Biological Influences on Learning: Belongingness

Biological Influences on Learning: Belongingness

In the early days of learning theory, investigators widely believed that animals (both humans and others) are capable of connecting any CS to any US in classical condition-ing, and of associating virtually any response with any reinforcer in instrumental con-ditioning. A child could be taught that a tone signaled the approach of dinner or that a flashing light or a particular word did. Likewise, a rat could be trained to press a lever to get food, water, or access to a sexually receptive mate.

But a great deal of evidence speaks against this idea; instead, each species seems pre-disposed to form some associations and not others. The predispositions put biologicalconstraints on that species’ learning, governing what the species can learn easily andwhat it can learn only with difficulty. These associative predispositions are probably hardwired and likely to be a direct product of our evolutionary past (Rozin & Kalat, 1971, 1972; Seligman & Hager, 1972).

TASTE AVERSION LEARNING

A central example of the biological constraints on learning comes from studies of tasteaversion learning. These studies make it clear that, from the organism’s viewpoint, somestimuli belong together and some do not (Domjan, 1983, 2005; Garcia & Koelling, 1966).

To understand this phenomenon, we need to begin with the fact that when a wild rat encounters a novel food, it generally takes only a small bite at first. If the rat suffers no ill effects from this first taste, it will return (perhaps a day or two later) for a second helping and will gradually make the food a part of its regular diet. But what if this novel food is harmful, either because of some natural toxin or an exterminator’s poison? In that case, the initial taste will make the rat sick; but because it ate only a little of the food, the rat will probably recover. Based on this experience, though, the rat is likely to develop a strong aversion to that particular flavor, so it never returns for a second dose of the poison.

This sort of learning is easily documented in the laboratory. The subjects, usually rats, are presented with a food or drink that has a novel flavor—perhaps water with some vanilla added. After drinking this flavored water, the rats are exposed to X-ray radiation—not enough to injure them, but enough to make them ill. After they recover, the rats show a strong aversion to the taste of vanilla and refuse to drink water flavored in this way (Figure 7.30).

This learned taste aversion is actually based on classical conditioning. The flavor (here, vanilla) serves as the CS, and the sensation of being sick serves as the US. This is, however, a specialized type of classical conditioning that is distinct from other forms in the sheer speed of learning: One pairing of a taste + illness is all it takes to establish the connection between them. This one-trial learning is obviously much faster than the speed of ordinary classical conditioning. What’s more, this form of condi-tioning is distinctive in its timing requirements. In most classical conditioning, the CS must be soon followed by the US; if too much time passes between these two stim-uli, the likelihood of conditioning is much reduced (see Figure 7.10). In taste aversion learning, in contrast, conditioning can be observed even if several hours elapse between the CS and the US.

Learned taste aversions are also remarkable for their specificity. In one early study, thirsty rats were allowed to drink sweetened water through a tube. Whenever the rats licked the nozzle of this tube, a bright light flashed and a loud clicking noise sounded. Thus, the sweetness, bright light, and loud noise were always grouped together; if one was presented, all were presented. One group of these rats then received an electric shock to the feet. A second group was exposed to a dose of X-rays strong enough to cause illness.

Notice, then, that we have two different USs—illness for one group and foot shock for the other. Both groups also have received a three-part CS: sweet + bright + noisy. The question is: How will the animals put these pieces together—what will get associ-ated with what?

To find out, the experimenters tested the rats in a new situation. They gave some of the rats sweetened water, unaccompanied by either light or noise. Rats that had received foot shock showed no inclination to avoid this water; apparently, they didn’t associate foot shock with the sweet flavor. However, rats that had been made ill with X-rays refused to drink this sweetened water; they associated their illness with the taste (Figure 7.31).

Another group of rats were tested with unflavored water accompanied by the light and sound cues that were present during training. Now the pattern was reversed. Rats that had become ill showed no objection to this water. For them, the objectionable (sweet) taste was absent from this test stimulus, and they didn’t associate their illness with the sights and sounds that were present during the test. However, rats that had been shocked earlier refused to drink this water; in their minds, pain was associated with bright lights and loud clicks (Garcia & Koelling, 1966).

For the rat, therefore, taste goes with illness, and sights and sounds go with exter-nally induced pain. And for this species, this pattern makes biological sense. Illness in wild rats is likely to have been caused by harmful or tainted food, and rats generally select their food largely on the basis of flavor. So there’s survival value in the rats being able to learn quickly about the connection between a particular flavor and illness; this will provide useful information for them as they select their next meal, ensuring that they don’t resample the harmful berries or poisoned meat.

Using this logic, one might expect species that choose foods on the basis of other attributes to make different associations. For example, many birds make their food choices from a distance, relying on the food’s visual appearance. How will this behavior affect the data? In one study, quail were given blue, sour water to drink and were then given a low dose of poison—enough to make them ill, but not enough to harm them. Some of the birds were later tested with blue, unflavored water; others were tested with water that was sour but colorless. The results showed that the quail had developed a strong aversion to blue water but no aversion to the sour water. They learned which water was safe based on its color rather than its taste (Wilcoxin, Dragoin, & Kral, 1971).

Clearly, what belongs with what depends upon the species. Birds are predisposed to associate illness with visual cues. Rats (and many other mammals) associate illness with taste. In each case, the bias makes the animal more prepared to form certain asso-ciations and far less prepared to form others (Seligman, 1970).

We should also mention that taste aversion learning, as important as it is, is just one example of prepared learning (Figure 7.32). We mentioned a different example in the Prologue: Humans in one experiment were shown specific pictures as the CS and received electric shocks as the US. When the pictures showed flowers or mushrooms, learning was relatively slow. When the pictures showed snakes, learning was much quicker. The impli-cation is that humans (and many other primates) are innately prepared to associate the sight of a snake with unpleasant or even painful experiences (Öhman & Mineka, 2003; Öhman & Soares, 1993; also Domjan, Cusato, & Krause, 2004).

These results may help us understand why so many people are afraid of snakes and why strong phobias for snakes are relatively common. Perhaps it’s not surprising that many cultures regard snakes as the embodiments of evil. All these facts may simply be the result of prepared learning in our species—our innate tendencies toward making certain associations but not others.

BIOLOGICAL CONSTRAINTS ON INSTRUMENTAL  CONDITIONING

Prepared learning can also be demonstrated in instrumental conditioning because, from an animal’s viewpoint, certain responses belong with some rewards and not oth-ers (Shettleworth, 1972). For example, pigeons can easily be taught to peck a lit key to obtain food or water, but it’s extremely difficult to train a pigeon to peck in order to escape electric shock (Hineline & Rachlin, 1969). In contrast, pigeons can easily be taught to hop or flap their wings to get away from shock, but it’s difficult to train the pigeon to produce these same responses in order to gain food or water.

Once again, this pattern makes good biological sense. The pigeon’s normal reaction to danger is to hop away or break into flight, so the pigeon is biologically prepared to associate these responses with aversive stimuli such as electrical shock. Pecking, in con-trast, is not part of the pigeon’s innate defense pattern, so it’s difficult for the pigeon to learn pecking as an escape response (Bolles, 1970). Conversely, since pecking is what pigeons do naturally when they eat, the pigeon is biologically prepared to associate this response with food or drink; it’s no wonder, then, that pigeons easily learn to make this association in the psychologist’s laboratory.