Студопедия

Главная страница Случайная страница

Разделы сайта

АвтомобилиАстрономияБиологияГеографияДом и садДругие языкиДругоеИнформатикаИсторияКультураЛитератураЛогикаМатематикаМедицинаМеталлургияМеханикаОбразованиеОхрана трудаПедагогикаПолитикаПравоПсихологияРелигияРиторикаСоциологияСпортСтроительствоТехнологияТуризмФизикаФилософияФинансыХимияЧерчениеЭкологияЭкономикаЭлектроника






Rectifying the track to the aim: short cutting and detours as elements of cognitive mapping in animals






 

In Tolman’s concept of cognitive map one of key elements is the ability to construct new short-cutting routes. This behaviour is not so simple and requires ability to link local scenes without any experience of the direct rout from one point to another.

Besides demonstrating new short cuttings themselves, another possible experimental strategy that can demonstrate the operation of a cognitive map is a so called detour test. Detour behaviour demonstrates the ability of an animal to reach a stimulus (goal) when there is an obstacle between the subject and the stimulus. Detour tests allow researchers to integrate experimental approaches aimed to studying how animal intelligence is displayed in time and space. Indeed, the detour test is of particular interest for comparative cognitive research as a peculiar and quite naturalistic example of “delayed response” (see Chapter 10). In the absence of any local orienting cue emanating from the goal, detour performance should require the maintenance of a “memory” of the location of a disappeared object. As we will see in Chapter 13, developmental psychologists have thoroughly investigated these abilities in human infants (Piaget, 1954). In order to test how animals master the detour task, experimenters first show a goal to an animal and then place a subject behind a barrier some distance away. If the subject selects the correct, shortest route to the goal, then this may be because it possesses a cognitive map of the problem area and uses it to determine how to respond.

Short-cutting. Tolman’s interest in novel short-cutting as an essential part of cognitive mapping has been developed in many experiments. It is worth to note that in this local field of studying cognitive mapping, the general problem of parsimonious explanation reproduces itself at a small scale: some authors claim the results as providing clear evidences of cognitive maps while others suggest simpler alternatives for explanation of behaviour. Many examples of novel short cutting have been described but it is still discussible whether they indicate unambiguously cognitive maps or could be explained as movements towards familiar landmarks seen from a new angle.

For example, a useful demonstration of short cutting behaviour is reported by Chapius and Scardigli (1993). In one of their experiments, a hamster was placed into Chamber A from the apparatus shown in Fig. IV-2. An animal had to pass along the route identified by the dotted line to obtain food in Chamber E. Locked doors prevented the hamster from deviating from this route. After a number of sessions of this training, all the doors in the apparatus were unlocked except the one that allowed the animal to leave Chamber A on its normal route to food. If hamsters are able to calculate the shortest detour to food, then they should pass along the route marked by the solid line in Fig. IV-2. For the three test trials that were conducted, hamsters selected this route significantly more frequently than would be expected on the basis of chance. It is important to note that during the training phase, the maze was rotated relative to the room from one trial to the next. In addition, the chamber that served as the start of the route (Chamber A), was varied from trial to trial. The success of the animals in solving this problem was thus not due to them orienting towards landmarks that lay beyond, or within the maze. A more plausible explanation is that the hamsters formed a map of the shape of their route during the training stage, and then used this map to plot their course to the goal on the test trial (Pearce, 2000).

Bennett (1996) argues that in this and other experiments path integration could be use to perform the short-cut. As it was described above, path integration, or dead reckoning, is the mechanism of integration of distance and direction while moving that allows an animal to make a straight-line return to the starting point. Several experiments have shown that it can be performed without access to any previously seen landmarks, acting through either optic flow or internal acceleration detectors. Path integration then does not require memory, special or otherwise, of previously seen landmarks. Thus, the possibility that animals make short-cuts using path integration must be eliminated before one can conclude from novel short cutting that an animal has a special form of landmark memory in the form of a cognitive map.

One more example concerning novel short cutting in honey bees demonstrates importance of methodological details for travelling in this field of cognitive mapping. The question of whether bees can take novel short-cuts between familiar sites has been central in the discussion about the existence of cognitive maps in these insects. The failure of bees to show this capacity could be a result of the training procedure, because extensive training to one feeding site may eliminate or weaken memories of other sites that were previously learned. Menzel et al. (1998) investigated this problem by rewarding honey bees at two feeding sites, one (S m ) at which they could eat in the morning, and the other (Sa ) at which they could feed in the afternoon (Fig. IV-3). The bees were then displaced to S a in the morning and to S m in the afternoon either from the other feeding site or from the hive. Bees were also displaced to two novel sites, one at a completely unfamiliar location (S 4) and another that was located halfway between the two feeding sites (S3) Bees displaced from either of the feeding sites never took novel short cuts; instead, they used the homeward directions that would have been correct had they not been displaced. Bees caught at the hive entrance, however, chose the correct homeward direction not only when displaced to both feeding sites, but also when displaced to S3, although not from S 4. Control bees that had been trained to only one of the feeding sites were not able to travel directly home from S3 excluding the possibility that bees use landmarks close to the hive.

