* The person behind this article and its illustrations is Lev Tankelevitch, a current Neuroscience PhD student at Oxford University who is really excited about science communication. Scientific study references that back up his points can be clicked on in the text to open the original reports in a separate tab.
When I was twelve I visited my grandmother in Germany. She lived in Konstanz, a small university city lying along a picturesque bank of Lake Constance, a name that coincidentally alludes to the point of this story. On one of the islands dotting this lake, we happened upon a quaint restaurant offering variations on the traditional German meat-and-potatoes fare. Although I foolishly did not remember the name of the restaurant, the meal left such an indelible impression on me that when I returned to Konstanz six years later, I knew that I had to find this place at all costs. When I stepped onto the island again, I began to search the narrow cobbled streets. To my surprise, I had managed to find the place rather quickly, using a mysteriously intuitive sense of direction which I had somehow retained from my initial visit, years earlier. What did this navigational sense really consist of, and how did I learn it?
In an attempt to answer these questions, one set of researchers based in Paris staged a fascinating, Inception-like version of my own learning experience. Instead of the island on Lake Constance, they used a circular open field, about one metre in diameter. Instead of the hearty German grub, they provided electrical brain stimulation. Likewise, the protagonists in their story were not hungry adolescent boys, but rather a group of mice. The heart of their study, though, lies in the idea that the mice likely had no awareness of their own learning process, for the simple fact that most of it occurred as they soundly slept in their cages.
There is a seahorse-shaped structure lodged deep in the brain of mice and humans alike called the hippocampus. Brain cells, or neurons, in the hippocampus are thought to represent a kind of map of the space around us. Work conducted in the 1970’s by John O’Keefe (which recently earned him the 2014 Nobel Prize in Physiology and has been explored by countless others since), has shown that when a mouse is in a particular location, certain neurons get excited and fire away. That is, each location in the surrounding area corresponds to a set of neurons which represent that location, as a sort of coordinate. Considered together, these “place cells” are thought to make up a map of the environment, presumably allowing the animal to navigate to the mouse equivalent of homely German restaurants.
Although it is accepted that place cells form a map of space, it is less clear whether mice actually use these maps to navigate and how they might do this. This uncertainty arises because previous studies have looked at place cells at the same time as mice are navigating their environments. This is a paradoxically frustrating situation: a mouse’s current location always matches the firing of the associated place cells. This is what’s expected, of course, but it also leads to an alternative and relatively more boring explanation that place cells reflect a simple “you are here” signal, rather than a full map that animals actively use to find their way around. It is the age-old problem haunting science: correlation is not the same as causation. To determine that place cell maps have a causal influence on mouse navigation, it is necessary to first interfere with place cell activity, and then see if navigation changes accordingly. To do this, the Paris team decided to interfere with place cells as the mice slept, a time during which they would certainly not be navigating their environments – except maybe in their dreams.
This last point is critical to the Paris study. Earlier work has shown that place cells spontaneously activate during sleep in a rhythmic pattern. The presumption is that they are “replaying” the spatial map that the animal has learned while awake, and thereby solidifying it for future visits. The team in Paris manipulated this replay process in real time as the mice slept, and then observed whether this would have any effect on where the mice navigated once they awoke. In essence, they wanted to test if this intervention could implant into a mouse’s mind a ‘memory’ of visiting a particular place.
But first, they had to find some place cells to work with. To do this, they recorded from cells in the hippocampus as mice ran around a small circular field, and observed the bustling neuronal activity. If a cell reliably fired when the mouse was in a particular place in the circular field, the team could be confident that this was a place cell which represented that particular location.
Next, they had to find a way to reliably influence the carefully orchestrated replay process that these place cells would engage in when the animals would later fall asleep. Since it is unclear exactly how place cells work together during the replay process, directly tinkering with them may be too messy of an affair, and could lead to uninterpretable results. Instead, the Paris team made use of the brain’s natural teacher and reinforcer – the dopamine system – to bias the place cell map toward an arbitrary but real location (that is, to “convince” these neurons that one place was special). It is as if you became certain, overnight, that precisely down the street there exists a fantastical German restaurant serving the best food you’ve never tried.
To achieve this, the mice were implanted with stimulating electrodes in a bundle of nerve fibres in the brain creatively named the medial forebrain bundle, or MFB for short. Electrical stimulation of the MFB causes the release of dopamine. Often simplistically and misleadingly called the brain’s “pleasure” chemical, what dopamine really does, at least in part, can be more accurately defined as reinforcing behaviour. Events or actions that lead to rewarding outcomes (like eating food!) cause the release of dopamine, which acts to reinforce those events or actions in the brain and thereby ensures that they are marked as important for the future. Here’s a classic demonstration from the 1950s: if given a chance to press a button which triggers MFB stimulation, and therefore dopamine release, animals learn to press this button without end. What this illustrates is that a rewarding outcome like a delicious German meal is not actually necessary to learn anything, as long as you have the associated release of dopamine.
In the current study, mice didn’t have to press a button to trigger stimulation of the MFB. Instead, stimulation was dependent on the spontaneous activation of place cells that the researchers had selected earlier when the mice were exploring their circular field. That is, as the mice slept, every time a specific place cell fired, stimulation of the MFB would occur immediately after. It’s as if flipping through a photo album on repeat, you would highlight one particular photo over and over again, marking it as especially memorable.
Now, if a place cell represents a particular location in a mouse’s environment, and its replay of this particular spot during sleep can be reinforced through stimulation of the MFB, then perhaps it is possible that mice would come to prefer this particular location once they wake up and return to the field? That is exactly what the Paris team found. Before the stimulation trick, the mice explored each location of the field with equal curiosity or boredom. They showed no preference for any given location, as they had no good reason to do so. But after pairing the place cell activity during sleep with MFB stimulation, the mice immediately darted for the location which was represented by the chosen place cell, and spent more time there than anywhere else in the field. Something in each mouse’s brain, presumably some communication between its place cells and dopamine system, was telling it that this location was especially important. The team in Paris had not only demonstrated that mice indeed rely on their place cell maps to navigate, but had also provided further evidence that replay during sleep is important for reviewing and cementing these maps for future use.
Of course, after a few trips to that memorable location, becoming aware of the lack of good German restaurants in the neighbourhood and no longer receiving dopamine reinforcement caused the mice to quickly abandon their preference. This only goes to further demonstrate the continuously evolving nature of place cell maps. And this is what makes my own memory of that restaurant’s location so fascinating: I had learned it after one visit, and retained it for years later. What enables such persistence? Perhaps it is the fact that the continued and detailed existence of our memories so often relies on external memory storage in the form of photos and stories that we share with others.