Absence of an Interaction Between Navigational Strategies Based on Local and Distal Landmarks

Understanding the role of distance, direction and cue salience in an associative model of landmark learning

Navigation and spatial memory relies on the ability to use and recall environmental landmarks relative to important locations. Such learning is thought to result from the strengthening of associations between the goal location and environmental cues. Factors that contribute to the strength of this association include cue stability, saliency and cue location. Here we combine an autoregressive random walk model, that describes goal-directed swimming behaviour, with an associative learning model to provide an integrated model of landmark learning, using the water maze task. The model allows for the contribution of each cue, the salience and the vector information provided (both distance and directional) to be separately analysed. The model suggests that direction and distance information are independent components and can influence searching patterns. Importantly, the model can also be used to simulate various experimental scenarios to understand what has been learnt in relation to the cues, thereby offering new insights into how animals navigate.

Interactions of spatial strategies producing generalization gradient and blocking: A computational approach

We present a computational model of spatial navigation comprising different learning mechanisms in mammals, i.e., associative, cognitive mapping and parallel systems. This model is able to reproduce a large number of experimental results in different variants of the Morris water maze task, including standard associative phenomena (spatial generalization gradient and blocking), as well as navigation based on cognitive mapping. Furthermore, we show that competitive and cooperative patterns between different navigation strategies in the model allow to explain previous apparently contradictory results supporting either associative or cognitive mechanisms for spatial learning. The key computational mechanism to reconcile experimental results showing different influences of distal and proximal cues on the behavior, different learning times, and different abilities of individuals to alternatively perform spatial and response strategies, relies in the dynamic coordination of navigation strategies, whose performance is evaluated online with a common currency through a modular approach. We provide a set of concrete experimental predictions to further test the computational model. Overall, this computational work sheds new light on inter-individual differences in navigation learning, and provides a formal and mechanistic approach to test various theories of spatial cognition in mammals.

Transfer of Spatial Behaviour Controlled by a Landmark Array with a Distinctive Shape

In two experiments, rats swam to a submerged platform in one corner of a rectangular or kite-shaped array created by four identical landmarks attached to the walls of a circular pool. After training in the rectangular array, rats expressed a preference for the corner in the kite-shaped array that was geometrically equivalent to where the platform was located previously. After training in either array, the removal of two landmarks from the rectangular array, or the landmark at the apex of the kite-shaped array, did not affect the control over searching exerted by the remaining landmarks. The results imply that rats use local rather than global spatial representations when searching for a hidden goal with reference to an array of landmarks.

Effects of landmark distance and stability on accuracy of reward relocation

Although small-scale navigation is well studied in a wide range of species, much of what is known about landmark use by vertebrates is based on laboratory experiments. To investigate how vertebrates in the wild use landmarks, we trained wild male rufous hummingbirds to feed from a flower that was placed in a constant spatial relationship with two artificial landmarks. In the first experiment, the landmarks and flower were 0.25, 0.5 or 1 m apart and we always moved them 3-4 m after each visit by the bird. In the second experiment, the landmarks and flower were always 0.25 m apart and we moved them either 1 or 0.25 m between trials. In tests, in which we removed the flower, the hummingbirds stopped closer to the predicted flower location when the landmarks had been closer to the flower during training. However, while the distance that the birds stopped from the landmarks and predicted flower location was unaffected by the distance that the landmarks moved between trials, the birds directed their search nearer to the predicted direction of the flower, relative to the landmarks, when the landmarks and flower were more stable in the environment. In the field, then, landmarks alone were sufficient for the birds to determine the distance of a reward but not its direction.

Simple gaze analysis and special design of a virtual Morris water maze provides a new method for differentiating egocentric and allocentric navigational strategy choice

We present a novel method of combining eye tracking with specially designed virtual environments to provide objective evidence of navigational strategy selection. A simple, inexpensive video camera with an easily built infrared LED array is used to capture eye movements at 60Hz. Simple algorithms analyze gaze position at the start of each virtual maze trial to identify stimuli used for navigational orientation. To validate the methodology, human participants were tested in two virtual environments which differed with respect to features usable for navigation and which forced participants to use one or another of two well-known navigational strategies. Because the environmental features for the two kinds of navigation were clustered in different regions of the environment (and the video display), a simple analysis of gaze-position during the first (i.e., orienting) second of each trial revealed which features were being attended to, and therefore, which navigational strategy was about to be employed on the upcoming trial.

Features enhance the encoding of geometry

Successful navigation within an environment requires that the traveler establish the correct heading--a process referred to as orienting. Many studies have now shown that humans and non-human animals can use the geometric properties of an enclosure to orient. In the present study, two groups of Clark's nutcrackers (Nucifraga columbiana) were trained, in a reference memory task, to find food hidden in one of four containers arranged to form a rectangular array. One group had unique objects placed next to each of the containers, whereas the second group had identical objects placed next to each of the containers. Here, I show for the first time that for the Clark's nutcracker, the distinctive properties of these objects enhanced the encoding of the array's geometry compared to the learning of geometric properties from an array of identical objects, which remained at chance after substantial amounts of training. Subsequent transformation tests showed that an object not associated with reward, but sharing the same geometric properties as the correct object, may have had inhibitory qualities. Furthermore, by systematically removing objects from the array, I show that although nutcrackers encoded the geometry of the array, they did not encode a complete featural representation of the objects within the array.

Multiple Systems for Spatial Learning: Dead Reckoning and Beacon Homing in Rats

Rats homed with food in a large lighted arena. Without visual cues, they used dead reckoning. When a beacon indicated the home, rats could also use the beacon. Homing did not differ in 2 groups of rats, 1 provided with the beacon and 1 without it; tests without the beacon gave no evidence that beacon learning overshadowed dead reckoning (Experiment 1). When the beacon was at the home for 1 group and in random locations for another, there was again no evidence of cue competition (Experiment 2). Dead reckoning experience did not block acquisition of beacon homing (Experiment 3). Beacon learning and dead reckoning do not compete for predictive value but acquire information in parallel and are used hierarchically.