Discrimination Learning

Simultaneous visual discrimination learning is another area of cognitive behavior in which elephants have been compared to standards set by other mammals.

From: Evolution of Nervous Systems , 2007

Primates

J. Bachevalier , in Evolution of Nervous Systems, 2007

4.31.3.3 Learning and Memory Abilities in Infancy

Two separate lines of evidence strongly support the role of androgens in the development of sex differences in learning and memory abilities in nonhuman primate infants.

Acquisition of an object discrimination reversal task, known to depend on the integrity of the orbital frontal cortex in the adult monkey, is significantly more rapid in male infant monkeys than in females (Clark and Goldman-Rakic, 1989). Postnatal injections of testosterone propionate (TP) in the females enhance their performance to the level of normal infant males. When orbital prefrontal cortex is removed in infancy, intact male monkeys and androgenized female monkeys are as impaired as adult monkeys with the same lesions, whereas treated infant females do not differ from untreated age-matched females. The data suggest that the orbital frontal cortex matures earlier in male than in female monkeys.

Conversely, acquisition of a concurrent visual discrimination learning task ( Figure 1), known to depend on the integrity of the inferior temporal cortex, is significantly more rapid in female infant monkeys than in males (Bachevalier and Hagger, 1991), and this sex difference is positively correlated in 3-month-old male animals with circulating levels of testosterone (the higher the level of testosterone, the poorer the score on the task), but not with estradiol levels. Neonatal orchiectomy, which reduced plasma testosterone levels, hastens performance on visual discrimination learning in male infant monkeys, whereas treatment of androgens (DHT) in neonatally ovariectomized female infant monkeys delays their performance. Finally, early postnatal inferior temporal cortex lesions affect performance of female but not of male infant monkeys, though male and female adults with the same lesions are equally impaired. The data suggest that this temporal cortical area is functionally more mature in female infant monkeys than in males.

Figure 1. Summary of rate of acquisition of 20 concurrent discrimination problems in 3-month-old monkeys (Macaca mulatta). Normal females learned at a faster rate, and required fewer trials than normal males to reach the learning criterion. Young female monkeys were impaired after lesions to the temporal cortical area TE (lesion female), requiring more trials than did females that did not have surgery (normal female). By contrast, males with the same lesions (lesion male) and normal males learned at the same rate. Newborn males that were orchiectomized (ORX male) learned faster than normal males and at the same rate as normal females. By contrast, newborn females (OVX female   +   DHT) that were ovariectomized and treated with DHT (2   mg   kg−1 three times a week) learned at a slower rate than normal females.

The two sets of findings lead to intriguing conclusions. On the one hand, they suggest that sex differences in the development of learning and memory abilities are due to the influence of gonadal steroid hormones on the maturation of the specific cortical regions underlying these abilities. On the other hand, they also indicate that the directionality of such hormonal influences on structure and function varies from one cortical area to the other. Specifically, whereas the orbital frontal cortex appears to mature earlier in males than in females, inferior temporal cortical area TE appears to mature earlier in females than in males. Although the specific biological mechanisms for such cortical effects of androgens are still unknown, it has been hypothesized (Bachevalier and Hagger, 1991) that the actions of TP and DHT might be mediated by different receptors in the cortex. It has been shown in nonhuman primates that TP, which is aromatized into estrogens at a neural level, has a high affinity for estrogen receptors, whereas DHT, an androgen not significantly aromatized in the primate brain, has higher affinity than TP for the androgen receptors (Michael et al., 1986). Since both estrogen- and androgen-binding systems have been located in all cortical areas of the primate brain, there might be local cerebral metabolic differences (estrogenic vs. androgenic action) that underlie some of the different behavioral effects of gonadal steroids in primates. Such regional metabolic differences were underscored by the work of Michael et al. (1989).

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B012370878800166X

Theories, Development, Invertebrates

R. Menzel , ... M. Giurfa , in Evolution of Nervous Systems, 2007

1.26.3.3 Invertebrate Composite Operant Conditioning

Of course, in freely moving animals, a CS rarely occurs independent of all the animal's behaviors, just as a US will very rarely be contingent on a behavior independent of any other stimulus present. Such situations (i.e., with a BH, a CS, and an US present) are called composite operant conditioning. Drosophila, Aplysia, and the honeybee are also the most widely used model systems for studying composite operant conditioning. Exploiting their unique technical advantages, one paradigm each in both Drosophila and Aplysia has been explicitly developed to combine operant and classical components in a single preparation. If one adds coloration of the (unpatterned) arena as a CS to the operant yaw torque learning paradigm described above for Drosophila, one creates an instance of composite operant conditioning (Heisenberg et al., 2001). Whenever the yaw torque of the fly is in one of the two domains (roughly corresponding to left or right turns, respectively), the arena surrounding the animal is illuminated in one color; if the yaw torque moves into the other domain, the coloration changes as well. Thus, yaw torque and color become equivalent predictors of the punishing heat. This switch (sw)-mode learning causes a larger after-effect than yaw torque learning. This increased effectiveness of composite over purely operant or classical conditioning was observed previously, comparing classical conditioning to flight-simulator (fs) mode (Brembs and Heisenberg, 2000). In this setup, the angular velocity is calculated from the on-line torque signal that this momentum would give the fly. But instead of turning the fly, the yaw torque is made to turn the patterned arena around the fly in the opposite direction. This arrangement enables the fly to stabilize the arena, that is to fly straight with respect to the patterns on the arena wall, and to choose 'flight directions' with respect to these patterns. Before the training, the fly shows a moderate fixation of the patterns without a striking preference for one or the other. During training, the punishing heat beam is applied whenever the fly chooses a flight direction toward one of the pattern types. If after a few minutes of training the heat is permanently switched off the fly still prefers flight directions toward the previously 'cold' patterns.

