We are constantly forgetting. For a moment, consider what you have done today. It might be relatively straightforward, for example, to recall your journey to work. Similarly, you might recall easily the people you encountered and conversations you had, even with a fair degree of clarity. We keep a good record of these ordinary, everyday events. But what about memories of similar, everyday activities that occurred a week ago? Or a month or a year ago? Unless something extraordinary occurred, it is unlikely that you can recall much of what happened, and certainly not in any great detail. This reflects the fact that while we are continuously encoding our experiences, the vast majority of these encoded experiences (or memories) are ‘cleared away’ and only a small portion ultimately retained. The hippocampus is thought to be the automatic encoder, with the cortex serving as the final repository for the fraction of memories that are successfully consolidated (Wang and Morris, 2010). But how are memories cleared from the hippocampus?

Recent work has identified one likely clearance mechanism. In the hippocampus, new cells are continuously generated in the subgranular zone of the dentate gyrus. Most of these new cells differentiate into granule cells and migrate into the granule cell layer, where, after a few weeks, they synaptically integrate into the hippocampal circuitry. There has been plenty of interest in how these newly generated neurons might facilitate the formation of new memories (eg, by increasing the mnemonic capacity or facilitating certain types of computations carried out by the dentate gyrus, such as pattern separation) (Christian et al, 2014). However, as new cells integrate into the hippocampus they necessarily remodel existing circuitry. This remodeling may degrade memories already stored in those circuits (or at least render them difficult to access) (Deisseroth et al, 2004; Weisz and Argibay, 2012). We recently provided experimental support for this prediction (Akers et al, 2014). Voluntary exercise increases hippocampal neurogenesis in adult mice. We found that running-induced increases in neurogenesis led to forgetting of established contextual fear and spatial memories. While running induces a number of physiological changes, the forgetting effects appeared to depend on elevated neurogenesis, since genetically attenuating this consequence of running prevented forgetting. Furthermore, pharmacological (eg, memantine, fluoxetine) and genetic (conditional deletion of p53 from neural progenitors) interventions that artificially elevate hippocampal neurogenesis, when introduced after training, similarly weakened existing hippocampus-dependent memories, suggesting that running-induced forgetting is mediated by a neurogenic mechanism.

There are two important implications of these findings. First, not only do they tell us about how forgetting normally occurs, but perhaps additionally they hint at an important functional consequence of ongoing neurogenesis in the adult hippocampus. Established memories interfere with encoding of new memories, especially when the new and established memories are in conflict with one another. By continuously clearing hippocampal memories, ongoing neurogenesis may serve to minimize this form of proactive interference (Frankland et al, 2013). Second, these findings identify a mechanism that could be targeted in memory-related disorders. For example, inefficient neurogenesis-mediated clearance may contribute to human disorders characterized by problems with memory interference (eg, in old age and Alzheimer’s disease) or rumination (eg, in PTSD and depression). Interestingly, stress may compound these conditions by further lowering the rates of ongoing neurogenesis.

FUNDING AND DISCLOSURE

The authors declare no conflict of interest.