No full-text available
To read the full-text of this research,
you can request a copy directly from the author.
Long delay learning in the T-maze
Abstract
Rats were trained to go to one side of a T-maze with delays of reward lasting 1, 20, or 60 min in Expt 1 and 1 or 60 min in Expt 2. Mediation by secondary reward was prevented by administering the same delay treatment regardless of whether the response was correct or incorrect: after a response, the rat was removed from the choice alley and placed in its home cage to spend the delay. Feedback for the response was given in the startbox after the delay interval ended. The rats learned and there were no significant differences in performance among groups trained with different delays. These results had been expected on the basis of Revusky's (1971) hypothesis that removal of the rat from the learning situation to spend the delay elsewhere facilitates long delay learning by reducing associative interference. In Expt 3, this notion was tested explicitly by varying the amount of a 2-min delay to be spent in the experimental situation. Different groups of rats were left in the choice alley after the response for 0, 15, or 60 sec; then the rats were removed to spend the remainder of the 2-min delay in the home cage As predicted, the level of performance decreased as the length of time in the choice alley was increased.
... The elegance of Grice's work convinced most psychologists that the problem of delayed reinforcement had at last been resolved, with the result that little further research was Copyright 1979 by the American Psychological Association, Inc. 0097-7403/79/0503-0224$00.75 carried out in this area over subsequent decades (but see Lawrence & Hommel, 1961). Recently, however, Lett (1973Lett ( , 1974Lett ( , 1975 reported dramatic evidence that even with secondary reinforcement controlled, reinforcement may still be effective after delays of 1 hr. The rationale for Lett's research was based on an analysis of delayed reward proposed by Revusky (1971). ...
... It was this prediction that Lett successfully tested. When rats were removed from a Tmaze immediately after making a choice response, Lett found that they could still solve visual discriminations even when reinforcement was delayed for 1 min (Lett, 1974) and could still solve spatial discrimination even after delays of 1 hr (Lett, 1975). Lett's results thus provided impressive confirmation for Revusky's predictions, providing support not only for the general concepts of an interference analysis but also for the specific principle of situational relevance. ...
... If so, then any responses made in the side arm during the delay interval would have been more likely to be reinforced than responses in the choice box, and hence to interfere with learning. The poorer learning shown by subjects in this group, therefore, could simply be a reflection of the greater proportion of the delay interval they spent in the side arm (see, e.g., Lett, 1975). ...
In 4 experiments, male or female PVG hooded rats were trained on spatial discriminations in which reward was delayed for 1 min. Exp I tested B. T. Lett's hypothesis that responses made in the home cage during the delay interval are less likely to interfere with learning than responses made in the maze. Experimental Ss were transferred to their home cages during the delay interval, and controls were picked up but then immediately replaced in the maze. Contrary to Lett's hypothesis, both groups learned. Further experiments suggested that handling following a choice response was the crucial variable in producing learning: No learning occurred when handling was delayed (Exp II) or omitted (Exp III). One possible explanation for the fact that handling facilitated learning is that it served to mark the preceding choice response in memory so that Ss were then more likely to recall it when subsequently reinforced. In accordance with this interpretation, learning was found to be just as strong when the choice response was followed by an intense light or noise as by handling (Exp IV). The implication of marking for other phenomena such as avoidance, quasi-reinforcement, and the paradoxical effects of punishment is also discussed. (20 ref)
... Marking was first introduced into the animal learning literature by Lieberman et al. (1979) to explain a series of experiments in which rats learnt a spatial discrimination in aT-maze with delayed reinforcement. The initial aim of their research was to replicate Lett's (1975) finding that rats could learn such discrimination, provided that they were removed immediately from the maze after each choice response and returned to the home cage for the duration of the delay. Lett (1975) proposed that her findings supported Revusky's situational relevance theory (e.g., Revusky, 1971), that organisms will easily associate events that occur in the same physical environment, but not those occurring in different environments. ...
... The initial aim of their research was to replicate Lett's (1975) finding that rats could learn such discrimination, provided that they were removed immediately from the maze after each choice response and returned to the home cage for the duration of the delay. Lett (1975) proposed that her findings supported Revusky's situational relevance theory (e.g., Revusky, 1971), that organisms will easily associate events that occur in the same physical environment, but not those occurring in different environments. For example, if a rat is given food in a T-maze, it would be likely to associate that food with previous responses in the maze. ...
... If, however, the rat is moved outside the maze, it would be less likely to associate the food with any subsequent responses. Thus, in Lett's (1975) study, the rats would be able to learn the responsereinforcer association because home cage removal reduced the interference created by other responses during the delay. In other words, any home cage responses made during the interval would be unlikely to become associated with food and hence would not interfere with learning. ...
p>It has long been known that the impairment to discrimination learning caused by brief delays to reinforcement can be counteracted by the response-contingent presentation of a conditioned reinforcer during the delay interval following a correct response (Spence, 1947). More recently, it has been shown that reinforcement delay can also be overcome using response-marking procedures, in which the same stimulus contingently follows both correct responses and errors (e.g. Lieberman, McIntosh & Thomas, 1979). This thesis examined the effects of response-marking procedures on human learning of conditional discrimination tasks with delayed reinforcement.
Experiments One to Three employed single case experimental designs (alternating treatments) to evaluate the effect of response marking during matching-to-sample tasks with delayed reinforcement, using children with autism as participants. Experiment One showed that for both marking and conditioned reinforcement supported acquisition of conditional discrimination performance over a 5 s delay, although the latter appeared more efficient. Experiment Two, however, showed that – with more effective techniques – both procedures were equally effective, and that both were more effective than a control in which no response-contingent stimuli occurred during the delay. Experiment Three compared the standard marking procedure with a novel marked-before procedure in which all sample stimuli were marked before a matching response was made. Both procedures produced very similar acquisition rates, and both were more effective in establishing conditional discrimination than a delay only control.