These results support the hypothesis of taking a novel short-cut by activating two vector memories simultaneously and meet the basic requirement to prove the existence of a cognitive map. However, the authors rigorously estimate their results. They support the notion that what their bees do is an instance of basic cognition, that is, an activating of separately acquired memories and their application in a novel adaptive sense. Whether such a memory organisation is classified as a cognitive map depends on how rich the map is under the concrete circumstances. The experimenters suggest that the inference of a map would be premature if bees’ map only allows them to navigate in one direction (home), and only in some motivation state (hive arriving but not feeder-related motivation). Indeed, there are supportive data concerning bees and wasps which memorise landscape hallmarks when searching for a way home and do not use this information when they search for food source; these insects have to learn afresh under the new circumstances and another motivation (Mazokhin-Porshnyakov and Kartsev, 1979, 2000; Kartsev, 1990; Kartsev et al., 2005). More data are needed to demonstrate whether multiple vectors between landmarks, feeding sites and the hive could be stored together with the particular sensory properties of these sites. Anyway, a window of opportunity is still open for honey bees together with other insects to demonstrate that their mentality is rich enough to construct a cognitive map.

Detour tests. Detour tests can be considered parental for a theory of cognitive map. Detour behaviour has been widely studied in many species (review in Chapuis, 1987; Vallortigara, 2004). In this issue we will consider several concrete examples of experimental investigations.

Kö hler (1925) was one of the first to study detour problem. In his experiments a subject was placed behind the bars and could see food being placed on the other side. When dogs were tested, their behaviour depended very much on the distance between the food and the barrier. If the dog was some distance away, then it would immediately run in a smooth loop around the barrier to collect the food. When the food was placed behind a fence, the dog would run directly towards it and stay there even though the food was inaccessible. Kö hler tested chickens in a similar way, but they rarely solved the problem. Instead, they persistently tried to reach the food by pushing through the barrier. The chickens were successful only when their attempts to pass through the fence led them by chance to pass around it. Kö hler suggested that chickens are not capable of solving detour problems. However, subsequent studies have challenged this view. For example, Etienne (1973) presented 6-day-old chicks with a mealworm that disappeared behind one of two screens. She found that, after repeated testing, all chicks developed searching behaviour behind either screen and a minority of the birds (24%) spontaneously learnt to orient their delayed response directly to the correct screen.

During last decades it has been demonstrated that it was not so much cognitive deficit in animals as deficit of motivation and biological significance of stimuli in early studies of animals’ detour behaviour that was responsible for their failures. For instance, young chicks provide experimenters with the unique possibility to test a highly motivated animal that is eager to follow a moving object that was imprinted a few hours after its birth as a “mother”. This may be a ball, a cupboard triangle, or something similarly far from a real bird (see details in Chapter 24). Regolin et al. (1995) tested 2-day-old chicks in a detour situation requiring them to abandon a clear view of a desired goal (a small red object on which they had been imprinted) in order to achieve that goal. The chicks were placed in a closed corridor, at one end of which there was a barrier with a small window through which the goal was visible. Two symmetrical apertures placed midline in the corridor allowed the chicks to develop strategies to pass round the barrier. After entering the apertures, chicks showed searching behaviour for the goal and appeared able to locate it, turning either right or left depending on their previous direction of turn. Thus, in the absence of any local orienting cue emanating from the goal, chicks showed an ability to understand the continuing existence of an object that had disappeared and to represent its spatial localisation in egocentric coordinates.

Recent comparative study in three species of birds, belonging to different eco-ethological niches, allows for a better understanding of the cognitive mechanism of such detour behaviour (Zucca et al., 2005). Young quails (Coturnix sp.), herring gulls (Larus cachinnans) and canaries (Serinus canaria), were tested in a detour situation requiring them to abandon a clear view of a biologically interesting object (their own reflection in a mirror) in order to approach that object. Birds were placed in a closed corridor, at one end of which was a barrier through which the object was visible. After entering the apertures, birds could turn either right or left to re-establish social contact with the object in the absence of any local sensory cues emanating from it. Quails appeared able to solve the task, though their performance depended on the type of barrier used, which appeared to modulate their relative interest in approaching the object or in exploring the surroundings. Young herring gulls also showed excellent abilities to locate spatially the out-of-view object. Canaries, on the other hand, appeared completely unable to solve the detour task, whatever barrier was in use. It is suggested that these species differences can be accounted for in terms of adaptation to a terrestrial or aerial environment.