Is visual pattern discrimination learning operant? The fly learns to associate the patterns (CS) with heat and 'no-heat' (US), meeting the definition of classical (Pavlovian) conditioning. Could the operant behavior in fs-learning merely accompany a learning process which in essence is classical? The standard experiment to answer this question is a 'yoked' control ( Brembs and Heisenberg, 2000). Exact sequences of pattern position and heating periods are recorded during operant training and are played back to a naïve fly. The replay control is a purely classical conditioning experiment, since the fly has no influence on the stimuli presentation. This kind of training is considerably less effective than the training operantly controlled by the fly. Thus, just as the sw-mode learning is more effective than the yt-learning, the training in fs-learning is more effective than in replay control. In other words, in both cases the three-term contingency (CS   →   US, BH   →   US) is more efficient than the two-term contingency (BH   →   US in yt-learning; CS   →   US in the yoked control of fs-learning). The interesting difference is that in fs-learning the behavior is not directly modified (Brembs and Heisenberg, 2000), whereas in sw-learning the spontaneous yt-distribution is altered just as in yt-learning. This allowed investigating which of the components are learned in the three-term contingency (Heisenberg et al., 2001). These studies showed that in both fs- and sw-learning only the classical but not the operant association is separately accessible, ruling out that the two act additively. Not summation but rather interaction between the operant and classical components characterizes the three-term contingency. Apparently, composite conditioning is so effective because learning about sensory stimuli is enhanced once the fly can control them by its own behavior, whereas behavioral modifications tend to be avoided (Heisenberg et al., 2001). Experiments with robots have yielded similar results, supporting the notion of a general synergistic mechanism (Verschure et al., 2003).

Similar questions can now be asked in a newly developed preparation in Aplysia. It was described above how the feeding behavior of Aplysia can be conditioned classically and operantly (Nargeot et al., 1997, 1999a, 1999b, 1999c; Lechner et al., 2000a, 2000b; Brembs et al., 2002, 2004; Brembs, 2003a, 2003b; Mozzachiodi et al., 2003). Taking advantage of the greater physiological accessibility, a reduced preparation of the isolated buccal and cerebral ganglia was used. This preparation still produces spontaneous nervous activity, which can be recorded as patterned motor output at selected motor nerves. These neuronal patterns can be related to feeding behavior in the intact animal. Stimulation of sensory nerves serves as CS and US input. Thus, all elements of the three-term contingency are present: BH, CS, and US. The preparation has been shown to produce robust learning effects; it will be interesting to see if results can be found that are analogous to those in Drosophila and, if so, what their neural basis is (Brembs et al., 2004).

Susswein and colleagues have developed another promising composite paradigm using Aplysia feeding behavior (Susswein and Schwarz, 1983; Schwarz and Susswein, 1984, 1986, 1992; Susswein et al., 1986; Schwarz et al., 1991; Chiel and Susswein, 1993; Botzer et al., 1998; Katzoff et al., 2002). Food touching the lips of Aplysia initiates biting, which causes food to enter the mouth. Swallowing is triggered by food within the mouth. Food or nonfood objects in the mouth can also trigger rejection (Kupfermann, 1974). In this aversive procedure, animals learn to avoid biting on inedible food items. Food is made inedible by wrapping it in a plastic net that the animals can neither swallow nor break. During training, the netted food touches the lips and the animal tastes the food through holes in the net. The animal bites, food enters the mouth and elicits swallows, which fail to convey the tough food to the gut. The food eventually is rejected. The food continues to stimulate the lips and elicit bites, which again lead to failed swallows. As training proceeds, the response of the animal to the food gradually changes. Food stays within the mouth for progressively shorter periods, eliciting fewer swallows and more rejections. Animals eventually stop responding to the food (Susswein et al., 1986).

