A model of working memory capacity in the radial-arm maze task

The radial arm maze is one of the most commonly used tests for assessing working memory in laboratory animals. However, to date, there exists no quantitative method of estimating working memory capacity from performance on this task. Here, we present a mathematical model of performance on the radial arm maze from which we can derive estimates of capacity. We derive explicit results for the two most commonly used measures of performance as functions of number of arms in the maze and memory capacity, assuming a uniform random search. We simulate random non-uniform search strategies. Comparing our model to previous experiments, we show that our model predicts a working memory capacity in the range of 3-9 at the level of performance observed in these experiments. This estimate is within the typical estimate of human working memory capacity. Performance of rats on large mazes (e.g. 48 arms) has been used as evidence that the working memory capacity of rats may be significantly larger than that of humans. We report that memory capacity in the range 3-9 is sufficient to explain the performance of rats in very large radial mazes. Furthermore, when we simulate non-uniform random search strategies observed in the experiments, the resulting estimates do not differ significantly from those assuming a uniform random search. We conclude that a list-based model of working memory with modest capacity is more powerful than previously expected.     

Radial arm maze performance of mice: acquisition and atropine effects

CD-1 mice were successfully trained in a six-arm radial maze in which half of the arms were baited, a procedure which had been used to differentiate between reference and working memory. Stable performance was achieved following eight daily training sessions, as measured by decreasing running time and number of errors. This finding strengthens the foraging hypothesis as a basis for the performance of rodents in the radial maze. Acute subcutaneous administration of the cholinergic antagonist atropine sulfate (1-6 mg/kg) to trained mice produced dose-related increases in running time and working memory errors, with a slight decrease in reference memory errors. This is in agreement with other studies on the role of the cholinergic system in memory processes. The peripheral cholinergic blocker, atropine methyl nitrate (4 mg/kg), somewhat increased running time without decreasing performance accuracy. In contrast to the prolonged pharmacological and physiological effects of atropine, behavioral decrements disappeared within 3 hr. It is concluded that mice trained in the radial arm maze may be used for screening of the effects of drugs on cognitive function.     

A test of the reward-value hypothesis

Rats retain source memory (memory for the origin of information) over a retention interval of at least 1 week, whereas their spatial working memory (radial maze locations) decays within approximately 1 day. We have argued that different forgetting functions dissociate memory systems. However, the two tasks, in our previous work, used different reward values. The source memory task used multiple pellets of a preferred food flavor (chocolate), whereas the spatial working memory task provided access to a single pellet of standard chow-flavored food at each location. Thus, according to the reward-value hypothesis, enhanced performance in the source memory task stems from enhanced encoding/memory of a preferred reward. We tested the reward-value hypothesis by using a standard 8-arm radial maze task to compare spatial working memory accuracy of rats rewarded with either multiple chocolate or chow pellets at each location using a between-subjects design. The reward-value hypothesis predicts superior accuracy for high-valued rewards. We documented equivalent spatial memory accuracy for high- and low-value rewards. Importantly, a 24-h retention interval produced equivalent spatial working memory accuracy for both flavors. These data are inconsistent with the reward-value hypothesis and suggest that reward value does not explain our earlier findings that source memory survives unusually long retention intervals.     

Selective fimbria and thalamic lesions differentially impair forms of working memory in rats.

Several series of experiments were designed to compare the effects of selective lesions of the fimbria or of thalamic nuclei on three different tasks involving working memory in rats: object recognition, place recognition, and the radial arm maze test. The main effects of fimbria lesions were as follows: they produced deficits in the radial maze; object recognition was spared or even facilitated, whereas place recognition was impaired. Electrolytic lesions of either centromedian-parafascicularis (CM-Pf) or dorsomedialis (DM) nuclei produced highly significant deficits in the radial maze test but spared object and place recognition. Ibotenate lesions of the CM-Pf had no effect on any test, which means that the critical structure in the effects of the electrolytic lesions of the CM-Pf was the fasciculus retroflexus (FR). These data may contribute two main points to animal models of hippocampal and thalamic amnesia: (1) different forms of working memory in rats might have different neural bases and (2) the FR may be involved in learning and memory processes.     

A comparison of four corvid species in a working and reference memory task using a radial maze

Birds were tested in an open-room radial maze with learned spatial locations that varied from trial to trial (working memory) and locations that remained spatially stable (reference memory). Three of the species, the Clark's nutcracker (Nucifraga columbiana), pinyon jay (Gymnorhinus cyanocephalus), and Western scrub jay (Aphelocoma coerulescens) store food to varying degrees. The other species, the Eurasian jackdaw (Corvus monedula) does not. Pinyon jays and scrub jays performed better than the nutcrackers and jackdaws in both working and reference memory components of the maze. The pinyon jay and jackdaw performed as would be expected on the basis of their natural history and previous research, but the scrub jay and nutcracker did not. Results are consistent with phylogenetic relationships among the 4 species, but could also be explained by differences in response strategies or interference in processing both types of memory components of the maze.     