Experiments four to Seven employed group comparison designs to compare marking against conditioned reinforcement, delay and immediate reinforcement using adult humans in a laboratory version of the matching-to-sample task. Marking effects were found only in Experiment Seven, when the confounding effects of verbal behaviour were adequately controlled.
Overall, the findings indicated that response-marking procedures may be effective with human participants but that their effects are more reliable in applied settings with children than in laboratory settings with adults.</p
... The reward was again presented in the start area of the maze, and subjects spent the delay interval in the home cage. A later experiment (Lett, 1975, Experiment 3) evaluated the importance of placing subjects in the home cage for the delay interval. Some subjects were confined to the chosen arm of the T maze for the first 15 or 60 sec of a 2-min delay interval before being placed in the home cage for the rest of the interval. ...
... Although the findings of Lett (1973Lett ( , 1974Lett ( , 1975 are consistent with the concurrent interference theory, the interpretation of these results has been called into question by subsequent investigators. With certain refinements of the procedures that had been used by Lett (1974), Roberts (1976) failed to find convincing evidence that rats can learn a black-white visual discrimination with 1-min delayed reinforcement. ...
... Where they spent the delay interval was much less, if at all, important. These results suggest that the poor learning Lett (1975) obtained when she allowed subjects to remain in the chosen arm of the maze during part of the delay interval was not due to remaining in the maze but was caused by the fact that the subjects were not handled immediately after making a choice response. ...
This chapter reviews the empirical and theoretical analyses of the phenomena that have taken place because of the issue of biological constraints. Recent studies of biological constraints on instrumental and classic conditioning have stimulated the modifications of previously accepted ideas about learning. They have greatly increased the range of factors that are considered to be relevant to predicting learning in various situations. However, such studies have not required abandonment of the pursuit of general theories of learning. Recent experiments discussed in the chapter illustrate that a detailed empirical investigation of a biological constraint on learning often leads to the formulation of new general principles of behavior. Although positive reinforcement is remarkably effective in increasing the probability of a variety of responses, notable exceptions to this general rule have been evident from the inception of research on instrumental conditioning. Some of the most dramatic examples of constraints on instrumental conditioning are found in avoidance learning. Depending on the kind of response that is required to prevent aversive stimulation, avoidance learning may occur very rapidly, at an intermediate rate, or very slowly.
... Further, it seems obvious that memories must be subject to interference outside of the apparatus if the extra-apparatus situation is at all like the experimental apparatus. This notion has recently been explored by Lett (1975) in an appetitive position response task in the T-maze (see also Lett, 1973Lett, , 1974. Lett (1975) trained rats in the T-maze under conditions of long delay of reinforceraent (up to 60 rainutes in Exp. ...
... This notion has recently been explored by Lett (1975) in an appetitive position response task in the T-maze (see also Lett, 1973Lett, , 1974. Lett (1975) trained rats in the T-maze under conditions of long delay of reinforceraent (up to 60 rainutes in Exp. I) and found no differences between groups given 1, 20, or 60 minutes delay of reinforcement. ...
... The Lett (1975) study (Exp. III) is of particular importance to the present arguments. ...
ACKNOWLEDGMENTS I wish,to acknowledge,my gratitude,to Dr. Dennis,Clark,Cogan my major,advisor,and,the,director,of this,dissertation. ,Dr. Cogan has,been,a continuing,source,of inspiration,and,guidance,throughout the,past,four,years,and,I owe most,of my,thinking,about,Psychology and,professional,activities,to him. ,Thanks,are,also,due,to Drs. Robert Bell, Bill Locke, Philip Marshall, and William Portnoy for their,service,as members,of the,committee. 11 TABLE OF CONTENTS ACKNOWLEDGMENTS,ii LIST OF TABLES ,v LIST OF FIGURES ,vi CHAPTER I. INTRODUCTION,1 The Experiment ,8 Experimental Design ,i 9 Extinction Predictions ,9 Aftereffects Position ,10
... Fourteen of the 16 subjects fixated on the arm to which they went when first shocked, which seems consistent with elicitation theory. Lett (1973Lett ( , 1975Lett ( , 1979 has demonstrated what appears to be long-delay learning in the T-maze, and she has used this finding to argue against traditional theories of instrumentallearning (e.g., Grice, 1948). The obvious way to account for Lett's results is to postulate that reinforcement can be effective after much longer delays than had heretofore been thought. ...
... The apparatus was modeled closely after the T-maze used by Lett (1975). It was an enclosed T with a gray stem 15.75 cm high, 9.8 cm wide, and 58.3 cm long. ...
If rats in a T-maze made an arbitrarily designated correct response, that response was the terminal trial of the day. If, instead, they made the other incorrect response, they were shocked on the next trial whenever they failed to choose either arm of the maze within 3 sec. All rats quickly learned to choose one arm, but the incorrect arm was chosen as often as the correct. Thus the original hypothesis that escape from shock in conjunction with the tendency to alternate would mediate the learning of the correct response was not supported. However, the animals that learned the incorrect response had a much weaker tendency to alternate on shock trials than had the learners of the correct response. Fourteen of the 16 subjects fixated on the arm to which they went when first shocked, which seems consistent with elicitation theory.
... This result also caused a stir, because the animals choose accurately even if removed from the maze for several hours after entering the first four arms-a delay much longer than that sustainable under conventional delay-of-reinforcement procedures. In fact, the rats' ability to remember across a long delay is not as puzzling as it may seem, given the lengthy mtertrial interval in these experiments as well as other differences between this procedure and conventional delay-ofreinforcement procedures (Lett, 1975;Staddon, 1983Staddon, , 1985. Thus, the biggest puzzle posed by radial-maze experiments is why the animals are so accurate in the first place. ...
