. The term fear renewal was first introduced by Bouton and Bolles (1979). ABA-renewal is one of the most well-studied context-induced fear renewal effect, which entails fear acquisition in context A, extinction in context B and return of the conditioned fear response when reinstated to context A and presented with the CS (Bouton & Bolles, 1979). Commonly used context manipulations include setting the computer background screen to an environmental context (Dibbets et al., 2012) or changing the light color in the laboratory room (Effting & Kindt, 2007). If IR indeed works over UCS devaluation, IR would have greater impact on CR modulation after a context-switch than exposure, as it allows generalization of therapy exercises to other environments and stimuli (Arntz, 2012; Dibbets et al., 2012). In order to investigate whether UCS devaluation truly encompasses the underlying working mechanism of IR, the present series of pilots originally aimed to test whether IR can inhibit fear renewal within the framework of a 3-day ABA-context-conditioning paradigm.

1.4 Testing of IR in Context Fear Conditioning Paradigms

Despite the body of research supporting IR as an effective technique to treat anxiety disorders, there is little research that investigates whether UCS devaluation may constitute an underlying

(5)

working mechanism of IR (e.g. Arntz, Tiesema, & Kindt, 2007). Specifically in the field of contextual fear conditioning, there is scarcely any research that investigates this. A recent pioneering study by Dibbets serves an inspiration for the current study: Dibbets et al. (2012) compared IE with added IR to IE alone on UCS expectancy within the framework of a 1-day ABA-fear conditioning paradigm: In this framework, acquisition took place in context A, extinction in context B, and renewal in context A. They showed that an intervention of IE plus IR resulted in subsequent less renewal of UCS expectancy ratings than IE alone when reinstated to the original context (A). Furthermore they provided evidence for UCS devaluation as underlying working mechanism of IR, by showing that the reduced fear renewal resulted from diminished UCS valence, as indicated by changes in mental representations of the UCS (Dibbets et al., 2012). Despite the fact that this study paved a first way for experimental research on IR within a context-switch paradigm, the study has a few shortcomings. Five main

shortcomings were identified and serve as an inspiration for the present study: 1) Firstly, the study was conducted on one day, which does not allow memory consolidation to take place, a process which has shown to require approximately 24 hours (e.g. Nader et al., 2003). 2) Furthermore, UCS expectancy instead of fear responses was measured, and furthermore no physiological fear measurements (e.g. fear-potentiated startle) were taken. 3) The two contexts were two different landscapes which were set as background screen on the computer. Whereas the landscape image represents a good approach to mimic an environmental context, it is questioned whether a computer screen background is enough to function as a global situation-setter, as observed in real-life trauma formation. 4) Lastly, IR was implemented in addition to extinction to represent the experimental condition, and was therefore not investigated as a stand-alone intervention. 5) Furthermore, in comparing IR + extinction to mere exposure, it is not controlled for what effect mental imagery may have on intervention effectiveness. Furthermore, IR and exposure together constituted a longer training than exposure alone; therefore it cannot be disentangled whether a difference between interventions stemmed from length of treatment or the actual effects of IR.

1.5 Present Study

The present study aims to test the effects that IR may have on a subsequent fear renewal test within an improved fear conditioning paradigm: The aim is to resolve the shortcomings as in Dibbets et al. (2012) in the following manner:

1) The study will consist of 3 days, instead of 1 day, of testing: fear acquisition will take place on day 1 in context A and extinction on day 2 in context B. On day 3, a switch from context A to context B during unreinforced CS+ presentations will take place, which will allow testing of fear renewal. The context consists of full-surround light-color that the whole room will be lit up in. A context-switch thus entails the switch from one room light color to the other. The 3-day procedure was specifically implemented to ensure memory consolidation between each experimental phase. To the best of our knowledge, this is the first study to test the effects of IR on fear renewal within the framework of a 3-day paradigm.

(6)

2) Instead of a single measure of UCS expectancy ratings as in Dibbets et al. (2012), the present study will assess fear responses. The study will not only assess subjective distress ratings, but furthermore a physiological fear response, namely fear-potentiated startle. Recent research in

pharmacological disruption of fear memory reconsolidation has shown that propranolol erased fear-potentiated startle responses, but left expectancy of the fear stimuli intact (Kindt et al., 2009). As such, Kindt et al. (2009) showed that reconsolidation of spider fear memory led to a reduction in behavioral, but not self-reported spider fear. Therefore, it is evident that pharmacological interventions to reduce fear memories target distinct processes, exerting differential effects on physiological versus cognitive expressions of fear memory. In contrast to pharmacological alterations of fear memory, it is therefore of interest to investigate whether the psychological technique of IR targets primarily cognitive (subjective distress), behavioral (startle), or both forms of fear expressions equally. To assess this potential dissociation, the present study measures both subjective distress as well as physiological fear responses in order to disentangle the IR intervention effects on both types fear expression.

3) The context will be manipulated by changing room light colors in the laboratory instead of using computer background screen as a context. Recent research supports room light color as a more ecologically valid form of context (Effting & Kindt, 2007): Room light color may represent a more global and engulfing situation-setter, as it represents a 3-dimensional form of context as in real-life trauma, compared to the 2-dimensional and less impacting context that a computer screen provides. 4) The present study investigates IR as a stand-alone measure, and not as an add-on treatment to extinction, as in Dibbets et al (2012). Therefore, the expected reduction in fear renewal can be assumed to stem directly from the effets of the IR intervention. The present study therefore assesses the clinical relevance of IR as a stand-alone therapeutic technique to cure anxiety disorders.

5) Lastly, a control imagery group will be implemented, to control for the general effects that mental imagery may have on fear renewal. The control condition will consist of a rehearsal of the aversive movie that participants will have watched (exposure). By comparing IR to such a control imagery condition, any effects of IR can be attributed to the content of rescripting, rather than mental imagery effects.

In the present study, it was hypothesized that an intervention of IR will result in a decrease of fear renewal on both subjective distress and fear potentiated startle response in comparison to the control condition: for the control condition it is expected that there will be a robust fear renewal response on both subjective distress and startle. Prior to launching the study, a pilot needed to test whether fear renewal occurs in the control condition. This is a direct test of whether the context-switch works in renewing previously extinguished fear responses. Only when the context-switch as a

technique is able to produce fear renewal in the control condition, IR rescripting can be applied within this framework to test its expected capacity to reduce fear renewal. Contrary to the expectations, pilot testing of the control condition (exposure) resulted in no fear renewal after the context-switch. In order to improve the paradigm, a series of five pilot studies was conducted within the control condition.

