Animal Navigation, Neural Networks

Animals show a remarkable ability to navigate through their environment. For example, many animals must cover large regions of the local terrain in search of a goal (food, mates, etc.), and then must be able to return immediately and safely to their nesting spot. From one occasion to the next, animals seem to use varied, novel trajectories to make these searches, and may enter entirely new territory during part of the search. Nonetheless, on obtaining the goal, they are typically able to calculate a direct route to return to their home base.

For many species, this ANIMAL NAVIGATION is thought to be based on two general abilities. The first, called "dead reckoning" (or "path integration"), uses information about the animal's own movements through space to keep track of current position and directional heading, in relation to an abstract representation of the overall environment. The second, landmark-based orientation, uses familiar environmental landmarks to establish current position, relative to familiar terrain.

Over the last few decades, some insight has been gained about how NEURAL NETWORKS in the mammalian brain might work to provide the basis for these abilities. In particular, two specialized types of cells have been observed to possess relevant spatial signals in the brains of navigating rats.

"Place cells," originally discovered in the rat HIPPOCAMPUS (O'Keefe and Dostrovsky 1971), fire whenever the animal is in one specific part of the environment. Each place cell has its own, unique region of firing. As the animal travels through the environment, these cells seem to form a maplike representation, with each location represented by neural activity in a specific set of place cells.

Remarkably, these cells also seem to use the two general abilities mentioned above: dead reckoning and landmark-based orientation. Evidence that place cells use landmarks to establish the place-specific firing patterns came from early studies in which familiar landmarks were moved (e.g., O'Keefe and Conway 1978; Muller and Kubie 1987). For example, in one experiment (Muller and Kubie 1987), rats foraged in a large, cylindrical apparatus, equipped with a single, white cue card on its otherwise uniformly gray wall. When this single orienting landmark was moved to a different location on the wall, this caused an equal rotation of the place cell firing fields. Evidence that the cells can also use dead reckoning, however, was obtained from studies in which the landmarks were removed entirely (e.g., Muller and Kubie 1987; O'Keefe and Speakman 1987). In this case, the place cells were often able to maintain their firing patterns, so that they continued to fire in the same location, as the animal made repeated, winding trajectories through the environment. It was reasoned that this ability must be based on a dead reckoning process because in the absence of orienting landmarks, the only ongoing information about current position would have to be based on the animal's own movements through space. Further support for the dead reckoning (path integration) process came from a study in which artificial movement-related information was given directly to the animal while it navigated (Sharp et al. 1995). Animals foraged in a cylinder with black and white stripes, of uniform width. Both activation of the vestibular system (indicating that the animal had moved through space), or rotation of the vertical stripes (as would happen due to the animal's own movement in relation to the stripes) could sometimes "update" the place cell firing fields, so that they shifted the location of their fields, as though the animal had actually moved in the way suggested by the vestibular or optic flow input. Thus movement-related inputs directly influence the positional setting of the hippocampal place cells.

The second type of navigation-related cells, known as "head direction cells" (Taube, Muller, and Ranck 1990a), complement the place cells by signaling the animal's current directional heading, regardless of its location. Each head direction cell fires whenever the animal is facing one particular direction (over an approximately 90 degree range). Each has its own, unique, directional preference, so that each direction the animal faces is represented by activity in a particular subset of head direction cells. These cells were initially discovered in the postsubiculum (a brain region closely related to the hippocampus), and have since been discovered in several other anatomically related brain regions.

Although it might be thought that these directional cells derive a constant, earth-based orientation from geomagnetic cues, this seems not to be the case. Rather, like place cells (and the animal's own navigational behavior), head direction cells seem to use both landmark orientation and dead reckoning (Taube, Muller, and Ranck 1990b). For example, in a familiar environment, these cells will rotate the preferred direction when familiar landmarks are rotated, using the landmarks to get a "fix" on the animal's current directional heading. When, however, all familiar landmarks are removed, the cells retain the animal's previously established directional preference. And, again like place cells, head direction cells use both vestibular and optic flow information to update their locational setting (Blair and Sharp 1996).

