Travel into the Past: delayed response behaviour

Delayed response behaviour is a behaviour that does not only depend on the current stimuli, but also depends on stimuli the subject had as input in the past. This pattern of behaviour cannot be described by direct functional associations between current input and output (Griffin, 2001).

In order to make situations in which animals display their ability to operate with past and future fathomed, the experimental technique was elaborated by Hunter (1912) and later transformed by Tinklepaugh (1928, 1932) and Harlow with co-authors (Harlow et al., 1932). This scheme was called a “ delayed reaction test “(“ delayed response procedure ”, “ delayed reaction time task ”). The simplest task required the animals to discriminate between two caps placed in different locations, one of which had previously been baited with food.

The set up described in Griffin (2001) after Tinklepaugh (1932) is working as follows (Fig.III-2). Separated by a transparent screen (at position p0), at each of the two positions p1 and p2 a cup (upside down) and a piece of food can be placed. At some moment (with variable delay) the screen is raised, and the animal is free to go to any position. Within this experiment the animal is able to observe the position of food, cups, screen and itself.

The following situations are considered:

1. Situation 1. At both positions p1 and p2 an empty cup is placed.

2. Situation 2. At position p1 an empty cup is placed, and at position p2

a piece of food, which is visible for the animal.

3. Situation 3. At position p1 an empty cup is placed and at position 2 a

piece of food is placed, after which a cup is placed at

the same position, covering the food. After the food

disappears under the cup the animal cannot sense it

anymore.

4. Situation 4. At position p1an empty cup is placed and at position p2 a cup and a piece of food are placed, in such a manner that

the animal did not see the food.

After the screen has been taken away, in situation 1 the animal, not having observed any food, will not show a preference for either position p1 or p2; it may even go elsewhere or stay where it is. In situation 2 the animal, observing food at position p2 and that no screen is present, will go to position 2, which can be explained as pure stimulus-response behaviour. In situation 3 the immediate stimuli are the same as in situation 1: no food is observed, and it is observed that no screen is present. Animals that react in a strictly stimulus response manner will respond to this situation as in situation 1. Animals that show delayed response behaviour will go to p2, where food can be found, as has been observed in the past. In situation 4 animals will behave as in situation 1; however they can find food and eat it, and therefore be reinforced in successful behaviour.

This scheme of experiments has been applied to many species such as rats, dogs, cats, racoons, macaques, chimpanzees and human infants. It turned out that rats, dogs and cats orient their bodies towards the cup with food and keep themselves axial during the delay, that is, follow their own noses. Being disturbed by an experimenter, they fail to find food. Racoons are able to stand the test without resorting to “following the nose” if the delay is short. Primates demonstrate delayed response within tests with relatively long delays, besides, they do not need to keep themselves axial, and moreover, they perform successfully even being isolated in other room during delay.

Tinlepaugh (1932) found that monkeys were very accurate in this task, but the accuracy declined when the task was made more complex by increasing the number of cup pairs. Two young chimpanzees coped with the complex task within the following scheme. The animals were led through six rooms and in each they were shown how the food was placed under one of two boxes. Then the animals were returned into initial point and allowed to go through all rooms but raise only one from two boxes in every room in order to obtain food. If the animal failed in one room, it was led into the next room without any reward. The accuracy was demonstrated from 88% to 92% successes.

The same subjects came through even the more complex test. An animal sat on a chair in a centre of a room where 16 pairs of boxes were placed around. The animal could see how the food was placed by the experimenter under one box from each pair. Then the animal was led in turn to each pair of boxes and allowed to raise one of them. In a case of success, the animal returned to the chair, ate the food and then approached the next pair of boxes and so on, clockwise. The accuracy was between 78% and 89% and this was higher than in human infants in similar tests. The chimpanzees remembered quite well what kind of food was placed under the box and demonstrated signs of disappointment and indignation when the preferred food was replaced by another kind of items or big item was replaced by a small bit.

One of modifications of the delayed reaction test, known as delayed-matching-to sample (DMTS), has proved extremely useful for the study of animal memory. The procedure is the following.

Subjects are presented with one of two stimuli at the beginning of a trial. In the case of pigeons this could be the illumination of a response key by either red or green light. After a while this sample stimulus is turned off and nothing is presented for a period known as retention interval. At the end of this interval, two different response keys are illuminated, one with red, the other with green light. These are referred to as the comparison stimuli. To gain reward, the pigeon must peck the comparison colour that is the same as the sample presented on the particular trial. Pecks to the other colour result in both keys being darkened and no reward was presented. After the completion of the trial there is a period in which nothing happens- the inter-trial-interval (ITI) - before the sample is presented for the next trial. In order to gain a reward, the subject must store information at the time the sample is presented and use it to select the correct response when the comparison stimuli are subsequently available (see: Pearce, 2000).

DMTS seems to be a rather difficult task for most animals. Subjects must first be trained to a simpler, matching-to sample test with the single sample and two comparison stimuli presented simultaneously. Once they have learned to peck the comparison that matches the sample, the comparison stimuli are presented as soon as the sample is turned off. After considerable training, this 0-second retention interval is gradually extended, but rarely to long intervals. For example, most researchers use delays of 5 to 10 seconds when pigeon used as subjects. In one study reasonably accurate performance was obtained by pigeons on DMTS with a 1-minute interval. This was achieved at the expense of considerable effort by both experimenter and subjects, as it required some 17000 trials (Grant, 1976). In monkeys accurate retention was possible for 2 minutes, and with careful training this can be extended to 9 minutes (D’Amato and Worshman, 1972). In experiments with dolphins, accurate DMTS has been achieved with a retention interval of 4 minutes (Herman and Thomson, 1982).

