Search

Supplemental material, sj-pdf-3-pss-10.1177_09567976231170878 for Generalization of Costly Pain-Related Avoidance Based on Real-Life Categorical Knowledge by Eveliina Glogan, Peixin Liu and Ann Meulders in Psychological Science

Supplemental material, sj-docx-2-pss-10.1177_09567976231170878 for Generalization of Costly Pain-Related Avoidance Based on Real-Life Categorical Knowledge by Eveliina Glogan, Peixin Liu and Ann Meulders in Psychological Science

Supplemental material, sj-pdf-4-pss-10.1177_09567976231170878 for Generalization of Costly Pain-Related Avoidance Based on Real-Life Categorical Knowledge by Eveliina Glogan, Peixin Liu and Ann Meulders in Psychological Science

The Best Possible Self (BPS) intervention is one of the currently most widely used Positive Psychology Interventions, with a large body of empirical evidence suggesting its effectiveness on several positive outcomes, such as optimism, well-being or positive future expectancies (Peters , Flink , Boersma & Linton, 2010; Meevissen, Peters, Alberts, 2010). The intervention combines a writing exercise where the participant imagines and writes about the best version of him- or herself in the future, followed by daily visualization of the “best possible self”. More specifically, in a first session participants think about their best possible self in terms of personal, relational and professional domains and are then asked to write about this mental image. During the following week, participants engage in a daily mental visualization of what they have written about their best possible self. The BPS has been found to exert beneficial effects regardless of duration, length and intensity of the intervention. However, the intervention has so far only been tested with focus on the self, thus concentrating on individual participants, and effects. The effectiveness of such an intervention on relationship outcomes has never been investigated before. We want to explore whether couples would benefit from visualizing the best version of their relationship in terms of sexual, relational as well individual/personal outcome variables. To examine the specific effects of a “best possible self” versus a “best possible us” intervention on individual and relational outcome variables, we will conduct an online study in couples. In the experimental condition, participants visualize the best possible version of their relationship (thus including their partner) in the future = the Best Possible Us intervention (BPU). As in the BPS, participants are asked to write down what they have imagined; this content is used for daily visualization exercises during the following week. The effectiveness of the BPU on various outcomes (e.g. relationship satisfaction, optimism, well-being) will be investigated and compared with the effects of the “standard” BPS intervention, which will also be implemented in both partners of a couple. Further, both of these will be compared to a active control condition (imagining a route to a certain location; RTL). Thus, in this study we will investigate whether the BPU has beneficial effects on various markers of optimism and relationship satisfaction beyond those that could be expected by practicing a BPS or active control (RTL).

Avoidance behavior is a hallmark symptom of anxiety disorders and chronic pain. Exposure-based treatments targeting fear reduction have been successfully used to treat fear and avoidance in these conditions. Although exposure is relatively successful, relapse often occurs, so there is still ample room for improvement. One type of relapse is renewal, i.e. the return of conditioned fear responding typically observed when individuals are tested outside of the context (e.g. daily life environment) in which extinction took place (e.g. therapy context). This phenomenon has been widely documented in the human fear conditioning literature but has rarely been examined with regard to avoidance behavior. A first set of studies by Urcelay, Symmons and Prével (under review) had addressed this gap and found renewal of avoidance when the feared stimulus was presented outside the extinction context. However, they reported substantial inter-individual differences with some participants not showing extinction of avoidance in the first place. Hence, there remains an urgent need to better understand the processes underlying extinction and renewal of avoidance behavior itself, and not merely as a proxy of fear. This study aims to conceptually replicate the study by Urcelay et al. and to investigate the role of individual differences in avoidance learning, extinction, generalization, and renewal using an aversively motivated task. Specifically, we will examine whether high and low trait anxious individuals show different learning patterns, and whether elevated arousal during extinction facilitates persistent avoidance behavior. We will adopt the design by Urcelay et al. but invite participants based on pre-assessed trait anxiety scores (high vs low anxious), thus employing a mixed design. It must also be emphasized that the generalization phase is new compared to Urcelay et al. To better model real-life therapy approaches, the extinction phase will be changed to an “extinction with response prevention” procedure, that is, participants will not be able to perform avoidance behavior during the extinction phase, so that they cannot falsely attribute safety to their own behavior (i.e. protection from extinction). The findings of this experiment are clinically relevant as they may underline the importance of tackling avoidance behavior in people living with anxiety disorders, obsessive-compulsive disorder or chronic pain alike, and point out potential obstacles. More specifically, the findings of this experiment may provide more insight into one of the major relapse mechanisms of avoidance behavior: renewal. By implementing these theoretical insights in the clinical practice, the long-term outcomes of exposure therapy may be improved.