(7)

Across these pilots, adjustments were made to the paradigm with the aim of improving the effects of context change on subsequent fear renewal within the framework of the recently developed fear conditioning paradigm (Kunze et al., 2015). These pilots investigated the post-hoc hypothesis that light color as context can elicit fear renewal after a final context-switch. However, none of the pilots showed successful fear renewal after the context-switch on either subjective (distress) or

psychophysiological measures (EMG). Therefore, hypothesis testing was not feasible within the current project. In the present report, the pilot studies will be described and important issues concerning the interpretation and validation of the results will be discussed.

Method 2.1 Participants

Participants were in total 39 healthy undergraduate students recruited via the online research platform of the University of Amsterdam. The study was granted approval by the Ethics Committee. The pilots vary in the number of participants, with the smallest pilot consisting of 4, and the largest pilot of 9 participants. Prior to begin of the study, all participants underwent a screening of history of physical and/or sexual abuse, current mental and/or physical illness, and (prescribed) medication and/or drug intake at the time of testing). All participants signed an informed consent. Participants were randomly allocated to the experimental condition or the control condition. Reimbursement for participation was given either in form of money (20 euro) or course credits.

2.2 Measures

2.2.1 Primary Outcome Measures

Fear-potentiated startle. The psychophysiological measure of fear-potentiated startle (FPS) was used to assess conditioned fear responses with electromyography of the left orbicularis oculi muscle. The FPS was elicited by a startle probe (104dB), which consisted of a 40 ms burst of broadband white noise over headphones (binaurally), with near instantaneous rise time (e.g., Kindt, Soeter, & Vervliet, 2009; Sevenster, Beckers, & Kindt, 2013; Van Ast, Vervliet, & Kindt, 2012). Stronger potentiation of eyeblink to the startle noise at CS presentation indicates greater fear response (Kindt et al., 2009). FPS is known to be an effective and reliable measure of learned fear responses (e.g., Kindt, Soeter, & Vervliet, 2009; Sevenster, Beckers, & Kindt, 2013; Van Ast, Vervliet, & Kindt, 2012).

Subjective distress. For each CS presentation, the participant indicated their distress levels on a visual analogue scale (VAS), which appeared on the right side of the screen for 5 seconds after 2.5 seconds of stimulus onset. The VAS was a vertical red-to-green colored bar, which ranged from very anxious or distressed (100) to not at all anxious or distressed (0).

2.2.2 Secondary Outcome Measures (Questionnaires)

STAI-T / STAI-S. The State-Trait Anxiety Inventory (STAI-T/STAI-S; Spielberger, Gorsuch, & Lushene, 1970) is a commonly used instrument to assess state and trait anxiety levels. The

(8)

items include “I am tense” (state anxiety), and “I am a steady person” (trait anxiety). Ratings are made on a 4-point scale, ranging from “Almost Never” to “Almost Always”, with higher scores reflecting greater anxiety. Cronbach’s alpha for the instrument has been found to range from .86 to .95 (Spielberger et al., 1983). The instrument disposes over high construct and concurrent validity (Spielberger, 1989).

PANAS. The Positive and Negative Affect Schedule (PANAS; Watson, Clark, &

Tellegen,1988) assesses mood changes resulting from the aversive film and the fear conditioning. The instrument has shown to be a valid and efficient measure of the two dimensions of positive and negative effect. There are 10 items for each dimension, asking how one feels in the very moment, e.g. “enthusiastic” (positive affect) and “irritable” (negative effect), which can be rated from 1= very slightly or not at all, to 5= extremely. The instrument disposes over high internal consistency, with Cronbach’s alpha ranging from .86 to .90 for the Positive Affect Schedule, and .84 to .87 for the Negative Affect Schedule (Watson, Clark, & Tellegen,1988).

Emotion-Mood Visual Analogue Scales. The researchers of the studies created a 6-item measure to assess emotions and mood. Participants were instructed to rate on Visual Analogue Scales to what extent they felt following moods and emotions at this very moment: 1. ashamed 2. scared 3. sad 4. angry 5. disgusted, and 6. in control.

SAM. The Self-Assessment Manikins (SAM; Bradley & Lang, 1994) is a commonly implemented, non-verbal pictorial instrument that has shown to be reliable and valid in assessing valence and arousal to a broad range of stimuli (Bradley & Lang, 1994). In the present study, the SAM were used to assess valence and arousal towards the CS. The valence scale has demonstrated internal consistency reliability ranging from .63 to .82, and the arousal scale from .93 to .98. The instrument disposes over high construct validity (Backs et al., 2005).

QMI. The Questionnaire of Mental Imagery (QMI) will be administered to assess vividness of imagery and imagery ability across a range of modalities (Dibbets et al., 2012). The instrument

consists of 35 items with ratings on a 7-point Likert scale ranging from 1 (“perfectly clear, as if it were real”) to 7 (“I think about it, but I cannot imagine it”) and has demonstrated high reliability of

(Sheehan, 1967).

Contingency awareness. Contingency awareness checked whether participants knew which CS was followed by the UCS. Therefore participants were presented with both pictures of the CS+ and CS- printed out on the questionnaire, and participants were asked to make a cross next to the picture that was followed by the UCS.

Manipulation Check. Manipulation checks will assess moods and emotions that participants experienced in regard to the aversive movie, film clip and the IE intervention. Items include how intense, (un)pleasant, and close they experienced the movie and film clip. For the IE intervention, the manipulation check assessed how alive, (un)pleasant, and negative/positive they experienced it. Ratings were made on Likert-scales ranging from -5 (not at all) to 5 (very much).

(9)

Post-Experimental Questionnaire. This questionnaire consisted of several sub-questionnaires which assess emotion, mood changes, and perception about the film, the IE intervention, and the entire experiment.

2.3 Materials

Film Stimuli. The aversive film was a compilation of scenes from the film: “Salo, or the 120 Days of Sodom” (Pasolini, 1975). The movie depicts physical and sexual abuse. The movie has been shown to be an effective measure in producing fear acquisition (see also Kindt, Van den Hout, & Buck, 2005; Kunze et al., 2015).

Conditioning Stimuli. There were two conditioned stimuli (CS): The reinforced conditioned stimulus (CS+) depicted a picture of the face of a man who appeared as one of the main characters in the aversive film. The unreinforced conditioned stimulus (CS-) consisted of a random man’s face (unrelated to film). The UCS was a 3 second film clip taken from the aversive movie, where a girl is forced to bite into a piece of cake with nails and screams (peak 85 dB) (Kunze et al., 2015). In the reinforced CS presentations (fear acquisition), the CS+ was followed by the UCS (100 % contingency rate) and in the unreinforced trials (fear renewal test), the CS+ was never followed by the UCS. The CS- was never followed by the UCS in any of the testing phases.

