Unexpected GPS Relativity

How new data changes our understanding of how neurons work

The first details of the GPS “built into the brain” began to appear in the 1970s. In the laboratories of University College London, John O'Keefe and his student Jonathan Dostrovsky recorded the electrical activity of neurons in the hippocampus of freely moving rats. They found a group of neurons activated only when the rat was in a certain place [ 1 ]. They called these cells " place neurons ."

Based on these early discoveries, O'Keefe and his colleague Lynn Nadel suggested that the hippocampus contains an invariant representation of space, independent of mood or desires. They called it the " cognitive map " [ 2 ]. From their point of view, all neurons of a place in the brain represent the whole environment of an animal as a whole, and the activation of a certain cell indicates its current location. In other words, the hippocampus works like a GPS. It tells you where you are on the map, and this map does not change, no matter whether you are hungry and are looking for food, or if you want to sleep and are looking for a bed. O'Keefe and Nadel suggested that the absolute location, the idea of ​​which is stored in the neurons of the place, provides a mental platform that the animal can use to orient in any situation - to search for food or places to rest.



In the 40 years that followed, other researchers — including the married duet of Edward and May-Britt Moser — supported the idea that the contours of the hippocampus function as an embedded GPS [ 3 ]. For their pioneering work, O'Keefe and Moserz were awarded the 2014 Nobel Prize in Physiology or Medicine. It could be decided that the role of the hippocampus in the orientation of animals in space was unraveled.

But the study of the brain never goes so straightforward. The 2014 Nobel Prize, like a match that ignites the fuse, caused an explosion of experiments and ideas, some of which began to object to the early interpretation of O'Keefe and Nadel. The new work suggested that in the case of spatial navigation, the hippocampus contour represents not absolute location information, but relative and modifiable under the influence of experience. The study of the hippocampus, apparently, stumbled upon an old philosophical argument.

For centuries physicists have been struggling with the question of whether it is absolutely or relatively space, before leaning on the side of relativity. But it was only in recent years that they began to ask similar questions when studying the brain. For many years, in neuroscience has been in charge of absolute space. For example, it has long been assumed that there are two channels in the visual system for the flow of information. [ 4 ] The first is the “what” channel, which transmits information about the identity of an object observed by an animal. The second is the “where” channel containing information about the absolute location of the object. It was believed that the channel "that" does not contain any positional information. However, a recent paper demonstrated that, although the channel does not contain information about the absolute location of the object, it does contain information about the relative location. [ 5, 6 ] This information on the relative location is likely to be very important for object recognition.

Such discoveries serve as a foothold for the idea that relative information is also important to the brain. This point of view is reinforced by the recently begun synthesis of neuroscience with computer science and AI. Work at the junction of these disciplines has shown that a brain that uses an absolute, unchanging model of the world needs more computing resources to live in a constantly changing environment than a brain that uses relative information. Understanding where and when the brain uses absolute and where relative information can shed light on the work, flexibility and speed of its subsystems and our behavior. In particular, the hippocampus may be one of the first milestones in this investigation.

A key objection to the interpretation of the absolute representation of the location from O'Keefe and Nadela appeared in the work of last year, sponsored by Kimberly Stachenfeld, Matthew Botvinnik and Samuel Gershman. [ 7 ] These researchers associated with Google DeepMind, Princeton University, University College London and Harvard University, suggested that the hippocampus represents not the absolute location of the animal, but where the animal is most likely to go in the near future. This presentation takes into account preferred movements and learned habits. From this point of view, the hippocampus is a predictive, not an absolute map.

Previous studies have shown that the activity of neurons in a place is constantly decreasing when an animal moves away from the center of some place of interest. O'Keefe and Nadel decided that this is a sign that the neurons of the place represent the current location of the animal. But within the framework of the platform proposed by Stachenfeld and her colleagues, it is proposed to consider the degree of neuron activity as a presentation of how likely the animal will be at the center of the place of interest in the next moment. If it is already in the center, then the probability that it will be there at the next moment is rather high, therefore the cell activity is also high. If it has gone so far from the center that it cannot return there at the next moment, then the neurons of the place are inactive.

Theories of O'Kif and Stachenfeld may seem similar, and they both seem to explain the basic properties of the activity of the neuron site. However, they make different assumptions about the nature of the spatial map in the hippocampus, and only ingenious experiments and tests on computational models will help to separate one from another. Stachenfeld, Botvinnik and Gershman achieved this by re-analyzing data from previously published works, and finding that some of them can explain their model of the hippocampus, but not traditional models. The most striking of these examples are the data from a study conducted by Ellis Alwernet and her colleagues from Marseille, France. [ 8 ] These researchers used the “maze with Tolman branches” in which the rat needs to run along a single path from beginning to end. In some situations, the path was closed, which caused the animal to bypass an obstacle along one of two C-shaped corridors.

According to the interpretation of the activity of neurons by O'Keefe with his cognitive map, the neuron that was active when the rat was at a fork between the direct path and the bypass path should be activated the same way, regardless of whether this path is blocked or not. But in the experiment there was a different picture. This cell behaved differently, depending on the presence of blocking paths. The degree of neuron activity was influenced by previous experience of the rat. The absolute map should not work that way. Moreover, Stachenfeld and his colleagues did computer simulations to show that the activity of the neurons of the place that Alverner and his colleagues observed in their experience coincides with their hypothesis of the predictive map much better than the hypothesis of O'Kif on the cognitive map.

Stachenfeld’s argument against O'Kif-Nadel’s interpretation was that the neurons of a place do not encode an absolute position, but only an arrangement relative to the history of movements, experience, and behavioral preferences. Just a few months later, another set of studies showed that the location of other animals of the same species also affects the activity of the neurons of the site. [ 9, 10, 11 ] In the papers published this year, Necham Ulanovsky from the Weismann Institute in Israel and Shigeyoshi Fujisawa from the RIKEN Brain Research Institute in Japan trained animals to move in a given area by showing them races that other individuals of their species performed. . At the same time, Ulanovsky used bats, and Fujisawa used rats. When animals followed the prescribed path, their neurons in the place expectedly intensified. And the surprise was that the subset of these neurons in the site also became more active when the animals watched the races of other individuals. The researchers called these neurons "social neurons of the place."

The results again diverge from the initial interpretation of the activation of the neurons of the site, which connected them with the absolute location in space. The representation of a place in the hippocampus is not just different from the absolute - it seems that it can be influenced by observing others.

The activity of the neuronal site proved to be much more complex than previously thought. The classic look at the role of the hippocampus contours in spatial navigation, awarded the Nobel Prize, was not a complete description of what is happening, and the hippocampus performs much more functions than the simple invariant representation of the subject's location in space.

The idea of ​​the predictability of neurons in space and the influence on them of learning and the behavior of other animals can facilitate the construction of a concept capable of describing both the role of the hippocampus in spatial orientation and the generally accepted role in learning and memory formation. Since the discovery that the removal of the hippocampus can lead to the impossibility of the formation of new memories, it has been studied as one of the most important brain regions responsible for memory. [ 12 ] And although from the very first experiments of O'Kif and Dostrovsky it was known that the hippocampus plays a major role in spatial navigation, how and why this tiny piece of the brain is capable of storing both spatial maps and complex memories in itself remains poorly understood . The emerging understanding of the relativity of our spatial maps and the effect on them of memory and behavior makes it easier to understand the dual function of the hippocampus. Fifty years after the first observations of O'Kif and Dostrovsky, we begin to understand more clearly how this key area of ​​the brain forms our personalities.