The spreading, or generalization, of fear and avoidance is an adaptive mechanism, which allows organisms to apply the once-learned predictive value of a certain stimulus to perceptually similar future instances, thus sparing organisms from the need to always learn anew (Dymond et al., 2015). However, unrestrained fear generalization results in undue fear towards safe stimuli, causing unnecessary distress. For example, people with chronic pain have been shown to excessively generalize pain-related fear, compared to healthy controls (e.g. (Meulders et al., 2014; Meulders, Harvie, et al., 2015; Meulders, Jans, et al., 2015)). Furthermore, where fear unarguably causes discomfort, avoidance is a direct pathway to disability (Krypotos et al., 2015; Meulders, 2019). Yet, previous literature mainly focused on fear instead of avoidance, leaving the mechanisms underlying avoidance generalization poorly understood (Krypotos et al., 2015). In addition to generalizing based on perceptual similarity, humans possess the ability to generalize based on higher-order reasoning, such as conceptual similarity and category membership (e.g. (Dunsmoor & Murphy, 2015)). For example, following a painful movement during an exercise class, a person may generalize pain-related fear of that specific movement, to a range of different sports-related activities, even if on a proprioceptive/haptic level these behaviors are very distinct (Glogan et al., 2018; Meulders et al., 2017). Previous research showed generalization of fear of movement-related pain based on artificial categories (Bennett et al., 2015) established in a matching-to-sample (MTS) task (Skinner, 1950). More recently, also avoidance (of aversive stimuli) was shown to generalize based on artificial stimulus categories learned in an MTS task (Bennett et al., 2020). Concept- or category-based generalization introduces another level of complexity to understanding how people learn to fear safe stimuli and situations (Dunsmoor & Murphy, 2015), and may further complicate treatment. However, investigations into this conceptual pathway of fear generalization in the pain domain are scarce, and especially so concerning avoidance generalization. Yet, these are crucial considerations, given that excessive generalization of avoidance is one of the main disabling factors in chronic pain (Meulders, 2019). Therefore, we aimed to investigate the possibility that, similarly to fear of movement-related pain, pain-related avoidance behavior can also generalize based on artificial categorical relationships between responses (movements), established within a MTS task. Specifically, participants will be required to perform an arm-reaching task in a 2-dimensional horizontal movement plane whilst operating a robotic arm. The reaching task consists of moving a “green ball” from a starting point to a target location, on an LCD screen. Participants will be able to choose between 3 movement trajectories, visualized as arches on the LCD screen, to reach the target location on each trial. These will be either T1-3 for acquisition, and G1-3 for generalization. T1 and G1 will be the shortest, most direct trajectories to the target, T2 and G2 will be in the middle, and T3 and G3 will be the longest trajectories to the target. There will be two groups, differing in the equivalence categories they are taught. Specifically, for MTS-incongruent, the equivalence categories will be T1 = G3, T2 = G2, T3 = G1, and for MTS-congruent they will be T1 = G1, T2 = G2, T3 = G3. The equivalence categories will be taught to the participants at the beginning of the experiment, during the "MTS task". During this task, participants will learn their respective equivalence classes by means of trial and error. They will receive feedback following each trial, e.g. messages "Correct" or "Wrong" depending on whether they matched the two movements correctly or incorrectly, respectively. Subsequently, participants will advance to the "Acquisition phase", during which they will learn the experimental movement-pain contingencies. Specifically, during this phase, only the acquisition trajectories (T1-3) will be available, and participants can freely choose which movement to perform on each trial. However, T1 will be paired with an 80% chance of the participant receiving a painful electrical stimulus, T2 will be paired with a 40% chance of the electrical stimulus, and T3 will never be paired with the electrical stimulus (0% chance). After the acquisition phase, there will be a "Generalization phase". This phase will be similar to the acquisition phase, except that, only the generalization trajectories (G1-3) will be available, and none of these will be paired with the electrical stimulus.

