Spatial Peak Shift and Generalization in Pigeons
Department of Psychology, Macquarie University
Marcia L. Spetch and Michael Johnston
Department of Psychology, University of Alberta
Pigeons were presented a square at a particular location on a monitor
screen (the S+ location). With the square at the S+ location, the first
peck anywhere on the screen after 5 s was rewarded. In Experiment 1, birds
were trained with the S+ location only, and then tested occasionally with
the square at a range of horizontal locations. The generalization gradient
peaked at the S+ and was approximately Gaussian in shape. During training
of Experiments 2 and 3, S+ trials were interspersed with S- trials in which
the square was presented at a location horizontally displaced from the S+
location (S- location) and no reward was dispensed. In Experiment 2, horizontal
generalization tests showed an asymmetric gradient, with the S+ side falling
less steeply. In Experiment 3, in which the S- was closer to the S+, the
horizontal generalization gradient was again asymmetric, and peak of responding
shifted initially in the direction away from S- (peak-shift effect). In
both cases, the asymmetry diminished with continued testing. Experiment
4 examined generalization along the vertical dimension for birds trained
in Experiments 2 and 3. The gradients were approximately exponential in
shape. This is the first demonstration of generalization and peak shift
in the spatial domain.
Distances and Directions are Computed Separately
by Honeybees in Landmark-based Search
Department of Psychology, Macquarie University
Honeybees were trained with two landmarks at some angle, e.g., 120 degrees,
apart from the target. On crucial unrewarded tests, only a single landmark
was present. If distances and directions to landmarks are computed separately
(independent averaging), the search distance to the landmark should equal
the landmark-target distance found in training. If entire vectors are averaged,
the search distance should be much shorter. Three experiments with short
target-landmark distances found results in between the predictions of the
two hypotheses. A fourth experiment used longer target-landmark distances
and isolated double peaks on single-landmark tests: one predicted by the
independent averaging hypothesis, and one very close to the landmark. The
near peak is interpreted as arising from approach and exploration of a landmark
in a new location, and not from searching.
Landmark-based spatial search in honeybees: II.
Using gaps and blocks
Department of Psychology, Macquarie University
Honeybees were trained to find sugar water at a fixed location in front of a rectangular block or a gap in a row of rectangular blocks aligned east-west. Bees learned to use both blocks and gaps to pinpoint a target place. In approach, the bees tended to head directly towards the block but not the gap. In approaching the gap, they tended to approach the wall, and then fly parallel to the wall until the gap was encountered. In approaching the block, they approached from varying directions. When the width of the block or gap was doubled, bees still searched at the same perpendicular distance from the landmark. When the height of the block was doubled, they searched farther away but not as far as double the distance on control tests, with variations across individuals. Near the target area, the bees tended to face almost parallel to the gap or block, turned slightly towards the landmark by 0--30°. In all setups, bees showed a tendency to search near the landmark, and to search on average closer than the training distance. The results confirm the basic processes identified for landmark-based search in bees, but show variations in how the steps are carried out.
How honeybees find a place: Lessons from a simple mind
Department of Psychology, Macquarie University
Foraging honeybees find their way from their hive to their food in a
stereotypical manner using up to four place-finding servomechanisms in sequence.
1. They first fly a vector (straight-line distance and direction) from their
home to the vicinity of the target. Direction is determined by the sun-compass
and by distant landmarks, while distance is estimated by visual flow. 2.
They then beacon in on a landmark near the target location. 3. Enroute toward
the landmark, they may adopt a sensorimotor trajectory that takes them toward
the target. 4. Lastly, near the expected target location, they attempt image
matching, which consists in trying to put surrounding landmarks at the correct
positions on their eyes. In doing image matching, they fly facing a stereotypical
direction, a strategy that makes it unnecessary to translate retinal coordinates
into another coordinate system.