Another well known technique for studying delayed reaction is WGTA (Wisconsin Test Apparatus) set elaborated by Harlow and co authors (Harlow et al., 1932; Harlow and Bromer, 1938). Since that time, numerous versions of the WGTA have been utilized, including recent computerized model operated by animals using touch screens or joysticks. This apparatus standardises situations and gives a possibility for mapping brain activity during trials. WGTA is one of the basic techniques based on stimulus discrimination, which is still actually used for a battery of tests, from simple discrimination to categorisation. As far as delayed response procedure is concerned, the basic task is the same as in Hunter’s (1912) tests: an animal must remember a cue stimulus over a delay period and then make a behavioural response based on the cue.

Despite the many modifications over the years, including increased automation, certain elements of the WGTA design remain standard. It is essentially a testing apparatus with: (1) a stimulus tray, (2) a food reward delivered under 1-3 objects, (3) the option to permit or deny a subject the opportunity to observe placement of the food reward under an object, (4) an observation interval in which subjects can see but not displace objects, (5) a subsequent interval in which subjects can obtain objects and food rewards, and (6) a one-way screen for experimenter observation. This basic arrangement allows both the researcher and the subject to manipulate objects in a safe and controlled environment.

In a typical scheme, which is used nowadays (see, for example,

O’ Donnell et al., 1999; Buckmaster et al., 2004) a subject, for example, a monkey observes through a clear barrier while one of the two lateral wells, of the WGTA stimulus tray is baited with a food reward. The wells are then covered with plaques identical in appearance and the barrier is raised immediately to permit a response. Many trials are provided per daily test session. The left/right position of the reward is balanced within a session. Usually animals are tested until they reach a 90% correct criterion (9 or fewer errors in 9 consecutive 10-trial blocks). Testing subsequently continues in the same fashion except that 1-s retention interval is imposed by lowering the opaque door of WGTA between the baiting and response phase of each trial. When the performance criterion has been re-achieved, the memory demands of the task are made progressively more challenging by imposing successively longer delays.

It is worth noting that a concept of secondary reinforcement (see Chapter 6) is used in many experimental schemes being implemented on WGTA. This is a so-called “ go-signal ”. For example, if a required response is key-pressing, then, after a delay period a bridge stimulus is presented for an animal, say, white light is presented on the key as a go signal and the animal is required to press the key within 1 s. A typical “go-no-go” procedure is illustrated by the Fig. 2-1 where a pigeon is acting within a “go-no-go” conditioning chamber in Huber’s laboratory.

Recent studies based on a technique of delayed- matching-to sample have helped to make the distinction between implicit and explicit processes in animals. For example, Hampton (2001) developed a new method by which two rhesus monkeys were trained in a match to sample procedure to report, by pressing the appropriate image on a touch-sensitive video monitor; whether they did or did not remember one of four visual patterns they had seen a short time ago. In this test touching the correct image yielded the food, but pressing the wrong image produced no food. Before the apparatus presented four images, of which one had been seen previously, the monkey had the opportunity to press either of two different images, one of which caused the test stimuli to appear, whereas the other avoided the test and yielded a food item that was less preferred than what would be delivered after a correct choice in the test. The monkeys learned to avoid the test when they did not remember the images well enough to believe they could make a correct choice. When the four images had been seen only 15 to 30 sec previously they almost always chose to take the test, touched the correct image, and receive a food reward. But when the interval was 2 to 4 min they made much more errors if they chose to take the test. These animals certainly seemed to know when they did and did not remember a visual pattern.

Another Hampton’s (2003) experiment was based on the following scheme. At the start of each trial the animal studies a picture, and a delay follows. On 2/3 of trials the monkey then gets to choose between taking a memory test and escaping from the test. On 1/3 of trials the monkey is only given the option of taking the test. Both monkeys tested were more accurate when they chose to take tests than when they had no choice about taking tests, indicating that they adaptively escaped from trials on which they had forgotten the picture. In order to rule out the use of cues other than the absence of a memory for the picture in controlling the escape response, Hampton presented monkeys with occasional probe trials on which no picture was presented for study. Since no picture entered the animals’ memory on these trials, they should treat them like trials on which they had forgotten the picture. The monkeys were much more likely to escape on trials where no picture had been presented than they were to escape from normal trials. These results demonstrate that monkeys know whether or not they remember a recently seen stimulus - a form of explicit memory.

Many reports of observed delayed response behaviour in experiments of different types, from simple to complex, enabled researchers to assume that animals of the type studied maintain internal (mental) representations of the world state on the basis of their sensor input, and that they make use of world state model (in addition to the actual sensory input that is used) to determine their behaviour (see Vauclair, 1996; Griffin, 2001).

Training subjects in more natural settings (e.g. using biological stimuli as discriminative cues) might yield more impressive results. It is well known that in more natural situations animals often amaze their trainers with their capability, speaking in anthropomorphic manner, to keep in mind many events and imaginations for a long time. Kö hler and Yerks gave many examples of how chimpanzees accepted friendly their mates after from 4 to 18 months of being separated. They also recognised their human friends after long intervals. For example, one ape after four years recognised a man who once restored it to health and felt on his neck affectionately.