Pain-related avoidance is adaptive when averting bodily harm, i.e. during acute pain. However, excessive spreading (overgeneralization) of avoidance to safe movements/behaviors can culminate in chronic pain disability (Meulders, 2019; Vlaeyen & Linton, 2012). According to the fear-avoidance model of chronic pain, pain-related avoidance is motivated by pain-related fear. Thus, pain-related fear has received much attention in the literature, at the expense of avoidance due to the assumption that avoidance would align. However, increasing evidence suggests that there is no 1-to-1 relationship between fear and avoidance (Glogan, Gatzounis, Meulders, & Meulders, 2020). Furthermore, avoidance, as opposed to fear, leads to disability due to directly decreasing daily activity, and pursuit of valued life goals (Meulders, 2019). Thus, achieving a better understanding of costly pain-related avoidance, and its generalization specifically, is of utmost importance. In differential Pavlovian fear generalization procedures, fear is first conditioned by pairing one stimulus (CS+) with the unconditioned stimulus (US), and by not pairing another (CS-) with the US, resulting in the CS+ signaling threat and the CS- signaling safety. Generalization is subsequently tested to novel stimuli (GSs) lying on a similarity continuum ending at the CS+ and CS-, respectively (Dymond, Dunsmoor, Vervliet, Roche, & Hermans, 2015). Previous research has shown that healthy participants show selective fear generalization (Meulders, Jans, & Vlaeyen, 2015), meaning that their pain-related fear responses decline with decreasing similarity to a movement CS+. Such declining fear responses are referred to as the fear generalization gradient, and are often taken as a measure of the degree of generalization (Lissek et al., 2008). In contrast to healthy controls, people with chronic pain exhibit less selective generalization of pain-related fear (Meulders et al., 2015), i.e. compared to healthy participants, chronic pain patients' fear levels remain high also to those movements resembling the CS-. Furthermore, a common phenomenon in differential fear generalization procedures is the "peak shift" effect, i.e. a shifted generalization gradient beyond the CS+ in the direction opposite to the CS- (Lee, Hayes, & Lovibond, 2018). Thus, peak responding occurs to a GS, which is an "extreme" version of the CS+. Furthermore, the "area shift", in which responses are simply elevated towards novel stimuli on the side of the CS+ away from the CS- (Honig & Urcuioli, 1981), often co-occurs with the peak shift. However, the area shift can also be observed even in the absence of a significant peak shift. Peak, and area shifts have commonly been explained as a consequence of interactions between conditioned excitation and inhibition to the CS+ and CS-, respectively. Although a common finding in the animal literature, in humans, the peak shift effect remains elusive and has produced inconclusive results. Furthermore, the peak shift effect has never been investigated in human avoidance behavior before, nor has it been studied with the incorporation of response-costs. Unlike fear, avoidance is conditioned instrumentally, i.e. behavior (the response, R; e.g. a movement) followed by a negative outcome (e.g. pain) decreases (i.e. punishment) (Burrhus Frederic Skinner, 1953). On the other hand, one may learn that another behavior (i.e. avoidance) leads to the omission of pain (the outcome, O) (i.e. negative reinforcement) (B. F. Skinner, 1953). Therefore, in a robotic arm-reaching paradigm (Meulders, Franssen, Fonteyne, & Vlaeyen, 2016), we will first instrumentally induce avoidance behavior in an Experimental Group by using two different arm-movements: one low-cost, pain-associated arm-movement (T+; 80% punished), and another high-cost, safe arm-movement (T-; never punished). The cost will be operationalized in terms of effort to overcome the resistive force and the deviation required to perform the arm-movement. Thus, during the Acquisition phase, participants in the Experimental Group will be able to avoid pain by choosing to perform the more effortful arm-movement. Another group of Yoked controls will receive pain on the same trials as their Experimental Group counterpart, but this will not be contingent with the movements they choose to perform. Thus, avoidance will be operationalized as the maximal deviation from the shortest possible movement, and as movement-choice behavior. During a subsequent Generalization phase, in the absence of pain, three intermediate generalization arm-movements (G1-3) will be introduced between the two original arm-movements (T+ and T-), as well as an extreme version of the safe high-cost arm-movement (G4; extreme in terms of required deviation and force, thus a peak shift movement). Pain-related fear and pain-expectancy, will be measured throughout the experiment, as self-reported indices of fear learning. Furthermore, eye-blink startle (EMG) amplitudes will be measured in response to the different movement trajectories, as a psychophysiological measure of fear. Thus, additional experimental phases will be added, during which participants will be instructed to perform T+ and T- (Startle acquisition test) and G1-4 (Startle generalization test), in order to collect startle EMG responses for each movement trajectory, respectively.