Not a member yet? Register for full benefits!

Artificial Memories Placed in an Organic Brain

The quest to place artificial memories in an organic brain has been an extremely long one. In pop culture the idea dates back to the Matrix films, and Neo/Trinity learning new skills by direct upload of the memories of studying all the ins and outs. In research the quest dates back far further, almost to the beginning of neurology itself.

We've long known that false memories are possible, because individuals have falsified their own memories in a great many cases. Especially well-documented in the legal system are cases where convictions have been made based on the testimony of the victim and witnesses, yet overturned later on when impartial evidence such as DNA evidence has surfaced, proving those memories were at least in part, false.

To even begin to truly understand the mechanisms of false memory placement, the brain's memory storage and retrieval systems have been studied in ever increasing detail, with new technologies re-purposed to the same old task. Over the last few years in particular, many studies have increasingly pinpointed where in the hippocampus memories have stored, and investigated the proteins involved in the storage process.

Episodic memories memories of experiences are made of associations of several elements, including objects, space and time. These associations are encoded by chemical and physical changes in neurons, as well as by modifications to the connections between the neurons, and are collectively known as memory engrams.

If we understand when a memory engram is false, it will aid in legal cases where the integrity and validity of a memory needs to be checked. Just as importantly, if we are able to understand the memory formation process, then we can actually choose which memories are remembered meaning that we could realistically actually download knowledge directly into the brain.

It's not the end-game yet, but a major breakthrough has been achieved by MIT researchers, in that for the first time, this is exactly what has been done; artificial memories transferred directly into the hippocampus of a living being.

Not quite ready for prime-time in humans just yet, the new memories have been implanted into a mouse hippocampus, using an optogenetic neural prosthesis, and a variation on the standard gene therapy to activate the gene for channelrhodopsin was applied. This renders the neurons sensitive to light so an optogenetic neuroprosthesis can actually connect to them, and command them when to and not to fire.

The image above is of a mouse hippocampus. The areas outlined in red are areas where engram encoding / retrieval actually occurs.
Credit: Steve Ramirez, Xu Liu, MIT

The variation had a single specific purpose: to have the cells only produce channelrhodopsin (and thus be sensitive to light) when the c-fos gene was turned on in the brain. This gene is naturally activated whenever a new memory engram is stored, across all mammalian brains including humans. It functions like an I/O begin command for the brain's 'hard drive' as it were. So, with this modification when the natural gene turned on to initiate a read/write session, the engineered gene also turned on to make the neurons sensitive to light and interface directly with the optogenetic neuroprosthetic. In this way, the neuroprosthetic became part of the standard engram I/O cycle.

The mice were then conditioned by exposing them to a particular chamber (A) in which they received an electric shock. Every time the memory was accessed during or immediately after the shock, the neuroprosthetic also turned on, becoming part of the network of neurons necessary to encode that particular memory.

After several sessions the mice were accustomed to the fact they would be shocked in chamber A, but not in any other chambers. The mice were dissected and the neuroprosthetics removed.

A second set of mice enter the picture here. They were implanted with the same optogenetic neuroprosthetics as the first set of mice, and underwent the exact same genetic therapy. These mice were not however, initially placed in chamber A. Instead, they were placed in chamber B.

In B, where the mice were subjected to a low voltage, enough to tingle their feet, but not enough to shock them. Every time the mild tingle started, so their memory I/O would activate, and the researchers activated the neuroprosthetic as well, with it's job this time being output rather than input. Rather than becoming part of a memory engram being encoded, it was now encoding the surrounding neurons during the time the memory area was accessible for I/O because the c-fos gene was active.

After a few sessions like this, the mice were placed in chamber A. These mice had never been in chamber A, yet they all reacted as if they expected a major shock, and their hippocampal cells generated the memory engram of having been badly shocked in this chamber before, despite their never having been anywhere near it.

The memory from the first batch of mice had been successfully transferred over to the second batch of mice, where it was as real to them as it would have been had they lived through it themselves to their brains, they had lived through it themselves.

The experiment would not have worked without the optogenetic neuroprosthetic, which was the means through which the memories were transferred from one mouse to another. Unlike with neurons, it had the entire engram encoded within it, because all of the cells surrounding it in the original mice had been made light sensitive, so it could detect when each neuron involved in the engram was made part of that storage network.

In the new mice, it could do the opposite; use its knowledge of the locations and pulse frequencies of the original engram, to activate the now-receptive-to-light local neurons into picking up the exact same activation pattern.

The result was a perfectly preserved memory, taken from one individual and implanted in another.

Now, obviously this is a very crude experiment, and it will be some time yet before we learn enough about what is going on to be able to transfer specific memories, much less encode new knowledge into engram form for implantation in a new host.

However, as a proof of concept, it works extremely well, and marks the first time a successful memory transplant has been performed from one brain into another. With the basic concept proven to be viable, it reaffirms a great many hoped-for possibilities in encoding memories and direct experiences directly into the brain, are indeed possible.


Neuroscientists plant false memories in the brain

Optogenetic stimulation of a hippocampal engram activates fear memory recall (first paper)

Creating a False Memory in the Hippocampus (second paper)

Staff Comments


Untitled Document .