Shepard's universal law supported by honeybees in spatial generalization
Department of Psychology, Macquarie University
An animal that is rewarded for a response in one situation (the S+) is
likely to respond to similar but recognizably different stimuli, the ubiquitous
phenomenon of stimulus generalization. Based on functional analyses of the
probabilistic structure of the world, Shepard (1987) formulated a universal
law of generalization, claiming that generalization gradients, as a function
of the appropriately scaled distance of a stimulus from S+, should be exponential
in shape. The law was tested in spatial generalization in honeybees. Based
on theoretically derived scales, generalization along the dimensions of
the distance from a landmark and the direction to a landmark both followed
Shepard's law. Support in an invertebrate animal increases the scope of
the law, and suggests that the ecological structure of the world may have
driven the evolution of cognitive structures in diverse animals.
Self control in honeybees
Ken Cheng, Jennifer Peña, Melanie A. Porter, Julia D. Irwin
Department of Psychology, Macquarie University
Self control means choosing a large delayed reward over a small immediate reward, and impulsiveness is the opposite. The metabolic hypothesis states that the amount of self control across species correlates negatively with metabolic rate (Tobin & Logue, 1994). Foraging honeybees have high metabolic rates; the metabolic hypothesis would predict little self control in bees. But foraging bees work for the long term good of their hive, conditions that seem to require self control. In three experiments, we gave bees the choice between 1) a sweeter delayed reward and a less sweet immediate reward, and 2) a large delayed reward and a small immediate reward. Bees showed much self control, inconsistent with the metabolic hypothesis.
Spatial generalization and peak shift in humans
Ken Cheng* and Marcia L. Spetch#
*Department of Psychology, Macquarie University, Sydney, Australia, #Department of Psychology, University of Alberta, Edmonton, Canada
Using a computer betting game, five experiments tested university students
on spatial generalization and peak shift. On each trial, one location was
marked and the subject was invited to bet 0 to 4 points. At the winning
location (S+), bets won four times the points betted. At nearby losing locations
(S-s), points betted were lost. Generalization gradients were exponential
in shape, supporting Shepard's (1987) law (Experiment 1). With peak shift
manipulations, three kinds of peak shift or area shift were found. 1) Subjects
betted more on the S+ side than on the S- side (Experiments 2-4). 2) When
asked if a location was the winning location, subjects responded "yes"
more often to locations on the S+ side than locations on the S- side (Experiments
3-5). 3) When asked to point to the winning location on the screen, subjects'
errors indicated peak shift (Experiment 5).
Navigating desert ants (Cataglyphis fortis) learn to alter their search patterns on their homebound journey
Ken Cheng* and Rüdiger Wehner°
*Department of Psychology, Macquarie University, Sydney, Australia, °Department
of Zoology, University of Zurich, Zurich, Switzerland
In navigating home, desert ants first run off a global vector estimated on their outbound journey, and then engage in systematic search consisting of ever increasing loops interrupted by returns to the starting point of search. Desert ants (Cataglyphis fortis; Wehner, 1983) were trained to travel 6 m down a channel to a food source. Different groups of ants were trained to return home in another channel, from distances of 6 m (control), 9 m, or 12 m. Ants at the feeder were then tested in a long test channel. The measure of where the ants first turned back on a test gave an estimate of the length of the global vector calculated on their outbound trip. The median distance of search on a 5-min test gave an estimate of the centre of the search pattern. Relative to controls, the experimental ants did not increase their estimated length of global vector, but changed their search patterns, searching on average further from the start than the controls. Tests of the outbound journey, however, revealed no differences between groups. Desert ants can learn to modify their search pattern based on experience.
Is There a Geometric Module for Spatial Orientation?
Squaring Theory and Evidence
Ken Cheng, Macquarie University, Sydney
Nora S. Newcombe, Temple University
There is evidence, beginning with Cheng (1986), that mobile animals may use the geometry of surrounding areas to reorient following disorientation. Gallistel (1990) proposed that geometry is used to compute the major or minor axes of space, and suggested that such information might form an encapsulated cognitive module. Research reviewed here, conducted on a wide variety of species since the initial discovery of the use of geometry and the formulation of the modularity claim, has supported some aspects of the approach, while casting doubt on others. Three possible processing models are presented that vary in the way in which (and the extent to which) they instantiate the modularity claim. The extant data do not discriminate among them. We propose a modified concept of modularity for which an empirical program of research is more tractable.
