Tackling maladaptive memories through reconsolidation: From neural to clinical science

Accepted Manuscript Tackling maladaptive memories through reconsolidation: From neural to clinical science James W.B. Elsey, Merel Kindt PII: DOI: Reference:

S1074-7427(17)30038-2 http://dx.doi.org/10.1016/j.nlm.2017.03.007 YNLME 6648

To appear in:

Neurobiology of Learning and Memory

Received Date: Revised Date: Accepted Date:

17 January 2017 7 March 2017 8 March 2017

Please cite this article as: Elsey, J.W.B., Kindt, M., Tackling maladaptive memories through reconsolidation: From neural to clinical science, Neurobiology of Learning and Memory (2017), doi: http://dx.doi.org/10.1016/j.nlm. 2017.03.007

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Tackling maladaptive memories through reconsolidation: From neural to clinical science Elsey, James W.B.1, & Kindt, Merel.2,3

1

PhD candidate in Experimental and Clinical Psychology at the University of

Amsterdam, 129B Nieuwe Achtergracht, 1018WS, Amsterdam, Netherlands. 2

Professor of Experimental and Clinical Psychology at the University of Amsterdam,

129B Nieuwe Achtergracht, 1018WS, Amsterdam, Netherlands. 3

Corresponding Author. [email protected]

Keywords: Translational research, memory reconsolidation, anxiety disorders, PTSD, clinical applications of reconsolidation, prediction error.

Running header: From neural to clinical science of reconsolidation

1

Abstract Behavioral neuroscience has greatly informed how we understand the formation, persistence, and plasticity of memory. Research has demonstrated that memory reactivation can induce a labile period, during which previously consolidated memories are sensitive to change, and in need of restabilization. This process is known as reconsolidation. Such findings have advanced not only our basic understanding of memory processes, but also hint at the prospect of harnessing these insights for the development of a new generation of treatments for disorders of emotional memory. However, even in simple experimental models, the conditions for inducing memory reconsolidation are complex: memory labilization appears to result from the interplay of learning history, reactivation, and also individual differences, posing difficulties for the translation of basic experimental research into effective clinical interventions. In this paper, we review a selection of influential animal and human research on memory reconsolidation to illustrate key insights these studies afford. We then consider how these findings can inform the development of new treatment approaches, with a particular focus on the transition of memory from reactivation, to reconsolidation, to new memory formation, as well as highlighting possible limitations of experimental models. If the challenges of translational research can be overcome, and if reconsolidation-based procedures become a viable treatment option, then they would be one of the first mental health treatments to be directly derived from basic neuroscience research. This would surely be a triumph for the scientific study of mind and brain.

2

Tackling maladaptive memories through reconsolidation: From neural to clinical science

Behavioral neuroscience has greatly informed how we understand the formation, persistence, and plasticity of memory. Research has demonstrated that memory reactivation can induce a labile period, during which previously consolidated memories are sensitive to change, and in need of restabilization. This process is known as reconsolidation. These findings have advanced not only our basic understanding of memory processes, but also hint at the prospect of harnessing these insights for the development of a new generation of treatments for disorders of emotional memory. However, even in simple experimental models, the conditions for inducing memory reconsolidation are complex: memory labilization appears to result from the interplay of learning history, reactivation, and also individual differences, posing difficulties for the translation of basic experimental research into effective clinical interventions. In this paper, we review a selection of influential animal and human research on memory reconsolidation to illustrate key insights these studies afford. We then consider how these findings can inform the development of new treatment approaches, with a particular focus on the transition of memory from reactivation, to reconsolidation, to new learning, as well as highlighting possible limitations of experimental models. If the challenges of translational research can be overcome, and if reconsolidation-based procedures become a viable treatment option, then they would be one of the first mental health treatments to be directly derived from basic neuroscience research. This would surely be a triumph for the scientific study of mind and brain.

3

The past century has seen radical shifts in the way we conceptualize and treat mental disorders. Dominant models of the mind and its maladies that have held sway at different periods have included psychoanalytic, behaviorist (which in its most radical form rejected the mind as an object of study), cognitive, and, more recently, neuroscientific perspectives. Within these broad delineations have been a plethora of variations on the major themes of each school of psychology, with corresponding treatment implications. While some treatments have been plainly ineffective or even detrimental, controlled trials point to the efficacy of a host of different treatment approaches, but also significant room for improvement (Butler, Chapman, Forman, & Beck, 2006; Cuijpers et al., 2013; Leichsenring, 2001; Shedler, 2010). Even in cognitive behavioral therapy (CBT), currently the most widely recognized evidencebased treatment, research suggests that a substantial portion of patients may fail to achieve significant improvements, and a large proportion of successfully treated patients go on to relapse (Durham, Higgins, Chambers, Swan, & Dow, 2012; Hofmann & Smits, 2008; Loerinc et al., 2015). In light of the imperfection of current treatments and the ever expanding arsenal of different approaches, several authors who stand at the interface of research and practice have called for a move away from the relatively simple question of which of the many therapeutic approaches work to an understanding of why (Kazdin, 2001; McNally, 2007).1 If reached, a clear understanding of the underpinnings of mental disorder and mechanisms of change could transform research in mental health treatment from a more descriptive science – cataloguing what works – to an explanatory one – explaining why such treatments are effective – or even a predictive one, proposing what will work best in novel situations given an understanding of the 1

The term ‘relatively’ is sincerely meant: designing studies that do justice to the different therapeutic approaches under investigation, in a controlled yet clinically relevant setting from which meaningful conclusions can be drawn, is of course very difficult.

4

root problems and the means through which they can be tackled. Interdisciplinary dialogue, and particularly the translation of findings from different levels of analysis, will prove essential in this endeavour. At present, only basic and behavioral neuroscience research (by which we mean in vitro and in vivo animal models, rather than functional imaging in humans) can deconstruct the low level neural networks that underpin mental disorders, but only in combination with human experiments and clinical trials can the relevance of this basic knowledge be assessed and its value for clinical practice realised. Current work in models of exposure therapy highlights the potential utility of such a translational approach. Exposure is a technique commonly used in the treatment of anxiety disorders, in which patients are confronted with their fears. This may be in vivo (i.e. confrontation with external stressors in real-life), imaginal (i.e. exposure to feared situations or memories through the imagination), interoceptive (i.e. exposure to feared internal sensations), or, more recently, in virtual reality (i.e. using virtual reality technology to construct analogues of real-world situations) (McNally, 2007; Powers & Emmelkamp, 2008). This line of treatment developed out of basic studies of learning in animals. Animals that had been taught to fear a particular stimulus by pairing it with an aversive event (conditioning) could be made to display less fearful responding by re-exposing them to the newly feared stimulus in the absence of the aversive event (extinction). Likewise, patients can be made less fearful of feared objects or situations through exposure to them. Yet, more research in animal models of learning has indicated that extinction training does not cause unlearning of the original fear memory, but rather creates a new, inhibitory memory trace that competes with original learning for control over behavior (Bouton, 2002). Changes in context, aversive events, or the simple passage of time can lead to the resurgence of the old

5

memory. As the underlying mechanisms of change are thought to be analogous in exposure therapy, such findings help explain relapse in clinical settings. More intriguingly, they also point to ways of improving treatment outcomes. As Craske and colleagues have emphasized, if exposure therapy operates through the formation of an inhibitory memory trace, then clinicians should aim to optimize this inhibitory learning (Craske, Treanor, Conway, Zbozinek, & Vervliet, 2014; Craske et al., 2008). For example, rather than simply ‘habituating’ patients to a feared stimulus (exposing them until their fear levels drop), treatments ought to focus on violating the patient’s expectations about negative outcomes that might occur upon exposure to the feared stimulus, as large prediction errors are thought to generate strong inhibitory memories. Insights into treatment mechanisms also have somewhat counterintuitive implications. Rodent and human experimental research indicates that providing occasional reinforcement (negative outcomes) during extinction might actually reduce spontaneous recovery (Bouton, Woods, & Pineño, 2004; Gershman, Jones, Norman, Monfils, & Niv, 2013; Woods & Bouton, 2007). Occasional reinforcement or gradual extinction may reduce the disparity between new learning and the original memory, potentially acting directly on the original memory or meaning that later negative outcomes don’t automatically activate the maladaptive memory trace because the new, more adaptive one also contains (and can therefore be more easily activated by) aversive experiences. Again, conceiving of change through the lens of inhibitory learning and competing memory traces could serve to enhance treatment outcomes. In summary, understanding mechanisms of change in treatment, and the underpinnings of the disorders we wish to treat, are crucial goals for mental health research if the aim of this is to optimize available as well as novel treatments.