2.4 Data Analysis

2.4.1 Difference Scores and Effect Sizes of Key Time Points

Due to the small sample size in each pilot, traditional significance tests could not be applied. It is not uncommon that pilots with small samples therefore rely on effect sizes to interpret data and to draw results. See e.g. Kearney et al. (2013) for a specific example on a pilot on meditation for post-traumatic stress disorder and Moore et al. (2011) for a description of recommendations for planning pilot studies in clinical and translational research. Therefore, the data analysis relied on visual inspection of mean data points generated into point plots and effect sizes of difference scores of time points important for the interpretation of the results. For EMG data, within-subject z-scores were calculated using the mean and standard deviation across all EMG measurement points for each participant. The participant ratings for each CS+ and CS- presentation for each testing point served as the subjective distress measure. The mean values were inserted in a table and point plotted to produce a graphical representation of the data. For both startle and distress, the following difference scores and respective effect sizes were calculated:

1) CS+ versus CS-, for all of the following key time points: first acquisition, last acquisition, first extinction, last extinction, and first renewal test score.

2) CS+ versus CS+ (from one key time point to another) to determine the evolution of the fear ratings and startle response toward CS+. These key comparisons entail (with denotations): first acquisition to last acquisition (CS+ACQ.first to CS+ACQ.last), first extinction to last extinction (CS+EXT.first to CS+EXT.last), and last extinction to first renewal test (CS+EXT.last to CS+TEST.first). The same was done for CS- ratings respectively.

(10)

2.4.2 Post-Hoc Power Analysis

A post-hoc power analysis was performed to determine the power for the pilot with the smallest effect size of differentiation on last extinction trial, largest effect size of CS+EXT.last to CS+TEST.first, and largest effect size of differentiation on first renewal test. There was not one pilot that fulfilled all of these qualifications. The subjective distress scores of Pilot 4 come closest to these criteria in terms of a medium-large effect size of d=.42 from CS+EXT.last to CS+TEST.. Yet, even on this pilot, a large

differentiation between CS+ and CS- was found on the last extinction trial (d=1.42), and an even smaller differentiation on the first trial of the renewal test (d=1.20). As pilot 4 consisted of 8 participants, a power analysis with an alpha of 0.05 for effect size .42 reveals a power of 0.27 for CS+EXT.last to CS+TEST.first, thus representing an indadequate power. If the sample had comprised 40 participants, a power of 0.81 would have been obtained, thereby fulfilling the international power standard of 0.80 (Cohen et al., 2003).

2.5 Procedure

A series of five pilots was conducted. Fear responses were measured throughout the entire experiment with both subjective distress ratings and fear potentiated startle. Whereas all five pilots differ to various extents in their procedure and method, they all dispose over the general 3-context ABBA differential fear conditioning structure that each participant goes through: acquisition, intervention, extinction, fear renewal test. These stages occur in three phases, which in pilot 1 also represent the individual testing days: Phase 1= acquisition (A), Phase 2= intervention (B), Phase 3= extinction (B) + fear renewal test (A). A and B stand for the different light color contexts: The light context was a full-surround light, which was diffused equally throughout the entire room. Context A constituted a red light (RGB color code: 255,0,0), and context B a blue light (RGB color code: 0,0,255).

As the original aim of the procedure was to test light-color context induced fear renewal, the context mechanism itself needed to be proven effective before the IR intervention was implemented, whose aim was to reduce fear renewal. Therefore, the procedure of the pilots was adjusted to solely consist of a control condition intervention, which consisted of an intervention of imagery exposure (IE). All participants were subjected to the same experimental procedure, as outlined below. 2.5.1. Preparation and Screening

Before any testing took place, participants underwent a medical screening, where any prior incident of physical or sexual abuse functioned as exclusion criteria. Then participants filled out the STAI-T and the QMI. Afterwards, participants were brought into the room where the study took place and the researcher attached the electrodes.

2.5.2 Testing Phase 1: Habituation and Acquisition (Context A)

Participants were asked to fill out the Emotion-Mood VAS, STAI-S1, PANAS-1, and SAM-1. Then, participants were presented with 8 startle probes, to habituate participants to the noise.

(11)

blink startle reflex was assessed in response to both stimuli to determine baseline levels of responding. Once habituation ended, the researcher returned to the room and instructed participants that they would be presented with two pictures, one of which would depict a man who was in the movie they were about to watch. Participants were explicitly told that a short film clip would always follow the picture of the man who starred in the movie. Also, they were told that the other picture, a picture of a man who did not appear in the movie, would never be followed by a short film clip. Participants were asked to repeat the instructions to ensure they comprehend the difference between the pictures and their consequences. Participants then watched the aversive 12-minute movie, and after the movie, two habituation startles were presented again. Then fear acquisition started. The acquisition entailed three presentations of each CS+ (reinforced), CS- (unreinforced) and NA. After conditioning, participants were asked to fill in questionnaires again in the respective order to assess emotion and mood changes due to the conditioning procedure: Emotion-Mood VAS, STAI-S2, PANAS-2, contingency awareness, and a manipulation check for the aversive movie and film clip.

2.5.2 Testing phase 2: Control Condition Intervention (Context B)

This phase started with participants completing the Emotion-Mood VAS, STAI-S3, PANAS-3. Then, IE began, which consisted of a 6-minute audio rehearsal of the aversive movie. After IE, extinction took place. The subsequent extinction was implemented because IE represents a form of exposure as in the extinction itself, and therefore the IE condition was not assumed to have any effects on the conditioning results, hence why exposure was implemented as a control condition. Furthermore, the IE condition allowed to control for general effects of mental imagery that may affect subsequent fear renewal. The control condition was implemented instead of IR (experimental condition) for the purpose of testing whether color context-switch can elicit fear renewal. After the testing phase, participants again filled out Emotion-Mood VAS, STAI-S4, PANAS-4, a manipulation check for IE, in the respective order.

2.5.3 Testing phase 3: Extinction (Context B) and Renewal (Context A)

This testing phase began with completion of Emotion-Mood VAS, STAI-S5, PANAS-5. Then participants underwent a classical extinction procedure to extinguish all remaining fear responses. The extinction was administered prior to fear renewal in order to flatten out all fear responses from CS+. Only if all fear responses are extinguished, fear renewal can be measured reliably and its effects can be attributed to context-switch and thereby be distinguished from spontaneous recovery. After the

extinction, the color context was changed to context A during unreinforced CS+ presentation, marking the beginning of the renewal test. Between extinction and the final fear renewal test the light was always set out for 1000 ms to mark a dinstinction between the two contexts. Emotion and mood changes were assessed with subjective distress ratings and startle responses again, in order to

determine whether the context change brought about the expected fear renewal on the CS+. After this testing phase the questionnaires STAI-S6, PANAS-6, SAM2, and the Post-Experimental

(12)

debriefed and receive their compensation, either monetary (25 euros) or participation points (2,5). Order of trial type (i.e., CS+, CS-, NA) and color context (ABA/BAB) will be counterbalanced within and across conditions.