An instrument more suitable than the fs for using the powerful genetic techniques in Drosophila for operant conditioning is the heat box (Wustmann et al., 1996; Wustmann and Heisenberg, 1997; Putz and Heisenberg, 2002). In the tiny, dark chamber, every time the fly walks into a designated half, the whole chamber is heated. As soon as the animal leaves the punished half, the chamber temperature reverts to normal. Even if the heat is switched off after a few minutes, the animals still restrict their movements to only one half of the chamber. Because it is completely dark in the chamber, the animal most likely relies on idiothetic cues to orient itself, thus minimizing the contamination with potential classical predictors. It can be shown that the operant memory consists of two components, a spatial component and a "stay-where-you-are" component (Putz and Heisenberg, 2002). One of the mutants found using the heat box is the 'ignorant' gene (Putz, 2003; Putz et al., 2004). Interestingly, it appears that 'ignorant' has very different effects on operant and classical conditioning. The original mutant allele (ignP1) shows a sexual dimorphism in the heat box, where males are impaired but females appear normal (Putz, 2003; Putz et al., 2004). Both males and females of that line are statistically indistinguishable from the wild-type controls in olfactory classical conditioning (Bertolucci, 2003). The null mutant (ign58/1) shows decreased learning and memory in the classical case (Bertolucci, 2003), but is normal in the heat box (Putz, 2003; Putz et al., 2004). Finally, several partial deletions of the ignorant gene are defective in the heat box (Putz, 2003; Putz et al., 2004), but these lines have not yet been tested for classical conditioning. Apparently, different mutations of the 'ignorant' gene have different effects on operant and classical conditioning, indicating fairly well-separated mechanisms for both forms of learning.

Another composite paradigm in honeybees and bumblebees is visual discrimination learning in the Y-maze. The freely flying bees enter a triangular decision chamber on one side and can see two target objects at the ends of the two arms of the 'Y' attached to the other two sides of the triangular chamber. Only one of the two targets contains a sucrose reward and thus animals can be trained to associate one of the two targets with the reward. Repeating the procedure with the targets at both arms in random sequence establishes a visual discrimination memory that is independent of the location of the target and hence of the turning maneuvers needed to reach it. In this situation, bees learn to associate a given target with reward and the alternative target with the absence of reward such that the former is excitatory and the latter inhibitory (Giurfa et al., 1999). Furthermore, bees are rewarded for every correct choice such that operant and classical associations certainly drive their choice within the maze. Several variants are known of this kind of visual learning in bees: from the simple visual target conditioning in which a bee has to land on a single rewarded stimulus, to training with a complex maze made from several connected boxes, at the end of which they get rewarded with sucrose solution (Zhang et al., 1999), bees efficiently learn this kind of task which has allowed further insights into higher-order forms of learning (see Section 1.26.4).

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B012370878800183X

Specialized Lesions: The Split-Brain Technique

COLWYN TREVARTHEN , in Methods in Psychobiology: Specialized Laboratory Techniques in Neuropsychology and Neurobiology, 1972

I HISTORICAL INTRODUCTION

Surgical section of the corpus callosum has been employed in physiological and psychological studies since near the turn of this century (review: Bremer et al., 1956). Particularly notable are the experiments of Bykov (1925) who demonstrated that section of the callosum reduced irradiation of reflexes conditioned to touch from one side of the body to the other.

Nevertheless, the split-brain technique first became a significant tool for the study of higher brain function with the experiments of R. E. Myers in the early 1950s. Myers showed that cutting the cerebral commissures dissociates two cortical mechanisms of perception and learning. He did this within the framework of experiments on interocular transfer of learning in fish, and on cortical connectivity functions in visual pattern perception in cats, in R. W. Sperry's laboratory in Chicago.

Myers (1955) found that midline sectioning of the optic chiasm in cats did not stop interocular transfer of visual discrimination learning. When, however, he added complete surgical division of the corpus callosum and anterior commissure, each subject behaved as if experiences obtained through the two eyes were entirely separate ( Myers, 1956, 1961; Sperry et al., 1956). The chiasm-callosum sectioned animal, which learned as two individuals, became known as a split-brain preparation (Fig. 1).

FIG. 1. Left: Cat brain, with major nerve tracts stretched, showing visual projection from left visual field to midbrain and cerebral cortex. Primary projection = black arrows and stippled areas. Secondary projection (through pulvinar), visual association cortex and commissural connections = white arrows and crosses. Right: Split-brain of baboon, left cerebral cortex removed. Percept of an object felt by the left hand or seen in the left visual field is integrated only within the right hemisphere.

Soon, further tests showed that both split-brain cats and monkeys had divided mechanisms for discrimination by vision, or by touch with hand, foot or paw. Commissural and intrahemispheric pathways were compared in transfer of visual learning between the eyes, and the principal region of the callosum responsible for interhemispheric communication was located in the splenium. The anterior commissure was also shown to be capable of a certain degree of visual learning transfer in the monkey. Bihemispheric control of motor patterning remained after callosum sectioning and a variety of experiments were aimed at analysing the pathways of perceptomotor integration. Unilateral ablations were made to find the minimal amount of cortical tissue which could still sustain a visual or tactile memory trace. Eventually, testing procedures were devised which brought out functions which were not completely divided by total cerebral (forebrain) commissurotomy and which appeared to involve midbrain or other brain stem cross-connections as well as bilateral duplication of functions in the hemispheres.

In 1961 Sperry reviewed the large amount of work already done and described the potentialities of the method for further exploration of cerebral organization in the control of behaviour (Sperry, 1961a, b). Since that time, a number of reviews of continued work with cats and monkeys, or symposia drawing on this work, have appeared. (Downer, 1962; Myers, 1962; Sperry, 1964, 1967; Ettlinger, 1965; Trevarthen, 1968; Ettlinger and Blakemore, 1969; Gazzaniga, 1970).

