Learning preparedness: some associations can be built more readily than others

There are many examples in classic and modern ethological literature that some species demonstrate the so-called innate recognition of certain stimuli whereas others do not. In human studies a concept of “preparedness” was introduced by Seligman (1970, 1971) to explain why fears and phobias are so much more likely with certain stimuli (e.g. snakes, spiders) than with others (see: Davey, 1995, for a review). Seligman (1970) summarized evidence from animal instrumental-learning paradigms to suggest that for a particular organism, certain behaviours differ in their potential to be successfully conditioned. He also analysed in details protocols of “The Little Albert Study” by Watson and Rayner, (1920) and concentrated on the ease with which Albert was conditioned, the durability of his reactions, and on several other details of the experiment (see a description of the experiment in Chapter 2). In particular, Seligman (1971) noted that the experimenter did not get fear conditioning to a wooden duck, even after many pairings with a startling noise. Although Seligman seemed to misinterpret some details of Watson’s experiment (see Harris, 1979), he, however, was able to create a viable concept. In general, the concept of “preparedness” includes many aspects of readiness of organisms to form certain associations easier than other.

There are many evidences of that what animals can learn is often biologically influenced, that is, in many (may be in most) cases learning is possible within the boundaries set by instinct. Several examples were described in previous chapters. It is now clear that some animals have innate predisposition towards forming certain associations. For example,

if a rat is offered a food pellet and at the same time is exposed to X rays (which later produces nausea), the rat will remember the taste of the food pellet but not its size. Conversely, if a rat is given a food pellet at the same time an electric shock is delivered (which immediately causes pain), the rat will remember the size of this pellet but not its taste. Similarly, pigeons can learn to associate food with colours but not with sounds; on the other hand they can associate danger with sounds but not with colours. These examples of learning preparedness demonstrate that what an animal can learn has adaptive significance. The seed a pigeon eats may have a distinctive colour that the pigeon can see, but it makes no sound the pigeon can hear.

Let us consider in details one of the most significant and representative sets of examples concerning predator recognition in animals. The idea that some birds possess an innate tendency to escape when they perceive a configuration of a bird of prey goes back to Spalding (1873). Up to now, some studies have suggested that predator recognition is innate in animals whereas others have stressed the importance of learning. Several different nonexclusive possibilities of how animals acquire predator recognition have been suggested: animals may have a genetically programmed ability to recognise predators without earlier experience with that very predator, or they may be sensitised to learn about particular stimuli very quickly. There may also be age-related maturation of the ability to express a recognition response. Furthermore, learning from conspecifics through social learning has been shown to exist in many animals (review in: Kullberg and Lind, 2002). Here we are interested in the first two sources of animal knowledge about possible danger, namely, innate predator recognition and predisposition for quick learning.

Tinbergen (1951) conducted an elegant experiment examining the reaction of naï ve chicks to a mounted silhouette model of a flying bird. The wings were symmetrical; at one end there was a short protuberance, at the other end a longer one. When a model was pulled along a guide-wire in one direction, it appeared to have a long neck and a short tail, like a goose. Flown the other way, it resembled a short-neck bird of prey. In the first case, it was ignored by young chickens, turkeys, ducks and geese on the ground. When they viewed it being pulled along the other way, however, they tried to escape. The results of this experiment are still discussible; moreover the matter concerns such intricate problem as “innateness” of complex releasers. The interpretation of Tinbergen’s results has been questioned since the birds under study were reared in a natural environment in which they have been more habituated to geese than to hawks.

Subsequent studies on several species of fishes, birds, mammals and spiders have shown clear evidence for an innate component of predator recognition (see reviews in: Kullberg and Lind, 2002; Caro, 2005).

For example, Veen et al. (2000) compared predator recognition responses in two isolated but genetically similar Seychelles warbler (Acrocephalus sechellensis) populations, only one of which had experience with the egg predating Seychelles fody (Foudia sechellarum). Individuals in the predator-free population significantly reduced nest guarding compared to individuals in the population with the predator, which indicates that this behaviour was adjusted to the presence of nest predators. However, recognition responses (measured as both alarm calls and attack rates) towards a mounted model of the fody were equally strong in both populations and significantly higher than the responses towards either a familiar mounted non-predator or a novel mounted non-predator bird species. Responses did not differ with a warbler's age and experience with the egg predator, indicating that predator recognition is innate in this species.

Another type of reaction of naive birds to predators has been revealed in great tits by Kullberg and Lind (2002). Experiments in an aviary showed that 30-day-old naï ve great tit fledglings (Parus major) do not respond differently to a model of a perched predator than to a similarly sized model of a non-predator. Although chicks showed distress responses such as warning calls and freezing behaviour, they did not differentiate between the stimuli. In contrast, wild-caught first-year birds and adults responded differently to the two stimuli. Lack of recognition of a perched predator might be one explanation for the high mortality rate found in newly fledged great tits. Researchers suggest that the presence of parents emitting alarm calls is of importance to reduce predation risk during the time when fledglings are most vulnerable.

In principle, there are different scenarios of acquiring predator recognition, from cultural transmission in one species to innate fear based on characteristic features of predator’s image in others. In many species juveniles come into the world for the second time after their birth when they first emerge from a quiet, dark, natal burrow, or a hollow, a lodge, or a nest. A trade-off should be reached by juveniles between exploring and vigilance behaviours. Innate predisposition to distinguish between really dangerous and neutral objects can be adaptive.

