Implanting False Memories

Like so many other psychological phenomena once so mysterious as to be treated almost as supernatural, memory is without doubt the result of the biological, chemical and physical activities of the brain. Another step forward in our understanding of how these mechanisms work came to light in the July 26, 2013 issue of Science. The research was a product of Susumu Tonegawa's lab at MIT (Tonegawa received the 1987 Nobel Prize in Physiology or Medicine for his discovery of the mechanisms generating antibody diversity), and was authored by Steve Ramirez (a graduate student), Xu Liu (a post-doc), and 5 others.

In this paper, the MIT group claims to have implanted false memories in mice by activating carefully chosen cells in the brain. More specifically, by manipulating selected cells in the hippocampus associated with the memory of fear of one environment, the MIT team claims that they can make mice fear a different environment they've never been in before. The most generous interpretation of their result is that the animal's brains were implanted with the false memory that the new environment was dangerous. That statement may be a little beyond what they show, largely because we don't know what the mice actually "remember." But that's largely irrelevant since, at that level, the psychology is ambiguous. (What, unambiguously, is a memory to an organism of a different species?) Expressing the result in the context of conditioning, which is unambiguous (and, despite the title, how the authors actually couch their result) makes the work fairly compelling.

Here's an outline of their evidence. They manipulated the mice genetically so that memory-associated hippocampal cells, when activated (i.e., when they send electrical signals to other parts of the brain), would express a protein called channelrhodopsin-2 (ChR-2). It turns out that a hippocampal cell which expresses ChR-2 will also send signals when exposed to light. This light-dependent activation can be switched off by spiking the animals' food with a chemical called Dox (short for doxycycline).

Figure 1. Location of the human hippocampus. This is where the structure is in the human brain; the research reported on here was done with a transgenic strain of mouse.

The cells chosen for specific study were in two regions of the hippocampus—the dentate gyrus (DG) and a region called CA1 (Figs. 1 and 2). Before the experiment proper began, the MIT team surgically implanted lights to shine on these regions of the brain. The researchers then began the experiment by exposing the mice to environment A to allow hippocampal cells activated by A to produce the light-sensitive protein, ChR-2. The location of these activated cells were mapped for a reason I'll explain in a moment. The mice were then fed Dox for the remainder of the experiment to turn ChR-2 expression off. So now the only cells that will respond to light are the same cells that were activated in environment A. Next, the mice were exposed repeatedly (presumably) to a different environment, B, and scared with (mild) electrical shocks to their feet. When "scared," the animals would freeze (not move). While the mice were being conditioned to freeze in environment B, the researchers also turned on the lights in the hippocampus to activate the ChR-2-expressing cells that had been activated by environment A. They mapped the cells activated by environment B like they did for A and showed that these environments "turned on" a recognizably different set of memory cells. So, now they have a group of mice that are conditioned to fear environment B but not A. Now it's time to test the effect of light-activating environment A memory cells while fear-conditioning the animals to environment B. Would they fear A even though they were not taught to fear A? Would they fear all environments?

Figure 2. Histological regions of the rat hippocampus. Again, this is the wrong species, but close enough. The regions manipulated were the dentate gyrus (DG) and the area marked CA1.

To find out, the researchers put the conditioned mice back in environment A, where they tended to "act scared" (freeze) even though they were given no shocks. When moved to a novel environment the mice had never seen before (environment C), they seemed much less fearful (rarely froze). Therefore, the memory of their fear of environment B seems to have been transferred to their memory of environment A. The mice were not simply taught to fear all environments. There's more to the paper, mostly focusing on eliminating potential confounding effects (like the result cannot be due to the way the mice were genetically manipulated, and so on) and nuances associated with other parts of the brain, but this is the meat of it.

So, what does this mean in the larger context? First of all, imagine a child who witnessed a leopard kill and eat his or her brother. (Sorry—I know it's kind of graphic, but it needs to be truly scary.) Whatever other fear of leopards we humans may be born with, this poor child has learned to fear leopards. The memory of this event and the fear it elicited was somehow coded by the activity of certain cells in the child's hippocampus. It would be surprising in no way for this child to also have learned to be afraid of jaguars since, to a nonspecialist, leopards and jaguars look similar. Therefore, sighting a jaguar would probably activate the same hippocampal cells, and therefore the memory of the fear, of leopards. This paper has little to say about that. Instead, if these researchers' interpretations are correct and extend from mice to humans (which is not at all unlikely), then in principle they could terrorize this child with a ham sandwich. All they would have to do is give the child a ham sandwich, discover which hippocampal cells organized the memory of the sandwich, and activate those same cells when the child saw its sibling being devoured by the leopard. The memory of the ham sandwich apparently would be conflated with the memory of the fear, resulting in the "false memory" of a horrifying ham sandwich. Whether the child would somehow falsely remember a ham sandwich devouring its sibling is neither answered by this paper nor really relevant (yet).

Now, it might seem flip to nonchalantly associate a ham sandwich with an event that, if it were to happen in reality, would be literally life altering. But I deliberately chose a disturbing image because, as interesting and exciting as this result is (and as many well-deserved accolades and rewards its authors will reap if the result passes scrutiny over the next decade or so), part of me finds it quite disturbing. The ability to deliberately create false memories offers great therapeutic potential but also has already been greatly abused in the past. Although clearly important, we should tread lightly with this technology.