Cognitive behaviorist E. C. Tolman (1932) proposed many years ago that rats and men navigate with the aid of cognitive maps, but his theory was incomplete. Critic E. R. Guthrie (1935) pointed out that Tolman's maps lack a rule for action, a route finder. We show that a dynamic model for stimulus generalization based on an elementary diffusion process can reproduce the qualitative properties of spatial orientation in animals: area-restricted search in the open field, finding shortcuts, barrier learning (the Umweg problem), spatial “insight” in mazes, and radial maze behavior. The model provides a behavioristic reader for Tolman's cognitive map.
... To examine these capabilities, a valuable learning experiment from neuroscience tests the cognitive capabilities of rodents in a Tmaze. The T-maze can be described as an enclosed structure that takes the form of a horizontally-placed T (Lett, 1975;Wenk, 1998;Dudchenko, 2001;Deacon and Rawlins, 2006;Engelhard et al., 2019), with the maze beginning at the base of T and ending at either side of the arms, see Figure 1. The rodent moves down the base of the maze and chooses either side of the arms. ...
We propose that in order to harness our understanding of neuroscience toward machine learning, we must first have powerful tools for training brain-like models of learning. Although substantial progress has been made toward understanding the dynamics of learning in the brain, neuroscience-derived models of learning have yet to demonstrate the same performance capabilities as methods in deep learning such as gradient descent. Inspired by the successes of machine learning using gradient descent, we introduce a bi-level optimization framework that seeks to both solve online learning tasks and improve the ability to learn online using models of plasticity from neuroscience. We demonstrate that models of three-factor learning with synaptic plasticity taken from the neuroscience literature can be trained in Spiking Neural Networks (SNNs) with gradient descent via a framework of learning-to-learn to address challenging online learning problems. This framework opens a new path toward developing neuroscience inspired online learning algorithms.
... To examine these capabilities, a valuable learning experiment from neuroscience tests the cognitive capabilities of rodents in a T-maze. The T-maze can be described as an enclosed structure that takes the form of a horizontally-placed T (50)(51)(52)(53)(54), with the maze beginning at the base of T and ending at either side of the arms, see Figure 1. The rodent moves down the base of the maze and chooses either side of the arms. ...
We propose that in order to harness our understanding of neuroscience toward machine learning, we must first have powerful tools for training brain-like models of learning. Although substantial progress has been made toward understanding the dynamics of learning in the brain, neuroscience-derived models of learning have yet to demonstrate the same performance capabilities as methods in deep learning such as gradient descent. Inspired by the successes of machine learning using gradient descent, we demonstrate that models of neuromodulated synaptic plasticity from neuroscience can be trained in Spiking Neural Networks (SNNs) with a framework of learning to learn through gradient descent to address challenging online learning problems. This framework opens a new path toward developing neuroscience inspired online learning algorithms.
... To examine these capabilities, a valuable learning experiment from neuroscience tests the cognitive capabilities of rodents in a T-maze. The T-maze can be described as an enclosed structure that takes the form of a horizontally-placed T [50,51,52,53,54], with the maze beginning at the base of T and ending at either side of the arms, see Figure 1. The rodent moves down the base of the maze and chooses either side of the arms. ...
We propose that in order to harness our understanding of neuroscience toward machine learning, we must first have powerful tools for training brain-like models of learning. Although substantial progress has been made toward understanding the dynamics of learning in the brain, neuroscience-derived models of learning have yet to demonstrate the same performance capabilities as methods in deep learning such as gradient descent. Inspired by the successes of machine learning using gradient descent, we demonstrate that models of neuromodulated synaptic plasticity from neuroscience can be trained in Spiking Neural Networks (SNNs) with a framework of learning to learn through gradient descent to address challenging online learning problems. This framework opens a new path toward developing neuroscience inspired online learning algorithms.
... The first of these two alternative accounts-the Mnemonic Association Hypothesis 19 (see also 6 )-suggests that jays' strategic caching behaviour may be the result of long-delay learning, guided by what the birds have learned at the time of cache recovery 23 . At the time of the outcome (e.g., at retrieval of previously made caches), jays may have recalled their previous actions, and thus created a positive association with the specific action that resulted in a more beneficial outcome, i.e., the action of caching a specific type of food in that specific location at a specific time. ...
Previous research reported that corvids preferentially cache food in a location where no food will be available or cache more of a specific food in a location where this food will not be available. Here, we consider possible explanations for these prospective caching behaviours and directly compare two competing hypotheses. The Compensatory Caching Hypothesis suggests that birds learn to cache more of a particular food in places where that food was less frequently available in the past. In contrast, the Future Planning Hypothesis suggests that birds recall the 'what-when-where' features of specific past events to predict the future availability of food. We designed a protocol in which the two hypotheses predict different caching patterns across different caching locations such that the two explanations can be disambiguated. We formalised the hypotheses in a Bayesian model comparison and tested this protocol in two experiments with one of the previously tested species, namely Eurasian jays. Consistently across the two experiments, the observed caching pattern did not support either hypothesis; rather it was best explained by a uniform distribution of caches over the different caching locations. Future research is needed to gain more insight into the cognitive mechanism underpinning corvids' caching for the future.
... The Mnemonic Association Hypothesis ( [17]; see also [6]) suggests that jays' strategic caching behaviour may be the result of long-delay learning, guided by what the birds have learned at the time of cache recovery [21]. At the time of the outcome (e.g. at retrieval of previously made caches), jays may have recalled their previous actions, and thus created a positive association with the specific action that had resulted in a more beneficial outcome, i.e. the action of caching a specific type of food in that specific location at a specific time. ...
