ABSTRACT:It is often assumed that navigation implies the use, by animals, of landmarks indicating the location of the goal. However, many animals (including humans) are able to return to the starting point of a journey, or to other goal sites, by relying on self-motion cues only. This process is known as path integration, and it allows an agent to calculate a route without making use of landmarks. We review the current literature on path integration and its interaction with external, location-based cues. Special importance is given to the correlation between observable behavior and the activity pattern of particular neural cell populations that implement the internal representation of space. In mammals, the latter may well be the first high-level cognitive representation to be understood at the neural level.

During short foraging excursions away from their home, central place foragers update their position relative to their point of departure by processing signals generated by locomotion. They therefore can home along a self-generated vector without using learned references. In rodents and other mammals, this path integration process (dead reckoning) can occur on the basis of purely internal signals, such as vestibular or proprioceptive (re)afferences. We report here that hamsters are also capable of proceeding to a previously learned feeding site through vector information from locomotion only. The subjects compute the direction and distance to the goal by subtracting their current-position vector from the stored nest-to-goal vector. This computation pertains to locations per se and therefore occurs in absolute space, independently of landmark objects. If available, prominent visual cues merely serve to confirm the path planned through the addition of self-generated vectors, whereas visual as well as nonvisual references confirm that the subject has arrived at the goal site.

No abstract

During short excursions away from home, some mammals are known to update their position with respect to their point of departure through path integration (dead reckoning) by processing internal (idiothetic) signals generated by rotations and translations. Path integration (PI) is a continuously ongoing process in which errors accumulate. To remain functional over longer excursions, PI needs to be reset through position information from stable external references. We tested the homing behaviour of golden hamsters (Mesocricetus auratus W.) during hoarding excursions following a rotation of the arena and nest. In continuous darkness, the hamsters returned to their point of departure at the rotated nest, and therefore depended on PI only. In other trials, the animals were briefly presented with visual room cues during or at the end of the outward trip, visual cues being pitted by 67°or 98°against the animal's current selfgenerated position vector. After a fix, the animals headed for the usual (unrotated) nest location, as defined by room cues, independent of the timing of the fix. These results were obtained in two different geometrical settings and showed that, after the fix, the animals update their position, and not merely their head direction or internal compass, in a new reference frame. Thus, episodic fixes on familiar external references reset the PI and therefore greatly enhance the functional signification of navigation that is based on feedback information from locomotion.

No abstract

When hoarding food in an experimental arena (ø = 2.20 m), golden hamsters tend to return along a direct path from the food source at the centre of the arena to their peripherally located nest. Under infra-red light the animals' homing behaviour is exclusively controlled by 'internal' cues which have been generated during the outward journey to the feeding place. The paper examines the limitations on the registration and computation of directional cues which originate from an active or a passive outgoing trip. The compensation of the angular component of the outward journey was examined by inducing the subjects to walk (actively) around the centre of the arena, or by rotating the animals (passively) on a platform which contained the food source. The greater the number of rotations, the less precise the homing performance. After three to five active rotations, and after two to three passive rotations of 360°, the animals ceased to yield significant homing vectors. Unidirectional rotations induced a systematic ipsidirectional bias, an indication of undercornpensation; no bias was observed after rotations which occurred in equal extents in both directions. Special importance was given to the compensation for passive translations. The experiments involved either the shift of the subjects from their nest exit to an unfamiliar, adjacent arena, or the combination of an active and a passive outward leg during the outward journey to the centre of the animals' own arena. During their return to the nest, either the hamsters did not take into account at all the direction of the passive translation, or if they did, the compensation was only limited. These results are discussed in relation to the animals' capacity to assess 1) an active outward journey through the availability of different categories of self generated cues, and 2) a passive outward journey predominantly on the basis of vestibular signals. A final series of experiments showed that under certain conditions, the hamsters commit systematic homing errors which can be attributed to their failure to initiate the registration and computation of path dependent cues at the right instant in time and space

When hoarding food under infra-red light, golden hamsters Mesocricetus auratus W. return fairly directly from a feeding place to their nest site by evaluating and updating internal signals that they have generated during the previous outward journey to the feeding place. To test more specifically the animals' capacity to evaluate the linear components of the outward journey, the subjects were led from their (cone-shaped) nest to a feeding place along a detour which comprised either 2 (experiment 1) or 5 (experiment 2) segments; adjoining segments were at right angles to each other. In these conditions, the subjects remained significantly oriented towards the nest and therefore were capable of assessing translations as well as rotations during the outward journey. In experiment 3, the nest was removed after the hamsters had started the direct outward journey to the feeding place and the hamsters were rotated during the food uptake. The animals were no longer oriented towards the starting point of their journey, but nonetheless covered, along a fairly straight path, the correct homing distance, and then changed over to a circular search path. These results confirm that mammals can derive the linear components of an outward journey from self-generated signals and therefore are able to judge the homing distance without relying on cues from the environment. For a number of detour outward journeys, our data yield an unexpectedly good fit to Müller and Wehner's (1988) model of dead reckoning in ants. However, this is no longer the case when the outward journey contains an initial loop which brings the subject back to the starting point. These findings are discussed in terms of the biological significance and limitations of an approximate form of path integration.

During locomotion, mammals update their position with respect to a fixed point of reference, such as their point of departure, by processing inertial cues, proprioceptive feedback and stored motor commands generated during locomotion. This so-called path integration system (dead reckoning) allows the animal to return to its home, or to a familiar feeding place, even when external cues are absent or novel. However, without the use of external cues, the path integration process leads to rapid accumulation of errors involving both the direction and distance of the goal. Therefore, even nocturnal species such as hamsters and mice rely more on previously learned visual references than on the path integration system when the two types of information are in conflict. Recent studies investigate the extent to which path integration and familiar visual cues cooperate to optimize the navigational performance.