Spatial navigation in complex and radial mazes in APP23 animals and neurotrophin signaling as a biological marker of early impairment

Impairment of hippocampal function precedes frontal and parietal cortex impairment in human Alzheimer's disease (AD). Neurotrophins are critical for behavioral performance and neuronal survival in AD. We used complex and radial mazes to assess spatial orientation and learning in wild-type and B6-Tg(ThylAPP)23Sdz (APP23) animals, a transgenic mouse model of AD. We also assessed brain content of nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), and neurotrophin-3 (NT-3). Performance was alike in wild-type and APP23 animals in the radial maze. In contrast, performance in the complex maze was better in wild-type than APP23 animals. Contrary to the wild-type, hippocampal BDNF levels decreased on training in APP23 animals. Hippocampal and frontal cortex NGF levels in APP23 animals correlated with the time to solve the complex maze, but correlated inversely with escape time in wild-type animals. NT-3 levels were alike in wild-type and APP23 animals and were unchanged even after training. Both types of mazes depend on hippocampal integrity to some extent. However, according to the cognitive mapping theory of spatial learning, the complex maze because of the increased complexity of the environment most likely depends more strongly on preserved hippocampal function than the radial maze in the working memory configuration applied here. Greater impairment in complex maze performance than in radial maze performance thus resembles the predominant affliction of the loss of hippocampal function in human AD. NGF and BDNF levels on maze learning are different in wild-type and transgenic animals, indicating that biological markers of AD may be altered on challenge even though equilibrium levels are alike.     

Vasopressin disrupts radial-maze performance in rats

The memory enhancing properties of vasopressin, observed in active and passive avoidance procedures, could derive from its influence on central systems, but may also be mediated by its endocrinological properties. Very little is known about the effects of vasopressin on behavior in procedures other than the active and passive avoidance paradigms. The present experiments were designed to assess the effects of vasopressin on behavior observed in the eight-arm radial maze. In Experiment I, male Wistar rats (N = 7), which had been extensively trained to collect food from all eight arms in a radial maze, were subcutaneously injected with different doses of vasopressin 5 min before the start of the session (0.00, 1.25, 3.75, and 6.25 micrograms/kg). In Experiment II, another group of male Wistar rats (N = 7) received the same doses of vasopressin after having been extensively trained to collect food from four of the eight arms. In both experiments, subjects spent more time in the maze as the dose of vasopressin was increased. Vasopressin also disrupted performance by preventing the subjects from visiting all of the baited arms in the maze. Performance thus decreased, not because of the fact that vasopressin interfered with memory processes, but because of the fact that it produced behavioral inhibition. Thus, if vasopressin affects memory processes, such effects are likely to be mediated through vasopressin's actions on endocrine and behavioral systems, rather than through a direct action on the neural substrate underlying memory functioning.     

Working memory deficits in transgenic rats overexpressing human adenosine A2A receptors in the brain

Adenosine receptors in the central nervous system have been implicated in the modulation of different behavioural patterns and cognitive functions although the specific role of A(2A) receptor (A(2A)R) subtype in learning and memory is still unclear. In the present work we establish a novel transgenic rat strain, TGR(NSEhA2A), overexpressing adenosine A(2A)Rs mainly in the cerebral cortex, the hippocampal formation, and the cerebellum. Thereafter, we explore the relevance of this A(2A)Rs overexpression for learning and memory function. Animals were behaviourally assessed in several learning and memory tasks (6-arms radial tunnel maze, T-maze, object recognition, and several Morris water maze paradigms) and other tests for spontaneous motor activity (open field, hexagonal tunnel maze) and anxiety (plus maze) as modification of these behaviours may interfere with the assessment of cognitive function. Neither motor performance and emotional/anxious-like behaviours were altered by overexpression of A(2A)Rs. TGR(NSEhA2A) showed normal hippocampal-dependent learning of spatial reference memory. However, they presented working memory deficits as detected by performance of constant errors in the blind arms of the 6 arm radial tunnel maze, reduced recognition of a novel object and a lack of learning improvement over four trials on the same day which was not observed over consecutive days in a repeated acquisition paradigm in the Morris water maze. Given the interdependence between adenosinic and dopaminergic function, the present results render the novel TGR(NSEhA2A) as a putative animal model for the working memory deficits and cognitive disruptions related to overstimulation of cortical A(2A)Rs or to dopaminergic prefrontal dysfunction as seen in schizophrenic or Parkinson's disease patients.     

