Brief Review: This is an excellent and well informed article by arguably the worlds leading expert on the topic. I have little to add except that Bolhuis (1991) and Horn's more recent review in Nature Reviews Neuroscience should be added to the list at the end to give some recent review readings. The only significant issue with the content of the article is that several years ago the official nomenclature of the avian brain was changed and, unfortunately, IMHV has had its name changed. The new nomenclature should be used for compatibility with other articles on the avian brain.
PATRICK BATESON Sub-Department of Animal Behaviour, University of Cambridge
Many birds and mammals will form strong and exclusive attachments to particular types of object after relatively brief exposures early in life to one of those objects. This is now known as "behavioural imprinting" in order to distinguish it from the carelessly labeled but nonetheless fascinating phenomenon of “genomic imprinting”. Usually the object of the attachment is a member of its own species. However, the nature of the rapid learning process is such that, if the young animal is reared under abnormal conditions, imprinting can result in the formation of bizarre attachments. Ducklings or domestic chicks that have been hand-reared for the first few days after hatching strongly prefer the company of their human “parent” to that of their own species. Early experience can also have long-lasting effects on sexual preferences, but the conditions are different from those in which the first filial attachments are formed.
The German term Prägung (translated as "imprinting") was first used by Oscar Heinroth, although Douglas Spalding (1873) had used a very similar metaphor, namely "stamping in". Konrad Lorenz (1935), who did so much to make the phenomenon famous, liked the image because it suggests, as he believed to be the case, an instantaneous, irreversible process. It also led to strong claims that imprinting is quite different from associative learning (e.g. Hess 1973). As more evidence became available, the claims were disputed (Bateson 1966; Sluckin 1972, Bolhuis 1991). The special flavour of the phenomenon, it was argued, comes from the particular biological job which the learning process has to perform and the context in which it occurs.
The biological function of imprinting is probably to enable the animal to recognise close kin. In the natural environment behavioural imprinting reliably results in the formation of a strong social bond between offspring and parent. The parent must recognise the offspring in order not waste time and energy caring for the young of others. The offspring must recognise its parent because it might be attacked and even killed by other adults of the same species that do not recognise it as their own. Filial imprinting ensures that a young animal can distinguish between its parent and other members of its own species. Sexual imprinting enables an animal to mate with an individual that is neither too closely nor too distantly related (Bateson 1979). Astonishing retention of sexual preferences is found in the face of considerable sexual experience with other objects in zebra finches fostered onto another species (Immelmann 1972). However, not all birds and mammals are so markedly affected by their early experience. Moreover, sexual preferences may be determined at a much later stage in development than filial preferences.
Operations involved in imprinting In order to demonstrate that imprinting takes place, one group of animals is exposed to an object (A) for a finite period of time and then given a choice between that object and another one (B). Since the animals might have preferred Object A over Object B without prior experience, it is necessary to expose a second group of animals to Object B before being given the choice test. If both groups prefer the object to which they have been exposed in the choice tests, then it is safe to assume that the difference in the preferences of the two groups is due to the period of exposure. The choice tests may be conducted so that the animals see both objects at the same time. Alternatively, the animals may be exposed first to Object A on its own, then to Object B; then, to control for any order effects, they are tested successively in the reverse order. A preference score may then be calculated by calculating the extent to which they respond to one object (A) divided by the extent to which they respond to both objects (A+B). If an animal has no preference its score will be 0.5. If it has strong preference for Object A, the score will be close to 1.0 and if it has a strong preference for Object B, the score will be close to 0.0. When precocial birds such as domestic chicks are used they are commonly tested in running wheels and the rotations of the wheels are scored as they attempt to approach each test object.
