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

Path integration, rather than being suppressed, is used to update spatial views in familiar environments with constantly available landmarks

This project tested three hypotheses conceptualizing the interaction between path integration based on self-motion and piloting based on landmarks in a familiar environment with persistent landmarks. The first hypothesis posits that path integration functions automatically, as in environments lacking persistent landmarks (environment-independent hypothesis). The second hypothesis suggests that persistent landmarks suppress path integration (suppression hypothesis). The third hypothesis proposes that path integration updates the spatial views of the environment (updating-spatial-views hypothesis). Participants learned a specific object's location. Subsequently, they undertook an outbound path originating from the object and then indicated the object's location (homing). In Experiments 1&1b, there were landmarks throughout the first 9 trials. On some later trials, the landmarks were presented during the outbound path but unexpectedly removed during homing (catch trials). On the last trials, there were no landmarks throughout (baseline trials). Experiments 2-3 were similar but added two identical objects (the original one and a rotated distractor) during homing on the catch and baseline trials. Experiment 4 replaced two identical objects with two groups of landmarks. The results showed that in Experiments 1&1b, homing angular error on the first catch trial was significantly larger than the matched baseline trial, undermining the environment-independent hypothesis. Conversely, in Experiment 2-4, the proportion of participants who recognized the original object or landmarks was similar between the first catch and the matched baseline trial, favoring the updating-spatial-views hypothesis over the suppression hypothesis. Therefore, while mismatches between updated spatial views and actual views of unexpected removal of landmarks impair homing performance, the updated spatial views help eliminate ambiguous targets or landmarks within the familiar environment.

A map of spatial navigation for neuroscience

Spatial navigation has received much attention from neuroscientists, leading to the identification of key brain areas and the discovery of numerous spatially selective cells. Despite this progress, our understanding of how the pieces fit together to drive behavior is generally lacking. We argue that this is partly caused by insufficient communication between behavioral and neuroscientific researchers. This has led the latter to under-appreciate the relevance and complexity of spatial behavior, and to focus too narrowly on characterizing neural representations of space-disconnected from the computations these representations are meant to enable. We therefore propose a taxonomy of navigation processes in mammals that can serve as a common framework for structuring and facilitating interdisciplinary research in the field. Using the taxonomy as a guide, we review behavioral and neural studies of spatial navigation. In doing so, we validate the taxonomy and showcase its usefulness in identifying potential issues with common experimental approaches, designing experiments that adequately target particular behaviors, correctly interpreting neural activity, and pointing to new avenues of research.

Bayesian causal inference: A unifying neuroscience theory

Understanding of the brain and the principles governing neural processing requires theories that are parsimonious, can account for a diverse set of phenomena, and can make testable predictions. Here, we review the theory of Bayesian causal inference, which has been tested, refined, and extended in a variety of tasks in humans and other primates by several research groups. Bayesian causal inference is normative and has explained human behavior in a vast number of tasks including unisensory and multisensory perceptual tasks, sensorimotor, and motor tasks, and has accounted for counter-intuitive findings. The theory has made novel predictions that have been tested and confirmed empirically, and recent studies have started to map its algorithms and neural implementation in the human brain. The parsimony, the diversity of the phenomena that the theory has explained, and its illuminating brain function at all three of Marr's levels of analysis make Bayesian causal inference a strong neuroscience theory. This also highlights the importance of collaborative and multi-disciplinary research for the development of new theories in neuroscience.