Conditioned reflex studies have been taken up again with callosumsectioned animals by a number of workers, and central as well as peripheral conditioned stimuli have been used (Doty and Rutledge, 1959; Doty and Giurgea, 1961; Voneida and Sperry, 1961; Meikle et al., 1962; Voneida, 1963, 1964; Majkowski, 1967; Mosidze and Rizhinashvili, 1968). Split-brain methods have been employed more frequently in recent years to explore the physiological mechanisms that regulate the cortical EEG. These studies have clarified the respective roles of hemispheric and subhemispheric circuits in the control of changes in sleep and wakefulness, and in the generation and transmission of paraoxysmal activity within the brain (Bremer et al., 1956; Bremer and Stoupel, 1957; Magni et al., 1960; Berlucchi, 1966; Batini et al., 1967; Majkowski, 1967; Giaquinto, 1969; Kevanishvili et al., 1969).

A highly important development has been the extension of testing procedures worked out for studying split-brain animals to human commissurotomy subjects (Geschwind and Kaplan, 1962; Geschwind, 1965; Sperry, 1967, 1968a; Sperry et al., 1969; Gazzaniga, 1970). In 1962, Gazzaniga, Bogen and Sperry reported the results of tests with the first of a number of patients of Drs. P. J. Vogel and J. E. Bogen of Los Angeles (Bogen and Vogel, 1962, 1963; Gazzaniga et al., 1962; Bogen et al., 1965). These patients were operated on to obtain control of intractable epilepsy (Bogen et al., 1969). Further intensive testing of these subjects has greatly clarified the role of the corpus callosum in cerebral processes in man and important differences in the psychological functioning of the two hemispheres have been described (Gazzaniga, 1967, 1970; Levy-Agresti and Sperry, 1968; Sperry, 1968a, b; Bogen, 1969; Sperry et al., 1969; Levy et al., 1972). New understanding of the relationships between cortical and subcortical mechanisms in perception and sensory-motor integration has been achieved (Trevarthen, 1970).

In recent papers Sperry has emphasized the significance of this research for conceptions of the nature of consciousness in man and its relation to brain design (Sperry, 1966, 1968a, b, 1969).

The split-brain technique which has led to the above spectacular advances is not merely a surgical operation, although development of specialized and refined methods of brain surgery have been essential. Equally important is a methodology of testing which has been in continuous evolution to bring out the highly complex patterning of psychological functioning that results from disconnection of the cerebral hemispheres. In order to test split-brain subjects, considerable ingenuity has been exercised in devising behavioural controls and special methods for presenting stimuli and channelling responses.

Overcoming the remarkable capacity of the brain, even when it is largely divided, for re-integrating its functions within a unified pattern has led us to make distinctions between different levels of integrative action both in perception and in the control of voluntary action. The anatomical extent of mechanisms performing sensory-motor integration and the part of centrencephalic or subhemispheric circuits have been clarified in experiments in which stimulus dimensions, orienting responses and the form of consummatory response have been controlled. Experience gained in experimentation with split-brain animals has proved invaluable in the development of tests for the complex aspects of the commissurotomy syndrome in man.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780125123020500134

Mammals

B.L. Hart , L.A. Hart , in Evolution of Nervous Systems, 2007

3.34.2 Behavioral Considerations

Here we briefly review relevant behavioral studies on elephants to provide a background for discussion of elephant brain anatomy, cytoarchitecture, and information processing.

3.34.2.1 Comparative Aspects of Cognitive Behavior

For this topic it is appropriate to compare elephants with chimpanzees (Pan troglodytes), the great ape most commonly studied with regard to cognitive behavior. Using sticks to fish termites out of underground nests or to reach inside bones for marrow (van Schaik et al., 1999; Matsuzawa, 2003) is a classic example of tool use in great apes. Perhaps the best example of complex tool use by chimpanzees is cracking nuts open by holding a nut against an anvil stone and hitting it with a smaller hammer stone (Humle and Matsuzawa, 2001).

Elephants engage in several types of tool use. Fly switching with branches, while not particularly complex, would appear, in fact, to be the first documented example of tool use among nonhuman animals, dating back to when a wildlife adventurer wrote in 1838 of seeing elephants emerging into an open glen, "bearing in their trunks the branches of trees with which they indolently protected themselves from flies" (Harris, 1838, p. 169). Elephant fly switching was even mentioned by Darwin (1871) in discussing the intelligence of beasts. We have documented the efficacy of fly switching in repelling flies (Hart and Hart, 1994) and the modification of branches to use as switches (Hart et al., 2001). Asian and African elephants engage in other types of tool use, including scratching with a stick and throwing stones at rodents competing for fruit (Hart and Hart, 1994; Kurt and Hartl, 1995; Wickler and Seibt, 1997). Despite the historical significance, the complexity of tool use of elephants pales in comparison with the rich repertoire of fast-action, highly coordinated tool use described for chimpanzees such as the hammer–anvil style of cracking nuts.