2.6 Results

The following results are divided into two sections: 1) A section dedicated to evaluation of fear acquisition and extinction, which represent the prerequisites to testing of the paradigm 2) a section on the fear renewal test. Fear renewal is the primary outcome measure of interest and is taken in the final context in Block 3 (ABBA), to test the hypothesis of whether a context-switch to the original context of fear acquisition (ABBA) elicits the previously extinguished fear responses. Fear acquisition and extinction are referred to as prerequisites as their successful fulfillment is required to conduct the fear renewal test and make a valid interpretation of it. Thus, these prerequisites need to be met to answer the question of study, but are not primary outcome measures of focus.

3.0 Pilot 1 3.1 Method Pilot 1 3.1.2 Participants

Pilot 1 consisted of 9 participants, from which participant 6 was identified as an outlier, due to providing distress ratings maximally up to 7 in all blocks following the extintion (distress: M= 1,95, SD= 1,87).

3.1.3 Procedure

In pilot 1, each of the three testing phases took part on separate days, resulting in a 3-day fear conditioning paradigm, with habituation and acquisition on day 1, control condition intervention on day 2, and extinction and renewal test on day 3. The extinction consisted of two blocks of each 8 trials, with a break of 10 seconds in between. Two extinction blocks were implemented in order to test whether simply the switch from one block to another, within the same context, thus without context-switch, may elicit spontaneous recovery. The three-day procedure was designed to ensure memory consolidation of the fear acquisition and intervention overnight. The color contexts comprised blue and red light in the laboratory. The experiment started and ended with colored light on in the laboratory. The computer background screen was always black.

3.2 Results Pilot 1 3.2.1 Prerequisite: Fear Acquisition

Subjective distress. Visual inspection of the data (see Figure 1a) indicated that acquisition was successful, namely that the CS+ received higher distress ratings than the CS- in all of its three

presentations. In line with the expectations, the mean difference scores of CS+ and CS- ratings diverged (incline on CS+, decline on CS-) from first stimulus presentation (first acquisition: x̄diffscore = 21,25; d=0.99) to last (last acquisition: x̄diffscore = 36,88; d=1.92). The effect sizes, especially the latter, exceed the convention for large effect size (d= .80, Cohen 1988), thus it can be concluded that the acquisition produced robust differences in distress ratings for CS+ and CS-. The mean difference

(13)

scores and medium effect sizes of CS+ACQ.first to CS+ACQ.last (x̄diffscore = 10,13; d=0.52) and CS+baseline to CS+ACQ.last (x̄diffscore = 35.38; d=1.66) further support the increase in distress. In light of this data it can thereby be concluded that the acquisition was successful.

Fear-potentiated startle. Upon visual inspection (see Figure 1b), the data suggest that stimuli were directly differentiated after the first acquisition trial, and remained stable for the rest of

acquisition: CS+ACQ.first to CS+ACQ.last (x̄z-diffscore = 0,07; d= 0.07) and CS-ACQ.first to CS-ACQ.last (x̄z-diffscore < 0,01; d < 0.01). The large visual differentiation between CS+ and CS- is supported by medium effect sizes on the last acquisition, (x̄z-diffscore = 0,29; d= 0.33). The startle data corroborate the subjective distress data that fear acquisition was successful.

3.2.2 Prerequisites: Extinction

Subjective distress, extinction block 1. The continuous differentiation between CS+ and CS- ratings from beginning to end of extinction indicate that no extinction took place: the CS+ lie above all CS- for each presentation, with medium to large effect sizes at the first presentation, CS-EXT.first to CS+EXT.first (d=0.66), and last extinction, CS+EXT.last to CS-EXT.last (d=0.78). Visual inspection (see Figure 1a) of the mean values in extinction block 1 indicate that there was no change in CS+ distress from beginning to end of extinction. Effect sizes of almost zero of mean differences furthermore support the lack of extinction, with CS+EXT.first to CS+EXT.last (x̄diffscore = 1,25; d=0.07). As expected for CS- evolution, there was no change from CS-EXT.first to CS-EXT.last (x̄diffscore = 0,875 d=0.049). Note that there was a small incline of CS+ ratings from last extinction block 1 presentation to first extinction block 2 presentation (d=0.30). This difference was examined to test whether there is an effect of simple within-context block switch on fear recovery. The small effect size illustrates a small effect of fear recovery due to block switch exclusively.

Subjective distress, extinction block 2. As in the previous extinction block, CS+ and CS- are differentiated throughout the entire extinction, from beginning (CS-EXT2.first and CS+EXT2.first; xdiffscore = 21 d=1,03) to end (CS+EXT2.last to CS-EXT2.last; (x̄diffscore = 8 d=0.40). The data therefore support that no extinction took place. Even though the decline in CS+ itself from beginning to end of extinction (CS+EXT2.first to CS+EXT2.last; d= 0.762) appear to suggest extinction, no difference between CS- and CS+ ratings at the end of the block is required in order to qualify as extinction. The CS- ratings remained relatively stable during extinction (CS-EXT2.first and CS-EXT2.last; d=0.05). Due to the CS differentiation it cannot be determined whether any fear renewal in the next block is due to context change or random recovery. Thus, it is concluded that the extinction cannot be considered a valid base to conduct a subsequent fear renewal test.

Fear-potentiated startle, extinction block 1. CS+ and CS- are somewhat differentiated in the first half of the extinction, but this differentiation seems to fade towards the end, CS+EXT.last to CS-EXT.last (x̄z-diffscore = 0,08; d=0.12), providing support for extinction. Even though CS+ and CS- both exhibit rather large fluctuations, the overall pattern indicates a decline on both CS+ and CS- (see Figure 1b). The effect sizes for difference scores on each CS+ and CS- from beginning to end of

(14)

extinction support the graphical indication: CS+EXT.first to CS+EXT.last (x̄z-diffscore = 0,72; d=1.07) and CS-EXT.first to CS-EXT.last (x̄z-diffscore = 0,1,14; d=1.13). These data furthermore suggest that extinction occurred on a physiological level. Interestingly, an incline from last extinction block 1 to first extinction block 2 can be seen on both CS+ and CS-, indicating an effect of simple within-context block switch on fear recovery: CS+EXT.last to CS+EXT2.first (x̄diffscore = 0,68; d=0.88) and EXT.last to CS-EXT2.first (x̄z-diffscore = 0,93; d=0.93).

Fear-potentiated startle, extinction block 1. The extinction block started with a small differentiation of CS- and CS+ (x̄z-diffscore = 0,33; d= 0.29) and ends in almost no differentiation ( x̄z-diffscore = 0.10; d= 0.14), indicating that extinction was successful. Furthermore, an overall decline can be seen on both CS+ and CS-, despite some fluctuations in between. Medium to large effect sizes support this visual indication (see Figure 1b): CS+EXT2.first to CS+EXT2.last (x̄z-diffscore = 0,67; d= 0.86) and CS-EXT2.first to CS-EXT2.last (x̄z-diffscore = 0,90; d= 0.94).

