Is it easy to distinguish between instinctive and learned behaviour?

The hopes for an easy distinction between instinctive and learned behaviour have been forsaken when Ethology entered the scene and when it was synthesised with other sciences of Animal Behaviour and Psychology. As Verplanck (1955) expressed this in his Psychological Review “Since learned behaviour is innate, and vice versa, what now? ” Early ethologists believed that behaviour is largely instinctive, or innate, as the product of natural selection. Soon it became clear that individual behaviour develops as a puzzle of environmental and genetic components, and very often it is difficult to find reliable signs to discriminate innate and learned behaviour. This problem has many aspects, from which we consider two in this issue: how innate behaviour can prevent organisms from learning although it is demanded by changes in the environment; and how conservative learned behaviour can prevent animals from learning something new.

Animals are pressed with innate rules. Let us consider several examples which demonstrate genetic basis of behaviour and provide evidence that in some cases innate rules can block new learning.

The most reliable methods that allow discriminating between innate and flexible components of behaviour are based on genetic approach. It is possible to select inbred lines of animals possessing certain specialised behaviours. In his famous experiment Tryon (1940) studied the ability of rats to find a way through a complex maze in order to obtain a food reward. Since some rats appeared to be fast learners, Tryon bred them with one another to establish a “maze-bright” colony, and he similarly bred the slow learners with one another to establish a “maze dull” colony. The offspring of maze-bright rats learned even more quickly than their parents had, while the offspring of maze dull parents were very poor at maze learning. After repeating this procedure over several generations, Tryon was able to produce two behaviourally distinct types of rats with very different maze-learning abilities. Clearly the ability to learn the maze was to some degree hereditary. It is important to note that genes that governed maze learning abilities were specific to this behaviour, as the two groups of rats did not differ in their ability to perform other tasks, such as running a completely different kind of maze. This study supports a hypothesis about the genetic basis of specific domains responsible for certain learning abilities.

Although recent research has provided much greater details of the genetic basis of behaviour, let us go to a classic example that brilliantly illustrates a battle between inherited behavioural pattern and learning. Dilger (1962) has examined two species of parrots, which differ in the way they carry twigs, paper, and other materials used to build a nest. Fischer's lovebirds (Agapornis personata fischeri) carry single strips of nest material in their beaks, while Peach-faced lovebirds (Agapornis rosecollis) carry material tucked under their flank feathers. The tucking behaviour was a leftover from ancestor species that lined their nests with small chips, which more easily stay put in the tail feathers. These two species can interbreed and produce sterile offspring. The hybrids showed a poorly organized mixture of the two strategies: they tuck nest material between their feathers but failed to go, pull it out again, and start over. After two years they become partly successful, managing to transport some material back to the next site, but not in a manner that resembles either parent species. Sometimes they would just turn their heads toward their rumps without tucking, and would then fly off with the material. In later mating seasons poor hybrids transported the material in their beaks but they performed head-turning behaviour each time they took an item to carry it to the nest. Parrots are known as fast learners but in this case innate predisposition to a certain behaviour blocked birds’ ability to learn a relevant way for nest building, and their attempts to improve practical skills were almost unsuccessful.

There is some more experimental evidence that innate rules may block learning. For instance, in experiments of Mazokhin-Porshnyakov and Kartsev (1984, 2000) honey bees and Polistes wasps were presented with the following task. Four small troughs with syrup were situated on a table covered with glass (see Fig. V-4). Troughs contained a small amount of food so that foragers had to visit all of them to be sated. Each trough was placed on an icon of a geometric form, and all of them were situated on the table in different orders, that is, diagonally, in corners of a square and so on. The troughs containing syrup alternated with the troughs containing salt, and this was a penalty for an insect to find its proboscis in salt. Wasps quickly learned to escape penalty and choose correct pictures. Bees insisted on choosing the nearest trough each time and thus alternated syrup with salt. It seems, as the authors explain this, that innate rules of searching are more flexible in wasps than in honey bees. In honey bees learning was blocked by their strong innate rule for searching, that is, “Remember how a bee plant looks like, take nectar and fly to the nearest one”. Since the task to fly troughs around simulates a natural situation, honey bees switched their behaviour to following one of their main searching rules. As this was described in Chapter 15, honey bees are able to solve very complex problems which demand capacity for abstraction and categorisation. Those tasks probably were so different with natural problems that innate rules did not prevent honey bees from learning.

Animals are pressed with learned rules. In his book, King Solomon's Ring, Lorenz (1952) describes how water shrews learn their paths in their home range. He considers this striking example of how learned novel behaviour may become highly routine and stereotyped. The water shrew is aquatic, but it also spends some time on land. Here it is nearly blind, and finds its way around mainly through its sense of touch and long whiskers. Once the shrew has learned a path it is bound to it, as Lorenz writes, as a railway engine to its tracks and is as unable to deviate from them by even a few centimetres. In order to examine what happens if there is an obstruction on the path, Lorenz experimented by moving a stone which had been on one of the shrew’s paths. This is what he found: " The shrews would jump right up into the air in the place where the stone should have been; they came down with a jarring bump, were obviously disconcerted and started whiskering cautiously right and left, just as they behaved in an unknown environment. And then they did a most interesting thing: they went back the way they had come, carefully feeling their way until they had again got their bearings. Then, facing around again, they tried a second time with a rush and jumped and crashed down exactly as they had done a few seconds before. Only then did they seem to realise that the first fall had not been their own fault but was due to a change in the wonted pathway, and now they proceeded to explore the alteration, cautiously sniffing the place where the stone ought to have been. This method of going back to the start and trying again always reminds me of a small boy who, on reciting a poem, gets stuck and begins again at an earlier place."

This bright example demonstrates that not only in laboratories but in nature learned behaviours may become “automatic”, and animals may follow these learned stereotypes for a long time even if such fidelity becomes completely useless. Chains of acts become “ritualistic” as we have already seen when considered behaviour of Pavlov’s chimpanzees who performed complex and long chains of behaviours in order to gain access to a cistern and scoop water there in order to put fire out. Apes did not surmise to scoop water from a surrounding lake and thus solve a problem by a single operation. Learned stereotype became a ritual just like in shrews jumping over a virtual stone on their way home.

Dog’s trainers know that it is necessary to vary a sequence of commands given to a dog during each training sessions. Should a trainer several times repeat two commands in the same sequence (“sit- lie”, and “sit-lie” again), and a dog will always perform these two commands one after another. The trouble is that it is much easier to teach a dog to perform something new than to teach it again and to force it to perform each command separately. The trainer now feels sorry that his dog is so fast a learner. Not being loaded with stereotypes so quickly wired, the dog would be more susceptible to a trainer’s request.