Mol. Cells

Forgetting Is Not the Mere Opposite of Learning

Additional article information


We always endeavor to remember something. Couples use D-Day mobile apps not to forget anniversaries, and students make to-do lists not to forget submission deadlines for overwhelming tasks. Forgetting those dates can be quite embarrassing for most people. On the contrary, we rarely try to forget something except for traumatic memories. We easily consider forgetting a natural process as time passes, so do not put effort into trivial memories. However, did you know that neurons also have to actively work to forget encoded memories? For neurons, forgetting is considered another form of learning that involves engram plasticity, not just the reversal of learning.

Neurobiologists classify the learning process into three steps: (1) encoding, (2) consolidation, and (3) retrieval. At the encoding stage, we acquire new information from sensory experience and store it in our brain, which is very labile. At the consolidation stage, newly acquired memory makes connections with pre-existing memories and becomes solid. Neuroplasticity should be accompanied at this stage to create long-lasting memories. At the retrieval stage, the same neural circuit pattern is activated to represent the stored memory. The more frequently you recall, the easier it is to retrieve, i.e., the memory is reconsolidated. Engram is the trace of learning engraved in the brain at synaptic, cellular, or neuronal circuit levels, being shaped during the encoding stage of learning. Engram cells experience chemical and physical changes to stably store learned memories. They are selectively reactivated at the retrieval stage, resulting in the recall of the original memory (Tonegawa et al., 2018).

Engram studies have been accelerated thanks to the immediate-early gene tagging system combined with optogenetics and chemogenetics (Ryan et al., 2015; Tonegawa et al., 2015). Early studies were based on the cellular level of engram, but recent studies help us understand the mechanisms of memory at the molecular level. Beyond classic approaches to engram, scientists now examine three-dimensional genomic architecture to find the basis of the memory trace at a transcriptional level. Epigenomic change, a prerequisite for transcriptional change, also exhibits interesting patterns across the lifespan of memory formation and recall. During the encoding stage, the accessibility of enhancers increased without corresponding transcriptional changes. Next, consolidation leads to the spatial reorganization of chromatin segments and enhancer-promoter interactions, establishing a permissive chromatin environment. Subsequently, the reactivation of neurons is associated with de novo long-range interactions, where previously primed enhancers upregulate genes responsible for synaptic compartments (Marco et al., 2020).

What specifically occurs in engram cells during forgetting? The mechanisms of engram cell-mediated forgetting include receptor trafficking, spine instability, inhibition, synapse elimination, and neurogenesis (Ryan and Frankland, 2022). One or a combination of these mechanisms is applied to the process via epigenetic and transcriptional changes. Depending on cognitive engagement, forgetting can be divided into two categories: incidental and intentional forgetting. Incidental forgetting lacks cognitive control and includes biological decay (turnover of engram components) and retrieval interference (Davis and Zhong, 2017). The latter can be induced by accumulated similar memories: If you work in a biological lab, a timer is necessary. You may use the timer for mouse behavioral tests, plasmid preparation, etc., and every experiment occurs in different places. Therefore, the place of the timer differs every moment, so the memory of places accumulates and competes. Then you failed to clearly remember where you put the timer. Like this, retrieval interference occurs because previously learned information interferes with newly learned information, which is in a similar context, before the new information is converted into long-term and solid memory. Intentional forgetting involves cognitive control and includes erasing memory traces and inhibition. Erasing memory traces is related to morphological changes in a spine; a small protrusion from a neuron’s dendrite which receives signals from an axon. Rac1, a key modulator in intentional forgetting, regulates actin turnover and polymerization, which in turn remodels synaptic spines. Rac1, in conjunction with other postsynaptic proteins, receives dopamine from a presynaptic neuron and then generates a forgetting signal in that engram cell (Davis and Zhong, 2017). In mammals, GluA2-containing AMPA receptor endocytosis facilitates forgetting (Hardt et al., 2014). This seems natural because the AMPA receptor permits Ca2+ influx, which ultimately affects the gene expression and morphology of a cell. In both cases, memory traces in engram cells are removed through intracellular signaling and are no longer activated by the original stimulus. Alternatively, instead of erasing traces, intentional forgetting can strengthen inhibitory connections to engram cells while leaving the engram intact to prevent memory recall. Under suppression from inhibitory neurons, engram cells are no longer reactivated in response to retrieval cues. Ten Oever et al. (2021) gave a straightforward example of how an inhibition cue after acquisition induces forgetting. In the study phase, participants are presented with some words, each of which is followed by a “remember” or “forget” cue. This cue does not disturb the encoding phase; in other words, the participants are unaware at the moment of encoding whether a word is instructed to be remembered or forgotten. The “forget” cue modifies inhibitory connections to an acquired item after an initial memory trace has been built. Forgetting by inhibition resembles extinction learning, where the repeated presentation of the conditioned stimulus without the unconditioned stimulus reduces the conditioned response. Accumulated mismatches between expectations and the actual result decrease accessibility to engram cells, which makes memory forgotten. Extinction learning is an effective therapy for the well-known psychiatric disorder, post-traumatic stress disorder (PTSD). However, inhibiting the engram concerned with the traumatic memory cannot guarantee a person a normal life. Once heightened, responsiveness to stress remains in the mind and body long after trauma. Therefore, a person who suffered from PTSD becomes vulnerable to stressful daily events. This is why forgetting is another form of learning that requires much effort but can never be perfect.