Infrasound as a Cue for Seabird Navigation

Seabirds are amongst the most mobile of all animal species and spend large amounts of their lives at sea. They cross vast areas of ocean that appear superficially featureless, and our understanding of the mechanisms that they use for navigation remains incomplete, especially in terms of available cues. In particular, several large-scale navigational tasks, such as homing across thousands of kilometers to breeding sites, are not fully explained by visual, olfactory or magnetic stimuli. Low-frequency inaudible sound, i.e., infrasound, is ubiquitous in the marine environment. The spatio-temporal consistency of some components of the infrasonic wavefield, and the sensitivity of certain bird species to infrasonic stimuli, suggests that infrasound may provide additional cues for seabirds to navigate, but this remains untested. Here, we propose a framework to explore the importance of infrasound for navigation. We present key concepts regarding the physics of infrasound and review the physiological mechanisms through which infrasound may be detected and used. Next, we propose three hypotheses detailing how seabirds could use information provided by different infrasound sources for navigation as an acoustic beacon, landmark, or gradient. Finally, we reflect on strengths and limitations of our proposed hypotheses, and discuss several directions for future work. In particular, we suggest that hypotheses may be best tested by combining conceptual models of navigation with empirical data on seabird movements and in-situ infrasound measurements.

The Cognitive Ecology of Animal Movement: Evidence From Birds and Mammals

Cognition, defined as the processes concerned with the acquisition, retention and use of information, underlies animals’ abilities to navigate their local surroundings, embark on long-distance seasonal migrations, and socially learn information relevant to movement. Hence, in order to fully understand and predict animal movement, researchers must know the cognitive mechanisms that generate such movement. Work on a few model systems indicates that most animals possess excellent spatial learning and memory abilities, meaning that they can acquire and later recall information about distances and directions among relevant objects. Similarly, field work on several species has revealed some of the mechanisms that enable them to navigate over distances of up to several thousand kilometers. Key behaviors related to movement such as the choice of nest location, home range location and migration route are often affected by parents and other conspecifics. In some species, such social influence leads to the formation of aggregations, which in turn may lead to further social learning about food locations or other resources. Throughout the review, we note a variety of topics at the interface of cognition and movement that invite further investigation. These include the use of social information embedded in trails, the likely important roles of soundscapes and smellscapes, the mechanisms that large mammals rely on for long-distance migration, and the effects of expertise acquired over extended periods.

Integration of visual landmark cues in spatial memory

Over the past two decades, much research has been conducted to investigate whether humans are optimal when integrating sensory cues during spatial memory and navigational tasks. Although this work has consistently demonstrated optimal integration of visual cues (e.g., landmarks) with body-based cues (e.g., path integration) during human navigation, little work has investigated how cues of the same sensory type are integrated in spatial memory. A few recent studies have reported mixed results, with some showing very little benefit to having access to more than one landmark, and others showing that multiple landmarks can be optimally integrated in spatial memory. In the current study, we employed a combination of immersive and non-immersive virtual reality spatial memory tasks to test adult humans' ability to integrate multiple landmark cues across six experiments. Our results showed that optimal integration of multiple landmark cues depends on the difficulty of the task, and that the presence of multiple landmarks can elicit an additional latent cue when estimating locations from a ground-level perspective, but not an aerial perspective.

Optimal combination of environmental cues and path integration during navigation

Navigation is influenced by body-based self-motion cues that are integrated over time, in a process known as path integration, as well as by environmental cues such as landmarks and room shape. In two experiments we explored whether humans combine path integration and environmental cues (Exp. 1: room shape; Exp. 2: room shape, single landmark, and multiple landmarks) to reduce response variability when returning to a previously visited location. Participants walked an outbound path in an immersive virtual environment before attempting to return to the path origin. Path integration and an environmental cue were both available during the outbound path, but experimental manipulations created single- and dual-cue conditions during the return path. The response variance when returning to the path origin was reduced when both cues were available, consistent with optimal integration predicted on the basis of Bayesian principles. The findings indicate that humans optimally integrate multiple spatial cues during navigation. Additionally, a large (but not a small) cue conflict caused participants to assign a higher weight to path integration than to environmental cues, despite the relatively greater precision afforded by the environmental cues.