Theoretical models have been developed to simulate the spatial firing properties of these cells (e.g., Blair 1996; McNaughton et al. 1995; Skaggs et al. 1995; Redish, Elga, and Touretzky 1997; Samsonovich and McNaughton 1997). Most of these models begin with the idea that the place and head direction cells are respectively linked together to form stable attractor networks that can stabilize into a unitary representation any one of the possible places or directions. For example, in the head direction cell system, cells representing similar directional headings are linked together through predominantly excitatory connections, while cells representing different directional headings are linked through inhibitory connections. This reflects the basic phenomenon that, at any one time, cells within one particular portion of the directional range (e.g., 0 to 90 degrees) will be active, while all other head direction cells will be silent. Thus the stable attractor network, left on its own, will always settle into a representation of one particular direction (place). To reflect the finding that place and head direction cells can be "set" by environmental landmarks, most models equip the cells with sensory inputs that can influence which particular place or direction is represented. To reflect the finding that the navigation system can also be updated by movement related information, most models also incorporate an additional layer of cells (in some other, as yet unidentified brain region) that combines place or head direction information, along with movement-related cues, to feed back onto the place or head direction cells, permitting them to choose a new locational or directional setting in response to movement.

While these complementary place and directional representations are thought to guide the animal's overall navigational behavior (see O'Keefe and Nadel 1978), the mechanism is not yet clear.

Additional links

chapter3.doc

-- Patricia E. Sharp

References

Blair, H. T. (1996). A thalamocortical circuit for computing directional heading in the rat. In D. S. Touretzky, M. C. Mozer, and M. E. Hasselmo, Eds., Advances in Neural Information Processing, vol. 8, Cambridge, MA: MIT Press.

Blair, H. T., and P. E. Sharp. (1996). Visual and vestibular influences on head direction cells in the anterior thalamus of the rat. Behavioral Neuroscience 110:1-18.

McNaughton, B. L., C. A. Barnes, J. L. Gerrard, K. Gothard, M. W. Jung, J. J. Knierim, H. Kudrimoti, Y. Quin, W. E. Skaggs, M. Suster, and K. L. Weaver. (1995). Deciphering the hippo-campal polyglot: The hippocampus as a path integration system. Journal of Experimental Biology 199:173-185.

Muller, R. U., and J. L. Kubie. (1987). The effects of changes in the environment on the spatial firing of hippocampal complex-spike cells. Journal of Neuroscience 7:1951-1968.

O'Keefe, J., and D. H. Conway. (1978). Hippocampal place units in the freely moving rat: Why they fire where they fire. Experimental Brain Research 31:573-590.

O'Keefe, J., and J. Dostrovsky. (1971). The hippocampus as a spatial map: Preliminary evidence from unit activity in the freely moving rat. Brain Research 34:171-175.

O'Keefe, J., and L. Nadel. (1978). The Hippocampus as a Cognitive Map. New York: Oxford.

O'Keefe, J., and A. Speakman. (1987). Single unit activity in the rat hippocampus during a spatial task. Experimental Brain Research 68:1-27.

Redish, A. D., A. N. Elga, and D. S. Touretzky. (1997). A coupled attractor model of the rodent head direction system. Network 7:671-686.

Samsonovich, A., and B. L. McNaughton. (1997). Path integration and cognitive mapping in a continuous attractor neural network model. Journal of Neuroscience 17:5900-5920.

Sharp, P. E., H. T. Blair, D. Etkin, and D. B. Tzanetos. (1995). Influences of vestibular and visual motion information on the spatial firing patterns of hippocampal place cells. Journal of Neuroscience 15:173-189.

Skaggs, W. E., J. J. Knierim, H. S. Kudrimoti, and B. L. McNaughton. (1995). A model of the neural basis of the rat's sense of direction. In G. Tesauro, D. S. Touretzky, and T. K. Lean, Eds., Advances in Neural Information Processing Systems, vol. 7, Cambridge, MA: MIT Press.

Taube, J. S., R. U. Muller, and J. B. Ranck, Jr. (1990a). Head direction cells recorded from the postsubiculum in freely moving rats: 1. Description and quantitative analysis. Journal of Neuroscience 10:420-435.

Taube, J. S., R. U. Muller, and J. B. Ranck, Jr. (1990b). Head direction cells recorded from the postsubiculum in freely moving rats: 2. Effects of environmental manipulations. Journal of Neuroscience 10:436-447.

Animal Navigation, Neural Networks

See also

Additional links

References

Further Readings