6

However, in the push to realise the potential of treatment approaches, lessons learned from lower levels of analysis (which are not always easily translatable to the clinic and frequently impose limitations on the scope of applications) are not necessarily taken on board. Research pushing the boundaries of our current knowledge through the attempted translation of lab work into therapeutic approaches is necessary for the advancement of any novel treatment approach, but efforts in this direction must be informed by the most relevant available research. In this paper, we focus on attempts at disrupting reconsolidation of maladaptive memories in anxiety and trauma-related disorders through pharmacological means, and particularly the use of propranolol in humans. Many of the issues we highlight, however, would most likely apply to other approaches inspired by the idea of reconsolidation, such as performing extinction after a brief reminder cue (Monfils et al., 2009), or to the use of other pharmacological agents aiming to block reconsolidation, and the overarching message that clinical interventions should aim to understand mechanisms of change and take account of the most relevant research is of course applicable to all mental health treatments. It is also conceivable that reconsolidation might be harnessed so as to enhance certain adaptive memories, though this will not be explored in the present review. In the following sections, we briefly outline current research into the pharmacological disruption of memory reconsolidation and attempts to translate this into clinical interventions. Then, we draw upon insights from experimental research on reconsolidation in humans and animals to make some empirically grounded suggestions for reconsolidation-based treatments. We also consider limitations of current experimental models, and suggest several avenues that can be pursued in future research.

7

Memory reconsolidation The dominant model of memory formation proposes that memories transition from a short-term and relatively unstable trace to a more persistent long-term form (McGaugh, 2000). This transition from short-term memory to long-term memory is known as consolidation, and is most commonly thought to be mediated by protein synthesis dependent synaptic changes (Kandel, Dudai, & Mayford, 2014, though see discussion of this below). Protein synthesis inhibitors (PSIs) prevent the expression of long-term memory when administered shortly after learning (Schafe & LeDoux, 2000). Once consolidated, memories appear insensitive to protein synthesis inhibition, and can prove highly recalcitrant to attempts at modification (LeDoux, Romanski, & Xagoraris, 1989). However, it has been found that reactivation of a memory can render it vulnerable to amnestic interventions once more: protein synthesis inhibition shortly after reactivation can also prevent the later expression of long-term memory (Nader, Schafe, & LeDoux, 2000). It is now thought that, under certain conditions, a consolidated memory can be brought into a labile state by reactivation, during which the memory trace can be modified or even disrupted, and that this labile state requires restabilization in a manner similar to consolidation. Similarly to consolidation, this reactivation-induced period of lability is temporary, with amnestic interventions having no effect when administered 6 hours after reactivation (and probably at shorter time delays – the exact time at which the reconsolidation window closes is currently unclear, and likely varies depending on the type of memory and organism studied) (Nader et al., 2000). Due to parallels with initial consolidation, this process has become known as reconsolidation (Nader & Hardt, 2009a; Przybyslawski & Sara, 1997).

8

Amnesia following protein synthesis inhibition that is timed to coincide with the putative process of post-reactivation memory reconsolidation has now been demonstrated in a host of animal models, including sea snail Aplysia Californica, crabs, fish, as well as rats and mice (Chen et al., 2014; Eisenberg & Dudai, 2004b; Nader et al., 2000; Pedreira & Maldonado, 2003; Ryan, Roy, Pignatelli, Arons, & Tonegawa, 2015). Unfortunately, commonly used protein synthesis inhibitors such as anisomycin and cycloheximide have a range of unpleasant side effects that make them unsuitable for use in humans. Crucially for the prospect of translating animal findings into human interventions, Dębiec and LeDoux (2004) showed that amnesia for auditory fear conditioning could be induced by the systemic administration of propranolol, timed to coincide with memory reactivation. The blockade of betaadrenergic receptors (β-ARs) by propranolol is believed to indirectly inhibit protein synthesis by halting noradrenaline-stimulated CREB phosphorylation in the amygdala (Kindt, 2014; Thonberg, Fredriksson, Nedergaard, & Cannon, 2002). The feasibility of using propranolol to affect fear memory reconsolidation in humans has since been demonstrated in several experimental studies. In the first of these, Kindt, Soeter, and Vervliet (2009) used Pavlovian fear conditioning, in which one of two spider pictures was paired with an unpleasant electric shock. Conscious expectancy of shock and startle responding were used as indices of learning. The startle response is a defensive reflex elicited in response to the sudden onset of an intense stimulus across mammalian species (e.g. a loud noise for acoustic startle) (Davis, 1984). Unlike skin conductance (another commonly used measure in fear conditioning), potentiated startle responding does not seem to reflect simple conscious expectation of aversive events, as participants can display startle potentiation without awareness of picture-shock contingencies, and potentiated startle responses are

9

typically retained even after instructions indicating that there will be no more shocks (Sevenster, Beckers, & Kindt, 2012a; 2014a; Weike et al., 2005; Weike, Schupp, & Hamm, 2007). Rather, the reflex appears to track the emotional valence of stimuli, as it can be both attenuated in the face of appetitive or positive stimuli and potentiated for negative, aversive stimuli (Andreatta & Pauli, 2015; Bradley, Cuthbert, & Lang, 1999). As expected, on the first day of conditioning participants developed a potentiated startle to the stimulus that was paired with shock (the CS+) relative to the unpaired stimulus (CS-), known as differential fear potentiated startle (FPS). On the second day of the experiment, participants received propranolol (tmax = 1-2 h) or placebo (double blind) 90 minutes before memory reactivation, to allow for the drug to reach peak plasma levels. Participants then received a single, unreinforced presentation of the CS+ to reactivate their memories for the conditioning procedure. A third group received propranolol without memory reactivation, to test for nonspecific effects of the drug. On a third and final testing day, participants were again exposed to the CS+ and CS- for several unreinforced trials. Participants still retained their declarative memories for the conditioning procedure, and those who had received placebo + memory reactivation or propranolol alone still displayed potentiated startle responding to the CS+ relative to the CS-. In contrast, differential FPS was abolished in the propranolol + memory reactivation group. These findings suggest that the negative affective valence of the CS+ was effectively neutralized, while declarative memory for the fear-conditioning episode (as indexed by conscious expectancy of shock) was unaffected. In this study and later ones from the same lab, attempts were made to probe the reversibility of these effects on differential FPS, including the presentation of unsignalled electrical shocks (reinstatement), testing memory up to one month after