Human spatial representation: what we cannot learn from the studies of rodent navigation

Studies of human and rodent navigation often reveal a remarkable cross-species similarity between the cognitive and neural mechanisms of navigation. Such cross-species resemblance often overshadows some critical differences between how humans and nonhuman animals navigate. In this review, I propose that a navigation system requires both a storage system (i.e., representing spatial information) and a positioning system (i.e., sensing spatial information) to operate. I then argue that the way humans represent spatial information is different from that inferred from the cellular activity observed during rodent navigation. Such difference spans the whole hierarchy of spatial representation, from representing the structure of an environment to the representation of subregions of an environment, routes and paths, and the distance and direction relative to a goal location. These cross-species inconsistencies suggest that what we learn from rodent navigation does not always transfer to human navigation. Finally, I argue for closing the loop for the dominant, unidirectional animal-to-human approach in navigation research so that insights from behavioral studies of human navigation may also flow back to shed light on the cellular mechanisms of navigation for both humans and other mammals (i.e., a human-to-animal approach).

Wild rufous hummingbirds use local landmarks to return to rewarded locations

Animals may remember an important location with reference to one or more visual landmarks. In the laboratory, birds and mammals often preferentially use landmarks near a goal ("local landmarks") to return to that location at a later date. Although we know very little about how animals in the wild use landmarks to remember locations, mammals in the wild appear to prefer to use distant landmarks to return to rewarded locations. To examine what cues wild birds use when returning to a goal, we trained free-living hummingbirds to search for a reward at a location that was specified by three nearby visual landmarks. Following training we expanded the landmark array to test the extent that the birds relied on the local landmarks to return to the reward. During the test the hummingbirds' search was best explained by the birds having used the experimental landmarks to remember the reward location. How the birds used the landmarks was not clear and seemed to change over the course of each test. These wild hummingbirds, then, can learn locations in reference to nearby visual landmarks.

How You Get There From Here

How do people combine their sense of direction with their use of visual landmarks during navigation? Cue-integration theory predicts that such cues will be optimally integrated to reduce variability, whereas cue-competition theory predicts that one cue will dominate the response direction. We tested these theories by measuring both accuracy and variability in a homing task while manipulating information about path integration and visual landmarks. We found that the two cues were near-optimally integrated to reduce variability, even when landmarks were shifted up to 90°. Yet the homing direction was dominated by a single cue, which switched from landmarks to path integration when landmark shifts were greater than 90°. These findings suggest that cue integration and cue competition govern different aspects of the homing response: Cues are integrated to reduce response variability but compete to determine the response direction. The results are remarkably similar to data on animal navigation, which implies that visual landmarks reset the orientation, but not the precision, of the path-integration system.

Reinforcement probability modulates temporal memory selection and integration processes

We have previously shown that rats trained in a mixed-interval peak procedure (tone=4s, light=12s) respond in a scalar manner at a time in between the trained peak times when presented with the stimulus compound (Swanton & Matell, 2011). In our previous work, the two component cues were reinforced with different probabilities (short=20%, long=80%) to equate response rates, and we found that the compound peak time was biased toward the cue with the higher reinforcement probability. Here, we examined the influence that different reinforcement probabilities have on the temporal location and shape of the compound response function. We found that the time of peak responding shifted as a function of the relative reinforcement probability of the component cues, becoming earlier as the relative likelihood of reinforcement associated with the short cue increased. However, as the relative probabilities of the component cues grew dissimilar, the compound peak became non-scalar, suggesting that the temporal control of behavior shifted from a process of integration to one of selection. As our previous work has utilized durations and reinforcement probabilities more discrepant than those used here, these data suggest that the processes underlying the integration/selection decision for time are based on cue value.

Arena geometry and path shape: when rats travel in straight or in circuitous paths?