Simultaneous visual discrimination learning is another area of cognitive behavior in which elephants have been compared to standards set by other mammals. Under seminatural conditions, Asian elephants may learn a black/white or large/small discrimination but the performance of even the fastest-learning elephants is unremarkable compared with other mammals ( Nissani et al., 2005). Tests of insight behavior, exemplified by pulling a cord to obtain a desirable object, also reveal disappointing behavior by elephants which do not perform in a manner consistent with that of chimpanzees, rhesus monkeys, and even several species of birds (Nissani, 2004).

Finally, in the way of classical tests of cognitive behavior, there is the test of self-recognition, as studied in the well-known mirror experiment in which the self-recognizing animal touches a mark on its head in front of a mirror. In contrast to chimpanzees that perform well (Povinelli et al., 1997), Asian elephants fail in self-recognition (Povinelli, 1989; Nissani, 2004).

3.34.2.2 Elephants and Long-Term, Extensive Memory

Now we turn to an aspect of behavior for which elephants have legendary abilities ("memory like an elephant"; "an elephant never forgets"). Quantitative field studies actually support this popular perspective. One would expect long-term memory about details of the environment regarding food and water resources to be essential for survival of herbivores like elephants, whose digestive system is adapted to handle large volumes of low-quality forage. Among the studies where movements are documented by individual animal recognition and/or radio-tagged individuals are those of desert elephants that occupy a huge home range during the dry season and visit water holes spaced more than 60   km apart (Viljoen, 1989). One cannot help but be impressed by the ability of an elephant family, led by a matriarch of some 30-plus years, to head unerringly toward an isolated water hole after a stressful 4 days without water; especially when particular water holes may only be visited every 8 months or so. The same desert elephants travel annually from their home ranges to new forage grounds in response to localized rainfall almost 200   km away (Viljoen, 1989). The fact that elephants may arrive at such distant locations as soon as 3 days after the start of rainfall, and without prevailing winds to carry chemosensory cues (Viljoen, 1989), suggests they are responding to one or more sensory cues of distant rainfall. One possibility is that seismic waves, which are detectable by elephants (O'Connell-Rodwell et al., 2005), and which are produced by lightning strikes and the accompanying thunder vibrations (O'Connell-Rodwell et al., 2001), may be sufficient for elephants to detect over long distances. Thus, travel to distant foraging grounds could be triggered by seismic signals from distant storms with direction coordinated by long-term memory of where vegetation is likely to appear.

The fitness value of the long-term spatial–temporal information retained by long-living matriarchs was vividly illustrated in a study conducted during a prolonged drought in Tarangire National Park in Tanzania (Foley, 2002). Clans in which families were led by older matriarchs left the park to forage in nonpark areas. However, a clan in which families were led by only young matriarchs (due to poaching) stayed in the park and sustained severe losses from insufficient water and forage: infant mortality and all-age mortality were more than double that of the clans with older matriarchs. To have experienced the most recent severe drought, where matriarchs could have remembered where to go, they would have had to be at least 35 years old and the matriarchs of the clan not leaving the park were apparently not this old.

Social memory is another area in which elephants show exceptional ability. An aspect of social behavior involves chemosensory communication and urine could theoretically allow an animal to identify particular conspecifics even decades after the last encounter. During sexual encounters, urine is typically orally investigated by adult male elephants, as in other ungulates, through the process of flehmen behavior that involves the transport of fluid materials from the mouth to the vomeronasal organ (Hart et al., 1989). Because long-lived male elephants leave their natal group, an ability to identify the urine of mothers would be important in avoiding inbreeding. This appears to be the case, as revealed in a controlled laboratory study in which adult males identified maternal urine decades after being separated from their mothers (Rasmussen and Krishnamurthy, 2000).

Acoustic stimuli represent another area in which elephants may recognize individuals. Elephants can recognize individual calls of 100 or more elephants at 1   km or more. Because there is uneven attenuation of the various frequencies of the contact calls at 1   km away, this means that such extensive recognition can occur with just a fraction of the acoustic signature that is otherwise present at close range (McComb et al., 2000, 2003). Discriminating between familiar and unfamiliar elephants is important in coordinating interactions at water holes and avoiding conflicts. One study revealed that families with older matriarchs are better at discriminating between familiar and unfamiliar individuals than families with younger matriarchs, and the age of the matriarch was a significant predictor of the number of calves successfully produced in the family per female (McComb et al., 2001).

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B0123708788003438

The rat's not for turning: Dissociating the psychological components of cognitive inflexibility

Simon R.O. Nilsson , ... Peter G. Clifton , in Neuroscience & Biobehavioral Reviews, 2015