3.2.3 Fear Renewal Test

Subjective distress. To determine whether fear renewal took place, a difference score between the last extinction CS+ presentation (CS+EXT.last ) and the first CS+ presentation after the final context change (CS+TEST.first ) was calculated. The resulting difference score of 2.88 yields a very small effect size of d=0.15. Thus, the data indicate that the context change did not work. The small effect size of the difference score of CS+TEST.first and CS-Test-first (x̄z-diffscore = 4,25; d=0.206) rating furthermore support that fear renewal was not successful, as the stimuli should have been greatly differentiated again if the context change had worked. The very small difference score and effect size of EXT.last and CS-TEST.first (x̄z-diffscore = 0,87; d=0.04) showed no effect of context change of CS- on distress ratings, as expected.

Fear-potentiated startle. A medium incline was seen on CS+ from extinction to fear renewal, (CS+EXT2.last and CS+Test.first: x̄z-diffscore = 0,30; d=0.44), providing support for fear renewal. Furthermore, an incline was seen on CS-: (CS-EXT2.last and CS-Test.first : x̄z-diffscore = 0,32; d=0.41). Nevertheless, there is almost no differentiation between CS+ and CS- on the first renewal test (x̄z-diffscore = 0,12; d=0.14), which might indicate non-differential fear renewal. Yet, this differentiation is necessary in addition to the observed CS+ context-switch incline in order to conclude that fear renewal took place. The startle data are in line with the indications of the distress ratings that fear renewal did not take place. 3.2.4 Figures Pilot 1

(15)

Figure 1b. Pilot 1 EMG

3.3 Discussion Pilot 1

The acquisition was successful as indicated by distress and startle data. In light of the fact that the average subjective distress ratings of the CS+ were higher than those for the CS- across the entire study, there are two important implications: 1) the present acquisition represents a very powerful differential fear conditioning procedure and 2) the differentiation of CS+ and CS- throughout the whole extinction phase might indicate that extinction did not take place. The startle data on the other hand indicated fear extinction. In contrast to the expectation, the data of both measures revealed that no fear renewal took place. The author put forth the following explanation for the unexpected findings: fear renewal may not have taken place because participants may have failed to make an association between the color context of acquisition on day 1 and the final color context of the fear renewal test at the end of day 3 (see 3.2 for a thorough discussion of this argument). Therefore, the pilot was

modified to consist of a 1-day procedure (pilot 2), with the aim of producing a stronger association between the acquisition and the final context, and hence greater fear renewal. Lastly, the extinction in pilot 2 was reduced down to one block for practical reasons, to accommodate for the reduction to a 1-day procedure.

4. 0 Pilot 2 4.1 Method Pilot 2 4.1.1 Participants

Pilot 2 consisted of 5 participants. No participants were excluded. 4.1.2 Procedure

In pilot 2, the three blocks (ABA) took place consecutively on one single day. The reason for the reduction from a three-day to one-day procedure was that it is assumed that if the entire procedure takes place on one day that the context-switch is less subtle and thereby has a greater impact on the fear learning procedure. As such, participants may be more likely to make a connection between color context and the individual phases. Another change from pilot 1 to pilot 2 comprised the adjustment of trial number: pilot 2 extinction consisted of only one block of 8 trials, instead of two blocks. The color contexts comprised blue and red light as in pilot 1. Due to the procedure now taking place all on one day, a regular white room light was installed to appear at the beginning, between the individual contexts, and at the end of the experiment. The computer background screen was always black.

(16)

4.2 Results Pilot 2

Due to a technical failure, there are no startle data available for pilot 2. However, previous research using the present fear conditioning procedure has found that the subjective distress ratings are more susceptible to subtle changes in fear responding than fear-potentiated startle (e.g., Kunze et al., 2015). Therefore, the authors concluded that interpretation of the distress data would be sufficient to evaluate the pilot.

4.2.1 Prerequisite: Fear Acquisition

Subjective distress. Visual inspection of the data indicated that the acquisition worked (see Figure 2a): Mean difference scores and corresponding effect sizes support the diverging pattern of CS+ and CS- (first acquisition: x̄diffscore = 30,11; d=1.86 and last acquisition: x̄diffscore = 51,66; d=2.23), with a steady incline of CS+ and decline of CS-. From the visual inspection of first to last CS+ ratings, a small to medium but steady increase of distress was identified. The medium effect size of CS+ACQ.first to CS+ACQ.last (x̄diffscore =11,88 ; d=0.35) support this visual indication. Even though this effect is not large, the effect size for baseline to last acquisition support that there is a clear difference between the respective ratings: CS+baseline to last CS+ACQ.last (x̄diffscore =43,4 ; d=2.41). Thus, it can be concluded that the acquisition was successful. As in pilot 1, exhibited a decline, probably due to habituation, CS-ACQ.first to last CS-ACQ.last (x̄diffscore =9,66 ; d=0.54).

4.2.2 Prerequisite: Extinction

Subjective distress. The continuous differentiation between CS+ and CS- on all time points of extinction indicates that the extinction did not work and thus did not provide a base for subsequent fear renewal: CS+EXT.first to CS-EXT.first (x̄diffscore = 15,13; d= 2.39) and CS+EXT.last to CS-EXT.last (x̄diffscore = 25,85; d= 0.89). The graphical representation of means (see Figure 2a) for the CS+ evolution does not indicate a clear decline in CS+ ratings, but the difference scores of first to last CS+ (9,2) and its medium effect size (d=0.401), suggest that extinction took place. The large decline of CS+ from last acquisition to first extinction (x̄diffscore = 28,2; d= 1.56), which furthermore prevailed throughout the entire extinction, suggest that the switch from acquisition to extinction already flattened out all effects on CS+ already. Ultimately, despite the support of a decline on CS+ itself, the differentiation between CS+ and CS-, both at beginning and end of extinction, ultimately signify that extinction did not take place.

Subjective distress. The difference score of 7.6 from CS+EXT.last to CS+TEST.first yields a small effect size of d=0.32, suggesting small fear renewal. The large effect size of the difference score on the first renewal test presentation shows that the stimuli were differentiated (CS+TEST.first and CS-Test.first : x̄diffscore = 35,6; d=2.08). Nevertheless, this differentiation cannot be regarded as evidence for fear renewal, as the stimuli were still differentiated by the end of extinction (d= 0.89). In light of the fact that the prerequisite of a successful extinction did not take place prior to the fear renewal test, namely

(17)

no differentiation between CS+ and CS- at the end of the block, no valid conclusion can be drawn of whether fear renewal took place.

