Lane et al. propose that emotional arousal and memory reconsolidation mechanisms are integral to successful behavior change during psychotherapy. However, the neural mechanisms that integrate these processes to achieve the therapeutic goals are unknown. Here I discuss why this proposal presents a quandary to existing neurobiological accounts of emotional memory. Memory reconsolidation has been studied in nonhuman animals using predominantly pharmacologic and cellular neuroscientific techniques to identify the molecular pathways involved. Existing studies have focused on conditioned fear or conditioned reward memories, which have revealed a critical role for the amygdala in memory reconsolidation through engagement of intrinsic second messenger systems, protein synthesis, and a wide range of neuromodulatory influences (Diergaarde et al. Reference Diergaarde, Anton and DeVries2008; Nader & Hardt Reference Nader and Hardt2009). Recent extensions of this work to humans have confirmed enhanced amygdala activity during the reactivation of a conditioned fear memory (Agren et al. Reference Agren, Engman, Frick, Björkstrand, Larsson, Furmark and Fredrikson2012), which leads to less of a need for ventromedial prefrontal regulation during extinction training (Schiller et al. Reference Schiller, Kanen, LeDoux, Monfils and Phelps2013).

Conditioning may be a special case, as the amygdala itself may serve as a permanent site of storage of the fear memory. Attempts to translate memory reconsolidation mechanisms to other aspects of human memory, as well as pharmacologic manipulations of reconsolidated conditioned memories, have yielded mixed results to date (Schiller & Phelps Reference Schiller and Phelps2011). Although conditioning paradigms may provide a useful model for some aspects of anxiety disorders, many therapeutic efforts are focused on altering episodic memories of prior emotional experiences, which involve brain regions beyond the amygdala. McGaugh's memory modulation hypothesis proposes that the amygdala serves to enhance consolidation processes occurring in other memory systems, such as the hippocampus, through both direct neural interactions and indirectly through the release of stress hormones (McGaugh Reference McGaugh2000). This hypothesis, however, does not deal directly with reconsolidation processes, which are not necessarily synonymous with initial consolidation processes (Besnard et al. Reference Besnard, Caboche and Laroche2012). Nor is it known how the amygdala interacts with other brain regions at a neural systems level to support the reconsolidation of episodic emotional memories.

Neuroimaging studies of emotion regulation provide a further complication to incorporating the reconsolidation idea into a neural systems framework. Down-regulation of negative affect consistently reduces amygdala activity but increases activity in lateral prefrontal regions (Ochsner et al. Reference Ochsner, Silvers and Buhle2012). The degree of amygdala reduction is correlated with individual differences in cognitive abilities and is functionally coupled to enhanced activity in the ventrolateral prefrontal cortex (Winecoff et al. Reference Winecoff, LaBar, Madden, Cabeza and Huettel2011). This pattern of results is exactly opposite to that shown by the initial neuroimaging studies of conditioned fear memory reconsolidation discussed above. One challenge in integrating these research domains, beyond a difference in memory systems, is that laboratory studies of emotion regulation typically use novel stimuli whose representations are being actively altered in working memory rather than operating on a reactivated long-term memory trace. Nonetheless, similar ventrolateral prefrontal cortex results are found when regulating autobiographical memories (Kross et al. Reference Kross, Davidson, Weber and Ochsner2009).

A final issue is that most laboratory studies of emotion regulation do not investigate how the corticolimbic interactions change with prolonged practice. The therapeutic process is dynamic, and as therapy ensues, emotional arousal is reduced. Given that emotional arousal is a key factor in enhancing consolidation, the memory modulation hypothesis would predict less consolidation for later stages of therapy, with weaker memories being formed of the modified representations. Therein lies the full conundrum: If amygdala-dependent processes are key to (re)consolidation mechanisms, why would amygdala activation decrease as memories are being reworked, and how would the reconsolidated memories get encoded into long-term storage in the absence of high emotional arousal?

Importantly, Lane et al. propose that other factors contribute to the reworking of memories in the therapeutic context, including semantic elaboration processes and affect labeling. These processes are associated with ventrolateral prefrontal cortex function, which sometimes reduces amygdala activation but increases hippocampal activation to consolidate material into long-term memory (Dolcos et al. Reference Dolcos, LaBar and Cabeza2004; Lieberman et al. Reference Lieberman, Eisenberger, Crockett, Tom, Pfeifer and Way2007). Deep semantic processing of emotional material enhances memory, even in amygdala-lesioned patients (Phelps et al. Reference Phelps, LaBar and Spencer1997). Therefore, prefrontal-hippocampal pathways may provide a means by which new information integrated into the prior trauma episode can get consolidated into long-term memory even in the absence of amygdala-dependent, arousal-mediated (re)consolidation mechanisms.

Hayes et al. (Reference Hayes, Morey, Petty, Seth, Smoski, McCarthy and LaBar2010) provided initial empirical support for these putative frontolimbic interactions that integrate emotion regulation and memory consolidation processes. In this functional magnetic resonance imaging (fMRI) study, participants engaged in cognitive reappraisal or expressive suppression of emotional pictures, followed by a memory test for the regulated material and for passively viewed emotional and neutral pictures. Relative to passive viewing of emotional pictures, both forms of emotion regulation reduced amygdala activation and valence ratings. Despite the overall reduction in amygdala activity, the residual amygdala activation remained functionally coupled with the hippocampus to predict subsequent memory but only in the reappraise condition. The reappraise condition also uniquely engaged ventrolateral prefrontal cortex interactions with the hippocampus to predict later memory. The reappraised emotional pictures had the highest memory scores overall, likely due to this selective “double boost” in hippocampal function. These results are consistent with a depth-of-processing account of emotion regulation (Dillon et al. Reference Dillon, Ritchey, Johnson and LaBar2007), which argues that beneficial regulatory strategies, such as cognitive reappraisal, that foster semantic encoding of the reappraised material will enhance memory despite a reduction in arousal. By contrast, regulatory strategies that promote shallow processing of the regulated material, such as expressive suppression or attentional distraction, will impair memory.

Although these considerations provide some insights into the putative neural interactions involved, clearly more empirical research is needed to identify emotional memory reconsolidation mechanisms of the sort envisaged by Lane et al. In particular, there is a strong need for a broader neural-systems perspective on memory reconsolidation processes that go beyond intracellular amygdala-dependent mechanisms identified for conditioned learning. Future validating studies should integrate emotional memory, emotion regulation, and reconsolidation into a single paradigm that also accounts for the temporal dynamics that unfold over multiple sessions.