10

the intervention (spontaneous recovery), assessing the memory with different contextual cues (renewal), and re-conditioning participants (rapid re-acquisition) (Soeter & Kindt, 2010; 2011a; 2012a). FPS was not reinstated by unsignalled shocks, nor did it recover when stimuli were presented in a different experimental context or through the mere passage of time. Moreover, participants did not re-develop differential FPS any more quickly during re-conditioning than they did during initial learning, suggesting there were no ‘savings’ of the memory trace. Hence, the reduction in startle responding observed is probably not attributable to an inhibitory process such as extinction, as the above probes typically lead to the recovery of fear memory after extinction training (Bouton, 2002). Of course, the absence of fear responding can never prove that other procedures could not lead to memory retrieval, but does demonstrate that the effects of this reconsolidation-based procedure are stronger than normal extinction training. This modification of fear memory has since been replicated and extended in several studies. To address the concern that propranolol administered before memory reactivation might have affected retrieval, the effect of propranolol + memory reactivation on startle responding was replicated in studies that administered propranolol after reactivation (e.g. Sevenster et al, 2013; 2014b; Soeter & Kindt, 2012a; 2012b; 2015a). In line with the previously observed preservation of declarative memory, it was found that differential skin conductance, which is thought to reflect the propositional representation of associative fear memory (operationalized in contingency awareness or expectancy of shock) (Sevenster, Beckers, & Kindt, 2014a), remained intact alongside expectancy ratings (Soeter & Kindt, 2010). Further supporting the idea that the affective component of the memory is disrupted, self-

11

reported distress in response to the CS+ was also reduced by propranolol combined with memory reactivation (Soeter & Kindt, 2012a). These findings are exciting for both researchers and clinicians. From a purely academic perspective, they shed light on the underlying mechanisms of memory formation and change. Most intriguing, however, is the possibility that these new insights might be leveraged in the treatment of psychopathology. Maladaptive emotional memories are arguably a key feature in a range of mental disorders, from post-traumatic stress disorder (PTSD) – in which the memory for a trauma becomes so disturbing as to disrupt ordinary functioning – to addiction – where appetitive memories for drug consumption gain so much control over behavior that the drug is pursued even at great cost to the user and those around them (Brewin, 2001; Milton & Everitt, 2012). Should maladaptive memories in clinical conditions be susceptible to disruption through a similar approach, this could prove a great step forwards in their treatment. As has been discussed, the disruption of reconsolidation by pharmacological means has proven effective in both non-human animals and humans under controlled laboratory settings. Results from human subjects parallel those of the animal models upon which such research was based; translation of findings from simple learning paradigms – such as fear conditioning – in animals to simple paradigms in humans, has shown great promise. However, the aim of translational research is not merely the production of more experimental studies, but the understanding of the principles of memory modification and, ultimately, the development of empirically grounded interventions that can be derived from this understanding. Making this final step, from human experimental findings to effective clinical interventions, has thus far proven more difficult.

12

Brunet and colleagues (2008; 2011) found that propranolol administered so as to interfere with reconsolidation after exposure to traumatic imagery significantly attenuated physiological responding to the imagery relative to placebo (Brunet et al., 2008) and reduced PTSD symptoms in an uncontrolled trial (Brunet et al., 2011). In a subsequent study, however, Wood and colleagues (2015) found no differences between a memory reactivation + propranolol group and a propranolol only group in physiological reactivity to trauma imagery approximately one week after treatment. More recently, Kindt and van Emmerik (2016) presented data from three cases in which imaginal exposure to traumatic memories followed by propranolol produced a sharp decline in PTSD symptomatology, alongside one unsuccessful case. While these studies convey that a reconsolidation-based treatment of PTSD has potential, it is also clear that this promise has not yet been realised. Results have not been entirely consistent and all the studies thus far have used relatively small sample sizes and did not always control for alternative explanations (i.e. nonspecific effects of propranolol, or effects of exposure in the absence of propranolol). Regarding the treatment of specific phobias, Soeter and Kindt (2015b) found that propranolol given to highly spider fearful individuals after a brief exposure to a tarantula significantly reduced fear of spiders in all those in the active treatment group, and this reduction in fear was sustained at a one year follow up. Crucially, this study included both a placebo and a no-reactivation control. We have found such an approach to be effective in other cases of clinical specific phobia (Elsey & Kindt, in preparation). These results are encouraging, but it has also become apparent that certain changes in the preparation of participants and the specifics of what occurs in the reactivation session can affect treatment outcomes in unpredictable ways. Simply administering propranolol after frightening experiences is certainly no panacea, and a

13

deeper understanding of what occurs in such interventions will likely prove essential to their clinical utility. A key consideration is what we have learned from more basic science and experimental models that can help inform the development of this new approach to therapy, and how we understand the outcomes observed. The following sections address this. Reactivation and reconsolidation are not synonymous Though the adaptive purpose of reconsolidation is a matter of continued debate, one hypothesis is that it may allow for memories to be updated, thereby retaining their relevance and usefulness in a changing environment (Lee, 2009). From such a perspective, it might be predicted that reconsolidation would be most reliably induced by memory reactivations that in some way add to or indicate the need to update the memory. Findings from animal studies strongly suggest that this is the case (see Fernandez, Boccia, & Pedreira, 2016 for a review). In the first animal study to demonstrate this, Pedreira, Perez-Cuesta, and Maldonado (2004) used the crab Chasmagnathus, and found that a brief unreinforced, but not reinforced, exposure to conditioned stimulus rendered memory for a conditioned escape response susceptible to protein synthesis blockade. In a more recent, striking demonstration of the importance of prediction error using fear conditioning in rats, Alfei and colleagues (2015) found that memories could be made susceptible to the amnestic effects of protein synthesis inhibitors (PSIs) through different types of prediction error. If reinforcement occurred during reactivation, but at a different time to what had been learned during training (temporal prediction error), or if reinforcement did not occur (contingency prediction error) within the time frame that it would be expected based on learning, then memories for the conditioning

14

episode were susceptible to PSIs. Without these prediction errors, amnestic effects were not observed, despite a clear expression of memory retrieval, as indexed by freezing. Perhaps even more surprisingly, memory expression appears to not only be insufficient for inferring memory reconsolidation, but actually unnecessary (Barreiro, Suarez, Lynch, Molina, & Delorenzi, 2013; Mamou, Gamache, & Nader, 2006). Mamou and colleagues (2006) found that intra-amygdala infusion of NMDA receptor antagonists ifenprodil or D(-)-2-amino-5-phosphonovaleric acid (AP5) prior to memory reactivation in fear conditioned rats prevented the amnestic effects of a PSI, while having no effect on memory expression during reactivation. In contrast, prereactivation administration of the AMPA receptor antagonist 6-cyano-7nitroquinoxaline-2,3-dione (CNQX) blocked fear memory expression (freezing) during reactivation, but did not preclude later amnesia as a result of protein synthesis inhibition. The authors concluded that NMDA receptor activation was necessary for the labilization of a fear memory trace upon reactivation, whereas AMPA receptors were crucial for the expression of that memory, and that the two processes are dissociable. That prediction based on what has been learned could be integral to the initiation of reconsolidation in humans has been supported in a series of fear conditioning studies. In earlier experiments using propranolol as a means of disrupting reconsolidation, reactivation consisted of a single unreinforced presentation of the CS+ (Kindt, , & Vervliet, 2009; Soeter & Kindt, 2010b; Soeter & Kindt, 2011a). Sevenster, Beckers, and Kindt (2012b) similarly reactivated participants’ memories for the earlier fear conditioning episode, but included one group who did not have the electrodes that deliver the US attached to their wrists. Hence, for these