3 Behavioural analyses of discrimination learning

Discrimination learning occurs in response to dissociable reinforcements of stimuli, perceptual dimensions or contexts and the nature of subsequent reversal learning and attentional set-shifting processes largely depend on how these discriminations are acquired. Early views held discrimination learning to be a low-level comparative process where behaviour is controlled by relational features of stimuli and rules are acquired through trial-and-error processes. However, this account had difficulty with phenomena such as stimulus generalisation and reversal following transposition tests ( Spence, 1937). Spence suggested that discrimination learning depended on both excitation and inhibition (Spence, 1936). Positive reinforcements increase the excitatory strength of a stimulus and elicit approach, while non-reinforcements decrease the excitatory strength of a stimulus and make approach less likely. While Spence treated non-reinforcement as a non-aversive mechanism inhibiting the excitatory strength and approach tendency associated with a stimulus, others have considered non-reinforcement to result in aversive negative reinforcement (Amsel, 1958; Behar, 1961; Nissen, 1950; Terrace, 1971). However, Spence's conditioned excitatory–inhibition theory predicts that previously non-reinforced stimuli, or stimuli not correlated with reinforcement, will acquire excitatory tendencies at a similar or faster rate than neutral stimuli. The theory therefore fails to explain phenomena such as latent inhibition or learned irrelevance, in which such associations delay learning. The mechanisms of excitation and inhibition are also insufficient to explain phenomena such as the serial-reversal effect (Mackintosh et al., 1968) where later reversals are acquired at a faster rate than earlier reversals, or the overtraining reversal effect, where prolonged discrimination training can facilitate reversal learning (Lovejoy, 1966).

To account for these phenomena, it has been necessary to involve the additional mechanisms of attentional stimulus selection-processes and predictability (Mackintosh, 1983; Pearce and Hall, 1980). Attention is a composite term for processes ensuring appropriate and continued maintenance and selection of stimuli for goal-directed behaviour (Parasuraman, 1998). In the context of discrimination learning and cognitive flexibility, attention is thought of as a determinant of perception allowing stimuli predictive of reinforcement to gain excitatory or inhibitory conditioning while irrelevant information fail to interfere with these processes (Mackintosh, 1965). In attentional theory, it is suggested that subjects attend to those stimuli (Mackintosh, 1975), or stimulus dimensions (Sutherland and Mackintosh, 1971), that are the best predictors of the reward contingencies, and this attentional allocation subsequently drives responding towards the appropriate CS+ and CS−. This approach suggests that although all forms of discrimination learning and cognitive flexibility require attentional allocation, only extradimensional set-shifting require attentional relocation and will therefore be acquired at a slower rate. Overtraining reversal effects (Mackintosh, 1969) and serial reversal effects (Mackintosh et al., 1968) may also be explained by extensive training increasing attentional allocation to particular discriminative features.

In sum, although theoretical approaches to animal learning differ in the value they place on attentional factors and reinforcement contingencies throughout the process of discrimination acquisition, they explain differential responding to distinctive stimuli in terms of independent mechanisms of excitation versus inhibition and consequent changes in attention to the relevant stimuli.

Read full article

URL:

https://www.sciencedirect.com/science/article/pii/S0149763415001694

Festschrift in Honour of Jeffrey Gray - Issue 2: Schizophrenia and Consciousness

Urs Meyer , ... Benjamin K. Yee , in Neuroscience & Biobehavioral Reviews, 2005

Discrimination learning was not affected by maternal PolyI:C treatment at a dose of 5  mg/kg, but reversal was marginally facilitated. In spite of the limited size of the effect seen here, which might partly be attributed to the reduced sample sizes, this is suggestive of enhanced attentional set shifting in the treated subjects. The relevance of reversal learning to schizophrenia stems largely from its sensitivity to prefrontal lesions and dopaminergic manipulations (Weiner and Feldon, 1986; Mehta et al., 2001; Clark et al., 2004; Kruzich and Grandy, 2004; Salazar et al., 2004). Reversal learning is affected by hippocampal damage although effects in both directions have been reported (see O'Keefe and Nadel, 1978; Gray and McNaughton, 1983). Cytotoxic lesions of the entorhinal cortex have been reported to enhance reversal learning (Yee and Rawlins, 1998), and this brain region has also been implicated in the neuropathology of schizophrenia (Jakob and Beckman, 1989; Harrison, 1999). An overview of the reported effects of amphetamine has been summarized in Section 1.2.5.

Read full article

URL:

https://www.sciencedirect.com/science/article/pii/S0149763405000813

Foundations and Innovations in the Neuroscience of Drug Abuse

John A Harvey , in Neuroscience & Biobehavioral Reviews, 2004

A discrimination learning task was employed to test the hypothesis that prenatal exposure to cocaine had affected attentional processing. These are tasks in which attentional processing plays a more critical role than in simple acquisition of the CR. In the simplest discrimination learning task, two CSs are employed. One CS, designated the CS+, is consistently paired with the US whereas the other CS, designated the CS−, is never paired with the US. A commonly held view of discrimination learning is that successful discrimination involves learning to attend and respond to the CS+ and learning to ignore and not respond to the CS− [78]. If attentional processing was altered as a consequence of prenatal cocaine exposure, then one would expect to see some alteration in discrimination learning as well. Animals were trained with the relatively salient tone as the CS+ and the less salient light as the CS− (Fig. 9, upper panel). Although the additional attentional demands in this task slowed the rate of learning for both groups (compare with Fig. 8), that effect was more pronounced in the cocaine offspring so that there were no differences between them and saline offspring in the acquisition of CRs to the tone CS+. In the next experiment, conducted in separate groups of animals, the roles of the tone and light were switched. The light now served as the CS+ and the more salient tone served as the CS− (Fig. 9, lower panel). This discrimination task was more difficult than the previous one in that both groups required more sessions to achieve asymptotic performance when the more salient stimulus, the tone, was the CS−. More importantly, cocaine-exposed animals were significantly retarded in their ability to acquire CRs to the less salient, light CS+[79].