Imprinting, conditioning and perceptual learning The experimental operations by which imprinting is demonstrated differ in significant ways from classical conditioning and from learning in an instrumental situation involving the formation of associations between separately occurring neutral and significant external events. When an animal learns to use one event to predict the arrival of another, that is not the same as when it forms a new representation of something in the outside world. Learning to predict and control the environment subserves a different function from learning to categorise it. Moreover, the mechanisms differ. In uncovering causality, detecting order is crucial. If the supposed cause follows an event, then it is not a cause. By contrast, when establishing a category, temporal contiguity may be important, but the order in which the features occur is not (Bateson 1990). A bird that has to recognize its mother must gather information about her front, side and back views of its mother. Information from two separate arrays of features may be combined into a single representation when the two arrays occur in the same context or within a short time of each other. If imprinting objects are presented five or more minutes apart, the birds learns to discriminate between the two objects more quickly than those in the control group which had not been exposed to these two objects. However, when the two objects are presented 30 seconds or less apart, the imprinted birds takes longer than the control group to learn the discrimination. When they are treated as part of the same event, the order of presentation does not matter in the least. Imprinting differs both in its function and mechanism from widely studied processes of conditioning. Nevertheless, it provides a particularly striking case of the perceptual learning that occurs during exploration, latent learning and, indeed, virtually every perceptual transaction that a complicated animal has with its environment.
Predispositions The animal is predisposed to set the learning process in motion, actively searching for objects with particular features in terms of colour, movement and shape. Stimulation in other modalities, when presented concurrently with visual stimuli, can have a powerful motivating effect. In domestic chicks and mallard ducklings, the sounds most effective in eliciting pursuit of a moving visual stimulus are maternal calls of their own species. In forming a social attachment under natural conditions, auditory signals are very important in guiding the process. Nonetheless, studies of filial imprinting provide unambiguous evidence for the formation of visual stimulus representations. Initially, domestic chicks and domestic ducklings have relatively unstructured social preferences at hatching. Movement was regarded as essential in "releasing" the following response and hence in initiating the imprinting process. However, the effectiveness of the many visual stimuli used in the imprinting situation depends on such properties as their size and shape, as well as on the angle they subtend and the intensity and wavelength of light they reflect. Moreover, the rates at which these variables change are also important - hence the undoubted effectiveness of movement and flicker.
The bird clearly responds to a pattern of stimulation, and characterisation of the most effective stimulus must be cast in terms of clusters of features. Some features of the jungle fowl, the ancestral form of the domestic fowl, are particularly attractive to chicks. The head and neck is the crucial characteristic but the head and neck of a small mammal is as effective as that of a jungle fowl. In Japanese quail, the posture of a live adult female has a powerful motivating effect on the response to her by the chicks. Newly hatched domestic chicks at their first exposure to point-light animation sequences, exhibit a spontaneous preference for biological motion patterns over non-biological point-light sequences.
Sensitive periods Imprinting with a novel and conspicuous object usually occurs most readily at a particular stage of development. The word ‘imprinting’ suggests that a permanent irremovable image has been left by the impact of experience on the soft wax of the developing brain. It was supposed that the brain’s metaphorical wax is soft only during a particular stage in development and no impression can be left before or after the ‘critical’ period. However, the image of such a sharply delineated moment of imprinting is misleading, because the process is not so rigidly timed and may indeed be reversed under some conditions. This is why the term sensitive period is now most commonly used to refer to the phase during early development when the young animal most readily forms a social attachment.
Filial imprinting occurs just prior to the stage in the life-cycle when, for its own safety, the young animal needs to discriminate between its parents and other members of its own species which might attack it. In precocious species this is in the first few days after hatching or birth. In birds that are hatched blind, naked and helpless, such as swallows, the onset of imprinting occurs much later in relation to hatching than it does in the precocious ducklings, which are already feathered and active when they hatch. Changes in the animal’s abilities to perceive and deal with the external world play an important role in determining when a sensitive period starts. These abilities must themselves develop, but the rate at which they do so depends on the animal's experience.
The timing of behavioural imprinting can, within limits, be adapted to circumstances. Exposure to one object leads the individual to prefer it and reject anything perceived as different. Thus, experience of one sort prevents other types of experience from having the same impact. Such competitive exclusion is similar to what happens to visual development when one eye is covered up in early life. If the developing animal is deprived of the experience it would normally receive during the sensitive period, the imprinting process slows down and the sensitive period is lengthened. Hence, the sensitive period for imprinting is extended in domestic chicks if they are reared in isolation from other chicks. But rearing the animals in isolation does not simply delay a hypothetical internal clock from bringing the sensitive period to a close. It eventually leads to the formation of a preference for whatever the individual has experienced, no matter how unnatural. Accordingly, chicks that have been reared in social isolation in a pen with patterned walls will eventually form a memory for that pattern and will subsequently respond socially to a moving object if it bears the same pattern. This finding is important because it shows that sensory deprivation during the normal sensitive period for imprinting does not stop the nervous system from slowly settling into highly abnormal organisation. Despite its abnormality, the outcome reflects the same process that would lead under normal circumstances to the successful development of a social preference. In the natural environment the bird is virtually certain to be exposed to its mother sooner or later.