15

participants, there was nothing that could be learned regarding the contingencies of reinforcement. Although participants who did not have the shock electrodes attached still displayed memory retrieval, evident in their potentiated startle responses during CS presentation, propranolol had no effect on their later expression of conditioned responding. Hence, despite retrieval, the memory was not destabilized. Hence, retrieval is neither a necessary (Barreiro et al 2013) nor a sufficient (Sevenster et al 2012b) condition to trigger memory reconsolidation. Furthermore, Sevenster, Beckers, and Kindt (2013) demonstrated that it was prediction error, not the absence of reinforcement, which was necessary for the destabilization of conditioned fear memories. During learning, participants received either 100% or 33% reinforcement. At reactivation, participants who had undergone 100% reinforcement received either a reinforced (no prediction error) or unreinforced (negative prediction error) CS presentation, whereas those who had undergone 33% reinforcement received a reinforced trial (positive prediction error). All participants then received propranolol. At subsequent testing, it was found that both the positive and negative prediction error groups displayed amnesia for conditioned fear, as indexed by the loss of differential fear potentiated startle, whereas the no prediction error group did not. Hence, even when a shock was delivered, the presence of prediction error rendered the memory vulnerable to the amnestic influence of propranolol. These findings suggest that, for the translation of reconsolidation-based research into clinical practice, simply generating a fear response or reactivating a patient’s memory may not trigger reconsolidation of the target memory trace. An optimal reactivation session should involve some kind of prediction error. However, it may not be appropriate to simplistically extend findings from experimental studies to

16

therapeutic interventions. In experimental research, prediction error can be easily operationalized because the formation of the memory has been carefully controlled. This is not the case for clinical disorders, in which learning history is often unclear, and patients have a range of expectations, based on their fears, that may or may not be central to their disorder. Patients’ expectations in PTSD, for example, may be related to the emotions they will experience upon recall of a trauma, the magnitude of these emotions, or to the consequences of experiencing them (to name but a few). It may be beneficial to consider the most salient fears that a patient has in designing their reactivation sessions. If fear of being overwhelmed by trauma recall is most prominent, then recalling the most troubling parts of the trauma and seeing that one does not go insane, for example, may trigger reconsolidation. On the other hand, if an intense sense of hopelessness and defeat characterizes the traumatic memory, then it may be beneficial to change some elements of the trauma in the patient’s imagination as they recall it, such as imagining a loved one stepping in to help them (a technique known as imagery rescripting), to violate the expectations and feelings of inevitable defeat when trauma is recalled (Kindt & van Emmerik, 2016). It should also be noted that prediction errors could include not only the violation of expectations, but also the learning of additional information if learning has not reached asymptote. Hence, novel experiences with feared stimuli could also destabilize the target memory. At present, our tools for understanding prediction error and the destabilization of maladaptive memories are far from optimal. Prediction error and destabilization are internal events that are presumed to be occurring within the patient, but we often infer them based on external characteristics of the reactivation session that we assume ought to provide the requisite input to generate a prediction error and the subsequent modification of memory. Reliable means of not only inducing but also inferring

17

memory destabilization would prove hugely informative for clinical practice, and should be pursued in experimental investigations. Use of subjective reports from human subjects is likely to provide additional information that cannot be gleaned from animal studies in this domain. In animal studies, one must simply infer such a process from the observation that memory expression either was or was not affected by the experimental intervention. Sevenster and colleagues (2013) found that in the positive and negative prediction error groups, there was an increase and decrease, respectively, in their explicit expectation of shock on the final testing day, whereas the expectation of shock in the no prediction error group was unchanged. Such a change in expectancies provides an indicator of memory updating that is independent of observed deficits in later memory expression. However, as is discussed in the following section, the optimal amount of prediction error for inducing reconsolidation remains to be determined. Extinction learning in fear conditioning results from multiple prediction errors, and poses a boundary on reconsolidation – a boundary that may in fact be reached even before actual extinction occurs (Eisenberg et al., 2003; Merlo, Milton, Goozée, Theobald, & Everitt, 2014; Sevenster et al., 2014b). Prediction error, reconsolidation, and extinction Just as reactivation is not synonymous with reconsolidation, prediction error during reactivation is not always sufficient for the induction of reconsolidation. When prediction error occurs, the length of reactivation or extent of prediction error can determine whether reconsolidation or extinction is the result. For example, in a Medaka fish model of fear conditioning, Eisenberg et al. (2003) found that whereas a single unreinforced reactivation trial rendered memory vulnerable to PSIs, multiple unreinforced presentations of the conditioned stimulus resulted in extinction and precluded the disruption of the reactivated memory. Similarly, Lee, Milton and

18

Everitt (2006) found that infusion of an NMDA receptor agonist, which have been found to enhance learning, reduced or increased conditioned freezing after long versus brief memory reactivation lengths were used, respectively. In contrast, an NMDA receptor antagonist, which typically block learning, produced the opposite effect. These findings indicate that longer memory reactivations are likely to lead to extinction, which can be facilitated or blocked by NMDA agonists and antagonists respectively, whereas brief reactivation are more likely to trigger reconsolidation. Likewise, in Alfei and colleagues’ (2015) study of prediction error, it was found that ending memory reactivation shortly after a prediction error resulted in memory destabilization, whereas extending reactivation for another 4 minutes prevented a GABAA agonist (midazolam) from generating amnesia. With even longer time periods after prediction error, extinction occurred, and midazolam administration at this stage actually resulted in greater freezing at testing compared with animals who were administered saline, which can be understood as resulting from the midazolam having blocked the consolidation of extinction learning. Merlo et al. (2014) found that the transition, with extended CS exposure, from reconsolidation to extinction learning was paralleled by increasing levels of calcineurin (CaN) in the basolateral amygdala (BLA) of rats. Blocking the increase of CaN in the BLA prevented extinction from occurring, and administration of an NMDA agonist, which facilitated extinction, also resulted in increased levels of CaN. Merlo et al. (2014) also observed a transitional state in which neither reconsolidation nor extinction were the dominant active processes: interventions during this liminal or ‘limbo’ period could neither promote nor disrupt reconsolidation or extinction. Hence, it seems that depending on when reactivation is terminated, one may induce reconsolidation, extinction, or a liminal period between these two in which amnestic

19

agents have no effect, and which of these processes is triggered appears to be determined at the offset of the reactivation episode (Pérez-Cuesta & Maldonado, 2009). In line with these findings, Sevenster, Beckers, and Kindt (2014b) found that the extent of prediction error determined the fate of memory. In this study, participants underwent fear conditioning with 50% reinforcement. During reactivation, one group received a single unreinforced reactivation trial (no prediction error), a second group received 2 unreinforced trials (single prediction error), whereas a third group received 4 unreinforced trials (multiple prediction errors). All participants then received propranolol. Only the group that received a single prediction error displayed neutralization of conditioned startle responding on a subsequent testing day. These studies demonstrate that the induction of reconsolidation is a balancing act: without prediction error, memory is not labilized and made vulnerable to amnestic agents, but with extended reactivation and multiple prediction errors, destabilization does not take place and extinction may even occur instead. This matter is further complicated by earlier considerations regarding the learning history and temperament (including genetic influences) of the individual whose memory is being targeted. Taking into account these different factors, it begins to look unlikely that any single reactivation procedure will prove effective for all who undergo it, potentially undermining the use of very standardized reactivation procedures that may be pursued in clinical trials. The same reactivation session may trigger reconsolidation in one individual but result in the induction of a liminal period or extinction in another, and there seems to be a frustratingly narrow window within which the target process of reconsolidation occurs.