4.2.4 Figures Pilot2

Figure 2a. Pilot 2 Subjective Distress

CS+ and CS- were differentiated throughout every stage of the experiment, which can be attributed to a powerful acquisition. Whereas a powerful acquisition is good, the differentiation persisted throughout the entire extinction phase as well. These results suggest that context is not a predictor for the CS-UCS association, as has been shown in other context-dependent fear conditioning experiments (e.g. Kindt et al., 2009). In the general discussion it will be discussed why this might be the case, and what can be done to increase the context-dependency of the acquisition memory in order to test fear renewal. Due to the large CS differentiation at the end of extinction, no conclusion can be drawn on the effect of final context-switch on fear responses. The author suggests that the IE may have diminished the extinction and fear renewal even more, as the IE consists of rehearsal of the trauma movie. Therefore, the IE was taken out in pilot 3 and replaced with a control condition of a 25-minute waiting period with the aim of testing whether there would be a fear renewal without any intervention at all.

Pilot 3 remained in the same 1-day structure as pilot 2. The main difference to pilot 2 was the removal of the IE and its replacement with a 25-minute pause during which participants were asked to wait on their chair and were offered magazines to read. Furthermore, the extinction block was

extended from 8 to 20 trials to facilitate extinction. The aim of the latter was to make the extinction more powerful. Furthermore, the fear renewal test trials were reduced from 8 to 4 trials for practical reasons. Another change from pilot 2 concerned the light color for context: instead of blue and red light, the colors now constituted pink and green, which were previously shown to successfully induce

(18)

ABA-renewal for US-expectancy ratings (Leer et al., 2013). Just like in pilot 2, the regular room light was set at the beginning and end of the experiment and in between contexts. The background screen was always black.

Subjective distress. Visual inspection of the data (see Figure 3a) indicated a clear

differentiation between CS+ and CS-, namely that the CS+ received higher mean distress ratings than the CS-: Mean difference scores and corresponding effect sizes support the graphical diverging pattern of CS+ (incline) and CS- (decline): first acquisition: x̄diffscore = 30,11; d=1.86 and last acquisition: x̄diffscore = 51.66; d=2.23. Visual inspection and a medium effect size of CS+ from first to last acquisition furthermore support the increase in distress (CS+ACQ.first to CS+ACQ.last: x̄diffscore = 11.88; d=0.35). As in the previous pilots, a mild decline of CS- was seen, CS-ACQ.first to last CS-ACQ.last: x̄diffscore =9.66 ; d=0.54.

Fear-potentiated startle. Upon visual inspection (see Figure 3b), the data yielded roughly the expected acquisition pattern: an incline of CS+ and decline of CS-. It should be noted that for both stimuli the peaks of the respective inclines and declines are seen on the second acquisition and not the third. Still, effect sizes from the difference score of first to last extinction show the expected

evolutions for each CS+ (incline) and CS- (decline) respectively : CS+ACQ.first to CS+ACQ.last (x̄z-diffscore = 0,53; d= 0.54) and of CS-ACQ.first to CS-ACQ.last (x̄z-diffscore = 0,22; d= 0.34). Thus, the startle data

corroborate the distress data that fear acquisition was successful. 5.2.2 Prerequisite: Extinction

Subjective distress. Upon visual inspection of the graph (see Figure 3a), there seems to be a steady decline in distress from first to last CS+, as expected. A medium to large effect size supports this: CS+EXT.first to CS+EXT.last (x̄diffscore = 21,33; d= 0.65). The CS- furthermore experienced a decline (x̄diffscore = 7,22; d= 0.74), which can be explained by potential generalization of extinction on CS+. Despite indications of extinction when looking at each CS evolution by itself, visual inspection shows that CS+ and CS- are differentiated throughout the entire extinction, which ultimately signifies a lack of extinction. Effect sizes from beginning to end of extinction support this: CS-EXT.first and CS+EXT.first (x̄diffscore = 35,88; d=1.58), and CS+EXT.last to CS-EXT.last (x̄diffscore = 21,77; d=1.11). These large effect sizes, especially the latter, again illustrate the problem of not being able to attribute the observed fear renewal to a successful context-switch or spontaneous recovery. Therefore, as in the previous pilots, the extinction does not represent the optimal base for subsequent fear renewal interpretation. Fear-potentiated startle. The graphical data (see Figure 3b) indicate a decline of CS+ from first to last extinction (x̄z-diffscore = 0,77; d= 1.38) and no significant change in CS- (x̄z-diffscore = 0,11; d= 0.11). Even though these CS evolutions appear to suggest extinction, it cannot be regarded as enough evidence for successful extinction because the last extinction trial shows a large differentiation between CS+ and CS- (x̄z-diffscore = 0,66; d= 0.90). The startle data are thereby in agreement with the

(19)

distress data that the extinction did not provide a sufficient base for the fear renewal test. 5.2.3 Fear Renewal Test

Subjective distress. Visual inspection of the data (see Figure 3a) clearly indicated that fear renewal did not take place: Effect sizes support this: CS+EXT.last and CS+TEST.first (x̄diffscore = 5,33; d=0.15). Furthermore, the effect size for the difference score of CS+TEST.first and CS-Test-first (x̄diffscore = 25,44; d=1.10) rating shows that the stimuli are still differentiated, but again this cannot be regarded as evidence for fear renewal due to remaining differentiation at the end of extinction. The very small difference score and effect size of CS-EXT.last and CS-TEST.first (x̄diffscore = 1,66; d=0.18) shows no effect of context change on distress ratings, as expected.

Fear-potentiated startle. The data showed an incline for CS+ (x̄z-diffscore=0,21; d=0.44) and a large decline for CS- (x̄z-diffscore = 0,41; d=0.49), which would be expected for fear renewal. Yet, this pattern only provides support for fear renewal if the individual evolution of CS+ and CS- were examined in isolation. When supplementing this data with the other available data, namely the CS differentiation at the last extinction and first renewal test (x̄z-diffscore = 0,03; d=0.06), it becomes evident that the opposite pattern in CS differentiation took place than it should have: Fear renewal would be characterized by no differentiation upon last extinction and large differentiation upon final switch to the fear renewal test block, yet the data showed the reverse (Figure 3b).

5.2.4 Figures Pilot 3

Figure 3a. Pilot 3 Subjective Distress

(20)

As in the previous pilots, CS+ and CS- were continuously differentiated on subjective distress throughout the entire experiment, with the smallest difference (x̄diffscore = 17,66, d= 1.45) at the 17th extinction still reflecting a large differentiation. Both startle and distress data indicate that the extinction was not effective, whereby it can be assumed that extending the extinction trials from 8 to 20 did not improve extinction of the fear response. The distress and startle data suggest that no fear renewal took place, but this conclusion cannot be made due to the prevailing CS differentiation at the end of extinction for both startle and distress.

