How do memories work? That has been a burning question for over a century, since Richard Semon introduced the term “Engram” in the early 20th century as the fundamental unit of memory in the brain. I should say, that is how long the question existed in its modern neuroscientific form. Thinking about the nature of memory goes back to the early Greek philosophers. In fact we have evidence that they figured out pretty quickly that the brain was central to cognitive function:

Yet an important development in Hellenistic medicine should not remain unmentioned: the discovery, by the Alexandrian anatomists Herophilus and Erasistratus (late fourth/early third century BCE), of the nervous system, of the central role of the brain in cognition and locomotion and the observation, by Herophilus, of the various ventricles within the brain.

Of course, the ancient philosophers also thought that we had souls and that memory and thinking were fundamentally a process of the soul. But the soul had to live somewhere in the body, and to interface with the physical body, and early observations suggested it lived in the brain. This debate carried forward to Medieval times. In the time of the great physician, Galen, the primary debate was whether or not the memories of the soul lived in the heart (cardiocentric view) or the brain (encephalocentric view). Galen himself took the encephalocentric view, that memories and reason are in the brain, but felt that other functions, such as emotion, were in the heart. This is a belief that still resonates with our culture today (amazing how persistent ideas can be).

Let’s fast forward to Semon to the early 20th century. By now it was clearly established that the brain is the organ of memory and cognition. It was also accepted by then that the brain had different identifiable areas with specialized functions. Semon wanted to dig deeper, though, to identify the neuroanatomical correlates of memory itself. He hypothesized that memories are stored in brain cells, in how they connect and function, and that a single memory would correlate to some fundamental unit in the brain, that he called engrams. (On an unrelated side note, the term was stolen by L Ron Hubbard to refer to the memories of past-life trauma that are the true causes of all our psychological problems. It’s nice to see the term returning to is proper scientific use.)

Over the next century neuroscientists tried to identify the brain structures that would correlate with an engram, but did not make much progress, and research was mostly abandoned. This makes this story an interesting cautionary tale – do not necessarily interpret lack of progress as meaning that a theory is wrong, if we simply lack the technology or ability to properly test the hypothesis. There was never any real reason to reject engram theory, it was simply in scientific limbo, awaiting advances in technology that would allow us to properly study it.

Well, that day has come. Over the last 12 years or so scientists have been making incredible advances in understanding the nature of engrams, and this concept has returned to scientific utility. A recent paper reviews this research and our current understanding: Memory engrams: Recalling the past and imagining the future., by Sheena A. Josselyn1 and Susumu Tonegawa. Here are the key insights.

The research suggests that engrams are an ensemble of neurons that encode a specific memory and fire together when that memory is recalled. Further, the engram is part of a greater engram complex, that included neurons in the hippocampus and amygdala, which are already known to be critical to memory formation. Further still, memories can be recalled by activating overlapping related memories. There are two key study types supporting engram theory, as summarized by the authors:

The resurgence in research examining engrams may be linked to two complementary studies that applied intervention strategies to target individual neurons in an engram supporting a specific memory in mice. One study showed that ablating the subset of lateral amygdala neurons allocated to a putative engram disrupted subsequent memory retrieval (loss of function). The second study showed that artificially reactivating a subset of hippocampal dentate gyrus neurons that were active during a fearful experience (and, therefore, part of a putative engram) induced memory retrieval in the absence of external retrieval cues (gain of function). Subsequent findings from many labs used similar strategies to identify engrams in other brain regions supporting different types of memory.

So, if we kill the neurons that link a putative engram to the amygdala, the memory cannot be retrieved. But we can artificially induce memory retrieval by stimulating hippocampal neurons connected to a specific putative engram. Cool.

Further research has looked at how engrams form. They found that when a new memory is being formed, there is competition among available neurons to be recruited to the new memory. Those neurons that are more excitable tend to win this competition, which suggests that excitability (how easy is it to get those neurons to fire) is important to memory formation. Other research also shows that “synaptic plasticity” is also important. That is the ability to change the strength and nature of connections (synapses) between neurons.

A third type of study has found that completely artificial memories can be implanted in mice by manipulating neurons. This is a powerful proof of cause and effect. A fourth type of study has identified “silent” engrams – there are engrams that cannot, for some reason, be naturally retrieved.  External cues simply do not activate the engram. However, the engram can be artificially stimulated, which brings back the memory.

This is still early days. Good research like this introduces more questions than it answers. It is probably too early to say that the engram concept is established enough to be considered a scientific fact, but it’s getting there. The theory is holding up really well to recent research, which is now making rapid progress. Now that the research tools are in place, the research is taking off. But always, it takes decades for complex scientific research like this to mature. But for now engram theory is a good working theory that is consistent with existing research and is proving to be a fruitful theory, after languishing for about a century.