20

Findings regarding the extent of memory reactivation might have some relevance for previously discussed studies of PTSD. In their failure to replicate the earlier findings of Brunet and colleagues (2008), Wood et al. (2015) had participants not only perform the imagery generation procedure so as to reactivate their symptoms, but also conducted the clinician administered PTSD scale (CAPS) for the 90 minutes preceding this reactivation, and in some cases continued afterwards. It does not seem that the CAPS was administered in this fashion in Brunet et al. (2008). As Wood and colleagues note, the CAPS was not administered at an earlier intake session in order to avoid inadvertently reactivating trauma memories in the no-reactivation control group, and it is indeed possible that undergoing the CAPS assessment could lead to memory reactivation in addition to that provided by the traumatic script generation session. Such extended reactivation could cause a memory to enter the liminal phase between reconsolidation and extinction, or to enter extinction, rather than triggering reconsolidation. While we do not suggest that this is actually the explanation for why this study did not find a difference between the propranolol + reactivation and propranolol only groups (there were several other differences between the two studies), the possibility of such confounding influences highlights that attention should be paid to the design of reactivation sessions and surrounding assessments in order for strong conclusions to be able to be drawn. Care is warranted not only to prevent reactivation in nonreactivation groups, but also to control its nature and extent in those receiving active treatment. Ecological validity of experimental models While initial forays into clinical applications of these findings, such as in the treatment of specific phobia (Soeter & Kindt, 2015b), give reason for optimism, it is

21

not yet clear that there are sufficient parallels between true anxiety disorders and the experimental paradigms that have been investigated thus far. Almost all experimental research on the disruption of emotional memory has utilized relatively simplistic (e.g. tone-shock or picture-shock pairings), recent, and – from the perspective of participants in human studies – quite unimportant memories. Even specific phobias, which have often been considered to be the most simplistic type of anxiety disorder (hence, ‘simple phobia’ in the Diagnostic Statistical Manual of Mental Disorders, 3rd edition, APA, 1980), patients can have been suffering from their disorder for many years, have developed a complex web of meanings around their symptoms, and may find that it profoundly affects their lives. Although laboratory research has provided great insights into the development, maintenance, and treatment of anxiety disorders, it is not inevitable that the attempted pharmacological blockade of reconsolidation in full-fledged anxiety conditions will be as effective as it has been in experimental settings. Research gives reason to think that strength and age of memories may not pose absolute boundary conditions on reconsolidation. While some research on reconsolidation has proposed that older or stronger memories may prove invulnerable to disruption (Eisenberg & Dudai, 2004a; Milekic & Alberini, 2002), it now seems that even old and strong memories can be made more susceptible to destabilization if novel information is provided during reactivation (Winters, Tucci, & DaCostaFurtado, 2009). In humans, Soeter and Kindt (2011b; 2012b) found that administration of a noradrenergic receptor agonist resulted in particularly strong fear memories, which displayed delayed extinction curves and more rapid reacquisition. Nevertheless, these memories could be neutralized with propranolol. Finally, participants in Soeter and Kindt (2015b) were extremely afraid of spiders and had

22

been so for many years, yet they still benefitted from the reconsolidation-based intervention. Despite these findings, it cannot be said with certainty that especially strong fears, or other anxiety conditions, will be likewise amenable to modification. The complexity of real-world anxiety conditions could also make treatment more problematic. A study of second order fear conditioning (SOFC) in rodents provides one example of how the complexity of memories may pose a challenge to clinicians (Dębiec, Doyère, Nader, & LeDoux, 2006). In SOFC, animals are first trained using a standard fear-conditioning procedure, in which a tone is paired (CS1) with a shock (US). If a second conditioned stimulus (CS2) is repeatedly paired with the CS1, then the animal also develops a defensive response to the CS2+. Dębiec and colleagues (2006) found that if the CS1 was used for reactivation, followed by anisomycin, then conditioned responding to both the CS1 and CS2 was dramatically reduced. However, if the CS2 was the reactivation stimulus, then anisomycin only interfered with CS2 responding, leaving conditioned responding to the CS1 intact. Studies of the formation of first and second order conditioning memory traces shed some light on this. Tronel, Milekic, and Alberini (2005) found that the formation of a second-order conditioning memory trace was associated with molecular processes shown to be specific to consolidation, rather than reconsolidation, suggesting that second-order conditioning reflects the generation of a new memory trace, rather than the incorporation of new information into an old memory. Hence, it would seem unlikely that reactivating a second-order memory would necessarily destabilize the original memory trace. In clinical practice, it is not always easy to establish what type of stimulus would be most reflective of a patient’s core fears, versus those that are merely derivative of them. If something akin to a second order conditioned stimulus were

23

utilized, then one would expect the reduction in fear to generalize poorly to other stimuli in the fear network. It may not be necessary to make patients maximally fearful, but it would seem beneficial to choose a stimulus that strongly triggers the patient’s most central fears for optimal treatment outcomes. A final limitation of lab-based models, and most current studies in humans, is that although we can observe differences in neural networks and defensive behaviour in animals, or physiological responses in humans, as a result of the putative disruption of reconsolidation, remarkably little research has considered what the subjective experience of these changes is. Although we use the term ‘fear conditioning’ in describing animal research, it is not certain that animals undergoing such conditioning have the subjective experience of fear (LeDoux, 2014). Rather, we observe that the animals develop a defensive response to the conditioned stimuli and then might infer that this is correlated with a fearful or anxious subjective state. Likewise, in human experimental studies using propranolol as the means of reconsolidation blockade, studies have typically shown an effect on fear-potentiated startle, but only one study investigated the subjective experience of the participants (Soeter & Kindt, 2012a). This study found that participants who received propranolol + reactivation had reduced feelings of distress, relative to inactive groups, in response to the CS+ at follow up testing, but it remains to be seen how pronounced such subjective effects will be in patients. The subjective experience of fear in spider fearful participants in Soeter and Kindt (2015b) was significantly reduced along with their behavioral responses, but it was in general not completely nullified. Participants’ experiences of other emotions, such as disgust (which is common in specific phobias: Tolin, Lohr, Sawchuk, & Lee, 1997), or of more complex feelings such as guilt and shame after reconsolidation-based procedures are essentially untapped. As these other emotional

24

experiences might affect the course of recovery, they should be studied along with the subjective experience of fear and anxiety. Mechanisms of change and the demonstration of reconsolidation As well as considering several implications for reconsolidation-based treatments, it must of course also be established that reconsolidation is the most reasonable process through which an intervention operates (Elsey, van Ast & Kindt, submitted). Criteria for the demonstration of reconsolidation derived from animal studies that we suggest can be meaningfully applied to human research are: 1) the presence of an interaction between memory reactivation (e.g. a brief reminder cue) and the experimental manipulation (e.g. administration of propranolol), 2) time dependent effects of the manipulation (which should not have the same effect when administered outside of the putative reconsolidation window as it does when given inside), 3) memory specificity (the manipulation should not indiscriminately affect memory, but only the reactivated memory trace), and 4) a dissociation of immediate and delayed effects of the intervention (due to the preservation of a short-term form of memory that appears to persist after reconsolidation-based interventions in animals) (Elsey, van Ast, & Kindt, submitted; Nader & Hardt, 2009b; Tronson & Taylor, 2007). Research aiming to examine whether an intervention operates through reconsolidation should assess whether it meets these criteria for its demonstration. Here, experimental analogues of the clinical intervention are likely to be of great utility, as the assessment of these different criteria can require several different groups and assessment times that would not be easily practicable in a randomized clinical trial. Moreover, it is conceivable that some assessments, such as testing for the preservation of a short-term memory trace, could interfere with later treatment effects,

25

such that it might be more ethical to assess them in subclinical samples or healthy participants. However, it should also be stressed that if an intervention meets these criteria, it is strictly only consistent with, and not direct proof of, the possibility that the intervention works through reconsolidation. As we do not have any incontrovertible neural measure of whether reconsolidation has taken place in humans, we can only indirectly infer its presence based on such observations. Alternative processes may give rise to these same results. Hence, studies of the mechanistic underpinnings of reconsolidation-based interventions should also aim to make critical tests of alternative explanations. If these more parsimoniously explain the treatment effect then there will probably be different implications for how to optimise treatment outcomes. There are also possible reasons why these criteria might not be met even if reconsolidation is operating as a mechanism in treatment. Specifically, it is possible that some interventions might engage more than one process, such that an alternative process leads to an immediate effect of an intervention, while more delayed effects are the result of reconsolidation. In this case, the dissociation of immediate and delayed effects might not be observed, and researchers must carefully consider the most parsimonious explanation for the observed effects. These caveats point to the benefits that a clear and very specific indicator of reconsolidation would provide. At present, we are not aware of any such metric. Continued translation of low-level research to clinical implications As research continues, we can expect that new findings with implications for treatment will continue to come to light. Dialogue between researchers and clinicians must be maintained, and a critical and nuanced perspective on the precise mechanisms of treatment ought not to be abandoned as soon as reconsolidation-based procedures

