Gone for 60 seconds: Reactivation length determines motor memory degradation during reconsolidation

Abstract

When a stable memory is reactivated it becomes transiently labile and requires restabilization, a process known as reconsolidation. Animal studies have convincingly demonstrated that during reconsolidation memories are modifiable and can be erased when reactivation is followed by an interfering intervention. Few studies have been conducted in humans, however, and results are inconsistent regarding the extent to which a memory can be degraded. We used a motor sequence learning paradigm to show that the length of reactivation constitutes a crucial boundary condition determining whether human motor memories can be degraded. In our first experiment, we found that a short reactivation (less than 60 sec) renders the memory labile and susceptible to degradation through interference, while a longer reactivation does not. In our second experiment, we reproduce these results and show a significant linear relationship between the length of memory reactivation and the detrimental effect of the interfering task performed afterwards, i.e., the longer the reactivation, the smaller the memory loss due to interference. Our data suggest that reactivation via motor execution activates a time-dependent process that initially destabilizes the memory, which is then followed by restabilization during further practice.

Introduction

Animal research has convincingly demonstrated that reactivated stable memories are rendered transiently labile. These memories require subsequent restabilization, a process known as reconsolidation. In a landmark study, Nader, Schafe, and Le Doux (2000) showed that fear memories can be erased when protein synthesis is inhibited shortly after memory reactivation. Since then similar interference effects have been demonstrated across various memory domains indicating that reconsolidation is a universal process contributing to long-term memory formation (Besnard et al., 2012, Nader and Hardt, 2009, Nader and Einarsson, 2010).

Although reconsolidation has also been demonstrated in humans there are far fewer studies than in animals (Schiller & Phelps, 2011). Unlike in animal models where reconsolidation is blocked by injecting protein synthesis inhibitors directly into specific brain areas, human memories are interfered with either by acquiring a competing task (Chan and LaPaglia, 2013, Forcato et al., 2007, Hupbach et al., 2007, Walker et al., 2003) or by orally administered drugs like propranolol (Brunet et al., 2008, Kindt et al., 2009, Soeter and Kindt, 2011). In particular, the opportunity to erase pathological memories holds considerable clinical potential, e.g., in the treatment of chronically relapsing disorders caused by post-traumatic stress (Auber et al., 2013, Parsons and Ressler, 2013) or for weakening erroneous movement representations hampering the development of more efficient movement patterns during recovery from brain injury. However, previous studies in both humans and animals are inconsistent regarding whether a memory is truly degraded (e.g., in humans Kindt et al., 2009, Walker et al., 2003) or not (e.g., in humans Censor et al., 2010, Censor et al., 2014, Hupbach et al., 2007). These inconsistent findings are captured by two competing accounts (Lee, 2009): First, the destabilization theory posits that in order to add new information to an existing memory it is first destabilized, then modified, and finally restabilized generating a modified memory trace for future recall (Fig. 1A). Importantly, this hypothesis predicts that causing interference during the destabilization phase results in memory loss (Fig. 1B) (Chan and LaPaglia, 2013, Kindt et al., 2009, Nader et al., 2000, Walker et al., 2003). By contrast, the updating theory postulates that reactivating a stable memory opens a time window where the memory is modifiable, but importantly, no destabilization phase occurs (Fig. 1C). It predicts that interference blocks performance gains that would have been observed during uninterrupted memory formation (Fig. 1D), but it does not induce performance decrements (Censor et al., 2010, Censor et al., 2014, Rodriguez-Ortiz and Bermúdez-Rattoni, 2007).

One explanation for the divergent findings is that subtle boundary conditions constrain whether a memory can be experimentally interfered with upon reactivation (Rodriguez-Ortiz & Bermúdez-Rattoni, 2007). Whilst specific determinants of reconsolidation have been identified for animal models in the past (Auber et al., 2013, Bustos et al., 2009, Suzuki et al., 2004), they are currently not well understood in humans (Auber et al., 2013, Schiller and Phelps, 2011, Sevenster et al., 2013).

Here we used a motor sequence learning paradigm to show that the length of reactivation constitutes a crucial boundary condition determining whether a motor memory can be degraded. We show that reconsolidation is a dynamic time-dependent process that is initiated by memory reactivation and characterized by an initial destabilization phase followed by restabilization when reactivation is prolonged.