A comparison of the effects of fimbria-fornix, hippocampal, or entorhinal cortex lesions on spatial reference and working memory in rats: short versus long postsurgical recovery period.

Using a radial maze task and different postoperative recovery periods, this experiment assessed and compared the reference and working memory performances of adult Long-Evans male rats subjected to entorhinal cortex, fimbria-fornix, and hippocampus lesions. Sham-operated rats were used as controls. In order to see whether the duration of the postsurgical recovery period would influence acquisition of the complex radial maze task, training began 1 month following surgery (Delay 1) for half the rats in each group, while for the other half training was started 6.5 months following surgery (Delay 2). The results indicated that at both recovery periods the entorhinal cortex lesions failed to affect either working or reference memory in the spatial task. Conversely, both fimbria-fornix and hippocampus lesions impaired both reference and working memory. While the reference memory deficit was generally similar in both fimbria-fornix and hippocampal lesion groups, analysis of the results for working memory indicated that at the longer delay rats with fimbria-fornix lesions were still impaired but in animals that had the hippocampus removed, working memory did not differ from that of controls. These results suggest that there was some recovery in those rats with hippocampal lesions (e.g., on the working memory task) but both hippocampal and fimbria-fornix animals were still impaired compared to controls when training was delayed 6.5 months following the operations.     

Wistar rats with high versus low rearing activity differ in radial maze performance

Substantial work has shown that rats although identical in stock, sex, age, and housing conditions can differ considerably in terms of behavior and physiology. Such individual differences, which can be detected by specific behavioral screening tests, are rather stable, that is, they probably reflect a behavioral disposition or trait. Here, we asked whether and how such differences might affect performance in a task of spatial learning and memory, the radial maze. As in our previous work, we used the degree of rearing activity in a novel open field to assign male adult outbred Wistar rats into those with high versus low rearing activity (HRA/LRA rats). They were then tested in a plus-maze for possible differences in anxiety-related behavior. Finally, and most importantly, they were food deprived and underwent maze training using an 8-arm radial maze with four non-baited and four baited arms. One of these arms consistently contained a larger bait size than the other three. In the open field, HRA rats not only showed more rearing behavior, but also more locomotor activity than LRA rats. In the plus-maze, HRA rats again showed more locomotion, but did not differ in open arm time or percentage of open arm entries, that is, conventional measures of anxiety-related behavior. In the radial maze, HRA rats consistently needed less time to consume all pellets than LRA rats, which was due to faster locomotion on the arms and less time spent at the food pits (especially in baited arms) of HRA rats. During the initial days of training, they were also more efficient in obtaining all food pellets available. Furthermore, HRA rats visited more arms and made relatively less reference memory errors than LRA rats. This allowed them to forage food quickly, but was paralleled by more working memory errors than in LRA rats. In general, working memory errors were more frequent in the arm with the large bait size, but there were no indications that HRA and LRA rats responded differently dependent on reward size. Finally, LRA rats lost slightly more weight than HRA rats during the period of food deprivation. These results are discussed with respect to the role of cognitive and motivational mechanisms, which as subject-inherent factors can contribute substantially to inter-individual variability in the radial maze.     

Memory persistence of rats in a radial maze varies with training procedure

Different estimations of the time-course of spatial working memory have been reported. Some authors found working memory in the radial maze to be relatively long lasting, while others found more rapid exponential decay. In the present experiments it was attempted to account for the conflicting results by investigating the effects of different training procedures. Two types of training were examined under the same circumstances. A group of seven rats was given a series of delayed trials (5, 20, 60, and 120 min). Every delay was repeated at least four times and the delays were presented in an ascending order. The number of errors decreased at every delay except the last one (120 min), where error levels were constant (.50 errors/trial). The good performance was not based on use of intramaze cues or response chaining. In another group of seven rats the same delays were introduced in a quasi-random order and alternated with uninterrupted trials. The number of errors increased exponentially with the length of the delay. However, when this procedure was repeated, the number of errors decreased. These results suggest that training with delayed trials is a major factor to account for the differences in reports of memory persistence in the radial maze.     