In recent decades, artificial intelligence (AI) has attempted to mimic the human brain using “deep neural networks (DNN)”. Inspired by neuronal synapses, DNNs use multiple nodes and layers to process data, which enables parallel processing and dramatically increases computational speed. Nevertheless, DNNs still have limitations because they simply borrow the basic concept of how the human brain works. One of the biggest challenges for DNN is catastrophic forgetting (Kirkpatrick et al., 2017). As AI learns new things, older information is abruptly lost because the weight of the older one moves to learn new things. Scientists overcome this challenge by adding a new network when a new task arrives and integrating prior tasks into this network; interleaving data from various tasks during learning. As a result, partial damage does not lead to complete memory loss. This concept is analogous to the human brain, where the sound, color, and texture of a cat are stored separately in distinct brain areas and combined when needed. As DNN embodies the intricacy of the human memory system, we can realize that an active and elaborate forgetting process is required to remove an engram. Conversely, this implies that our brain can safely store data.

From now on, there is no need to blame yourself for having a poor memory. It is important to understand that your neurons actually employ sophisticated mechanisms to actively forget certain memories, and this process can be crucial for your overall mental well-being.

Article information

Mol. Cells.Aug 31, 2023; 46(8): 470-472.
Published online 2023-06-28. doi:  10.14348/molcells.2023.0072
Department of Life Sciences, Pohang University of Science and Technology, Pohang 37673, Korea
Received May 2, 2023; Accepted May 8, 2023.
Articles from Mol. Cells are provided here courtesy of Mol. Cells


  • Davis, R.L., Zhong, Y. (2017). The biology of forgetting-a perspective. Neuron. 95, 490-503.
  • Hardt, O., Nader, K., Wang, Y.T. (2014). GluA2-dependent AMPA receptor endocytosis and the decay of early and late long-term potentiation: possible mechanisms for forgetting of short- and long-term memories. Philos. Trans. R. Soc. Lond. B Biol. Sci.. 369, 20130141.
  • Kirkpatrick, J., Pascanu, R., Rabinowitz, N., Veness, J., Desjardins, G., Rusu, A.A., Milan, K., Quan, J., Ramalho, T., Grabska-Barwinska, A. (2017). Overcoming catastrophic forgetting in neural networks. Proc. Natl. Acad. Sci. U. S. A.. 114, 3521-3526.
  • Marco, A., Meharena, H.S., Dileep, V., Raju, R.M., Davila-Velderrain, J., Zhang, A.L., Adaikkan, C., Young, J.Z., Gao, F., Kellis, M. (2020). Mapping the epigenomic and transcriptomic interplay during memory formation and recall in the hippocampal engram ensemble. Nat. Neurosci.. 23, 1606-1617.
  • Ryan, T.J., Frankland, P.W. (2022). Forgetting as a form of adaptive engram cell plasticity. Nat. Rev. Neurosci.. 23, 173-186.
  • Ryan, T.J., Roy, D.S., Pignatelli, M., Arons, A., Tonegawa, S. (2015). Memory. Engram cells retain memory under retrograde amnesia. Science. 348, 1007-1013.
  • Ten Oever, S., Sack, A.T., Oehrn, C.R., Axmacher, N. (2021). An engram of intentionally forgotten information. Nat. Commun.. 12, 6443.
  • Tonegawa, S., Liu, X., Ramirez, S., Redondo, R. (2015). Memory engram cells have come of age. Neuron. 87, 918-931.
  • Tonegawa, S., Morrissey, M.D., Kitamura, T. (2018). The role of engram cells in the systems consolidation of memory. Nat. Rev. Neurosci.. 19, 485-498.

Figure 1

“Learning and Memory” is a fundamental process in our brain and can be divided into three steps: (1) encoding, (2) consolidation, and (3) retrieval. Forgetting, a seemingly opposite process, is indeed an active and sophisticated process that destabilizes connections between neurons. Depending on cognitive engagement, forgetting can be divided into two categories: incidental and intentional forgetting. Incidental forgetting lacks cognitive control and includes biological decay and retrieval interference. Intentional forgetting involves cognitive control and includes erasing memory traces and inhibition.