The major change from pilot 2 to pilot 3 was the removal of the IE condition to control whether context change works when there is no intervention at all. Despite the problem of not being able to attribute the lack of fear renewal with certainty to a failure in context mechanism (due to CS

differentiation in extinction), the following augmentation for pilot 4 was nevertheless implemented in hope of yielding a stronger fear renewal response: If it were assumed that no fear renewal took place, it can be suggested that removal of IE did not strengthen fear renewal, as predicted. If IE was thereby not responsible for the failed context-switch in previous pilots, it can be assumed that fear renewal also does not occur even when there is no intervention at all. Therefore, attention was directed to context-specific characteristics: It was suggested that participants did not associate the room light color enough with the stimuli presentations on the computer screen. Potential reasons for this failed association are discussed in the general discussion. Therefore, in the next pilot the background color of the screen and not the room light color constituted the context. It was expected that this might create a stronger association between the presented stimuli and the color context, thereby producing fear renewal.

Pilot 4 was the same as Pilot 3, with the exception that the color context now constituted the background color of the computer screen and the lab lights were set to regular white light. The background screen only turned black during the instructions and between the different blocks. Between extinction and fear renewal test block the background screen color did not turn black, meaning a direct switch from context A to B. The visual analogue scale was set from colored to light to dark grey.

Subjective distress. Visual inspection of the data (see Figure 4a) indicated a clear

(21)

declining pattern throughout aqusition: CS+ACQ.first to last CS+ACQ.last (x̄diffscore = 3; d= 0.15), CS-ACQ.first to last CS-ACQ.last (x̄diffscore =9,5 ; d= 0.74). As the data show, only the effect size of decline is large for CS-, and therefore the slight decline on CS+ is not viewed as problematic. Furthermore, CS+ and CS- are greatly differentiated at first (CS+ACQ.first to CS-ACQ.first: x̄diffscore = 39,5; d= 2.72) and even more so at last acquisition (CS+ACQ.last to CS-ACQ.last, x̄diffscore = 46; d= 2.61). Thus, even though there was a small effect of CS+ decline within the acquisition itself, its differentiation to CS- and its sharp incline from baseline level (x̄diffscore = 42,12; d=2.20) allows concluding that fear acquisition succeeded.

Fear-potentiated startle. Upon visual inspection (see Figure 4b), the data yielded roughly the expected acquisition pattern: an incline of CS+ and decline of CS-. Large effect sizes of difference scores support the visual indication: CS+ACQ.first to CS+ACQ.last (x̄z-diffscore = 1,18; d= 1.32) and CS-ACQ.first to CS-ACQ.last (x̄z-diffscore = 0,53; d= 1.06). Thus, the startle data corroborate the distress data that fear acquisition was successful.

Subjective distress. Upon graphical inspection (see Figure 4a) there seems to be a steady decline in distress from first to last CS+, as expected. A large effect size supports this indication: CS+EXT.first to CS+EXT.last (x̄diffscore = 24,12; d= 1.04). The CS- follows a similar pattern: CS-EXT.first to CS-EXT.last (x̄diffscore = 17,25; d= 2.25 ), but on average 26.81 points lower. Whereas CS+ and CS- viewed in isolation from each other appear to indicate extinction, the prevailing differentiation between them supports that extinction did not take place: CS+XT2.first and CS-EXT.first (x̄diffscore = 26,37 ; d= 1.54) and CS+EXT.last to CS-EXT.last (x̄diffscore = 19,5; d= 1.42).

Fear-potentiated startle. A rather strong decline can be seen from first to last extinction (see Figure 4b): CS+EXT.first to CS+EXT.last (x̄z-diffscore = 1,18; d= 1.77) and a modest change in CS- ratings: CS-EXT.first to CS-EXT.last (x̄z-diffscore = 0,18; d= 0.52 ). There is only a small to medium differentiation between CS+ and CS- at last extinction (x̄z-diffscore = 0,15; d= 0.37), and the visual inspection supports this indication as respective CS points seem to lie almost on top of each other. Thus, startle data would suggest that extinction was not optimal but did occur.

Subjective distress. Looking at the last extinction and first fear renewal score on CS+ (see Figure 4a), a small incline can be seen. A medium effect size supports this indication: CS+EXT.last to CS+TEST.first (x̄diffscore = 11,25; d=0.41). Even though the large effect size of the difference score of CS+TEST.first and CS-Test-first (x̄diffscore = 25,25; d=1.2) shows that the stimuli are largely differentiated, these data cannot be taken as evidence for fear renewal due to the prior differentiation in extinction. Fear-potentiated startle. The startle data of extinction actually suggest that a sufficient base for valid fear renewal data was provided (see Figure 4b). Therefore the fear renewal test data are of specific interest: No change can be seen on CS+ from last extinction to first renewal test (x̄z-diffscore = 0,04; d= 0.10), and furthermore not in CS-, as expected (x̄z-diffscore = 0,03; d= 0.07). Thus, in light of a

(22)

preceding sufficient extinction (on startle), it can be suggested that fear renewal did not take place on a physiological level.

6.2.4 Figures Pilot 4

Figure 4b. Pilot 4 EMG

Pilot 4 was the only pilot next to pilot 1, which disposed over a relatively sufficient extinction (on startle). Since extinction was not found on distress, only valid interpretations could be made for startle fear renewal. As the results show, fear renewal did not take place, as in pilot 1, even though all prerequisites were fulfilled. Nevertheless, the successful extinction on startle shed light on possible important mechanisms which may have rendered the extinction more successful: The background screen as color context may have provided more of a specific contextual cue, and thereby allowed a more direct association between CS and color context. The last pilot wanted to explore whether taking out color context completely and presenting subjects with a stimulus related to the UCS would

reactivate the conditioned response, by providing a reminder cue to the original fear memory. 7.0 Pilot 5

7.1 Method 7.1.1 Participants

(23)

7.1.2 Procedure

As there was no color context anymore, computer screen color was now black and the lab lights a simple white. The beginning 10 seconds of the movie were presented between extinction and the final block. Since a stimuli related to UCS was presented and not the actual UCS itself, this procedure is referred to by the present author as “return of fear” rather than traditional reinstatement. For this stimuli, the beginning first 10 seconds of the aversive movie were chosen. This excerpt represents a “compromise” between reexposure to the original UCS (i.e., reinstatement) and induction of a ‘reminder’ cue (such as a context manipulation). Thus, as the film clip represents both a reminder to the UCS but also reactivates the original memory representation, it was predicted that fear would return upon its presentation.

