Does exercise augment operant and Pavlovian extinction: A meta-analysis

Abstract

Background

Exposure therapy, a behavioral approach to reduce symptomology in fear, anxiety, and drug-related psychiatric disorders, is based on learning and memory principles of extinction, and is subject to relapse. As such, it is important to find ways to enhance outcomes. One such way is through exercise.

Objectives

Identify if exercise augments extinction behavior, and whether this depends on the experimental paradigm used (i.e. operant or Pavlovian) and/or stimulus (i.e. appetitive or aversive). Additionally, determine which moderating variables influence the effects of exercise on extinction learning.

Methods

A literature search was conducted and a Hedges’ g calculation was employed to conduct a meta-analysis (metaSEM) using a structural equation modeling approach. This approach was chosen because of its ability to account for dependencies in effect sizes.

Results

We found a significant effect of exercise as an augmentation over extinction alone (g = 0.37, p < 0.001), with extinction paradigm (but not stimulus) producing a moderating effect (B = 0.43, p = 0.030). Data were then split by extinction paradigm, with operant extinction models having a significant effect (g = 0.55, p < 0.001), and number of extinction sessions moderating aggregate effects. Pavlovian models did not have significant overall effects (g = 0.11, p = 0.3976), but were moderated by the number of animals housed together and exercise after extinction.

Conclusions

The effects of exercise on extinction learning are differentially modulated by the type of paradigm used, the number of extinction sessions, the timing of when exercise treatment was applied (after extinction), and the housing conditions.

Introduction

Clinicians working with psychiatric patients face the challenge of varying response rates, and the potential for relapse after treatment. Disorders such as anxiety affect approximately 30% of the U.S. population in a lifetime, and substance use disorders affected approximately 270,000 Americans in 2015 alone (Kessler et al., 2005, SAMHSA, 2015), underscoring the necessity to provide treatments that both work in the short-term, but are also resistant to relapse. Both anxiety-related and substance use disorders are thought to develop and be maintained through learning and memory mechanisms, and rely on behavioral principles of operant and Pavlovian conditioning (for review see Britton et al., 2011, Milton and Everitt, 2012). As such, they both allow for the potential use of standard extinction-based behavioral approaches to reduce levels of responding (G. B. Kaplan et al., 2011). Over the last few decades, exposure-based therapies (which possess elements that overlap with extinction) have been used in clinical populations (Foa and Mclean, 2016, Krijn et al., 2004). While these therapies have shown efficacy over control conditions, as well as above and beyond pharmacotherapies, there are still patients who do not show improvements (Barlow et al., 2000, Bystritsky, 2006, Foa et al., 2005, Schottenbauer et al., 2008, Simpson et al., 2013). In addition, patients treated with exposure-based therapies are still subject to extinction relapse (as evidenced by reinstatement, renewal, and spontaneous recovery; Bouton, 2002). Such phenomena have lead us to the understanding that while extinction can reduce levels of responding, it generally does not erase the initial memory, but rather leads to the formation of new inhibitory learning, which may compete with the initial memory (Pavlov, 1927; for review see Bouton, 2002, Bouton and Bolles, 1979, Quirk and Mueller, 2008).

With the knowledge that a new memory is formed during extinction/exposure training, additional research has focused on developing avenues to augment extinction efficacy using pharmacological and behavioral approaches, to improve behavioral/treatment outcomes and reduce the potential for relapse (for review see Fitzgerald et al., 2014). One pharmacological agent, D-cycloserine (DCS), has shown success across both extinction and exposure therapy (Ressler et al., 2004, Santa Ana et al., 2009, Walker et al., 2002); however, it has also been associated with persistence of responding under certain conditions (Bolkan and Lattal, 2014, Myers and Carlezon, 2012, Smits et al., 2013a, Weber et al., 2007). Additionally, work looking at methylene blue, a neurometabolic enhancing drug, found similar results to DCS in humans and animals. More specifically, when a session of extinction/exposure was successful, as measured by low levels of fear at the end of session, methylene blue was beneficial; however, when high levels of fear were displayed subjects were worse off or did not benefit at follow-up tests (Auchter et al., 2017, Telch et al., 2014). With these mixed results from augmentation through pharmacotherapy, there has been a push to find alternative approaches to facilitate exposure outcomes. Non-pharmacological treatment avenues have been considered, with patients indicating that they prefer behavioral, cognitive, and psychological approaches over pharmacotherapies (Arch, 2014, Mchugh et al., 2013, Roy-Byrne et al., 2003). In addition, both anxiety disorders and substance use disorders often develop in adolescent populations for which pharmacological interventions are controversial due to ongoing neurodevelopment, resulting in a limited range of approved pharmacotherapies (Ipser et al., 2009; G. Kaplan & Ivanov, 2011). One such behavioral approach is through exercise. Introduced independently, exercise has been shown to reduce anxiety, stress, and depression, as well as increase measures of well-being (Asmundson et al., 2013, Dunn et al., 2005, Manger and Motta, 2005, Muller et al., 2006, Stonerock et al., 2015), all of which are predictive of treatment success. When applied broadly to supplement varying behavioral therapy types, exercise has also been shown to be beneficial (Brown et al., 2010, Merom et al., 2008); however, as the articles in the present work will highlight, exercise appears to have mixed effects on extinction: positive (Lynch et al., 2010), negative (Peterson et al., 2014b), null (Sanchez et al., 2014) and even mixed findings within a single article (Zlebnik et al., 2014).

As the field of exercise to augment extinction/exposure has developed, researchers have used a wide variety of study designs to maximize overall effects. Specifically, some employ weeks (or continuous bouts) of exercise whereas others only use a single (or acute) bout. Others use forced exercise on treadmills whereas some experimenters allow voluntary access for an allotted amount of time. There is also a large range of timing of exercise, such as before, during, or after extinction, and even during the test session. This variation in exercise regimens alone warrants a better understanding of which parameters predict significant overall effects— in particular, those that produce a persistent reduction of responding over standard extinction, such that this analysis will focus on behavioral outcomes after extinction to determine treatment successfulness and the potential for exercise to reduce relapse behaviors. In the present analysis, we explore both the global effects of exercise on extinction, as well as determine whether exercise can have differential effects based on type of stimuli used (i.e. appetitive or aversive) or based on operant and Pavlovian extinction models. The overarching goal, through the use of a structural equation modeling approach to meta-analysis, is to determine not only if exercise can have a significant enhancement on extinction by producing a greater and persistent reduction in responding, but delineate what parameters are the strongest predictors of its success.