Small-scale spatial cognition in pigeons
Ken Cheng*, Marcia L. Spetch#, Debbie M. Kelly**, Verner P. Bingman##
*Centre for the Integrative Study of Animal Behaviour and Department of Psychology, Macquarie University, Sydney, NSW 2109, Australia
#Department of Psychology, University of Alberta, Edmonton, Alberta, Canada T6G 2E9
**Department of Psychology, University of Saskatchewan, Saskatoon, Saskatchewan, Canada S7N 5A5
##Department of Psychology and J.P. Scott Center for Neuroscience, Mind
and Behavior, Bowling Green State University, Bowling Green, Ohio 43403,
Roberts and Van Veldhuizen's (1985) study on pigeons in the radial maze sparked research on landmark use by pigeons in lab-based tasks as well as variants of the radial-maze task. Pigeons perform well on open-field versions of the radial maze, with feeders scattered on the laboratory floor. Pigeons can also be trained to search precisely for buried food. The search can be based on multiple landmarks, but is sometimes controlled by just one or two landmarks, with the preferred landmarks varying across individuals. Findings are similar in landmark-based searching on a computer monitor and on a lab floor, despite many differences between the two kinds of tasks. A number of general learning principles are found in landmark-based searching, such as cue competition, generalization and peak shift, and selective attention. Pigeons also learn the geometry of the environment in which they are searching. Neurophysiological studies have implicated the hippocampal formation in avian spatial cognition, with the right hippocampus hypothesized to play a more important role in the spatial recognition of goal locations. Most recently, single-cell recording from the pigeon's hippocampal formation has revealed cells with different properties from the classic 'place' cells of rats, as well as differences in the two sides of the hippocampus.
Honeybees (Apis mellifera) holding on to memories:
Response competition causes retroactive interference effects
Ken Cheng and Anne E. Wignall
Centre for the Integrative Study of Animal Behaviour
Five experiments on honeybees examined how the learning of a second task interferes with what was previously learned. Free flying bees were tested for landmark-based memory in variations on a paradigm of retroactive interference. Bees first learned Task 1, were tested on Task 1 (Test 1), then learned Task 2, and were tested again on Task 1 (Test 2). A 60-min delay (waiting in a box) before Test 2 caused no performance decrements. If the two tasks had conflicting response requirements, (e.g., target right of a green landmark in Task 1 and left of a blue landmark in Task 2), then a strong decrement on Test 2 was found (retroactive interference effect). When response competition was minimised during training or testing, however, the decrement on Test 2 was small or nonexistent. The results implicate response competition as a major contributor to the retroactive interference effect. The honeybee seems to hold on to memories; new memories do not wipe out old ones.
Behavioral ecology of odometric memories in desert ants: Acquisition, retention, and integration
Ken Cheng1, Ajay Narendra1,2, Rüdiger Wehner3
1 Centre for the Integrative Study of Animal Behaviour, Macquarie University
2 Department of Biological Sciences, Macquarie University
3 Department of Zoology, University of Zurich
Assuming that the acquisition and retention of memories have costs, properties of memories should fit the functional requirements for the system of memory. Based on a functional analysis of what path integration is meant to do, we predicted that odometric memories in desert ants should show 1) little improvement with repeated training: performance should be as good after 1 training trial as after 6 training trials, 2) decay of memory after 24 h, 3) performance based solely on the most recent outbound trip, with no integration over multiple memories. Desert ants (Cataglyphis fortis) travelled in narrow straight plastic channels to forage for cookie crumbs in a feeder at 6 or 12 m distance. Each ant was tested once by being taken from the feeder and released 2 m from the end of a 32-m channel to run home. The distance at which the ant first turned back (first turn) constituted the data. In acquisition, groups trained 1 or 6 times before being tested had unsystematic scatter that did not differ significantly. In retention, ants tested after 24 h delay showed larger unsystematic scatter than control animals tested after no delay. In integration, ants were trained 5 times at 6 or 12 m, and then tested at 12 or 6 m, respectively. No evidence of integration of multiple odometric memories was found. The results show that the properties of odometric memories are indeed tailored to what the memory system is used for.