Previous research reported that corvids preferentially cache food in a location where no food will be available or cache more of a specific food in a location where this food will not be available. Here, we consider possible explanations for these prospective caching behaviours and directly compare two competing hypotheses. The Compensatory Caching Hypothesis suggests that birds learn to cache more of a particular food in places where that food was less frequently available in the past. In contrast, the Future Planning Hypothesis suggests that birds recall what-when-where features of specific past events to predict the future availability of food. We designed a protocol in which the two hypotheses predict different caching patterns across different caching locations such that the two explanations can be disambiguated. We formalised the hypotheses in a Bayesian model comparison and tested this protocol in two experiments with one of the previously tested species, namely Eurasian jays. Consistently across the two experiments, the observed caching pattern did not support either hypothesis; rather it was best explained by a uniform distribution of caches over the different caching locations. Future research is needed to gain more insight into the cognitive mechanism underpinning corvids' caching for the future.
... Regardless of whether the choice was correct or incorrect, the rat was then removed and placed in its home cage for the appropriate delay interval. 18 Each rat repeated the trial six times. ...
Background
Hematoma is the chief culprit in brain injury following intracranial cerebral hemorrhage (ICH). Noninvasive hematoma clearance could be an option to prevent and alleviate early brain injury after ICH. Peroxisome proliferator-activated receptor γ (PPAR-γ) and nuclear factor-erythroid 2 related factor-2 (Nrf2) facilitate removal of hematoma in ICH. Monascin acts as the natural Nrf2 activator with PPAR-γ agonist, and the long-term effects of monascin following ICH have not been elucidated.
Methods
ICH in rats was induced by stereotactic, intrastriatal injection of type IV collagenase. Monascin was administered twice daily by gastric perfusion for 14 days after ICH induction. Long-term neurological scores (T maze, Garcia scales, rotor rod test, and Morris water maze), hematoma volume, as well as iron overload around hematoma and brain atrophy were evaluated at 7, 14, and 28 days after ICH.
Results
The results showed that monascin improved long-term neurological deficits, spatial memory performance, learning ability, and brain shrinkage after ICH. Monascin also reduced hematoma volume at 7 days and iron content at 7 and 14 days after ICH.
Conclusion
PPAR γ and Nrf2 play a crucial role in hematoma clearance after ICH in rat. As a dual agonist of PPAR γ and Nrf2, monascin improved long-term outcomes by facilitating hematoma clearance, and by attenuating iron overload and brain atrophy after experimental ICH.
... Revusky (1971Revusky ( , 1977 integrates these findings into what he describes as a more general associative interference theory. This theory inspired Lett (1973Lett ( , 1974Lett ( , 1975Lett ( , 1977 to demonstrate that a rat is capable of bridging a fairly large time interval between a discriminative stimulus and reinforcementand this with more conventional procedures. For this to happen, however, the situation must be designed so that the animal is urged to again "call to mind" the discriminative stimulus during reinforcement. ...
This manuscript is part of a special issue to commemorate professor Paul Eelen, who passed away on August 21, 2016. Paul was a clinically oriented scientist, for whom learning principles (Pavlovian or operant) were more than salivary responses and lever presses. His expertise in learning psychology and his enthusiasm to translate this knowledge to clinical practice inspired many inside and outside academia. Several of his original writings were in the Dutch language. Instead of editing a special issue with contributions of colleagues and friends, we decided to translate a selection of his manuscripts to English to allow wide access to his original insights and opinions. Even though the manuscripts were written more than two decades ago, their content is surprisingly contemporary. The present manuscript was originally published as part of a Liber Amicorum for Paul Eelen’s own supervisor, prof. Joseph Nuttin. In this chapter, Paul Eelen presents a modern view on Pavlovian learning. It appeared in 1980, at the heyday of cognitive psychology which initially dismissed conditioning. Paul Eelen’s perseverance in presenting learning principles as key to study human behaviour has proven correct and ahead of time.
First published as: Eelen, P. (1980). Klassieke conditionering: Klassiek en toch modern. In Liber Amicorum, Prof. J. R. Nuttin, Gedrag, dynamische relatie en betekeniswereld (pp. 321–343). Leuven: Universitaire Pers Leuven.
... In adulthood, the rats bred in those three settings were subjected to a T-maze test, which is a versatile test paradigm capable of identifying deficits in both cognition and emotion (e.g. Deacon and Rawlins, 2006;Roder and Roder, 1996;Graeff and Netto, 1998;Lett, 1975). It has been used to observe spatial learning (e.g. ...
Stress associated with social isolation in early life can lead to disturbances in the emotional regulation in adult rats. However, there are no reports on the impact of isolation from the mother while providing contact with peers. Under such conditions, young individuals have the opportunity to interact with others, are able to develop social behaviour, etc. Yet, there is no stimulation and care provided by the mother. We examined the relative impact of maternal contact and sibling contact in the rarely studied pre-juvenile (3rd and 4th week post birth) period on subsequent development. An experiment was designed to compare the impact of different social environments on the animals' behaviour in adulthood. There were three breeding conditions: young with mother, young with peers, and standard breeding conditions. Adult rats were subjected to a T-Maze test to measure the level of exploratory behaviour. Spatial learning was assessed by placing water bottles in the side corridors. The analysis revealed that a distorted environment during the development process has a negative impact on the rats' emotional regulation and a subtle effect on related aspects of adaptive behaviours (i.e. exploration). In the pre-juvenile period, to some degree, contact with peers may be complementary to the mother's influence.
... Such a planning mechanism does not work, however, with respect to past actions not directed towards the future reward: in this case, selecting the immediate reward. Then, the animal must already know that the selection event is future related, otherwise it could have associated any action, at any time, that failed to lead to the reward [11]. So, if mnemonic associations were used, the ravens must have known the item-outcome relationship before the self-control experiments, making our control valid anyway. ...
... One trial learning is possible with other methods, particularly when the US is intense (e.g., Mahoney & Ayres, 1976) which is perfectly consistent with general theories of learning (e.g., Rescorla & Wagner, 1972). The ability of taste aversion learning to survive a long delay is explained, at least in part, by the observation that in taste aversion no other relevant stimulus occurs between the flavor and illness during the long delay, and experiments have shown other forms of learning to survive a long delay between pairings when other potentially relevant stimuli are controlled (e.g., Lett, 1975). In the end, taste aversion learning led to huge advances in our understanding of a great many things such as hedonic shift, learned safety, within-compound associations, and the role of evolution in learning. ...