Pongracz et al. (2001) have exploited dogs’ attachment to their owners in order to estimate limits of social learning (see Part VIII) with the use of detour tests. The experimenters used V-shaped fence, 1 m high, with sides 3 m long. The task for the dogs was to get to a piece of food or their favourite toy by detouring along the fence. 30 dogs were divided between two equal groups that differed in the direction of detour necessary for reaching the target. For the inward detour group the experimenter placed the target behind the V-shaped fence near to the inner side of the angle. After six trials the position of the target and the dog were reversed (outward detour). In this seventh trial the dog was positioned inside the fence and the target was placed at the outer side of the intersecting angle. Dogs in the outward group were exposed to the same test procedure but in the reverse order. As one of the main characteristics, the dog’s latency to obtain the target was measured, this was defined as the time elapsed between the owner releasing the dog from the leash and the dog taking the target in its mouth.

The results showed that the dogs performed differently depending on their position with regard to the fence. Learning the detour seemed to be a difficult problem for the dogs from the inward group. In contrast, dogs in the outward group mastered this task much more easily; latencies were significantly shorter after the first trial. The direction of the detours along the fence in the outward group showed strong concordance with the direction of the first successive detour. The inward group lacked such a concordance. This could indicate the difficulty of this task for the dogs; these dogs might have got behind the fence by chance during the first trial, which made it more difficult for them to remember the direction in which they started.

The dogs displayed relatively conservative strategies in mastering detours. Even after several consecutive trials animals could not improve significantly if they started from outside the fence. Dogs were apparently unable to transfer their experience of mastering the task from inside the fence to a reverse detour. Pongracz et al. (2001) have demonstrated that the dogs were not flexible performing detour tests and they coped with the problem more easily starting from inside the V-shaped fence then when outside it. A possible explanation for this asymmetry lies in stored individual experience that is, in many respects, similar in domestic dogs. It could be, as the authors suggest, that dogs might more often encounter situation in which they had to get out from somewhere (e.g. from a garden), rather than get inside it. Furthermore, in the outward detour there is less ambiguity, that is, any exit is successful. It is also could be that dogs in the outward group had to walk mainly tangentially to the target, but the inward detour needed a long walk away from the target at first. One could also argue that the inward detour situation might generate higher levels of neophobia in the dogs than the outward detour task. However, the fence was an equally strange obstacle for the dogs in both situations. Furthermore, dogs in the inward detour group tried strongly to obtain the target, barking at it and sometimes trying to dig under the fence.

Indeed, this was not without reason that Thorpe (1950) estimated capability of Digger Wasp to solve detour task as higher than in dogs. These animals brave fences without wasting time for barking. The most astonishing thing is that the insects demonstrated ability to transpose a picture of their place from a “bird's” (wasp’s) -eye view to a “pedestrian”- eye view. In 1950 Thope did some experiments that suggested that digger wasps rely on a cognitive map of their large-scale environment to find their way to the vicinity of their nest from wherever they happen to capture the prey. He worked with the species of digger wasp that prey on caterpillars too large to be carried home in flight. A female of this species drags her prey home across the gravely ground. In his experiments Thorpe manipulated a wasp dragging its prey homeward. He placed obstacles in its path that forced the wasp to deviate from its course. As soon as the wasp had cleared each obstacle, it resumed its course. This means, as Gallistel (1998) gives this, that the wasp was not marching blindly along like a wind-up toy tank, without regard to where it’s marching was actually getting her, instead, when forced to deviate from the course, it corrected for the deviation it had made.

Extrapolation. Animals’ capacity to make detours could be considered a particular case of more complex ability to extrapolate trajectories of moving objects. Thompson Seton (1898) described a raven that lost a piece of bread over a river. Being entrained by a stream, the bread disappeared within a tunnel. After a short glance into the tunnel, the raven flied round it, waited till the bread appeared and picked it up. Every happy owner of a puppy or a kitten can add personal anecdotes to this story. Usually young dogs and cats do not attack a cupboard when a ball rolls under it; instead they prefer to wait for a ball at the other side extrapolating a trajectory of their favourite toy.