26

are convincingly demonstrated to have therapeutic applications. While we noted earlier that research into mechanisms of extinction learning and exposure therapy has provided a wealth of insights, it can be disappointing to find many clinicians still adhering to an out-dated model of exposure therapy focusing on the habituation of fear. This is understandable, as many clinicians have a heavy caseload and little time to pore over all the most relevant contemporary research, but it is by no means an acceptable state of affairs. At this early stage of translational research into reconsolidation, we can ensure that thought and practice surrounding reconsolidation does not stagnate, or become unnecessarily wedded to particular models of the phenomenon if these do not stand up to careful scrutiny and testing. Even foundational issues in learning and memory, such as the idea that memories are stored via long-term potentiation and depression of synapses, may need to be called into question, and the answers to such questions may have repercussions for the theory and practice of reconsolidation. For example, rather than being the site of memory engram storage, recent work in animal models suggests that learning induced changes in synaptic potentiation may play a key role in the retrieval of memories, with the engram stored through some other, as yet unknown, mechanism. By labelling networks of neurons activated during memory storage, it was possible to directly tag and even subsequently manipulate (enhance or suppress) engram cells using optogenetic techniques (Tonegawa, Liu, Ramirez, & Redondo, 2015). Ryan, Roy, Pignatelli Arons, and Tonegawa (2015) found that reconsolidation blockade of a contextual fear memory in mice reduced synaptic potentiation and dendritic spine density of engram cells to an essentially naïve state. This change was paralleled by amnesia for the fear memory in behavioral tests. Yet, engram cells still displayed greater connectivity with

27

each other than with non-engram cells. Optogenetically stimulating engram cells in the dentate gyrus triggered downstream engram cell activation and resulted in fear memory recall, restoring freezing behavior in otherwise amnesic mice. Experimental findings such as these challenge the supposition that the disruption of reconsolidation ablates the actual site of memory storage, and raise questions regarding the longevity of reconsolidation-induced symptom improvement in clinical settings. Strikingly, even after direct stimulation of the engram cells and concomitant memory recall, mice were unable to recall the memory under normal (non-stimulated) conditions, suggesting that synaptic changes are of exceeding importance for ordinary recall. Experimental studies in humans noted above also show that impairments in memory expression following a reconsolidation-based procedure are certainly more resistant to later retrieval than extinction-induced changes (Kindt et al., 2009; Soeter & Kindt, 2010; 2011a; 2012b). Nevertheless, it may be that the retention of protein synthesis independent changes subserving memory storage facilitates the reinstatement of synaptic potentiation and ultimately memory retrieval. Hence, it may be worth considering whether reconsolidation-based procedures should be supplemented with additional learning experiences to cement improvements in symptoms over the long term. Integrating models of extinction learning and the psychotherapeutic process more generally with insights from neuroscience may prove fruitful in this respect. For example, as has been discussed above, the generation of a new memory trace that may compete with and/or inhibit a maladaptive memory is now a well-established model of extinction learning. Likewise, Brewin (2006) posits that retrieval competition helps explain how newly learned ways of thinking can take over from more maladaptive patterns in successful cognitive behavioral therapy. Disruption of reconsolidation

28

could be used to significantly (sometimes even drastically) reduce the likelihood of maladaptive memory retrieval directly. Then, new inhibitory learning experiences and ways of thinking could be instantiated through supplemental behavioral and cognitive interventions, thereby further reducing the possibility of relapse. A therapist would likely experience much less resistance from patients, and sessions such as exposure could be much less taxing, once a patient has already had their acute maladaptive responses dampened. Therefore, cognitive and behavioral approaches need not be seen as mere competitors or alternatives to a reconsolidation-based approach, but rather as potential allies. Conclusion Understanding mechanisms of change is integral to the optimization of mental health treatments. Translational research spanning multiple levels of analysis, from low-level animal models up to clinical interventions, is already underway in the field of reconsolidation. If efforts at realising the promise of reconsolidation-based treatment are to be successful then it is crucial that the lessons learned from experimental studies are taken into account. Most notably, the importance of prediction error, and the careful balance that must be maintained so as to induce reconsolidation but avoid extinction learning, are of high importance: if the target memory trace is not destabilized then attempts at disruption will prove fruitless (Figure 1 conveys the multiple routes a memory may take after reactivation). However, there are significant limitations to experimental research, and ultimately only attempts at treatment can reveal the utility of a reconsolidation-based approach. Given the intricacy of the problem, researchers should not be overly discouraged if fledgling attempts at reconsolidation-based treatment are unsuccessful. However, we believe that experimental research and early efforts at clinical translation give cause

29

for optimism. Several important questions remain for the field that can be tackled both experimentally and in clinical settings, most pressing of which are: what are the optimal conditions for inducing reconsolidation under the variable conditions met with in therapy, and how can we reliably infer the induction of this process? Box 1 highlights a number of open questions for reconsolidation-based research. If we can answer these questions, and if reconsolidation-based procedures become a viable treatment option, then they would be one of the first mental health treatments to have been derived directly from the translation of neural to clinical science. This would surely be a triumph for the scientific study of mind and brain.

Figure 1. The fate of memory after reactivation: there are many possible outcomes once a memory is reactivated. How reconsolidation-based treatments can best navigate to the desired outcome is a topic of continued research.

References Andreatta, M., & Pauli, P. (2015). Appetitive vs. Aversive conditioning in humans. Frontiers in Behavioral Neuroscience, 9.

Alfei, J. M., Ferrer Monti, R. I., Molina, V. A., Bueno, A. M., & Urcelay, G. P. (2015). Prediction error and trace dominance determine the fate of fear memories after post-training manipulations. Learning & Memory, 22(8), 385-400. doi:10.1101/lm.038513.115

30

Barreiro, K. A., Suarez, L. D., Lynch, V. M., Molina, V. A., & Delorenzi, A. (2013). Memory expression is independent of memory labilization/reconsolidation. Neurobiology of Learning and Memory, 106, 283-291. doi:10.1016/j.nlm.2013.10.006

Bouton, M. E. (2002). Context, ambiguity, and unlearning: Sources of relapse after behavioral extinction. Biological Psychiatry, 52(10), 976-986.

Bouton, M. E., Woods, A. M., & Pineño, O. (2004). Occasional reinforced trials during extinction can slow the rate of rapid reacquisition. Learning and Motivation, 35(4), 371-390. doi://dx.doi.org/10.1016/j.lmot.2004.05.001

Bradley, M. M., Cuthbert, B. N., & Lang, P. J. (1999). Affect and the startle reflex. In M. E. Dawson, A. Schell, & A. Boehmelt (Eds.), Startle modification: Implications for neuroscience, cognitive science and clinical science (pp. 157183). New York, NY: Cambridge University Press.

Brewin, C. R. (2001). Memory processes in post-traumatic stress disorder. International Review of Psychiatry, 13(3), 159-163.

Brewin, C. R. (2006). Understanding cognitive behaviour therapy: A retrieval competition account. Behaviour Research and Therapy, 44(6), 765-784. doi://dx.doi.org/10.1016/j.brat.2006.02.005

Brunet, A., Orr, S. P., Tremblay, J., Robertson, K., Nader, K., & Pitman, R. K. (2008). Effect of post-retrieval propranolol on psychophysiologic responding during subsequent script-driven traumatic imagery in post-traumatic stress disorder. Journal of Psychiatric Research, 42(6), 503-506.