In order to illustrate extreme “innateness”, let us recall brush- turkey chicks (the megapodes) and consider their behavioural adaptations to the full measure of their suffering during childhood. As it was described earlier in this chapter, after a long incubation in a mound of leaf litter, the hatchlings dig their way out of the mound and strive to survive without any parental care (Fig VII-4, VII-5). Gö th (2001) tested the response of two-year old chicks to predators in the semi-natural setting of a large aviary. She exposed chicks to a range of potentially scary predators or stimuli: a real cat, a dog, a rubber snake and a mounted silhouette model of a hawk. Naive chicks responded by crouching and freezing to flying predators. Any aerial stimulus, within a certain size and speed probably causes chicks to crouch. It has also turned out that alarm calls of other rainforest birds may replace those of absent parents. Alarm calls of conspecifics elicit vigilant behaviour in chicks. It is likely that megapode chicks are equipped with innate generalised images of objects to fear basing on visual and acoustic stimuli and among them there are several specific features for which the hatchlings are tuned and this allows them to distinguish between dogs and cats, and between alarm calls of their own and alien species.

Let us now consider another scenario that includes sensitivity to quick learning. A good example here is an experimental study by Mineka, Cook and colleagues on shaping of fear in rhesus monkeys, a problem closely related to development of phobias in humans. Researchers have exploited fear of snakes as common behavioural characteristic among primates including humans (Ö hman and Mineka, 2003). Agras, Sylvester and Oliveau (1969) interviewed a sample of New Englanders about fears, and found that snakes are clearly the most prevalent object of intensive fear reported by 38% of females and 12% of males. According to a review of field data (King, 1997), 11 genera of primates showed fear-related response (alarm calls, avoidance, mobbing) in virtually all instances in which they were observed confronting large snakes. The hypothesis that this fear is adaptive in the wild has been supported by field reports of large snakes attacking primates (Goodall, 1986; Mineka and Cook, 1986). Experiments showed that rhesus and squirrel monkeys reared in the wild were far more likely than lab-reared monkeys to show strong fear responses to snakes (Mineka et al., 1980).

Further, series of experiments revealed strong “preparedness” in rhesus monkeys for development of fear of snakes as “fear-relevant” (FR) objects. First, important evidence has been obtained that lab-reared monkeys are not afraid of snakes, that is, they appeared not to be equipped by innate recognition of these FR objects. It turned out, then, that lab-reared rhesus monkeys can acquire a fear of snakes vicariously, that is, by observing other monkeys expressing fear of snakes. When non-fearful lab-reared monkeys were given the opportunity to observe a wild reared demonstrator displaying fear of live and toy snakes, they were rapidly conditioned to fear snakes, and this conditioning was strong and persistent. The fear response was learned even also in those cases when the fear-demonstrating monkey was shown on a videotape (Cook and Mineka, 1990). Then the most intriguing stage of experiments followed. Videos were edited so that identical displays of fear in the demonstrating monkey were modelled in response to toy snakes and flowers, or to toy crocodiles and toy rabbits. The lab-reared monkeys showed substantial conditioning to toy snakes and crocodiles, but not to flowers and toy rabbits. At the same time, both toy snakes and flowers served equally well for a group of naï ve monkeys as signals for food rewards. These results provide a strong support for selective (guided) learning which is probably based on a specialized behavioural module. This study also illustrates how sophisticated the interaction between components of behaviour may be and how difficult it is to completely rule out “non obvious” environmental contributions.

Another example of guided learning comes from experiments of Griffin et al. (2002) with Tammar wallaby (Macropus eugenii). Up to now tammars have only survived on predator-free Australian islands, but they evolved with a range of now extinct marsupial predators, such as the Thylacine. In their experiments researchers used taxidermically prepared models of predators like foxes and cats, and a model of a size-matched non-predator (a juvenile goat) as a control. Foxes and cats are likely to share convergent vertebrate morphologic features (such as frontally-placed eyes) with historically important predators, but tammars have no evolutionary experience with these introduced species, or with goats. In training procedures, models were presented to tammars paired with an aversive event, that is, a human simulating a capture procedure with a net. Humans with net reliably evoke alarm responses including fleeing and foot thumping in tammars. Researchers used simulated capture attempts as a standard fear evoking stimulus. It was revealed that training with a fox model together with a fear-evoking stimulus enhanced adequate reactions in wallabies. It was then sufficient to demonstrate a fox alone in order to observe fear reactions in wallabies. Animals were then tested in an array of other unfamiliar stimuli to determine the specificity of this change in behaviour. Training with a fox produced increased responses to another model predator, a cat, but not to a sized-matched non predator (a goat). This suggests that wallabies may not acquire fear response to any stimulus that is associated with an aversive event but rather might be predisposed to learn quite specifically about predators, at least, they have a bias to associate predators with frightening events.

Besides different variants of learning about a predator itself, there is also a gradient of preparedness in acquiring reactions to species-specific alarm calls of other animals. How and when alarm call responses develop varies between species. It is known, for example, that some young birds and primates can recognize and respond to conspecific calls upon first exposure whereas other naï ve juveniles display poor initial discrimination between alarm calls and other auditory stimuli (Sherman, 1977; Krebs and Dawkins, 1984). Concrete intermediary variants revealed in playback studies shed light on adaptive value of preparedness of these reactions in the context of animals’ life. For example, Mateo (1996) has demonstrated that ground squirrels Spermophilus beldingi develop responsivity to whistle alarm calls indicative of fast-moving predators earlier and more readily than to thrill alarm calls, associated with slow-moving predators.

It is worth noting that in some species animals, assimilating alarm calls of their own species, can then learn and “translate” heterospecific calls (see Chapter 31 for details).