Memory storage and retrieval
Where and how are memories physically stored in the brain, and how are they retrieved?
Summary
Why it is open
In 1904 Richard Semon introduced the term "engram" for the neural substrate of memory: an experience activates a subset of cells that undergo lasting chemical or physical changes, and reactivating them later brings the memory back [11]. Sheena Josselyn and Susumu Tonegawa write that recent findings "are beginning to define an engram as the basic unit of memory" [11]. Which change holds the information is less settled. Wickliffe Abraham, Owen Jones and David Glanzman note that most studies support synapses, the junctions between neurons, as a site of storage, "yet, despite the neuroscience community's best efforts, we are still without conclusive proof that memories reside at synapses" [10].
The problem is several questions. The first is the storage medium: the strength and pattern of synaptic connections, or molecular changes inside neurons [10]. The second is location over time: how memories that first depend on the hippocampus come to depend on the neocortex, on which "empirical inconsistencies and theoretical disagreements remain" [6]. The third is retrieval: whether some amnesia and some forgetting leave memories stored but unreachable by natural cues [7, 12]. How an active synapse is marked so that it can capture the proteins needed for lasting change is a related problem, the molecular identity of the synaptic tag [10].
No body sets a bar for a resolution. Josselyn and Tonegawa list four kinds of evidence that a set of cells in the rodent brain is an engram: the same cells are active during an experience and its retrieval, disrupting them impairs recall, activating them artificially produces recall, and an artificially created engram guides behaviour [11]. Clara Ortega-de San Luis and Tomás Ryan write that questions about the molecular and cellular mechanisms by which information is encoded, stored and recalled "remain largely unanswered" [13].
What is known
Engram cells in rodents (2009–2020). Lateral amygdala neurons with raised levels of the transcription factor CREB are preferentially activated when a fear memory is expressed; in mice, Jin-Hee Han, Josselyn and colleagues killed these neurons after learning, which blocked expression of that memory, and the loss was "robust and persistent", which they read as permanent erasure [2, 11]. Xu Liu, Steve Ramirez and colleagues in Tonegawa's laboratory made dentate gyrus neurons (part of the hippocampus) that were active during fear conditioning responsive to light; reactivating them made mice freeze in a different context, while cells labelled in a context not associated with fear did not [3].
Memory without the expected synaptic change (2014–2018). Ryan and colleagues produced amnesia in mice by blocking protein synthesis right after fear conditioning [5]. The engram cells then lacked the increases in synaptic strength and dendritic spine density seen in consolidated engrams, yet activating them with light still produced recall; the authors proposed that a specific pattern of connectivity among engram cells may store the information [5]. In cultured sensory and motor neurons of the marine snail Aplysia, Shanping Chen, Glanzman and colleagues found that serotonin training grew new presynaptic varicosities (synaptic swellings) [4]. Two treatments that disrupt memory returned their number to the pretraining value, but by retracting new and original varicosities apparently at random, and the memory for sensitisation persisted covertly after the same treatments [4]. In 2018 Alexis Bédécarrats, Glanzman and colleagues injected RNA from the nervous system of Aplysia given long-term sensitisation training (spaced tail shocks that strengthen a withdrawal reflex) into untrained animals, which then showed sensitisation [9, 13]. The effect required DNA methylation, and the RNA increased the excitability of cultured sensory but not motor neurons [9].
Stored but out of reach (2015–2022). In transgenic mouse models of early Alzheimer's disease, Dheeraj Roy and colleagues found that mice failed long-term memory tests with natural cues, but light activation of hippocampal engram cells produced recall, "revealing a retrieval, rather than a storage impairment" [7]. Josselyn and Tonegawa call such engrams "silent": artificial reactivation retrieves the memory, natural cues cannot [11]. Ryan and Paul Frankland suggest that all forms of forgetting involve circuit remodelling that switches engram cells from an accessible to an inaccessible state [12].
Where memories go with time (1997–2025). In the standard model of systems consolidation, information is stored at first in both the hippocampus and the neocortex, and the hippocampus guides a gradual reorganisation until the memory no longer depends on it; early evidence came from hippocampal damage that impaired recent memories but typically spared remote ones [6]. In 1997 Lynn Nadel and Morris Moscovitch argued that retrograde amnesia after damage to the hippocampal complex can encompass much of a lifetime, and that the hippocampus is involved in autobiographical episodic and spatial memories "for as long as they exist" [1]. Larry Squire and colleagues describe multiple-trace theory, later elaborated as the transformation hypothesis, as holding that detailed episodic memories stay hippocampus-dependent while consolidation into the neocortex leaves more gist-based, fact-like memories [6]. In mice, Takashi Kitamura and colleagues found that prefrontal engram cells formed during learning matured over weeks, while the hippocampal engram became silent but could still be reactivated with light [8]. In 2025 Sangyoon Ko, Josselyn, Frankland and colleagues reported in mice that time rewires hippocampal engram circuits, through the birth of new hippocampal neurons, as memories lose precision; they conclude that consolidation models "require updating" [15].