... To obtain evidence in support of this claim, Revusky ran a number of experiments on inter-trial learning and was encouraged by results obtained by Lett indicating that rats could acquire a discrimination, e.g., between black and white goal boxes, even when their choice of the correct box was rewarded 8 min later -as long as they were removed from the maze during this period (e.g., Lett, 1975). However, the results Revusky obtained from his own experiments were not impressive (Revusky, 1977, pp. ...
In a serial overshadowing procedure a target stimulus, A, is followed after an interval by a potentially interfering stimulus, B, and this is then followed by an unconditioned stimulus, US. Revusky (1977) proposed that the degree to which B overshadows conditioning of A depends on whether or not the two events take place in the same context. To test this proposal two experiments used a 1-trial long-delay conditioned taste aversion (CTA) procedure; sucrose served as the target taste (A) and dilute hydrochloric acid (HCl) as the overshadowing taste (B), with lithium chloride injection providing the US. In Experiment 1 these tastes were novel; weaker overshadowing by HCl of an aversion to sucrose was found when the two tastes were presented in different contexts. Experiment 2 tested whether the effect of pre-exposure to HCl, thereby rendering it less effective in overshadowing a sucrose aversion, was also context-dependent. In the conditioning session rats again received either context-same or context-different presentations of sucrose and HCl. However, for some rats HCl was pre-exposed in the same context to which it was later presented during conditioning (Consistent), while others were pre-exposed to HCl in a different context to the one in which it was presented during conditioning (Inconsistent). The Inconsistent group produced greater overshadowing than the Consistent group and thus confirmed that the latent inhibition effect was also context dependent. This study supports Revusky’s (1977) idea of situational relevance.
... In these cases, a novel taste could precede illness by hours, resulting in subsequent aversion to that taste. It was important that other candidate causes not interpose between the stimulus or response and that reinforcer (Lett, 1975; Williams, 1975), since the latter, more proximate event could interfere with the acquisition of control by the former. Contingency matters because it protects proximity: " The role of conditionality [contingency] in protecting a given response from being displaced by the reinforcement of some other response—a response perhaps more prevalent in the animal's repertoire—may be one of the most important factors [in performance] " (Jenkins, 1970, p. 101), a comment echoed by Mackintosh (1974, pp. ...
... Cheke and colleagues (2011b) argue that it is not clear why speed of learning should be used as evidence for type of learning; associative learning can occur within a single trial or take hundreds of trials to appear. By standard operant learning criteria, the action and reinforcer in this study (food choice and return of water) were not sufficiently temporally contiguous to allow associative learning, nor were conditions met for "long delay" associative learning (Lett, 1975). However, the results of this study must nonetheless be treated with caution since they present the behaviour of only two subjects. ...
Episodic Cognition (or “Mental Time Travel”) is the ability to mentally re-experience events from our personal past and imagine potential events from our personal future. This capacity is fundamental to our lives and has been argued to be uniquely human. The aim of this thesis is to use behavioural tasks developed in comparative cognition to integrate both the literature on different research subjects (animals, children, adults, patients) but also from different theoretical perspectives, with the hope of facilitating communication and comparison between these fields.
The backbone of the thesis is the behavioural tasks themselves, along with their origins in theory. Specifically, the “What-Where-When”, “Unexpected Question” and “Free Recall” episodic memory tasks and the “Bischof-Köhler” test of episodic foresight. Each of these tasks stems from different theoretical approaches to defining episodic cognition. Whilst extensively studied, these four tasks have never been undertaken by the same subjects and have never been directly compared. It is thus unclear whether these different theoretical perspectives converge on a single “episodic cognition” system, or a variety of overlapping processes. This thesis explores these issues by presenting these tasks to previously untested animal (the Eurasian Jay), developing children (aged 3-6), and a sample of human adults (Cambridge Undergraduates). Finally, these findings are applied in the assessment of episodic cognition in a population that is thought to have mild hippocampal damage – the overweight and obese.
It was predicted that if all these putative tests of episodic cognition were tapping into the same underlying ability, then they should be passed by the same animal species, develop at the same time in children, correlate in human adults and be impaired in those with damage to the relevant brain areas. These predictions were, to some degree, confirmed. While the novel animal model could not be tested on all paradigms, the jays performed well on Bischof-Köhler future planning test. However, the results of the What-Where-When memory test were equivocal. There was a relatively low degree of correlation between performance on all the tasks in human children, along with a suggestion that each had a distinct developmental trajectory. The study of human adults revealed that while performance on all the tasks were related to one another, this relationship was often nonlinear, suggesting the contribution of several different psychological processes. Finally, it was found that both memory and performance on the Bischof-Köhler future planning task were altered in individuals who are overweight. A potentially surprising theme throughout the results is that performance on the Bischof-Köhler tasks is in fact negatively related to performance on memory tests, and improves in patients thought to have mild hippocampal damage.
It is concluded that there may be a significant degree of overlap in the processes tapped by different putative tests of episodic memory, but that they can not be considered to be equivalent. Furthermore, it is suggested that episodic cognition is a fundamentally ineffective system with which to predict future motivational states, because it is biased by current feelings.
... After making a response, whether correct or incorrect, rats were removed from the experimental apparatus and placed in their home cages for a certain period before being returned to the apparatus, where they received food. Lett (1973) found that response-reward delays of up to 8 minutes did not impair performance, and in subsequent studies obtained a similar result with delays of up to 20 minutes using the T-maze (Lett, 1975) and up to 1 minute in a black-white discrimination (Lett, 1974). The key factor in all of Lett's experiments was the removal of animals from the apparatus to their home cages. ...