Fig. 9. Acquisition of discrimination tasks by rabbits that had been exposed prenatally to cocaine and their saline controls. Top panel, cocaine progeny acquired the discrimination at the same rate as controls when to salient tone stimulus was the CS+ and the less salient visual stimulus was the CS−. Bottom Panel, cocaine progeny required approximately 2 more weeks to acquire the discrimination when the less salient visual stimulus was the CS+ and the more salient tone stimulus was the CS− than did controls [79].

Read full article

URL:

https://www.sciencedirect.com/science/article/pii/S0149763403001477

Chemotherapy-induced cognitive impairments: A systematic review of the animal literature

A. Matsos , I.N. Johnston , in Neuroscience & Biobehavioral Reviews, 2019

3.3.3.1.7 Brightness discrimination learning

A brightness discrimination learning (BDL) task was used to assess whether chemotherapy influences the ability to discriminate between cues that signal an escape platform when they vary in brightness. MTX + 5-FU ( Winocur et al., 2012) and CMF-treated animals (Briones and Woods, 2011) required more trials to reach criterion. In contrast, similar studies have reported MTX + 5-FU does not significantly disrupt the rate of learning (Winocur et al., 2006, 2015). Under high interference conditions where the test cues had both light and dark features, MTX + 5-FU-treated mice performed significantly worse (Winocur et al., 2015).

Read full article

URL:

https://www.sciencedirect.com/science/article/pii/S014976341830945X

IBNS 2017 - Contemporary Contributions to Basic and Translational Behavioral Neuroscience Research

Anthony L. Riley , ... F. Scott Hall , in Neuroscience & Biobehavioral Reviews, 2020

3.2 Drug discrimination learning

In the DDL procedure, the interoceptive effects of the compound can be established and compared to other drugs with known abuse potential (see Ator and Griffiths, 2003; Colpaert, 1987; Järbe, 1986; Palmatier et al., 2005; Solinas et al., 2006 ). In this work, responding on one lever for a reinforcer (typically food) is preceded by an injection of a training drug, whereas responding on another lever for the same reinforcer is preceded by injection of the drug vehicle. Under such conditions, animals acquire discriminative control and distribute lever choice based on the subjective properties of the administered drug (or its vehicle) (for a review of drug discrimination learning, see Colpaert, 1987; Overton, 1991; for alternative DDL procedures, see Mastropaolo et al., 1989; Riley et al., 2016). Subsequent drug tests can be run to determine whether a novel drug substitutes for the training drug (and if it has similar interoceptive effects) and whether known pharmacological antagonists can attenuate or block discriminative control (providing some insight into possible receptor mediation of the drug's discriminative stimulus effects).

In one such examination, Gatch et al. (2013) trained male Swiss-Webster mice to respond on one lever for a food pellet following an injection of METH or cocaine, and a second lever following an injection of the drug vehicle. Once the discriminations were acquired, methylone (0.5–5 mg/kg), mephedrone (0.5–5 mg/kg) and MDPV (0.05–2.5 mg/kg) were assessed for their ability to substitute for METH and cocaine. As reported, each of the synthetic cathinones fully substituted for the discriminative stimulus effects of both METH and cocaine, suggesting common interoceptive effects (and presumably similar abuse potential; see also BBerquist and Baker, 2017; Dolan et al., 2018; Harvey et al., 2017; though see Varner et al., 2013 for a failure of a range of stimulants to substitute for mephedrone). Gatch et al. (2015a) extended their work to other synthetic cathinones, i.e., α-PVP, 4ʹ-MePPP and α-PBP, using male Sprague-Dawley rats trained to discriminate cocaine or METH from saline. Both α-PBP and α-PVP fully substituted for the discriminative stimulus effects of cocaine and METH, where 4ʹ-MePPP only fully substituted for the discriminative stimulus effects of METH. Under similar training conditions, Gatch et al. (2017) reported that α-PPP, α-PHP, α-PVT, MDPBP and ethylone all dose-dependently (and fully) substituted for the discriminative stimulus effects of METH, whereas only α-PPP, α-PHP and ethylone displayed complete substitution for the discriminative stimulus effects of cocaine. The authors suggested that the failure of MDPBP and α-PVT to substitute completely for cocaine may indicate lesser abuse vulnerability compared to the other compounds that were tested. The findings of these latter two studies together add to their previous work suggesting that common interoceptive effects among these stimulants may indicate similar abuse potential. (For other work involving second generation cathinones in rodent models, see Cheong et al., 2017; Javadi-Paydar et al., 2017; Naylor et al., 2015; for work in monkeys, see Smith et al., 2017.)