We show here that the global geometry of the environment affects the shape of the paths of travel in rats. To examine this, individual rats were introduced into an unfamiliar arena. One group of rats (n=8) was tested in a square arena (2 m × 2 m), and the other group (n=8) in a round arena (2 m diameter). Testing was in a total darkness, since in the absence of visual information the geometry is not perceived immediately and the extraction of environment shape is slower. We found that while the level of the rats' activity did not seem to differ between both arenas, path shape differed significantly. When traveling along the perimeter, path shape basically followed the arena walls, with perimeter paths curving along the walls of the round arena, while being straight along the walls of the square arena. A similar impact of arena geometry was observed for travel away from the arena walls. Indeed, when the rats abandoned the arena walls to crosscut through the center of the arena, their center paths were circuitous in the round arena and relatively straight in the square arena. We suggest that the shapes of these paths are exploited for the same spatial task: returning back to a familiar location in the unsighted environment.

How habitat features shape ground squirrel (Urocitellus beldingi) navigation

The purpose of this investigation was to determine whether Belding's ground squirrels (Urocitellus beldingi) from areas rich in beacons perform differently in a task of spatial memory compared with squirrels from beacon-thin areas. To assess the role of environmental experience in spatial memory, wild-born squirrels with several days of experience in the field were compared with squirrels born in a lab and with no experience in their original habitat. Over two summers, squirrels captured from beacon-dense and beacon-thin areas were tested in a radial maze interspersed with beacons, using number of trials to criterion as a measure of spatial memory. To evaluate the effect of landmark navigation, in year 2 juveniles were prevented from seeing outside the maze area. In both years squirrels from beacon-dense populations reached criterion faster than squirrels from beacon-thin populations, and a weak rearing effect was present in 1 year. Despite sex differences in adult spatial skills, no differences were found between males and females in the maze. This demonstrates variation in the navigation strategies of young U. beldingi, and highlights the need to evaluate spatial preferences as a function of population or ecology in addition to species and sex.

The synthetic approach to the study of spatial memory: have we properly addressed Tinbergen's "four questions"?

In 1963, Niko Tinbergen suggested that to truly understand the behavior of an animal, the ultimate causes (e.g., adaptive value, evolutionary history) as well as the proximate mechanisms (e.g., neurobiology, development) that result in the production of the behavior must be understood in an integrated framework. We examine whether the study of spatial memory in food storing birds has adequately addressed Tinbergen's questions and highlight the work of Sara Shettleworth, who has made a tremendous contribution to this area of study, and whom this issue honors. Our conclusion is that while the study of food caching and spatial memory in birds has been a very good model of a program of research that has addressed Tinbergen's questions, additional work remains.

Causal inference in perception

Until recently, the question of how the brain performs causal inference has been studied primarily in the context of cognitive reasoning. However, this problem is at least equally crucial in perceptual processing. At any given moment, the perceptual system receives multiple sensory signals within and across modalities and, for example, has to determine the source of each of these signals. Recently, a growing number of studies from various fields of cognitive science have started to address this question and have converged to very similar computational models. Therefore, it seems that a common computational strategy, which is highly consistent with a normative model of causal inference, is exploited by the perceptual system in a variety of domains.

Exploration and navigation in the blind mole rat (Spalax ehrenbergi): global calibration as a primer of spatial representation

The aim of this study was to uncover the process of initial spatial mapping of the environment. For this, blind mole rats (Spalax ehrenbergi), were tested in an unfamiliar square arena, in order to reveal how they construct a spatial representation. The mole rats first displayed a build-up phase, in which they gradually formed a path along the perimeter while travelling slowly, frequently pausing and repeating previously travelled segments of the path. This behaviour was followed by a free-travel phase, in which the mole rats appeared to locomote smoothly along the perimeter and through the centre of the arena while travelling faster with fewer stops or repetitions of path segments. Familiarity with the environment was reflected in local shortcuts at the arena corners and global shortcuts (crosscuts) through the arena centre. We suggest that scanning the perimeter throughout the build-up phase constitute a process of calibration, i.e. forming an initial representation of the size and perhaps the shape of the environment--a sort of basic global map. We further suggest that this calibration is later used for navigation, as indicated by the emergence of global crosscuts in the subsequent phase. Further investigation of the build-up phase, e.g. by manipulating environment size, might provide additional insight into the course of establishment of global environment representation (mapping).