31

Brunet, A., Poundja, J., Tremblay, J., Bui, E., Thomas, E., Orr, S. P., . . . Pitman, R. K. (2011). Trauma reactivation under the influence of propranolol decreases posttraumatic stress symptoms and disorder: 3 open-label trials. Journal of Clinical Psychopharmacology, 31(4), 547-550. doi:10.1097/JCP.0b013e318222f360 [doi]

Butler, A. C., Chapman, J. E., Forman, E. M., & Beck, A. T. (2006). The empirical status of cognitive-behavioral therapy: A review of meta-analyses. Clinical Psychology Review, 26(1), 17-31. doi://dx.doi.org/10.1016/j.cpr.2005.07.003

Chen, S., Cai, D., Pearce, K., Sun, P. Y., Roberts, A. C., & Glanzman, D. L. (2014). Reinstatement of long-term memory following erasure of its behavioral and synaptic expression in aplysia. Elife, 3, e03896.

Clark, L. A., Watson, D., & Mineka, S. (1994). Temperament, personality, and the mood and anxiety disorders. Journal of Abnormal Psychology, 103(1), 103.

Craske, M. G., Treanor, M., Conway, C. C., Zbozinek, T., & Vervliet, B. (2014). Maximizing exposure therapy: An inhibitory learning approach. Behaviour Research and Therapy, 58, 10-23. doi:10.1016/j.brat.2014.04.006

Craske, M. G., Kircanski, K., Zelikowsky, M., Mystkowski, J., Chowdhury, N., & Baker, A. (2008). Optimizing inhibitory learning during exposure therapy. Behaviour Research and Therapy, 46(1), 5-27.

Cuijpers, P., Sijbrandij, M., Koole, S. L., Andersson, G., Beekman, A. T., & Reynolds, C. F. (2013). The efficacy of psychotherapy and pharmacotherapy in

32

treating depressive and anxiety disorders: A meta analysis of direct comparisons. World Psychiatry, 12(2), 137-148.

Davis, M. (1984) The mammalian startle response. In R. C. Eaton (Ed.): Neural Mechanisms of Startle Behavior (pp. 287-351). New York, NY: Plenum.

Dębiec, J., & Ledoux, J. E. (2004). Disruption of reconsolidation but not consolidation of auditory fear conditioning by noradrenergic blockade in the amygdala. Neuroscience, 129(2), 267-272.

Dębiec, J., Doyère, V., Nader, K., & LeDoux, J. E. (2006). Directly reactivated, but not indirectly reactivated, memories undergo reconsolidation in the amygdala. Proceedings of the National Academy of Sciences of the United States of America, 103(9), 3428-3433.

Durham, R. C., Higgins, C., Chambers, J. A., Swan, J. S., & Dow, M. G. (2012). Long-term outcome of eight clinical trials of CBT for anxiety disorders: Symptom profile of sustained recovery and treatment-resistant groups. Journal of Affective Disorders, 136(3), 875-881.

Eisenberg, M., & Dudai, Y. (2004a). Reconsolidation of fresh, remote, and extinguished fear memory in medaka: Old fears don't die. European Journal of Neuroscience, 20(12), 3397-3403. doi:10.1111/j.1460-9568.2004.03818.x

Eisenberg, M., & Dudai, Y. (2004b). Reconsolidation of fresh, remote, and extinguished fear memory in medaka: Old fears don't die. European Journal of Neuroscience, 20(12), 3397-3403.

33

Elsey, J. W. B., van Ast, V. A., & Kindt, M. (2016). Memory reconsolidation in humans: A framework for evaluation and a critical review of the evidence. Unpublished manuscript.

Fernandez, R. S., Boccia, M. M., & Pedreira, M. E. (2016). The fate of memory: Reconsolidation and the case of prediction error. Neuroscience and Biobehavioral Reviews, 68, 423-441.

Gazendam, F. J., Kamphuis, J. H., & Kindt, M. (2013). Deficient safety learning characterizes high trait anxious individuals. Biological Psychology, 92(2), 342352. doi://dx.doi.org/10.1016/j.biopsycho.2012.11.006

Gershman, S. J., Jones, C. E., Norman, K. A., Monfils, M., & Niv, Y. (2013). Gradual extinction prevents the return of fear: Implications for the discovery of state. Frontiers in Behavioral Neuroscience, 7, 164. doi:10.3389/fnbeh.2013.00164

Hamm, A. O., Cuthbert, B. N., Globisch, J., & Vaitl, D. (1997). Fear and the startle reflex: Blink modulation and autonomic response patterns in animal and mutilation fearful subjects. Psychophysiology, 34(1), 97-107.

Hofmann, S. G., & Smits, J. A. (2008). Cognitive-behavioral therapy for adult anxiety disorders: A meta-analysis of randomized placebo-controlled trials. The Journal of Clinical Psychiatry, 69(4), 621-632. doi:ej07m03585 [pii]

Kandel, E. R., Dudai, Y., & Mayford, M. R. (2014). The molecular and systems biology of memory. Cell, 157(1), 163-186.

34

Kazdin, A. E. (2001). Progression of therapy research and clinical application of treatment require better understanding of the change process. Clinical Psychology: Science and Practice, 8(2), 143-151.

Kindt, M., Soeter, M., & Vervliet, B. (2009). Beyond extinction: Erasing human fear responses and preventing the return of fear. Nature Neuroscience, 12(3), 256258.

Kindt, M. (2014). A behavioural neuroscience perspective on the aetiology and treatment of anxiety disorders. Behaviour Research and Therapy, 62, 24-36.

LeDoux, J. E. (2014). Coming to terms with fear. Proceedings of the National Academy of Sciences of the United States of America, 111(8), 2871-2878. doi:10.1073/pnas.1400335111 [doi]

LeDoux, J. E., Romanski, L., & Xagoraris, A. (1989). Indelibility of subcortical emotional memories. Journal of Cognitive Neuroscience, 1(3), 238-243.

Lee, J. L. C. (2009). Reconsolidation: Maintaining memory relevance. Trends in Neurosciences, 32(8), 413-420. doi:10.1016/j.tins.2009.05.002

Lee, J. L., Milton, A. L., & Everitt, B. J. (2006). Reconsolidation and extinction of conditioned fear: inhibition and potentiation. The Journal of Neuroscience, 26(39), 10051-10056.

Leichsenring, F. (2001). Comparative effects of short-term psychodynamic psychotherapy and cognitive-behavioral therapy in depression: A meta-analytic approach. Clinical Psychology Review, 21(3), 401-419.

35

Loerinc, A. G., Meuret, A. E., Twohig, M. P., Rosenfield, D., Bluett, E. J., & Craske, M. G. (2015). Response rates for CBT for anxiety disorders: Need for standardized criteria. Clinical Psychology Review, 42, 72-82.

Mamou, C. B., Gamache, K., & Nader, K. (2006). NMDA receptors are critical for unleashing consolidated auditory fear memories. Nature neuroscience, 9(10), 1237-1239.

McGaugh, J. L. (2000). Memory--a century of consolidation. Science (New York, N.Y.), 287(5451), 248-251. doi:8182 [pii]

McNally, R. J. (2007). Mechanisms of exposure therapy: How neuroscience can improve psychological treatments for anxiety disorders. Clinical Psychology Review, 27(6), 750-759.

Merlo, E., Milton, A. L., Goozée, Z. Y., Theobald, D. E., & Everitt, B. J. (2014). Reconsolidation and extinction are dissociable and mutually exclusive processes: Behavioral and molecular evidence. The Journal of Neuroscience, 34(7), 24222431.

