Navigation neurons put monkeys on the right track

The term ‘cognitive map’ was first used by Edward Tolman in a classic 1948 paper, in which he described the now famous experiments he performed on rats:

In the typical experiment a hungry rat is put at the entrance of the maze…and wanders about…until he finally comes to the food box and eats. This is repeated…every 24 hours, and the animal tends to make fewer and fewer errors…and to take less and less time between start and goal-box until finally he is… running in a very few seconds from start to goal…[I] believe that in the course of learning something like a field map of the environment gets established in the rat’s brain…[and that] incoming impulses are usually worked over and elaborated…into a tentative, cognitive-like map of the environment…indicating routes and paths and environmental relationships.

Tolman’s theory of the cognitive map was highly influential, but he, and his contemporaries, had few research methods at their disposal. Today, researchers have a wide variety of techniques available to them; these techniques enable them to observe the human brain in action while it performs a cognitive task, or to probe the cellular and molecular bases of cognition. Using these methods, researchers have confirmed that organisms, including humans, do indeed generate cognitive maps of their environment and use them for spatial navigation. It is now well established that circuits in the hippocampus and parietal cortex are essential for encoding spatial information, but the cellular mechanisms by which they do so are still largely a mystery.

A new study, published in advance on the Proceedings for the National Academy of Sciences website, adds more to the little understanding we have of how individual neurons contribute to the production of a cognitive map.

Using single-unit electrophysiological recordings, Sato et al provide strong evidence for the existence of navigation neurons and movement-selective neurons in the medial parietal region of the Japanese macaque monkey (Macaca fuscata). Two monkeys were taught to use a joystick to navigate through a virtual environment consisting of a two-storey building with 8 rooms on the first floor and 7 on the second. As the monkeys navigated through the building, the researchers used microelectrodes to record the electrical activity of single cells in the medial parietal region (MPR, left).

Of the 580 cells from which recordings were made, 180 (or 31%) fired in response to the monkeys’ movements through the virtual environment. Of these, 77% were selective for specific movements in specific locations; these the researchers have termed ‘navigation neurons’; 14 of the cells (8%), named movement-selective neurons, fired in response to a specific movement such as a left or a right turn, irrespective of the location in which the movement was made. While these cells fired in response to unique combinations of movement and/or location, other cells (27 in total, or 15%) were non-selective. For example, one navigation neuron responded when the monkey made a left turn at a specific checkpoint in the building, but not when it made another movement at the same checkpoint or when it made a left turn at any other location.

The researchers then injected the GABA_A receptor agonist muscimol into the monkeys’ MPR. GABA (gamma aminobutyric acid) is the principal inhibitory neurotransmitter in the central nervous system, so activating GABA_Areceptors with compounds such as muscimol therefore inhibits neuronal activity. With the MPR ‘switched off’ upon injection of muscimol, the monkeys’ navigation was impaired, and they lost their way when attempting to reach their destination in the virtual building. This is consistent with studies of patients with damage to the medial parietal region, who can recognize landmarks in their surroundings but cannot determine which route they should take to reach their desired destination.

This film clip shows the performance of one of the monkeys on the navigation task; the monkey’s corresponding hand movements can be seen in the inset at the bottom right of the screen:

Investigations of spatial cognition have largely focused on the hippocampus, whose role in such processes has been extensively studied. The CA1 and CA3 subfields of the hippocampus are known to contain place cells, which are also involved in navigation. Place cells fire when an organism is in a specific location of its environment; the location to which an individual place cell responds is known as its ‘place field’.

For obvious reasons, researchers rarely get the opportunity to conduct such experiments in humans, and the vast majority of electrophysiological studies have been performed on animals. However, in 2003, Ekstrom et al, performed a study that was almost identical to that of the Japaneses team, on epileptic patients who were resistant to pharmacological treatment. Prior to surgery, the patients’ brains were invasively monitored with intracranial electrodes so that the surgeons could determine the loci of the seizures. During the monitoring, the patients played a computer game in which they navigated through a virtual town, and the responses of single neurons in the medial temporal and frontal lobes were recorded.

It was found that 11% of the cells examined were responsive to specific locations in the virtual town around which the patients moved. These cells were more prevalent in the hippocampus than in the other brain regions examined – 24% of all hippocampal cells whose activity was recorded turned out to be place-selective. The place fields of these cells were in locations of the virtual town that were traversed frequently, but fired independently of the direction in which movement was taking place. A small number of cells – the so-called ‘view cells’ -were location-independent but fired in response to specific objects, such as shop fronts.

Clearly, the cellular processes underlying the generation of cognitive maps are extremely complex, and there remains much to discover about how individual neurons encode spatial information. The distinction between place cells and similar cells remains unclear, and other cells encoding different properties of the environment will almost certainly be discovered in the future. Some of the cells studied by the Japanese team exhibit similar properties to hippocampal place cells, but further investigation is required to determine how the functions of navigation cells in the MPR differ from those of hippocampal place cells, and how these populations of cells interact. The study does, however, strongly support the notion that there is an abstract representation, at the cellular level, of what is called ‘route knowledge’ (the list of movements that have to be carried out in order to reach a specified location).