The reach-to-grasp-food task for rats: a rare case of modularity in animal behavior?

Humans and non-human animals make use of sensory hierarchies in "selecting" strategies for solving many cognitive and behavioral tasks. Often, if a preferred type of sensory information is unavailable or is not useful for solving a given task, the animal can switch to a lower-priority strategy, making use of a different class of sensory information. In the case of rats performing a classic reach-to-grasp-food task, however, prior studies indicate that the reaching maneuver may be a fixed action pattern that is guided exclusively by the food's odor plume until the point of contact with the food morsel [Whishaw IQ, Tomie JA. Olfaction directs skilled forelimb reaching in the rat. Behav Brain Res 1989;32(1):11-21; Metz GA, Whishaw IQ. Skilled reaching an action pattern: stability in rat (Rattus norvegicus) grasping movements as a function of changing food pellet size. Behav Brain Res 2000;116(2):111-22; Whishaw IQ. Did a change in sensory control of skilled movements stimulate the evolution of the primate frontal cortex? Behav Brain Res 2003;146(1/2):31-41]. We sought to confirm and extend these findings in several ways. In Experiment 1, using a GO/NO-GO variant of the classic task, we demonstrated that rats used the GO target's odor both to trigger and guide their reaches. In Experiment 2, we showed that rats deprived of (a) vision, (b) object-recognizing rostral whiskers and forearm sinus hairs, or (c) both, displayed no deficits in triggering and guiding their reaches. Finally, in a third experiment in which the GO target's location varied randomly across trials and only olfactory cues were available, we demonstrated that rats could determine the spatial endpoint of their reach without any loss of accuracy. Combined with results from a prior study in which bulbectomized rats never developed a new, successful reaching strategy despite extensive post-operative training [Whishaw IQ, Tomie JA. Olfaction directs skilled forelimb reaching in the rat. Behav Brain Res 1989;32(1):11-21], these results indicate that rats do not have a sensory hierarchy for solving the reach-to-grasp-food task, but rather, are guided by olfaction alone until their paw contacts the food morsel.

Taking the best for learning

Examples of how animals learn when multiple, sometimes redundant, cues are present provide further examples not considered by Hutchinson and Gigerenzer that seem to fit the principle of taking the best. "The best" may the most valid cue in the present circumstances; evolution may also produce species-specific biases to use the most functionally relevant cues.

Shape Parameters Explain Data From Spatial Transformations: Comment on Pearce et al. (2004) and Tommasi & Polli (2004)

In 2 recent studies on rats (J. M. Pearce, M. A. Good, P. M. Jones, & A. McGregor, see record 2004-12429-006) and chicks (L. Tommasi & C. Polli, see record 2004-15642-007), the animals were trained to search in 1 corner of a rectilinear space. When tested in transformed spaces of different shapes, the animals still showed systematic choices. Both articles rejected the global matching of shape in favor of local matching processes. The present authors show that although matching by shape congruence is unlikely, matching by the shape parameter of the 1st principal axis can explain all the data. Other shape parameters, such as symmetry axes, may do even better. Animals are likely to use some global matching to constrain and guide the use of local cues; such use keeps local matching processes from exploding in complexity.

What Do Rats Learn About the Geometry of Object Arrays? Tests With Exploratory Behavior

Six experiments using habituation of exploratory behavior tested whether disoriented rats foraging in a large arena encode the shapes of arrays of objects. Rats did not respond to changes in position of a single object, but they responded to a change in object color and to a change in position of 1 object in a square array, as in previous research (e.g., C. Thinus-Blanc et al., 1987). Rats also responded to an expansion of a square array, suggesting that they encoded sets of interobject distances rather than overall shape. In Experiments 4-6, rats did not respond to changes in sense of a triangular array that maintained interobject distances and angles. Shapes of object arrays are encoded differently from shapes of enclosures.