In his experiments with chimpanzees Kö hler (1925) simulated such a situation. For example, in one of experiments a basket containing fruits was fixed by a long rope. An experimenter set the basket in oscillatory motion in such a way that in a certain moment it passed by one of couple of rafters. As soon as the animal caught sight of this, it jumped on the rafter and then waited for the basket with its hands extended. In the second experiment a chimpanzee entered a room in which a window was closed by shutters. Before animal’s eye the experimenter opened the shutters, threw a fruit out and closed the shutters again. Instead of attacking the window, the chimpanzee immediately started to move in an opposite direction. It ran out into the garden and began searching for the fruit under the window. In both situations the chimpanzee’s behaviour could be explained by animal’s ability to grasp a problem and extrapolate trajectories of desirable things when they are moving.

The term “ extrapolation ” for explanation of animal behaviour was first suggested by Matthews (1955) who hypothesised that pigeons extrapolate the movement of the Sun for navigation. L.V. Krushinsky (1958, 1977) was the first to consider extrapolation a key element of reasoning in animals. Krushinsky and his colleagues and students carried out plenty of experiments on different species estimating their ability to extrapolate trajectories of objects disappearing from the animal view (corridor test, screen test, etc.; Krushinsky, 1990).

The tests of these series differ from the detour tests in which a subject can observe only a segment of the trajectory of the desired object (see Fig.IV-4). For example, in a series of “corridor” tests an animal owed to extrapolate the trajectory of a bowl with food moving in the opposite direction relatively to the animal view. First, on the path to a tunnel a bowl was open for following by and taking food from. Then it drove into the opaque tunnel which was immediately shut in order to prevent a subject from seeing the following movement of the bowl. These experiments were conducted on rabbits and several avian species. Rabbits, hens, ducks and pigeons persisted in searching for the bowl at the place where it had disappeared just before, while corvids manifested ability to extrapolate. It was clear at the second stage of experiment when the tunnel contained two ells with a gap between them. Magpies, after they caught site of the bowl in the gap, hurried to the end of the second ell where the bowl was expected to appear. Standing there in a tense posture, they were waiting for food.

The “screen”- tests were so devised as to provide the subject with information about the trajectory of a desirable stimulus (such as a trough with food or a toy) by a possibility to look at the stimuli through a narrow vertical slit in the centre of the opaque screen. When food served as a stimulus the animal was allowed to eat from the trough through the gap in the screen and then the filled trough was moving in one direction while an empty trough was moving in the opposite direction in order to prevent an animal to orient by sounds. These directions changed in different trials. The subject had to solve a problem based on its observations on how the trough moved when it could be seen through a narrow vertical slit. To solve this problem an animal must realise that the food bait, which has disappeared from its view, continues to move in the same direction as before, i.e. to perform extrapolation of the movement direction of the invisible food bait and, basing on this knowledge, to go round the respective side of the screen. Therefore, the animal had to decide from what side it should run round the screen in order to obtain food. Experimenters complicated the screen tests by added screens. In a simple case the added screen forced the animal to deviate from the course toward the target for moving athwart to the traffic rout and then to resume its course. In more complex cases the animal had to run round several added screens. This task requests not only ability to extrapolate but to sum vectors as well.

After the first tests on dogs, a wide range of species were studied.

Comparative studies revealed distinct tendencies concerning species specific abilities to extrapolate in animals. They ranked as follows: voles and mice displayed almost complete inability to solve even simple extrapolation tasks; then followed rats, rabbits, cats, dogs, foxes, corsacs, polar foxes, racoons, wolves, dolphins and apes (Krushinsky, 1965; Firsov, 1977). Among avian species only corvids manifested high level of capability while chickens, ducks, pigeons, kites and the honey buzzard did not cope with extrapolation problems. In some birds the number of correct choices gradually increased, that is, they become trained for performing these tasks. Surprisingly, reptiles (including three species of tortoises and one species of lizards) championed in many tasks. Similar experiments were conducted on human infants whose attention called to an underway toy. It turned out that two-year children behave a good deal like rabbits. They were eager to get a toy and refused to leave the place where it disappeared. Six-year-old children are able to solve the most complex tasks in majority of trials, and they reach 100% level of capability by 7, 5 years.

Recently detour tests for measuring animal’s ability to extrapolate trajectories of a shifting object has been implemented on a computer for those animals that could watch movements of a dot on a monitor and react in adequate manner. Both apes and monkeys displayed capability to extrapolate a trajectory of a dot on a monitor with the same level of accuracy as adult humans. It is interesting to note that when primates learnt to mentally pass ahead a dot moving on the monitor, say, from left to right, they easily transfer these acquirements for doing the same with the dot that moved top-down (Washburn and Rumbaugh, 1992).

 






© 2023 :: MyLektsii.ru :: Мои Лекции
Все материалы представленные на сайте исключительно с целью ознакомления читателями и не преследуют коммерческих целей или нарушение авторских прав.
Копирование текстов разрешено только с указанием индексируемой ссылки на источник.