Evidence for neocortical involvement in reference memory

Rats were trained on an eight-arm radial maze task using a procedure that provides for an assessment of both working and reference memory. Following training, rats received parietal cortex, medial prefrontal cortex, visual cortex, or nucleus basalis magnocellularis lesions. Rats with visual cortex lesions showed no change in performance on either working or reference memory. Rats with parietal cortex lesions displayed a temporary deficit in reference, but no deficit on working memory. Animals with medial prefrontal cortex lesions showed a temporary deficit on both working and reference memory. Rats with extensive lateral frontal and parietal cortex depletion of acetylcholinesterase following nucleus basalis magnocellularis lesions had a marked disruption only of reference but not of working memory. It is concluded that neocortex and possibly the cholinergic projections to neocortex play an important role in mediating reference memory.     

Different types of environmental enrichment have discrepant effects on spatial memory and synaptophysin levels in female mice

Environmental enrichment paradigms that incorporate cognitive stimulation, exercise, and motor learning benefit memory and synaptic plasticity across the rodent lifespan. However, the contribution each individual element of the enriched environment makes to enhancing memory and synaptic plasticity has yet to be delineated. Therefore, the current study tested the effects of three of these elements on memory and synaptic protein levels. Young female C57BL/6 mice were given 3h of daily exposure to either rodent toys (cognitive stimulation) or running wheels (exercise), or daily acrobatic training for 6 weeks prior to and throughout behavioral testing. Controls were group housed, but did not receive enrichment. Spatial working and reference memory were tested in a water-escape motivated radial arm maze. Levels of the presynaptic protein synaptophysin were then measured in frontoparietal cortex, hippocampus, striatum, and cerebellum. Exercise, but not cognitive stimulation or acrobat training, improved spatial working memory relative to controls, despite the fact that both exercise and cognitive stimulation increased synaptophysin levels in the neocortex and hippocampus. These data suggest that exercise alone is sufficient to improve working memory, and that enrichment-induced increases in synaptophysin levels may not be sufficient to improve working memory in young females. Spatial reference memory was unaffected by enrichment. Acrobat training had no effect on memory or synaptophysin levels, suggesting a minimal contribution of motor learning to the mnemonic and neuronal benefits of enrichment. These results provide the first evidence that different elements of the enriched environment have markedly distinct effects on spatial memory and synaptic alterations.     

Hierarchical structures: chunking by food type facilitates spatial memory

Three experiments assessed the ability of male Sprague-Dawley rats to organize the spatial locations of different food types in a hierarchical manner to maximize the efficiency of working memory. Independent groups were exposed, on a 12-arm radial maze, to baiting arrangements varying in the stability of the pattern and type of food used as bait. Training rats with stable, differentiable baiting arrangements produced increased accuracy in choice performance, hierarchically ordered patterns of choice selection, slower growth of proactive interference when trials were massed, and the learning of the geometrical relations among food types independent of other extramaze cues. Such findings are strong evidence of the rat's ability to encode and use local cues for navigation, based on properties of the reinforcer. The application of a chunking strategy may provide for more efficient use of working memory by facilitating information storage, recall, or resetting mechanisms.     

Memory systems in the rat: effects of reward probability,context, and congruency between working and reference memory

The interaction of working and reference memory was studied in rats on an eight-arm radial maze. In two experiments, rats were trained to perform working memory and reference memory tasks. On working memory trials, they were allowed to enter four randomly chosen arms for reward in a study phase and then had to choose the unentered arms for reward in a test phase. On reference memory trials, they had to learn to visit the same four arms on the maze on every trial for reward. Retention was tested on working memory trials in which the interval between the study and test phase was 15 s, 15 min, or 30 min. At each retention interval, tests were performed in which the correct WM arms were either congruent or incongruent with the correct RM arms. Both experiments showed that congruency interacted with retention interval, yielding more forgetting at 30 min on incongruent trials than on congruent trials. The effect of reference memory strength on the congruency effect was examined in Experiment 1, and the effect of associating different contexts with working and reference memory on the congruency effect was studied in Experiment 2.     

Effects of neocortical ectopias and environmental enrichment on Hebb-Williams maze learning in BXSB mice

Approximately 40-60% of BXSB mice have neocortical ectopias, a developmental anomaly characterized by migration of neurons into the neuron-sparse layer I of cortex. Previous studies have shown that ectopic BXSB mice have superior reference, but inferior working, memory on spatial tasks. Female BXSB mice were housed either in an enriched environment or in standard cages at weaning. Subsequently, these animals were tested on four of the Hebb-Williams mazes in a water-based version of this maze. Theoretically, two of the maze configurations placed greater emphasis on reference memory to find the goal, whereas the other two favored working memory. Ectopics reared in standard housing conditions were better than nonectopics on mazes that favored the use of reference memory, but poorer on mazes that favored working memory. In contrast, subjects raised in the enriched environment showed no ectopia differences. A comparison of enriched and standard housing conditions found that the enriched animals had better reference memory but poorer working memory. The latter effect may be because the enriched environment, although more stimulating, did not change in time or space; and other researchers have shown that daily replacement of stimuli in complex environments is correlated with better working memory.     