Behavioral test batteries are valuable methods which allow outcomes with varying characteristics and neurobiological bases to be assessed and compared in the same animals. This allows investigators to construct a profile of impairments produced by a pharmacological or toxicological challenge, and to propose mechanisms for further study based on those findings. This profile is valuable in the assessment of potentially hazardous substances, including environmental toxicants, drugs of abuse, and other neuropharmacologically active agents. Behavioral tests and batteries have been developed for a number of species, including a relatively recent and growing body of work with the zebrafish, Danio rerio. This chapter discusses the current zebrafish behavioral battery used in our laboratory, and some of the main factors that drove its development. The principal tests include a motility assay for larval fish (6 days post fertilization, dpf), and a battery intended for adolescent (2–3 months) and adult fish (5+ months), which assay sensorimotor, affective, and cognitive-like functions in these fish. Significant progress has been made in the areas of zebrafish neurobehavioral analysis, although further studies, refinements, and task development efforts will be needed to strengthen this approach in the future.
The nematode Caenorhabditis elegans (C. elegans) is a prevailing model commonly utilized in a variety of biomedical research fields, including neuroscience. Due to its transparency and simplicity, it is becoming a choice model organism for conducting imaging and behavioral assessment crucial to understanding the intricacies of the nervous system. Here, the methods required for neuronal characterization using fluorescent proteins and behavioral tasks are described. These are simplified protocols using fluorescent microscopy and behavioral assays to examine neuronal connections and associated neurotransmitter systems involved in normal physiology and aberrant pathology of the nervous system. Here, we aim is to make available to readers some streamlined and replicable procedures using the C. elegans models as well as highlighting some of the limitations.
Behavioral assessment is a critical component of neurotoxicological research as the consequences of neurotoxicant exposure frequently include changes in behavior. Behavior encompasses multiple domains, and numerous behavioral testing paradigms, ranging from simple to complex, are available for assessment within each such domain and described in the literature. Many laboratories adapt the simplest procedures, based on the assumption that these procedures would be both simple to implement and to interpret. Such assumptions fail to recognize that behavior is actually quite complex. For example, it is critical to consider potential confounding effects (stress, nutritional state, motivation, motor endurance, fear, pain sensitivity) inherent to common behavioral tests, i.e., to understand behavioral mechanisms of an observed behavioral change, because they may shift the interpretation of a change in behavior, particularly for cognitive/executive functions, such as learning. Additionally, this chapter addresses issues related to the use and misuse of test batteries and the influence of the behavioral history resulting from use of a test battery that may interact with the toxicant treatment. It also emphasizes the critical need to recognize that behavioral testing itself modifies the brain; hence, attempting to define biological mechanisms of a neurotoxicant from brains of organisms that have behavioral experience may prove misleading, as brain changes will reflect the effects of both the toxicant exposure and the behavioral experience. Finally, the chapter emphasizes the advantages of using the same behavioral paradigms across species, with appropriate confounds measured and ethologically relevant modifications made, to enhance translation.
Reward training may be summarized by the equation
$$S:R \to S^V$$
which designates that a specified response (R) performed in a given stimulus situation (S) results in presentation of a stimulus that is currently valued by the animal (Sv). Thorndike’s (1898) experiment in which hungry cats pulled a loop in a box to gain access to food illustrates one variety of reward training in which Sv is an appetitive stimulus. Non-appetitive forms of Sv include periods of freedom from an aversive stimulus (i.e., “escape training”), certain classes of exteroceptive stimuli (e.g., a change in ambient illumination level for rats), stimulus conditions which allow certain responses to be performed (e.g., presentation of a manipulable puzzle to monkeys), direct stimulation of certain brain areas (e.g., electrical stimulation of the hypothalamus), and classically conditioned stimuli previously paired with Sv (i.e., “conditioned reinforcement” training). When the parameters of training are properly arranged, an occurrence of S:R → Sv results in increased likelihood that R will occur when the animal subsequently encounters S. It is this result which is summarized in the Empirical Law of Effect (Chapter 1).
A constraint on a phenomenon is a limitation or boundary condition for its occurrence. Scientific investigations invariably lead to the identification of such boundary conditions. Studies of learning have documented numerous constraints, including, for example, the interval between conditioned and unconditioned stimuli (CS and US), the intensity and novelty of the CS and US, and the extent to which the unconditioned stimulus is surprising. Limitations on learning brought about by factors such as the CS–US interval and CS familiarity, for example, reliably occur in numerous situations in which classical conditioning is observed. The primary impetus for discussions about biological constraints on classical and instrumental conditioning was that certain learning phenomena were contrary to principles of learning widely held at the time, and being an exception t o the rule constituted a working definition of biological constraints.
General process approaches to learning presuppose that there are learning processes governed by principles general enough to apply to many species in wide varieties of learning situations. A parade example of a general process approach in action is B. F. Skinner’s teaching-machine method. This is a technology of teaching verbal materials to humans largely based on principles derived from conditioning repetitive feeding behaviors in rats and pigeons. Two leaps of faith are implicit: (1) that one can validly extrapolate from pigeons and rats to humans; (2) that one can validly extrapolate from teaching repetitive feeding behaviors to teaching verbal materials. Although Skinner’s approach is an extreme example, most experimental psychologists during the last century have considered learning to be the study of general processes which underlie a great diversity of phenomena.
The subject matter of learning theory and ethology derives from the comparative psychology of the late nineteenth century. Since that time, both disciplines have been transformed decisively. The study of the mental life in animals, which was supposed to provide insight into the evolution of human intelligence, gave way to the study of conditioned behavior. The original approach to the study of instinctive behavior, which amounted to little more than an exercise in taxonomy of instincts, has been replaced by analytic developmental studies which seek to reveal how a species’ genotype and particular experiences interact to produce various instances of species-specific behavior. During most of the twentieth century, conditioned behavior was explained by appeal to one or another variety of general learning theory which sought to explain all learned behavior, in all species, by appeal to a simple set of principles. During the past two decades, the goals of learning theory have become less comprehensive. Interest has focused primarily on associations between a cue and a biologically important event (S-S* associations) and between a response and a biologically important event (R-S* associations). Students of animal learning have also investigated various “biological constraints” that violate assumptions about the arbitrariness of the abstract stimuli and responses referred to by learning theory.
