top of page

Written by Brett Weiss

February 2019

Defining The Engram

Pinpointing the cellular basis of memory is not easy work, but many think it’s already been done.


Neuroscientists have long sought to pinpoint the elusive cellular mechanisms constituting the molecular basis of memory in the brain.  In 1949, Donald Hebb postulated that memories exist in cell assemblies, which he called the “engram”. Although untestable during his time, Hebb hypothesized that assemblies form through the strengthening or weakening of neural connections while a memory is being acquired.  Supporting his theory, in the 1970s and 1980s, neuroscientists discovered that the strength of particular neural connections could be enhanced upon stimulation - in other words, synapses are plastic (Poo et al., 2016). The discovery of synaptic strength was paradigm shifting and provided substantial credulity to Hebb’s concept, but the theoretical obstacle of integrating memory storage and synaptic plasticity remained (Tonegawa et al., 2015).  Hebb lacked the molecular biological tools to validate the existence of the engram, but the past several decades have beared witness to spectacular advances in neurotechnology. And within the past decade, researchers have started to think they’ve found hard evidence that Hebb was right-- memory exists in cell assemblies, and we can almost point them out cell by cell.

Proving the existence of the engram is not easy work, but many think it’s already been done.  The idea is this: take a group of neurons that are active while an animal is acquiring a memory, say they’re learning that every time they’re put in a particular cage, they’re going to get a foot shock.  At the exact moment the animal is experiencing the shock, you place a genetic marker in each neuron that was active during the shocking. Not easy work I said. What’s crucial is that this marker you placed can be controlled, artificially, so that each neuron receiving a marker can be activated upon command.  If you then place this animal in a new environment, where they haven’t ever been shocked, you’d expect them not to be too concerned about being shocked. But if you do this and artificially turn on the neurons that were active during the original foot shock, and the animal freezes in fear, well, you might then start to think you’ve made a very important discovery.


Want more Brain Science in your life? 

Got a dollar per month to spare?

Support Mind & Brain Illustrated today on Patreon.


Liu et al. (2012) performed an exciting study where they did just that.  From a slightly more technical perspective, here’s how you can approach this problem.  When neurons are activated, a particular set of genes are expressed. By creating artificial genetic constructs, researchers were able to link the natural expression of one of these genes to the expression of two others: First, a fluorescent gene, so neurons could be visualized, and second, a light-activated ion channel that causes neurons to fire upon stimulation with a blue light.   When neuron activity occurred, these genes became active, causing expression of the fluorescent gene and the light-activated ion channel. Theoretically, the neurons that expressed these genes would consist of neuronal populations that were active while the animals were doing something, say, encoding a memory.

To test this, mice were first placed in a particular environment, context B, where they received a mild foot shock.  When mice were returned to their home cage, then back into context B, they froze, presumably in fear of another foot shock.  The learning that occurred in context B theoretically entailed associating the foot shock with context B, which elicited freezing behavior.  Hence, a separate environment, context A, would not elicit freezing behavior, because it had never been paired with a shock. When experimenters placed the experimental mice in context A, as expected, no freezing behavior was observed.  However, when the animals were in context B, they weren’t just getting a foot shock. That genetic construct was being activated, distributing light sensitive ion channels to all the neurons that were active when the mice were learning to expect a foot shock.  So, when these animals were placed in context A, but this time while stimulating the light-sensitive receptors, ~35% of the mice displayed freezing behavior. This showed that engram cells were labelled with light-activated channels that could activate a learned behavior through artificial stimulation.


This study provides evidence for the sufficiency of a memory engram for behavioral expression of a fear memory.  At the same time, it does not necessarily provide evidence that this labelled cellular ensemble is necessary for recall.  The genetic protocol used in this study was focused on a very particular brain region, a sub-region of the famous memory-center called the hippocampus.  The group went on to theorize that during contextual fear conditioning, it’s likely that multiple contextual memory engrams get encoded in a series of regions in the hippocampus, not just the precise region they targeted.   Therefore, the possibility still exists that activation of other engrams in different regions may elicit the same freezing response (Liu et al., 2012).

Neuroscientists have an idea that Hebb was on the right track in associating memories with “cell assemblies”, as studies like the one mentioned here have provided strong evidence that the engram exists in this form.  Thus, it appears that neuroscientists have made great strides towards identifying the neural substrate of encoded memories. What this all means is that now is that instead of studying memory through inhibition or ablation of large-scale brain regions, scientists can study particular memory engram cells.  Engram targeted technologies have given scientists an unprecedented degree of precision in the study of memory.


  1. Liu, Xu, et al. "Optogenetic stimulation of a hippocampal engram activates fear memory recall." Nature 484.7394 (2012): 381.

  2. Poo M, Pignatelli M, Ryan T, Tonegawa S, Bonhoeffer T, Martin K, Rudenko A, Tsai L, Tsien R, Fishell G, Mullins C, Goncalves J, Shtrahman M, Johnston S, Gage F, Dan Y, Long J, Buzsaki G, Stevens C (2016).  “What is Memory? The present state of the engram.” BMC Biology.  14(40): 1-18.

  3. Tanaka KZ and McHugh T (2018).  “The Hippocampal Engram as a Memory Index.”  J Exp Neurosci.  12: 1-4.

  4. Tonegawa S, Pignatelli M, Roy DS, Ryan TJ (2015).  “Memory engram storage and retrieval.” Curr Opin Neurobiol.  35: 101-9.

bottom of page