Arthropod navigation: Ants, bees, crabs, spiders finding their way
Arthropods rely on path integration and landmarks to navigate. In path integration, the animal keeps track of the straight-line distance and direction as it travels. This requires computing and using metric information. The sky-compass is an important cue for telling direction, and in flying honeybees, optic flow is a dominant cue for estimating distance travelled. Landmarks are used to determine a direction of travel, to construct routes, help calibrate or 'reset' path integration, as beacons to head toward, and to pinpoint the exact location of a target in servomechanistic fashion. One important underlying mechanism for route-following behavior has been called local or sensorimotor vectors. Evidence concerning map-like navigation is also discussed. One theme stands out: stimulus conditions set the occasion for or bring on the operation of different place-finding servomechanisms.
Bayesian integration of spatial information
Ken Cheng, Sara J. Shettleworth, Janellen Huttenlocher, John J. Rieser
Spatial judgments and actions are often based on multiple cues. We review a multitude of phenomena on the integration of spatial cues in diverse species, to consider how nearly optimally animals combine the cues. Under the banner of Bayesian perception, cues are sometimes combined and weighted in a near optimal fashion. In other instances when cues are combined, how optimal the integration is might be unclear. Only one cue may be relied on, or cues may seem to compete with one another. We attempt to bring some order to the diversity by taking into account the subjective discrepancy in the dictates of multiple cues. When cues are too discrepant, it may be best to rely on one cue source. When cues are not too discrepant, it may be advantageous to combine cues. Such a dual principle provides an extended Bayesian framework for understanding the functional reasons for the integration of spatial cues.
Whither geometry? Troubles of the geometric module
In rectangular arenas, rats often confuse diagonally opposite corners, even when distinctive cues differentiate them. This led to the postulation that rats rely preferentially on the geometry of space, encoded in a dedicated geometric module. Recent research casts doubt on this idea. Distinctive featural cues such as entire walls of a distinct color can hinder or aid the learning of geometry. In one situation where using geometry would help greatly, rats had trouble learning the task. An associative model has been developed to capture these different learning processes. And view-based matching has been proposed as an alternative to the explicit coding of geometric cues. Considerations about how cues interact in learning are crucial in a recent theory of human spatial cognition.
Traveling in clutter: Navigation in the Central Australian desert ant Melophorus bagoti
Ken Cheng, Ajay Narendra, Stefan Sommer, Rudiger Wehner
The Central Australian desert ant Melophorus bagoti is the most thermophilic ant on the continent. It comes out to forage during the hottest part of the day in the summer months. The ant shares a cluttered, plant-filled habitat with other arthropods and uses a range of navigational strategies. We review recent studies on this species concerning its use of habitual routes, distant landmarks, landmarks around the nest, and path integration, which is keeping track of the distance and direction traveled from one’s starting point. Functional predictions concerning the acquisition, retention, and integration of memories of distances and of landmarks are also reviewed, illuminating the behavioral ecology of spatial cognition.
Categories and range effects in human spatial memory
Ken Cheng, Marcia Spetch, Andros Hoan
After learning a particular target stimulus, such as a location, humans’ judgements of whether a particular stimulus is the target or not is affected by the range of stimuli presented on tests. In such frequently found range effects, the peak of “yes” responses shifts toward the middle of the range of tested stimuli. Humans also code both the metric value and categorical information regarding a target stimulus, and use both forms of codes, such that responses are biased toward the category middle. Categorical codes should also affect range effects, with a test range crossing category boundaries producing less range effect than a test range within a category. We examined a set of past results presented in a review of range effects in humans (Thomas, 1993, Psychol. Rev. 100, 658-673) for functional explanations in light of categorical coding, and found that all results could be reasonably explained. Additional experiments comparing range effects across vs. within a category found limited supporting evidence, perhaps because the range effects were weak. The adaptive functions of using (in part) categorical coding accounts for many seemingly peculiar biases in human cognition.