Like learning theory, ethology went through an era of general theory construction which attempted to specify instinctive behavior on a broad level — notably the Lorenz-Tinbergen model of the innate releasing mechanism. In response to criticisms of such theories, ethologists have concerned themselves less with theory and more with research on the development of species-specific behavior. Such research has provided students of animal learning with many specialized examples of learning that, along with the literature on biological constraints, need to be accommodated by learning theory. It is suggested, however, that the main obstacles for the advancement of learning theory are problems from within. Specifically, until learning theorists are able to specify how the contents of learning are translated into performance, it seems doubtful that learning theory will be able to specify whether a particular biological constraint should be attributed to a) properties of learning a particular association, b) properties of performing some response, or both a) and b). It is also necessary for ethologists and learning theorists to develop a conceptual system that relates specialized instances of learning such as imprinting and song learning to more flexible varieties of learning.
Learning of a spatial discrimination in a T-maze with long delayed reward has been demonstrated by means of a procedure in which the rat is removed after a response to spend the delay in its home cage and later is returned to the T-maze for reward. This procedure was derived from Revusky’s (1971, in press) concurrent interference theory of long delay learning. According to this theory, the usual decrement in learning with a delay of reward is the result of concurrent interference produced when the response and/or reward become involved in extraneous associations with delay events. Removal of the rat to spend the delay outside of the T-maze minimizes concurrent interferences by reducing the associability of the delay events with respect to the events that occur inside the T-maze such as the response and the reward. The present experiments were concerned with a corollary of this theory: If events occurring in the home cage are not readily associated with events occurring in the T-maze, then a reward should be less effective when administered in the home cage than when administered in the T-maze. It was found that learning is retarded when reward is given in the home cage; however, reward given in the home cage as much as 2 min after a response produces learning.
Rats were rewarded for running down a runway if the previous trial, over 16 min earlier, had terminated in a white plastic goalbox, but were not rewarded if the trial had terminated in a larger black wooden goalbox (or vice versa). Overall, the rats learned the discrimination, albeit poorly. There were six groups of two rats each; the factors were 0-, 10-, and 30-sec delays in the startbox and whether the black or white goalbox was S+. Discrimination learning occurred under each of these conditions and, contrary to concurrent interference theory, was not noticeably affected by startbox delay.
The greater the separation in time between 2 events, A followed by B, the less likely they are to become associated. The dominant explanation of this temporal contiguity effect has been trace decay: During the interval between A and B, the trace left by A becomes too weak by the time B occurs for an association to be formed between them. Pavlov adopted this idea in the context of classical conditioning and Hull used it to account for the deleterious effect of delaying reinforcement on the acquisition of instrumental responses. By 1960 various studies supported the conclusion that animals could not learn to associate 2 events separated by more than around 45 s. Research on human skill acquisition with delayed feedback and later studies using causal or predictive judgment tasks indicated that explicit cognitive processing is generally involved when humans associate events separated by more than a few seconds. The discovery of long-delay taste aversion learning prompted Revusky's (1971) alternative analysis of contiguity effects in terms of interference: The greater the separation between A and B, the more likely that extraneous events compete for association with A and B. Although the analysis of overshadowing provided by associative learning theories provides a context for this account, none of these theories provide a satisfactory account of evidence on temporal contiguity from a wide range of animal studies. Alternative timing theories are arguably also unsatisfactory. (PsycINFO Database Record (c) 2014 APA, all rights reserved).
Rats can associate between events separated by intertrial intervals as long as several hours. For instance, they can learn that the type of food reward on one trial in a runway is correlated with whether running will be rewarded on the following trial. They also can learn that they will be rewarded if they alternate responses in a T-maze. In the otherwise similar experiments reported here, more traditional discriminative stimuli were used. Whether running down a runway was to be rewarded with sugar solution depended on whether the goalbox on the preceding trial, over 4 min ago, had been white or black. The rats learned this discrimination and retained it through an extinction phase.
Roberts's (1976) results should not be considered a failure to replicate the experiment by Lett (1974) demonstrating visual discrimination learning with a 1-min delay of reward. Although performance in Roberts's experiment was poor, the rats subjected to Lett's exact procedure showed learning by Roberts's own criterion of a significant-blocks effect. Moreover, by another criterion, the rats subjected by Roberts to three variants of Lett's procedure also exhibited learning. Thus, Roberts's results are consistent with Lett's.
Cue location has been an uncontrolled variable in food-aversion studies. While tastes are always attributes of the ingested object, visual, auditory and olfactory cues are often attributes of the food container or are located elsewhere in the conditioning chamber. A review of experimental studies indicates that cues which are attributes of the ingested object are almost invariably associated with both immediate and delayed illness, regardless of the sense modality of the cue and of the animal species involved. Cues which are attributes of the food container or conditioning chamber are associated with immediate but not delayed illness, again regardless of the sense modality and animal subject. Within the limits of present evidence, the same effects of cue location appear to occur when shock is the reinforcer. It is suggested that the association of attribute cues across delays is mediated by the conditioned behaviour, which is directed at the object of which they are attributes and which is biologically related to the subsequent reinforcement.
