Performance ofBetta splendens in a radial arm maze

Siamese fighting fish (Betta splendens) were tested in aquatic versions of radial arm mazes. In the first experiment, the fish were trained to find tubifex worms in an eight-arm maze in which the optimal strategy was to choose each arm once without repetition. After initial training, the fish entered approximately 6.63 different arms in eight choices, showing a strong tendency to choose sequences of adjacent arms, moving about the maze in a Stereotypic direction. This algorithmic response pattern was not, however, sufficient to predict the high performance level of the fish. In the second experiment, a delay of .5 or 5 min was interposed between the fourth and fifth choices. Similar Stereotypic patterns continued in Experiment 2, but choice accuracy following the longer delay declined to a level not significantly above chance. In the third experiment, different fish were tested in a three-arm maze, reinforced either for returning from the second arm to the arm in which they had most recently been fed (win-stay) or for visiting a third arm (win-shift). The fish were significantly faster at acquiring the win-shift contingency than the win-stay contingency. These results demonstrate that solution of spatial tasks depends on the interaction of appropriate behavioral strategies and cognitive capacities that may have little generality across species.     

Time-place learning in the eight-arm radial maze

Rats (n=4) searched for food on an eight-arm radial maze. Daily 56-min sessions were divided into eight 7-min time zones, during each of which a different location provided food; locations were randomized across subjects before training. The rats obtained multiple pellets within each time zone by leaving and returning to the correct location. Evidence that the rats had knowledge about the temporal and spatial features of the task includes the following. The rats anticipated locations before they became active and anticipated the end of the currently active locations. The rats discriminated currently active locations from earlier and forthcoming active locations in the absence of food transition cues. After the rats had left the previously active location, they visited the next correct location more often than would be expected by chance in the absence of food transition cues. The rats used handling or opening doors as a cue to visit the first location and timed successive 7-min intervals to get to subsequent locations.     

Spatial pattern learning in the radial arm maze

Rats experienced a spatial pattern of baited and unbaited arms in an eight-arm radial maze. The spatial pattern remained constant over trials, but the spatial locations that were baited varied unpredictably. Although there was no evidence of control by the spatial pattern during free choice training trials, the rats’ ability to locate baited arms in forced choice test trials was superior to that of animals in a control condition for which maze arms were not baited in a consistent spatial pattern. This is consistent with the results of experiments showing that spatial choices by rats in a pole box maze are controlled by abstract spatial patterns.     

Social effects on spatial choice in the radial arm maze

Social memory was investigated in the context of a spatial working memory task. Pairs of rats were tested in an eight-arm radial maze. Under most conditions, there was a tendency to choose maze locations that had been visited earlier by the other rat. The possibility that this tendency is produced by common preferences for particular maze locations was ruled out. An opposite tendency to avoid visits to locations that had been visited earlier during the trial by another rat was found only when the maze location contained two pellets (rather than an undepletable supply), the rats’ ability to see each other in the maze was restricted to the central arena, and the maze location had been previously visited by the focal rat. The amount of food available in maze locations did not otherwise modulate social influences on spatial choice. The results indicate that memory for a rat’s own previous choices is combined with memory for the choices made by another rat.     

Response strategies in the radial arm maze: Running around in circles

The effects of the size of the central arena on the use of response strategies by rats on an eight-arm elevated maze were examined. The size of the central arena had no effect on accuracy, but the use of adjacent arms increased significantly with a larger central arena, regardless of the size of arena to which rats were first exposed. These results are interpreted in terms of foraging efficiency.

     

Rats show preference for delayed rewards on the radial maze

Rats on an eight-arm radial maze chose between four arms on which a small reward could be obtained after a short delay and four arms on which a larger reward could be obtained after a longer delay. Experiments 1 and 2 showed that rats preferred the long-delay, large-reward arms over the short-delay, small-reward arms. This preference was particularly marked when the arms were made into enclosed alleys. Experiment 3 showed that this effect was not produced by a preference for staying in enclosed alleys. We argue that the rats endured longer delays to obtain larger rewards when fear of predation was minimized.     

Intratrial proactive interference in rats’ serial alternation performance in the radial maze

Rats acquired a serial alternation task in an eight-arm radial maze that was partitioned into four pairs of arms. Each pair was associated with a different distal stimulus. Rats were initially forced to the left or right arm in each pair (the study segment) before being exposed to both arms in each pair (the free-choice or test segment). Only the previously blocked arm of each pair remained baited. Following initial training, proactive interference (PI) was induced by presenting rats with a forced-choice (prestudy) segment containing arm positions opposite those in the subsequent study segment. Such trials generated poorer free-choice accuracy than did trials without a prestudy segment. Forcing rats to both arms in the pair in a prestudy segment produced only transient PI. A slight improvement in rats’ free-choice performance was obtained by forcing them to the same arm position, but only when the test segment was delayed by 30 min. Increasing the interval between the prestudy and study segments from 2 to 30 min eliminated PI, but only when free-choice testing was delayed by 2 min rather than by 30 min. These results suggest that intratrial PI in this preparation was primarily due to confusion about which arm position in each pair had been visited during the last forced-choice segment.     