The flexibility of the imprinting mechanism is important because it allows for naturally-occurring variations in env ironmental conditions. When the weather is warm, for example, the mother duck will lead her ducklings away from the nest within hours of hatching. When the weather is cold, however, the mother will brood her young for several days after they have hatched and the young may consequently see little of her until then because they are underneath her.
Underlying mechanism The detailed neural mechanisms underlying the imprinting process and the sensitive period have been analysed in the domestic chick (Horn 1985). A crucial problem was finding where in the brain the information about the mother (or her substitute imprinting object) is stored. An array of neurobiological techniques has implicated one particular region of the chick forebrain as the site where the neural representation of the imprinting object resides. This site is called the intermediate and medial part of the hyperstriatum ventrale, or IMHV.
It is not good enough simply to show that a particular part of the brain is active when the bird is learning about the imprinting object. This is because lots of other things happen during the imprinting process: the young bird is visually stimulated and aroused by the imprinting object (normally its mother) and it also tries to approach and follow the object. All these processes produce their own changes in brain activity. When experimental evidence is open to a variety of different interpretations, greater confidence in one particular explanation can be attained by tackling the problem in a number of different ways. A converging set of experiments established that increased biochemical activity in the chick’s forebrain roof is necessary for, and exclusively related to, the storage of information after imprinting. None of the experiments by itself ruled out all the alternative explanations. Each piece of evidence obtained by the different approaches was ambiguous, but the ambiguities were different in each case. Therefore when the whole body of evidence is considered, much greater confidence may be placed on the final interpretation. An analogy is the process of triangulation - locating on a map the position of a mountain top. One compass bearing is usually not enough; two bearings from different angles provide a much better fix; and three bearings give a reliable position.
Careful mapping of the chick brain has confirmed that a neural representation of what is learned during imprinting is stored permanently in the IMHV region. Chicks that have had both their left and right IMHVs removed surgically are unable to imprint. Moreover, if both IMHVs are damaged experimentally immediately after imprinting has taken place, the birds show no recognition of the imprinting object. Damaging the IMHVs erases their memory for the imprinting object. These experiments have demonstrated that the IMHV is needed both for imprinting to take place and for the memory of the imprinting object to be recalled after imprinting has occurred. Further studies have revealed that another neural representation is formed or confirmed in a different region of the brain about six hours after imprinting. This representation can be prevented from forming by surgically damaging the right IMHV soon after imprinting. In the critical IMHV regions, connections between neurons are both enlarged and diminished. The enlarged neural connections are specific to features of the particular object with which the bird has been imprinted, and the diminished connections are specific to features that are not present in that object (Horn 1998).
REFERENCES Bateson, P. P. G. 1966 The characteristics and context of imprinting. Biol. Rev. 4l, 177-220. Bateson, P. 1979 How do sensitive periods arise and what are they for? Anim. Behav. 27, 470-486. Bateson, P. (1990) Is imprinting such a special case? Philosophical Transactions of the Royal Society of London, B, 329, 125-131. Hess, E. H. 1973 Imprinting. New York: Van Nostrand Reinhold. Horn, G. 1985 Memory, Imprinting, and the Brain. Oxford: Clarendon Press. Immelmann, K. 1972 Sexual and other long-term aspects of imprinting in birds and other species. Adv. Stud. Behav. 4, 147-174. Lorenz, K. 1935 Der Kumpan in der Umwelt des Vogels. J. Orn. 83, 137-213, 289-413. Sluckin, W. 1972 Imprinting and Early Learning. 2nd edit. London: Methuen. Spalding, D. A. 1873 Instinct with original observations on young animals. Macmillan's Magazine 27, 282-293.
Additional Review: A clear, concise and well-informed article. I have nothing to comment except perhaps that in two sections the potential cost of lack of imprinting for the offspring is given as the possibility of aggression by other adults. In fact a sufficient potential cost would be that other adults would ignore the begging or following offspring. Simply adding the cost of being ignored would be sufficient pressure to imprint and recognise one's parent. A wonderful piece by perhaps the best informed scholar in the field.