Differential Effects of Global Ischemia on Delayed Matching- and Non-Matching-to-Position Tasks in the Water Maze and Skinner Box

In order to assess effects of global ischemia in tasks of spatial learning and working memory, male Wistar rats were subjected to four vessel occlusion (4 VO) for periods of 5, 10, and 20 min and compared with sham-operated controls over four test phases, from 6 to 54 weeks after surgery. Rats were assessed on acquisition in the water maze, a task that is sensitive to ischemic impairments, before testing in Skinner box and water maze working memory tasks, which both require the short-term storage of information, but make different demands on spatial information processing. Phases 1 and 3 assessed spatial learning in a standard water maze procedure (12 and 10 training days, 2 trials/day with a 10-min intertrial interval: ITI). Phase 2 involved training and testing in delayed non-matching-to-position task in the Skinner box, with delays of 2–10 s between the information and choice stages. Phase 4 examined working memory in a water maze delayed matching-to-position task with 4 trials/day, an ITI of 30 s, and a novel platform position on each day. Ischemic rats showed duration-related impairments in water maze acquisition and working memory, but not in the less spatially demanding Skinner box task. Since water maze acquisition deficits were seen both before and after testing in the Skinner box the lack of effect cannot be attributed to time or to prior training. Ischemic deficits were more marked in Phase 3 than in Phase 1 of acquisition, suggesting that impairment may be progressive. Histological assessment showed that cell loss was largely confined to the hippocampal CA1 field and was linearly related to duration of occlusion. At the maximal level of loss (5.7 mm before the interaural line) the 20-min group showed 90% loss, the 10-min group 60% loss, and the 5-min group, which did not differ from controls, less than 10% loss. Only the 20-min group showed significant damage beyond the CA1 field, ranging from 30–40% loss in the CA3 field to 5% loss in one striatal area. No cortical damage was seen. The extent of CA1 cell loss correlated modestly with water maze acquisition (Phase 3) and working memory scores, but not with trials to criterion in the Skinner box task. There were significant correlations between different measures both within and between water maze tasks, but not Skinner box tasks, suggesting that the two types of procedure engaged different cognitive processes. The results indicate that the intrahippocampal damage induced by 4 VO impaired tasks which required processing of allocentric spatial information, but did not impair the storage of limited spatial information in working memory.     

Preserved spatial memory in old rats survives 10 months without training

Aged rats with extensive prior training on the radial maze retain the capacity for accurate spatial working memory (WM) for at least 3 months without practice. To investigate the temporal limits of this influence of prior experience we compared the reacquisition of spatial WM by a group of experienced 21.5-month-old rats to the original acquisition by naive 3-month-old rats. The aged rats had received 225 radial maze tests between 3 and 11 months of age. Despite 10 months without practice the old rats rapidly reacquired critical performance. Their reacquisition was markedly superior to original learning by the young rats, even when delays as long as 5 h were imposed between the rats' fourth and fifth choices during the daily tests in the eight-arm maze. Additional tests showed that neither young nor old rats employed a response strategy to maintain accurate spatial WM performance. Experience clearly confers long-lived protection against the otherwise deleterious effects of aging on spatial WM, but the mechanism by which this influence arises is unknown.     

Working memory theory of hippocampal function needs qualification

To compare the predictive value of "cognitive map" and "working memory" theories of hippocampal function, the performance of rats with dorsal hippocampal lesions was compared to that of control rats in a series of experiments. In Experiment I, experimental rats learned a spatial alternation task with normal ease, but in Experiment II, they were significantly impaired on an elevated 8-arm radial maze. In Experiment III, the performance of the same experimental and control rats was compared on two versions of a 16-arm enclosed radial maze. In the first version, carpet inserts served as cues to mark eight unbaited arms and each of the remaining arms contained one food pellet. While both experimental and control rats successfully avoided the set of cued arms, experimental rats reentered uncued baited arms more frequently than did control rats. In the second version no intramaze cues were provided, but the spatial distribution of baited and unbaited arms remained the same as that used in the first version. In this uncued version, experimental rats both entered unbaited arms and reentered baited arms more frequently than did control rats, i.e., they were impaired in both "reference" and "working" memory. These findings are compatible with the hypothesis that hippocampal lesions result in an impaired capacity to form cognitive maps but they are not compatible with the working memory hypothesis. Furthermore, twelve separate evaluators classed experimental rats as using fewer mapping and more orientation strategies than control rats in the 8-arm maze.