Multiple-pattern learning by rats on an eight-arm radial maze

Rats were trained over a number of sessions on an eight-arm radial maze with eight trials on each session. Each of four arms on the maze contained a different pattern formed by sequences of reward (two pellets) or nonreward (no pellets) over successive trials within sessions. The patterns were single alternation, double alternation, and two patterns in which four rewards or four nonrewards were preceded and followed by two nonrewards or two rewards, respectively. The other four arms on the maze served as control arms and always contained one pellet. It was found that rats tracked all of these patterns when they were required to climb over barriers to enter and leave arms. However, rats showed no ability to extrapolate patterns beyond the training trials. These findings, and a further analysis of arm-choice stereotypy, led to the conclusion that rats tracked by using a trial-number strategy.     

Foraging for covered and uncovered food on a radial maze

Rats were trained to forage for food on a four-arm radial maze. Each arm of the maze was defined as a patch and contained four feeding stations. Each patch contained a total of 20 45-mg food pellets, with the first feeding station in each patch baited with 1 pellet and the remaining stations baited with 1, 5, or 13 pellets. In Experiment 1, one group of rats was tested with feeders open and food readily accessible, and another group was tested with metal covers on the feeders, which necessitated extra time to gain access to food. With open feeders, the rats visited each feeder in a patch in the order in which they encountered the feeders, from the center of the maze to the end of the arm. The rats in the group with the covered feeders often visited the feeders containing 5 or 13 pellets first and the feeders containing 1 pellet last. In Experiment 2, it was found that the rats switched readily between these two foraging strategies when tested with covered and open feeders on alternate sessions. The extra time and effort required to uncover food appeared to produce selective foraging in rats.     

Variations in radial maze performance under different levels of food and water deprivation

Four groups of rats were tested on an eight-arm radial maze under a free-choice procedure. The subjects were maintained at either 80% or 100% of their preexperimental free-feeding weights through restricted access to either food or water. Water-deprived subjects received water in the maze; food-deprived subjects received food. Water-deprived subjects learned the task faster than food-deprived subjects. The four groups developed different response patterns. These were measured by themean transition size, the average angular distance (in 45° units) between consecutively chosen arms. Rats foraging for food and water developed different search strategies, with water-deprived subjects exhibiting lower mean transition sizes. When the subjects were given three consecutive trials, 2 min apart, choice accuracy declined across trials, although performance on the last two trials improved across days. The groups’ mean transition sizes remained different, and were constant over trials and days. Thus, the test procedures differentially affected choice accuracy and response patterning.     

Evidence for a shift in the choice criterion of rats in a 12-arm radial maze

Rats were trained in a standard 12-arm radial maze task. Following training, each trial consisted of a sequence of 2, 4, 6, 8, or 10 choices, followed by a 15-min delay, which then was followed by a choice between a single arm and a response manipulandum mounted in the center of the maze. An arm visit was reinforced if the arm had not been visited prior to the delay, whereas a manipulandum response was reinforced if the arm had been visited. It was found that rats are relatively more likely to reject arms by responding to the manipulandum following a delay occurring late in the choice sequence. This indicates that the choice criterion used by rats in the radial maze becomes more strict as the choice sequence progresses. Such a process provides an alternative explanation for some of the data recently reported by Cook, Brown, and Riley (1985).     

Testing optimal foraging theory on the radial maze: The role of learning in patch sampling

A four-arm radial maze containing 10 feeders in each arm (patch) was used to study patch sampling in rats. In each of three experiments, rats foraged for 30 sessions. On each session, two randomly chosen patches were baited with food and the remaining two patches were empty. In Experiment 1, the number of baited feeders in baited patches (6) was varied from 1–10 over five groups of subjects. Mean visits to empty patches was an inverse function of 6, as predicted by an optimal foraging model. In Experiments 2 and 3, rats’ ability to discriminate between baited and empty patches was examined when food in baited patches was placed in fixed locations, either in clumps (Experiment 2) or distributed throughout the patch (Experiment 3). Rats in fixed-food-location conditions reliably visited fewer feeders in empty patches than did rats in randomly changing control groups. Examination of within-patch foraging patterns indicated that rats in fixed-food-location groups selectively sampled potentially baited locations and abandoned the patch if food was not found. It is suggested that processes of patch discrimination were responsible for these effects.     

Spatial configuration and list learning of proximally cued arms by rats in the enclosed four-arm radial maze

In an enclosed four-arm radial maze, after sampling three experimenter-selected baited arms (thestudy segment) and following rotation of the maze, rats had to find the fourth baited arm among all four unblocked arms (test segment). The rats learned this task with two sets of arm cues, objects at arms’ entrances and full arm inserts, each maintained in a fixed configuration. When we changed the configuration of one set of arms to itsmirror image and that of the other set to a moremixed variation by switching opposite and adjacent cued arms, the rats’ accuracy was similarly disrupted (Experiment 1). In Experiment 2, the same rats rapidly recovered their high search accuracy on four new configurations recombined from pairs of adjacent arms and pairs of opposite cued arms from the previous final two configurations. Their test segment search accuracy, however, was again disrupted when these configurations were varied either only over trials’ study segments or only over trials’ test segments. In Experiment 3, however, these rats attained accuracy as high on two sets of cued arms with constantly changing configurations as on two sets with constant configurations. Thus, the rats were able to separately represent four different spatially stable configurations, and then they could learn to represent two of these configurations as lists of spatially irrelevant items. We discuss these findings in terms of association theory and parallel map theory (Jacobs & Schenk, 2003).     