The wiring of an engram (2025). Marco Uytiepo and colleagues reconstructed connections from hippocampal area CA3 to CA1 in mice by three-dimensional electron microscopy, after labelling cells recruited during associative learning [14]. Those neurons "showed no strong preference for wiring with each other"; instead, their connections expanded through presynaptic terminals contacting several partners, alongside enlarged individual synapses [14].
Main approaches
Synaptic weights and wiring. Memory is held in the strength and pattern of connections among engram cells [5, 10]. Strongest evidence: most studies of the question support a role for synapses [10]; consolidated engram cells have stronger synapses and more spines [5]; and the synaptic model easily accommodates synapse specificity, the restriction of lasting change to particular synapses [10]. Main difficulty: recall survives in amnesic mice whose engram cells lack these changes [5], and memory persists covertly in Aplysia after learning-related synaptic growth is erased [4].
Molecules inside the cell. Information is stored within neurons; Abraham and colleagues consider epigenetic storage, particularly DNA methylation, the most likely candidate [10]. Charles Gallistel and colleagues argue that synaptic conductances cannot plausibly encode a number [10]. Strongest evidence: the covert memory in Aplysia [4]; the transfer of sensitisation by RNA [9], data that Abraham, Jones and Glanzman (an author of that study) call "difficult" and "perhaps impossible" to reconcile with the synaptic plasticity hypothesis [10]; and a ferret study in which the timing learned by a cerebellar Purkinje cell was intrinsic to the cell [10]. Main difficulty: how a nucleus could encode which of the neuron's thousands of synapses were strengthened [10]. Ortega-de San Luis and Ryan add that a clear physiological mechanism for handling information stored inside cells would be essential, and that the transferred RNA came from donors immediately after recall [13].
Lost access, not lost storage. Forgetting and some amnesias leave the engram in place but unreachable [12]. Strongest evidence: recall evoked by light in mice made amnesic by a protein-synthesis inhibitor, a result that later studies replicated and extended, one with genetic blockage of protein translation [5, 13], or modelling early Alzheimer's disease [7]. Main difficulty: for the first experiment, Abraham and colleagues noted that it was not known whether other synapses escaped the inhibitor, or how completely protein synthesis was blocked [10]; the extension to all forgetting is offered as a hypothesis [12].
Standard consolidation. Remote memories come to be held by the neocortex independently of the hippocampus [6]. Strongest evidence: in a 2005 study, five patients with hippocampal or larger medial temporal lobe lesions recalled early-life events in as much detail as controls and rated most as "remember" rather than "know" (87.1%, against 80.3% for controls) [6]; prefrontal engrams that mature as hippocampal ones fall silent in mice [8]. Main difficulty: extensive retrograde amnesia [1], and reorganisation within the hippocampus that consolidation models do not yet include [15].
Multiple traces and transformation. The hippocampus stays necessary for detailed episodic memories as long as they last [1, 6]. Strongest evidence: retrograde amnesia whose extent depends on the type of memory assessed [1]. Main difficulty: the 2005 patients' detailed "remember" recollections [6].
Open sub-questions
- Wiring after amnesia. In mice made amnesic by a protein-synthesis inhibitor, does electron microscopy of engram cells show retained engram-specific connections, as the connectivity proposal requires, given that normal engram cells showed no strong preference for wiring with each other [5, 14]?
- Steady-state RNA. Does RNA taken from trained Aplysia days after training, without a recall session, still transfer sensitisation, and which RNA species carries the effect [9, 13]?
- Specificity in a cell-intrinsic store. In Aplysia cultures with one sensory neuron and two motor neurons, does a memory reinstated after reconsolidation blockade return only at the trained branch [4, 10]? See Molecular identity of the synaptic tag.
- Detail at remote times. When hippocampal engram cells have become silent, do prefrontal engram cells carry the contextual detail of a memory or only its gist [8, 15]?
- The limit of retrieval. After natural forgetting in wild-type mice, at what delay does light activation of the engram stop producing recall [7, 12]?