Human cognitive capacities have so evolved that man is able to solve an extensive range of problems having very different properties. Comparative psychologists have endeavoured to throw light on the evolution and nature of this general intellectual capacity by exploring performance of non-human vertebrates in a variety of learning tasks, in the expectation of demonstrating superior intelligence in species more closely related to man. It has, however, proved difficult to establish that any observed difference in performance is due to a difference in intellectual capacity rather than to a difference in such contextual variables as, for example, perception or motivation. Three hypotheses that might account for the lack of experimentally demonstrable differences in intelligence amongst non-humans are discussed. The first proposes that the data currently available may have been misinterpreted: that, for example, the potential role of contextual variables has been exaggerated. According to the second hypothesis, the questions posed by comparative psychologists have been inappropriate: learning mechanisms are adaptations evolved for life in a specific ecological niche, so that mechanisms available to species from different niches are not properly comparable. It is argued that neither of these two hypotheses receives convincing empirical support. A third hypothesis proposes that there are, in fact, neither quantitative nor qualitative differences among the intellects of non-human vertebrates. It is argued that this null hypothesis is currently to be preferred, and that man's intellectual superiority may be due solely to our possession of a species-specific language-acquisition device.
Studied discrimination learning in 18 male Wistar albino rats using differential reinforcers as discriminative stimuli in both a runway and a -maze. In the initial experiment, Ss were administered a sequence of 3 runway trial outcomes repeated indefinitely: either milk (M), pellets (P), and nonreinforcement (N) or P, M, and N. A 15-min intertrial interval (ITI) separated each of the 9 daily trials. Results support the prediction that Ss would learn to run more slowly on N trials than on either M or P trials. In the 2nd experiment, Ss were trained to run to the left side of a -maze after an M trial and to the right side after a P trial or vice versa. An ITI of 3 min. was employed. Results indicate that each of the 6 Ss gradually learned to run to the correct side more frequently than the chance level of 50%. Results are interpreted as consistent with E. J. Capaldi's sequential hypothesis. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
"Rats were trained on a one-unit T maze to perform a temporal alternation with 15 min. between trials. The animals alternated significantly above chance even when the delays between trials were increased to several hours . . .. Traditional S-R mechanisms do not appear capable of explaining this behavior. Therefore, it seems reasonable to suppose that rats can remember where they went last, and that this memory serves as the cue for the correct response." (PsycINFO Database Record (c) 2006 APA, all rights reserved).
An audiovisual stimulus was made contingent upon the rat’s licking at the water spout, thus making it analogous with a gustatory stimulus. When the audiovisual stimulus and the gustatory stimulus were paired with electric shock the avoidance reactions transferred to the audiovisual stimulus, but not the gustatory stimulus. Conversely, when both stimuli were paired with toxin or x-ray the avoidance reactions transferred to the gustatory stimulus, but not the audiovisual stimulus. Apparently stimuli are selected as cues dependent upon the nature of the subsequent reinforcer.
This chapter discusses the sequential mode of theorizing along with several reward variables, such as magnitude of reward, delay of reward, and nonreward. Sequential theorizing is found to be very helpful in selective learning. This may also explain many phenomena not deducible from other hypotheses. Moreover, a generalization decrement interpretation sequential theorizing has been shown to be applicable to a wide variety of phenomena, currently explained in diverse terms, such as associational and motivational. Sequential and nonsequential theories normally emphasize different classes of independent variables. Considering partial reward (PR), for example, a nonsequential theory such as the dissonance hypothesis suggests that resistance to extinction (Rn) is regulated by total number of nonrewards regardless of their sequence or order. Where the dissonance hypothesis sees the sequence of nonrewards as irrelevant, the sequential hypothesis sees number of nonrewards as irrelevant.
The literature of specific hungers and bait shyness contains much indirect evidence of learning involving prolonged delay of reinforcement. Specific hunger refers to the selective feeding by animals as they learn to correct a specific dietary deficiency, such as thiamine deficiency, while bait shyness describes the rejection of poisoned baits by animals, which have survived a previous poisoning attempt. It seems unlikely that learning can take place at all with delays of more than a few seconds. Instances of learning with protracted delays of reinforcement are always cases where immediate secondary reinforcement occurs. It is generalized that delayed reinforcement is not effective except under elaborate training conditions. However, when the response is ingestion and the rewards or punishments are changes in the physiological state of the organism, this generalization appears to be incorrect. Most specific hungers can be explained in terms of learned associations involving delayed aftereffects; an exception is the specific hunger for sodium, which appears to be largely innate. If rats are subjected to sodium deficiency and recover from it by drinking salt water, they tend to drink an abnormally large amount of salt water after the deficiency is relieved.
Without the aid of secondary rewards to bridge the temporal gap, each of 15 rats learned to select the rewarded side of a T-maze although the reward was delayed until 1 min after the response was emitted. Similar results were obtained from another group of eight rats for which the length of the delay was 5 min. In a final experiment using the same basic procedure, five groups of rats were trained for 25 days with delays of 0.5, 1.0, 2.0, 4.0, or 8.0 min. The percentage of correct responses did not significantly differ among groups. According to prevailing psychological theory, these results are impossible.
Considers learning and memory within an adaptive-evolutionary framework, using an analysis of the role of learning in thiamine specific hunger. The demands of the environment on the rat, the contingencies in the natural environment, the importance of the novelty-familiarity dimension, and the realization of 2 new principles of learning, permit a learning explanation of most specific hungers. The 2 new principles, "belongingness" and "long-delay learning," specifically meet the peculiar demands of learning in the feeding system. An attempt is made to develop the laws of taste-aversion learning. It is argued that the laws or mechanism of learning are adapted to deal with particular types of problems and can be fully understood only in a naturalistic context. The "laws" of learning in the feeding system need not be the same as those in other systems. Speculations are presented concerning the evolution and development of learning abilities and cognitive function. It is concluded that full understanding of learning and memory involves explanation of their diversity and the extraction of common general principles. (108 ref.)
Two cues, either size or flavor of food pellet, were conditionally paired with either malaise induced by x-ray or pain induced by shock in four groups of rats. The combination of flavor and illness produced as conditioned decrement in consumption, but that of size and illness did not. Conversely, the combination of size and pain produced an inhibition of eating, but flavor and pain did not. Apparently, effective associative learning depends on central neural convergence of the paired afferent input.