Performance of honeybees in reversal and ambiguous-cue problems: Tests of a choice model

Individual honeybees visiting the laboratory regularly for food were trained in a variety of discrimination-reversal and ambiguous-cue problems to choose between two differently scented targets, of which one or the other contained sucrose solution. The training was simulated with a set of equations for predicting choice on the basis of associative strength, and the accuracy of prediction was indexed by the root-mean-square deviation of simulated data from obtained data. Developed earlier on the basis of performance in other problems (Couvillon & Bitterman, 1985), the model proved adequate for the new problems as well.     

Analysis of the overlearning-extinction effect in honeybees

In previous experiments with individual honeybees that visited the laboratory regularly to take sucrose solution from a target set on the shelf of an open window, the overlearning-extinction effect was found for high concentrations of sucrose but not for low. The purpose of the present experiment was to examine the possibility that declining resistance to extinction in the course of prolonged training with a high concentration of sucrose could be explained in terms of increasing nutritive level. Three groups of animals were extinguished on a distinctive target, one group after 6 visits to the target, a second after 18 visits to the target, and a third after 6 visits to the target that were interspersed among 12 visits to a different target. More rapid extinction in the second group than in the third, which had fewer training visits to the extinction target than the second but the same total number of training visits, rules out an explanation in terms of nutritive level and points instead to a frustration-like process evidenced also in earlier work on incentive contrast.     

Latent inhibition in honeybees

Aversive conditioning was studied in individual honeybees flying back and forth between the hive and the sill of an open laboratory window, where they took sucrose solution from a target so constructed that shock could be delivered while the proboscis was in contact with the solution. During feeding, a conditioned stimulus—substrate vibration or airstream—was paired with brief shock avoidable by interruption of feeding. In Experiment 1, unreinforced preexposure of the conditioned stimulus was found to retard acquisition (latent inhibition). In Experiment 2, which was designed to inquire into the stimulus specificity of the effect, differential conditioning was found to be impaired by unreinforced preexposure of the positive stimulus and facilitated by unreinforced preexposure of the negative stimulus. In Experiment 3, a summation experiment designed to test various alternative explanations of the effect, a preexposed stimulus was found to suppress response to an excitatory conditioned stimulus when the two stimuli were presented together.     

Intramodal blocking in honeybees

Foraging honeybees were trained in a concurrent blocking design with a compound stimulus (AX) reinforced and one of its components (A) either reinforced for a blocking group or nonreinforced for a control group. In Experiment 1, a compound of two colors was used; in Experiment 2, a compound of two odors was used; in Experiment 3, a color-position compound, with position defined in terms of proximity to a distinctive visual landmark, was used; and, in Experiment 4, an odor-position compound was used. In each of the first three experiments, the blocking group responded less than did the control group in a subsequent test with X; in the fourth experiment, the two groups did not differ. The results are in accord with expectations based on those of previous experiments with honeybees in which the independence assumption was found to hold for intermodal compounds but not for intramodal compounds.     

Analysis of choice in honeybees

Free-flying honeybees were trained in a set of four problems to choose between two differently scented targets, one or the other of which contained sucrose solution. The training was simulated quantitatively, always with the same simple linear equations for computing changes produced by reinforcement and nonreinforcement in the strength of association between each target and the sucrose, but with a diverse array of functions for predicting choice on the basis of relative strength. Accuracy of prediction was indexed by the root-mean-square (RMS) deviation of simulated data from real data. The results provide some good approximations of what is tentatively assumed to be the true choice function, setting the stage for further development of the associative features of the model to encompass more complex phenomena of honeybee learning in choice situations.     

Spatial localization of a goal: Beacon homing and landmark piloting by rats on a radial maze

We performed six experiments in order to examine the ability of rats to use moving beacons and landmarks as cues to the location of reward on an eight-arm radial maze. In Experiments 1–4, the cues and goals were moved before each trial, and groups in which a single beacon was placed on the rewarded arm, a single landmark indicated that reward was on the arm immediately to the left of a landmark, or two landmarks were placed on each side of the reward arm were compared. The rats rapidly learned to track the reward in the beacon condition, failed to find the reward sooner than chance expectation with a single landmark, and did only slightly better than chance with two landmarks. In Experiments 5 and 6, the rats were trained in five trials per day, with the landmark and goal locations constant over daily rewarded trials, and in two extinction trials that were inserted among the rewarded trials. The rats found the goal arm at substantially better than chance expectancy with both one and two landmarks. Our results, in agreement with data from recent swimming pool experiments (A. D. L. Roberts & Pearce, 1998), show that rats will use the relationship between moving landmarks and a goal in order to find reward.