Combining sky and Earth: Desert ants (Melophorus bagoti) show weighted integration of celestial and terrestrial cues
Eric Legge, Antoine Wystrach, Marcia Spetch, Ken Cheng
Insects typically use celestial sources of directional information for path integration, and terrestrial panoramic information for view-based navigation. Here we set celestial and terrestrial sources of directional information in conflict for homing desert ants (Melophorus bagoti). In the first experiment, ants learned to navigate out of a round experimental arena with a distinctive artificial panorama. On crucial tests, we rotated the arena to create a conflict between the artificial panorama and celestial information. In a second experiment, ants at a feeder in their natural visually-cluttered habitat were displaced prior to their homing journey so that the dictates of path integration (feeder to nest direction) based on a celestial compass conflicted with the dictates of view-based navigation (release point to nest direction) based on the natural terrestrial panorama. In both experiments, ants generally headed in a direction intermediate to the dictates of celestial and terrestrial information. In the second experiment, the ants put more weight on the terrestrial cues when they provided better directional information. We conclude that desert ants weight and integrate the dictates of celestial and terrestrial information in determining their initial heading, even when the two directional cues are highly discrepant.
Crucial role of ultraviolet light for desert ants in determining direction from the terrestrial panorama
Antoine Wystrach, Sebastian Schwarz, Aloys Tack, Jeanne Delor, Sabine
S. Nooten, Anne-Laurence Bibost, Cody A. Freas, Ken Cheng
Ants use the panoramic skyline in part to determine a direction of travel. A theoretically elegant way to define where terrestrial objects meet the sky is to use an opponent-process channel contrasting green wavelengths of light with ultraviolet (UV) wavelengths. Compared with the sky, terrestrial objects reflect relatively more green wavelengths. Using such an opponent-process channel gains constancy in the face of changes in overall illumination level. We tested the use of UV wavelengths in desert ants by using a plastic that filtered out most of the energy below 400 nm. Ants, Melophorus bagoti, were trained to home with an artificial skyline provided by an arena (experiment 1) or with the natural panorama (experiment 2). On a test, a homing ant was captured just before she entered her nest, and then brought back to a replicate arena (experiment 1) or the starting point (the feeder, experiment 2) and released. Blocking UV light led to deteriorations in orientation in both experiments. When the artificial skyline was changed from opaque to transparent UV-blocking plastic (experiment 3) on the other hand, the ants were still oriented. We conclude that UV wavelengths play a crucial role in determining direction based on the terrestrial surround.
weight and integrate the dictates of celestial and terrestrial
information in determining their initial heading, even when the two
directional cues are highly discrepant.
Steering intermediate courses: desert ants combine information from various navigational routines
Rudiger Wehner, Theirry Hoinville, Holk Cruse, Ken Cheng
A number of systems of navigation have been studied in some detail in insects. These include path inte- gration, a system that keeps track of the straight-line dis- tance and direction travelled on the current trip, the use of panoramic landmarks and scenery for orientation, and systematic searching. A traditional view is that only one navigational system is in operation at any one time, with different systems running in sequence depending on the context and conditions. We review selected data suggest- ing that often, different navigational cues (e.g., compass cues) and different systems of navigation are in operation simultaneously in desert ant navigation. The evidence sug- gests that all systems operate in parallel forming a heterar- chical network. External and internal conditions determine the weights to be accorded to each cue and system. We also show that a model of independent modules feeding into a central summating device, the Navinet model, can in prin- ciple account for such data. No central executive processor is necessary aside from a weighted summation of the dif- ferent cues and systems. Such a heterarchy of parallel sys- tems all in operation represents a new view of insect navi- gation that has already been expressed informally by some authors.
back to Ken's home page