Work on this in Papex
If you have a testable answer to one of the sub-questions above, you can draft it as a private hypothesis in Papex, or prepare the draft from Claude or ChatGPT with the Papex connector. You review the exact version, sign it and decide whether to publish it; a public record that names this problem is listed on this page. A Papex record shows who signed which version and when. It is not peer review and does not mean the hypothesis is correct.
Key references
- Nadel, L. & Moscovitch, M. (1997). Memory consolidation, retrograde amnesia and the hippocampal complex. Current Opinion in Neurobiology 7(2), 217–227. 10.1016/S0959-4388(97)80010-4 · Checked 16 Oct 2026
- Han, J.-H., Kushner, S. A., Yiu, A. P., Hsiang, H.-L., Buch, T., Waisman, A. et al. (2009). Selective erasure of a fear memory. Science 323(5920), 1492–1496. 10.1126/science.1164139 · Checked 16 Oct 2026
- Liu, X., Ramirez, S., Pang, P. T., Puryear, C. B., Govindarajan, A., Deisseroth, K. & Tonegawa, S. (2012). Optogenetic stimulation of a hippocampal engram activates fear memory recall. Nature 484(7394), 381–385. 10.1038/nature11028 · Checked 16 Oct 2026
- Chen, S., Cai, D., Pearce, K., Sun, P. Y.-W., 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. 10.7554/eLife.03896 · Checked 16 Oct 2026
- 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. 10.1126/science.aaa5542 · Checked 16 Oct 2026
- Squire, L. R., Genzel, L., Wixted, J. T. & Morris, R. G. (2015). Memory consolidation. Cold Spring Harbor Perspectives in Biology 7(8), a021766. 10.1101/cshperspect.a021766 · Checked 16 Oct 2026
- Roy, D. S., Arons, A., Mitchell, T. I., Pignatelli, M., Ryan, T. J. & Tonegawa, S. (2016). Memory retrieval by activating engram cells in mouse models of early Alzheimer's disease. Nature 531(7595), 508–512. 10.1038/nature17172 · Checked 16 Oct 2026
- Kitamura, T., Ogawa, S. K., Roy, D. S., Okuyama, T., Morrissey, M. D., Smith, L. M. et al. (2017). Engrams and circuits crucial for systems consolidation of a memory. Science 356(6333), 73–78. 10.1126/science.aam6808 · Checked 16 Oct 2026
- Bédécarrats, A., Chen, S., Pearce, K., Cai, D. & Glanzman, D. L. (2018). RNA from trained Aplysia can induce an epigenetic engram for long-term sensitization in untrained Aplysia. eNeuro 5(3), ENEURO.0038-18.2018. 10.1523/ENEURO.0038-18.2018 · Checked 16 Oct 2026
- Abraham, W. C., Jones, O. D. & Glanzman, D. L. (2019). Is plasticity of synapses the mechanism of long-term memory storage? npj Science of Learning 4, 9. 10.1038/s41539-019-0048-y · Checked 16 Oct 2026
- Josselyn, S. A. & Tonegawa, S. (2020). Memory engrams: recalling the past and imagining the future. Science 367(6473), eaaw4325. 10.1126/science.aaw4325 · Checked 16 Oct 2026
- Ryan, T. J. & Frankland, P. W. (2022). Forgetting as a form of adaptive engram cell plasticity. Nature Reviews Neuroscience 23(3), 173–186. 10.1038/s41583-021-00548-3 · Checked 16 Oct 2026
- Ortega-de San Luis, C. & Ryan, T. J. (2022). Understanding the physical basis of memory: molecular mechanisms of the engram. Journal of Biological Chemistry 298(5), 101866; correction of two grammatical errors in 299(8), 105070 (2023). 10.1016/j.jbc.2022.101866 · Checked 16 Oct 2026
- Uytiepo, M., Zhu, Y., Bushong, E., Chou, K., Polli, F. S., Zhao, E. et al. (2025). Synaptic architecture of a memory engram in the mouse hippocampus. Science 387(6740), eado8316. 10.1126/science.ado8316 · Checked 16 Oct 2026
- Ko, S. Y., Rong, Y., Ramsaran, A. I., Chen, X., Rashid, A. J., Mocle, A. J. et al. (2025). Systems consolidation reorganizes hippocampal engram circuitry. Nature 643(8072), 735–743. 10.1038/s41586-025-08993-1 · Checked 16 Oct 2026
Source
Open, as listed by Wikipedia · Unsolved problems in neuroscience. Checked 08 Oct 2026.
The status is the source’s. Papex does not decide whether a problem is solved.
Records on this problem
No public record addresses this problem yet.