Milekic, M. H., & Alberini, C. M. (2002). Temporally graded requirement for protein synthesis following memory reactivation. Neuron, 36(3), 521-525. doi:10.1016/S0896-6273(02)00976-5

Milton, A. L., & Everitt, B. J. (2012). The persistence of maladaptive memory: Addiction, drug memories and anti-relapse treatments. Neuroscience and Biobehavioral Reviews, 36(4), 1119-1139.

36

Nader, K., & Hardt, O. (2009a). A single standard for memory: The case for reconsolidation. Nature Reviews Neuroscience, 10(3), 224-234.

Nader, K., & Hardt, O. (2009b). A single standard for memory: The case for reconsolidation. Nature Reviews Neuroscience, 10(3), 224-234.

Nader, K., Schafe, G. E., & Le Doux, J. E. (2000). Fear memories require protein synthesis in the amygdala for reconsolidation after retrieval. Nature, 406(6797), 722-726. doi:10.1038/35021052

Pedreira, M. E., & Maldonado, H. (2003). Protein synthesis subserves reconsolidation or extinction depending on reminder duration. Neuron, 38(6), 863-869. doi:10.1016/S0896-6273(03)00352-0

Pérez-Cuesta, L. M., & Maldonado, H. (2009). Memory reconsolidation and extinction in the crab: Mutual exclusion or coexistence? Learning & Memory, 16(11), 714-721.

Powers, M. B., & Emmelkamp, P. M. (2008). Virtual reality exposure therapy for anxiety disorders: A meta-analysis. Journal of Anxiety Disorders, 22(3), 561569.

Przybyslawski, J., & Sara, S. J. (1997). Reconsolidation of memory after its reactivation. Behavioural Brain Research, 84(1), 241-246.

Ryan, T. J., Roy, D. S., Pignatelli, M., Arons, A., & Tonegawa, S. (2015). Engram cells retain memory under retrograde amnesia. Science, 348(6238), 1007-1013. doi:10.1126/science.aaa5542

37

Schafe, G. E., & LeDoux, J. E. (2000). Memory consolidation of auditory pavlovian fear conditioning requires protein synthesis and protein kinase A in the amygdala. The Journal of Neuroscience : The Official Journal of the Society for Neuroscience, 20(18), RC96. doi:20004501 [pii]

Sevenster, D., Beckers, T., & Kindt, M. (2012a). Instructed extinction differentially affects the emotional and cognitive expression of associative fear memory. Psychophysiology, 49(10), 1426-1435.

Sevenster, D., Beckers, T., & Kindt, M. (2012b). Retrieval per se is not sufficient to trigger reconsolidation of human fear memory. Neurobiology of Learning and Memory, 97(3), 338-345.

Sevenster, D., Beckers, T., & Kindt, M. (2013). Prediction error governs pharmacologically induced amnesia for learned fear. Science, 339(6121), 830833. doi:10.1126/science.1231357

Sevenster, D., Beckers, T., & Kindt, M. (2014a). Fear conditioning of SCR but not the startle reflex requires conscious discrimination of threat and safety. Frontiers in Behavioral Neuroscience, 8.

Sevenster, D., Beckers, T., & Kindt, M. (2014b). Prediction error demarcates the transition from retrieval, to reconsolidation, to new learning. International Journal of Neuropsychopharmacology, 17, 130.

Shedler, J. (2010). The efficacy of psychodynamic psychotherapy. American Psychologist, 65(2), 98–109.

38

Soeter, M., & Kindt, M. (2010). Dissociating response systems: Erasing fear from memory. Neurobiology of Learning and Memory, 94(1), 30-41.

Soeter, M., & Kindt, M. (2011a). Disrupting reconsolidation: Pharmacological and behavioral manipulations. Learning & Memory, 18(6), 357-366. doi:10.1101/lm.2148511 [doi]

Soeter, M., & Kindt, M. (2011b). Noradrenergic enhancement of associative fear memory in humans. Neurobiology of Learning and Memory, 96, 263-271.

Soeter, M., & Kindt, M. (2012a). Erasing fear for an imagined threat event. Psychoneuroendocrinology, 37(11), 1769-1779.

Soeter, M., & Kindt, M. (2012b). Stimulation of the noradrenergic system during memory formation impairs extinction learning but not the disruption of reconsolidation. Neuropsychopharmacology, 37(5), 1204-1215

Soeter, M., & Kindt, M. (2013). High trait anxiety: A challenge for disrupting fear memory reconsolidation. PLoS ONE, 8(11) (pagination), Arte Number: e75239. ate of Pubaton: 18 No 2013.

Soeter, M., & Kindt, M. (2015a). Retrieval cues that trigger reconsolidation of associative fear memory are not necessarily an exact replica of the original learning experience. Frontiers in Behavioural Neuroscience, 9, 122. doi: 10.3389/fnbeh.2015.00122.

Soeter, M., & Kindt, M. (2015b). An abrupt transformation of phobic behavior after a post-retrieval amnesic agent. Biological Psychiatry, 78(12), 880-886. doi:10.1016/j.biopsych.2015.04.006Thonberg, H., Fredriksson, J. M.,

39

Nedergaard, J., & Cannon, B. (2002). A novel pathway for adrenergic stimulation of cAMP-response-element-binding protein (CREB) phosphorylation: Mediation via α1-adrenoceptors and protein kinase C activation. Biochemical Journal, 364(1), 73-79.

Tolin, D. F., Lohr, J. M., Sawchuk, C. N., & Lee, T. C. (1997). Disgust and disgust sensitivity in blood-injection-injury and spider phobia. Behaviour Research and Therapy, 35(10), 949-953.

Tonegawa, S., Liu, X., Ramirez, S., & Redondo, R. (2015). Memory engram cells have come of age. Neuron, 87(5), 918-931.

Tronel, S., Milekic, M. H., & Alberini, C. M. (2005). Linking new information to a reactivated memory requires consolidation and not reconsolidation mechanisms. PLoS Biol, 3(9), e293.

Tronson, N. C., & Taylor, J. R. (2007). Molecular mechanisms of memory reconsolidation. Nature Reviews Neuroscience, 8(4), 262-275.

Weike, A. I., Hamm, A. O., Schupp, H. T., Runge, U., Schroeder, H. W., & Kessler, C. (2005). Fear conditioning following unilateral temporal lobectomy: dissociation of conditioned startle potentiation and autonomic learning. The Journal of Neuroscience, 25(48), 11117-11124.

Weike, A. I., Schupp, H. T., & Hamm, A. O. (2007). Fear acquisition requires awareness in trace but not delay conditioning. Psychophysiology, 44(1), 170-180.

40

Winters, B. D., Tucci, M. C., & DaCosta-Furtado, M. (2009). Older and stronger object memories are selectively destabilized by reactivation in the presence of new information. Learning & Memory, 16(9), 545-553.

Wood, N. E., Rosasco, M. L., Suris, A. M., Spring, J. D., Marin, M., Lasko, N. B., . . . Pitman, R. K. (2015). Pharmacological blockade of memory reconsolidation in posttraumatic stress disorder: Three negative psychophysiological studies. Psychiatry Research, 225(1-2), 31-39. doi:10.1016/j.psychres.2014.09.005

Woods, A. M., & Bouton, M. E. (2007). Occasional reinforced responses during extinction can slow the rate of reacquisition of an operant response. Learning and Motivation, 38(1), 56-74. doi://dx.doi.org/10.1016/j.lmot.2006.07.003

41

42

Highlights

•

Conditions for inducing memory reconsolidation are complex, posing difficulties for the translation of experimental research into clinical interventions

•

This review highlights key insights from experimental work in humans and animals that can inform the development of reconsolidation-based treatments

•

Reconsolidation-based treatments hold great potential, but there are several open questions that must be addressed before we witness a paradigm shift in clinical practice

43