7.2 Results 7.2.1 Prerequisite: Fear Acquisition

Subjective distress.Visual inspection of the data (Figure 5a) indicates the same pattern as in the previous pilots: There is a clear differentiation between both stimuli, CS+ACQ.first to CS-ACQ.first, (x̄diffscore = 20,5; d= 1.06) and CS+ACQ.last to CS-ACQ.last, (x̄diffscore = 36,62; d= 1.86). CS+ shows a peculiar pattern, by flaring up in the second acquisition and then declining back to the level of first acquisition:

CS+ACQ.first to last CS+ACQ.last (x̄diffscore = 0.75; d= 0.035). Even though there is only a very small increase in CS+, the large difference in baseline to last acquisition (x̄diffscore = 20,5; d= 0.89) and its

prevailing large differentiation to CS- suggest that the acquisition was nevertheless strong enough. Fear-potentiated startle. A clear acquisition pattern can be identified upon visual inspection

(see Figure Fb), with CS+ inclining and CS- declining from first to last extinction (CS+ACQ.first to last CS+ACQ.last : x̄z-diffscore = 0,32; d= 0.37 and CS-ACQ.first to last CS-ACQ.last: x̄z-diffscore =9,5 ; d= 0.74). Furthermore the stimuli are largely differentiated from beginning (x̄z-diffscore = 0,21; d= 0.76) to especially the end (x̄z-diffscore = 0,84; d= 3.04).

Subjective distress. Upon graphical inspection (Figure 5a) there seems to be a steady decline in CS+ distress from first to last extinction, as expected. A very large effect size supports this indication: CS+EXT.first to CS+EXT.last (x̄diffscore = 32,87; d= 1.37). The CS- follows a similar pattern: CS-EXT.first to CS-EXT.last (x̄diffscore = 14,62; d= 0.79), but on average 26,81 points lower. Yet, despite the individual favourable CS evolutions, there is a continuous differentiation between CS+ and CS- (note: the 12th extinction seems to identify as an outlier as CS- and CS+ are almost identical: x̄diffscore = 1,75; d= 0.06). Large and medium effect sizes support this pattern at beginning, CS+EXT.first and CS-EXT.first (x̄diffscore = 32,75; d= 1.84), and end of extinction CS+EXT.last to CS-EXT.last (x̄diffscore = 14,5; d= 0.59). Thus, the data indicate that extinction was not successful.

Fear-potentiated startle. The startle data show a large decline for CS+: CS+EXT.first and CS+EXT.last (x̄z-diffscore = 0,74; d= 1.24) and furthermore also a decline for CS-: CS-EXT.first and CS-EXT.last (x̄z-diffscore = 0,51; d= 0.79). Whereas these data would suggest extinction, the large differentiation

(24)

between CS+ and CS- at the end of extinction (x̄z-diffscore = 0,54; d= 2.49) again shows that the extinction was not successful (see Figure 5b), thus corroborating the respective distress results. 7.2.3 Fear Renewal Test

Subjective distress. A small incline can be seen in CS+ after the context change (see Figure 5a). The data supports this with a small effect size: CS+EXT.last to CS+TEST.first (x̄diffscore = 8,62; d=0.31). A large effect size of the difference score of CS+ and CS- right after the context change (CS+TEST.first and CS-Test-first, x̄diffscore = 22,75; d=1.13), shows that stimuli were differentiated, yet as in the previous pilots this cannot be seen as evidence for fear renewal because the extinction left off in differentiation (d= 0.59). No change in distress occurred for CS- as expected (x̄diffscore = 0.37; d=0.02).

Fear-potentiated startle. From last extinction trial to first renewal test, CS+ shows a small incline (x̄z-diffscore = 0,19; d=0.28) and CS- a large decline (x̄z-diffscore = 0,41; d=0.92). There is a large differentiation on first renewal test presentation (x̄z-diffscore = 0,75; d=2.69), but again this cannot be considered as evidence for fear renewal because of prior differentiation in the last extinction trial. 7.2.4 Figures Pilot 5

Figure 5b. Pilot 5 EMG

Due to differentiation for both distress and startle at the end of extinction, there is no sufficient base for interpretation of fear renewal on either measure. Pilot 5 implemented a rather simple fear conditioning paradigm (without context), thus the question is raised what might have caused the lack of extinction. As reinstatement has repeatedly been shown to be effective in eliciting return of the

(25)

conditioned response (e.g., Kunze et al., 2015; Soeter & Kindt, 2015), it was expected that this similar form of reinstatement would elicit return of fear.

8.0 General Discussion

8.1 Recapitulation of Results and Formulation of Discussion Points

The original aim of the current study was to investigate UCS devaluation as possible working mechanism of IR, by implementing the treatment technique as an intervention prior to a fear renewal test. Since the original study was never executed, and instead a series of 5 pilots on the control condition was conducted, the investigation shifted from hypothesis testing to establishing a reliable paradigm. The discussion sheds light on limitations and strengths of the present fear conditioning procedure.

The 5 pilots can be summarized as follows: All pilots disposed over a successful acquisition, supporting the novel paradigm’s powerful ability to induce a mini-traumatic fear memory. Yet, extinction and fear renewal proved to be less successful: From pilot 1 to 2, it was suggested that the 3-day procedure may have introduced too much of a disassociation between context and the CS-US association, and therefore no connection may have been made between the first context (acquisition) on day 1 and the last context (fear renewal) on day 3. When no fear renewal was observed after the adjustment to a 1-day procedure, it was suggested that IE inhibited the extinction on subjective distress too much (successful extinction on EMG), whereby IE was replaced with a 25-minute waiting period for the rest of the pilots. Another aim of the IE exclusion was to investigate whether context-switch occurs without any intervention at all. Nevertheless, when pilot 3 also did not show any fear renewal, it was suggested that the room light color did not sufficiently bind to the stimuli on the screen, failing to create a context-CS association. Therefore the computer screen background color was set as color context for pilot 4. With no success in fear renewal in pilot 4 either, the color context was taken out completely for pilot 5 and instead a small fragment from the aversive movie was inserted at the former time point of context-switch. This form of reactivation furthermore did not show success in eliciting fear renewal. Even though CS+ incline before and after context-switch considered in isolation appeared to suggest fear renewal, the prevailing differential CS ratings at beginning and end of extinction of all pilots on subjective distress did not allow conclusion of whether fear renewal, spontaneous recover, or neither took place. For EMG, extinction showed to be succesful on pilots 1 and 4, as there were no differential ratings on startle at the end of extinction, thus allowing for valid renewal test interpretation on the startle response. Yet, the pilot data indicated no increase of CS+ and CS differentiation after context-switch. Thus, whereas for most pilots it was generally not valid to draw conclusions about fear renewal because of lack of extinction, the EMG pilots 1 and 4 that did succeed on extinction did not show fear renewal. It will be discussed in section 8.3 below what may account for the discrepancy between successful extinction for half of the pilots that implemented fear-potentiated startle, versus lack of extinction in all pilots on subjective distress.