Consistent with the work by Gatch et al. (2017, 2015a, 2013) that synthetic cathinones can substitute for METH and cocaine in the DDL design, Fantegrossi et al. (2013) demonstrated the reverse. Specifically, in their work male NIH Swiss mice were trained to discriminate MDPV (0.3 mg/kg) from its vehicle and were then given various doses of MDMA and METH to assess their ability to substitute for MDPV's stimulus effects. During test sessions, a multiple-component cumulative dosing procedure was utilized in the absence of reinforcement which allowed for testing of four doses of each drug in a single session. MDMA dose-dependently and fully substituted for the MDPV training dose, with near exclusive choice of the MDPV-associated lever at a cumulative dose of 0.3 mg/kg. The interpolated ED50 for cumulative MDMA was identical to that of MDPV (0.03 ± 0.01 mg/kg). Substitution doses of MDMA dose-dependently suppressed response rates. Similarly, cumulative injections of METH dose-dependently and fully substituted for MDPV, with >80% of the total responses on the drug-associated lever at a cumulative dose of 0.3 mg/kg. The interpolated ED50 for cumulative METH was 0.08 ± 0.03 mg/kg. As with MDMA, METH dose-dependently suppressed response rates. Substitution assessments with negative controls, JWH-018 and morphine, resulted primarily in responding on the saline-associated lever. These results again suggest abuse potential and rewarding efficacy of MDPV that are similar to those reported for other known psychostimulants (see also, Gannon and Fantegrossi, 2016).

Assessments of drug mixtures have begun to be examined for discriminative stimulus effects using the DDL design as "bath salt" products are often either intentionally mixed, or otherwise adulterated, with other compounds prior to sale. For instance, Collins et al. (2016) used DDL to assess the effects of cocaine, MDPV and caffeine, alone and as binary combinations, in male Sprague-Dawley rats. A primary goal of the study was to determine the effect caffeine may have when used in tandem with another psychostimulant, as well as to determine if the effect these combinations of drugs that have either similar or dissimilar mechanisms of action are simply additive. Rats were trained to discriminate cocaine (10 mg/kg) from saline in a two-lever drug discrimination procedure, and binary drug combinations (cocaine:caffeine, MDPV:caffeine and cocaine:MDPV) were each prepared at three separate dose ratios (3:1, 1:1 and 1:3 with respect to their ED50 values). Single-drug assessments of their ability to induce dose-dependent cocaine-appropriate responding revealed that METH = MDPV > cocaine > caffeine, an indication of at least partially intersecting stimulus effects. Binary mixtures of cocaine:caffeine, MDPV:caffeine and cocaine:MDPV showed mostly additive effects, although there were some supra-additive effects seen in some combinations as well as some individual differences across subjects. The significance of these results comes from evidence of cocaine-like discriminative effects among these drug combinations and a demonstration of supra-additive interactions occurring between drugs that possess different mechanisms of action and different binding affinities for DAT and SERT (for another study assessing stimulant combinations with synthetic cathinones, see Harvey and Baker, 2016; for a study assessing the discriminative stimulus effects of MDPV, see Gannon et al., 2016; for a list of other DDL studies involving the synthetic cathinones, see Glennon, 2014).

Read full article

URL:

https://www.sciencedirect.com/science/article/pii/S0149763418303191

Landmark discrimination learning in the dog: effects of age, an antioxidant fortified food, and cognitive strategy

Norton W. Milgram , ... C.W. Cotman , in Neuroscience & Biobehavioral Reviews, 2002

The landmark discrimination learning test can be used to assess the ability to utilize allocentric spatial information to locate targets. The present experiments examined the role of various factors on performance of a landmark discrimination learning task in beagle dogs. Experiments 1 and 2 looked at the effects of age and food composition. Experiments 3 and 4 were aimed at characterizing the cognitive strategies used in performance on this task and in long-term retention. Cognitively equivalent groups of old and young dogs were placed into either a test group maintained on food enriched with a broad-spectrum of antioxidants and mitochondrial cofactors, or a control group maintained on a complete and balanced food formulated for adult dogs. Following a wash-in period, the dogs were tested on a series of problems, in which reward was obtained when the animal responded selectively to the object closest to a thin wooden block, which served as a landmark. In Experiment 1, dogs were first trained to respond to a landmark placed directly on top of coaster, landmark 0 (L0). In the next phase of testing, the landmark was moved at successively greater distances (1, 4 or 10  cm) away from the reward object. Learning varied as a function of age group, food group, and task. The young dogs learned all of the tasks more quickly than the old dogs. The aged dogs on the enriched food learned L0 significantly more rapidly than aged dogs on control food. A higher proportion of dogs on the enriched food learned the task, when the distance was increased to 1   cm. Experiment 2 showed that accuracy decreased with increased distance between the reward object and landmark, and this effect was greater in old animals. Experiment 3 showed stability of performance, despite using a novel landmark, and new locations, indicating that dogs learned the landmark concept. Experiment 4 found age impaired long-term retention of the landmark task. These results indicate that allocentric spatial learning is impaired in an age-dependent manner in dogs, and that age also affects performance when the distance between the landmark and target is increased. In addition, these results both support a role of oxidative damage in the development of age-associated cognitive dysfunction and indicate that short-term administration of a food enriched with supplemental antioxidants and mitochondrial cofactors can partially reverse the deleterious effects of aging on cognition.

Read full article

URL:

https://www.sciencedirect.com/science/article/pii/S0149763402000398