Ethology was originally defined as the study of animal behavior within the framework put forward by Lorenz (1937) and Tinbergen (1951). Later, ‘ethology’ was used more generally to describe any scientific study of the behavior of animals in relation to their natural environment. A modern inclusive term for the discipline is behavioral biology.
Classical ethological theory had surprisingly little to say about the development of behavior, in spite of the fact that Lorenz (1903-1989) himself had earlier (1935) published a landmark paper on the concept of imprinting. Tinbergen’s (1951) topic, for example, has only one short chapter on development, and only one paragraph on imprinting. Lehrman (1953), in his influential critique of ethological theory, pointed out this neglect of developmental questions, which subsequently led many ethological workers to consider problems of development (Kruijt, 1964). It also led Tinbergen (1963), ten years later, to reformulate his views on the aims of ethology. In this seminal paper, he stated that ethologists should aim to answer four major questions about behavior: its causation, function, evolution, and development. Although Tinbergen (1907-1988) emphasized that understanding behavior required addressing all four of these questions, we shall only discuss some of the major contributions ethologists and other behavioral biologists have made to the study of development. We shall see that many concepts and findings concerning behavioral development in animals have had important consequences for the study of human development.
Lorenz and the nature-nurture debate
In his early papers, Lorenz postulated that behavior could be considered a mixture of innate and acquired Full references for works cited in this entry which do not appear in the Bibliography can be found in one of the reference links in either Bolhuis & Hogan (1999), Hogan (2001), or Johnson & Bolhuis (2000).
Elements (Instinkt-Dressur-Verschrankung: intercalation of fixed action patterns and learning), and that analysis of the development of the innate elements (fixed action patterns) was a matter for embryologists. In reaction to Lehrman’s (1953) critique of ethological theory, Lorenz (1965) changed his formulation somewhat, and argued that the information necessary for a behavior element to be adapted to its species’ environment can only come from two sources: from information stored in the genes or from an interaction between the individual and its environment. This formulation also met with considerable criticism from many who insisted that development consisted of a more complex dynamic. Gottlieb (1997) discusses many aspects of this debate, and we have recently republished some of the original papers (Bolhuis & Hogan, 1999). Here, we will mention only two important aspects of the debate.
To begin with, Lehrman (1970) pointed out that he and Lorenz were really interested in two different problems: Lehrman was interested in studying the effects of all types of experience on all types of behavior at all stages of development, very much from a causal perspective, whereas Lorenz was interested only in studying the effects of functional experience on behavior mechanisms at the stage of development at which they begin to function as modes of adaptation to the environment. Hogan (1988, 2001) has argued that both these viewpoints are equally legitimate, but that Lorenz’s functional criterion corresponds to the way most people think about development. Nonetheless, it is essential not to confuse causal and functional viewpoints (cf. Hogan, 1994a; Bolhuis & Macphail, 2001). In this entry, we consider development from a causal perspective.
A second aspect of the debate is that even behavior patterns that owe their adaptedness to genetic information require interaction with the environment in order to develop in the individual. As Lehrman (1953) states: “The interaction out of which the organism develops is not one, as is so often said, between heredity and environment. It is between organism and environment! And the organism is different at each different stage of its development” (p. 345). We give examples later that illustrate this interactionist interpretation of development.
Imprinting: sensitive periods and irreversibility
Filial imprinting is the process through which early social preferences become restricted to a particular stimulus, or class of stimuli, as a result of exposure to that stimulus. This early learning phenomenon is often regarded as a classic example of a developmental process involving sensitive periods. The idea of a sensitive period has been extremely important (and controversial) in the study of behavioral development. Here we adopt the definition given by Bateson & Hinde (1987): “The sensitive period concept implies a phase of great susceptibility [to certain types of experience] preceded and followed by lower sensitivity, with relatively gradual transitions” (p. 20). Lorenz and other authors use the term ‘critical period’, borrowed from embryology, for this concept. However, Bateson & Hinde argue that the periods of increased sensitivity are not sharply defined, and consequently they suggested the use of the term ‘sensitive period,’ which is now widely (but by no means universally) used. We shall discuss recent evidence concerning sensitive periods in the context of filial imprinting.
Early imprinting researchers concluded that there was a narrow sensitive period within which imprinting could occur. This sensitive period was thought to occur within the first twenty-four hours after hatching in ducklings and chicks, and to last for not more than a few hours. Subsequent research has demonstrated that the sensitive period for filial imprinting is not so restricted in these species (Bateson, 1979; Bolhuis, 1991).
In the analysis of sensitive periods, it is important to distinguish between the onset and the decline of increased sensitivity, as there are often different causal factors for these two events. In filial imprinting, it is likely that the onset of the sensitive period can be explained in terms of immediate physiological factors, such as increases in visual efficiency and in motor ability in precocial birds some time after hatching. Different causal factors are thought to be involved at the end of the sensitive period for imprinting. Sluckin & Salzen (1961) suggested that the ability to imprint comes to an end after the animal has developed a social preference for a certain stimulus as a result of exposure to that stimulus. The animal will stay close to the familiar object and avoid novel ones; it will thus receive very little exposure to a novel stimulus and there will be little opportunity for further imprinting. This interpretation implies that imprinting will remain possible if an appropriate stimulus is not presented. Indeed, chicks that are reared in isolation retain the ability to imprint for longer than socially reared chicks. An apparent decline in sensitivity in isolated chicks can be explained as resulting from the animals’ imprinting to stationary visual aspects of the rearing environment (Bateson, 1964). Thus, it is the imprinting process itself that brings the sensitive period to an end.
The conventional view of the causes of sensitive periods in a number of disciplines is that they are due to some sort of physiological clock mechanism (Bornstein, 1987; Rauschecker & Marler, 1987). The evidence from filial imprinting studies, however, is not consistent with an internal clock model, but requires instead that experience with the imprinting stimulus is the causal factor for the end of the sensitive period. Bateson (1987) proposed just such a model: the competitive exclusion model. He pointed out that neural growth is associated with particular sensory input from the environment. His model assumes that there is a limited capacity for such growth to impinge upon the systems that are responsible for the execution of the behavior involved (e.g., approach, in the case of imprinting). Input from different stimuli ‘competes’ for access to these executive systems. Once neural growth associated with a certain stimulus develops control of the executive systems, subsequent stimuli will be less able to gain access to these systems. Furthermore, insofar as these early neural connections are permanent (Shatz, 1992; Hogan, 2001, pp. 263-269), this interpretation also explains the general irreversibility of many aspects of early learning. Evidence from recent studies of sexual imprinting (Bischof, 1994) and bird song learning (Marler, 1991; Nelson, 1997) is also consistent with an experience-dependent end to the sensitive period. These results all show that it is necessary to investigate the causes for both the beginning and end of any sensitive period before reaching conclusions about the mechanisms responsible.
Perceptual development as a continuous interactive process
In his classic paper on imprinting, Lorenz (1935) proposed the concept of ‘schema,’ which is a kind of perceptual mechanism that ‘recognizes’ certain objects (Hogan, 1988). In the development of a social bond between parents and young, Lorenz noted that in certain bird species (such as the curlew, Numenius arquata) the newly hatched chicks came equipped with a schema of the parent that he considered to be ‘innate,’ while in others (such as the greylag goose, Anser anser) the schema of the parent developed as a result of specific experience (imprinting). It is now known that the development of almost all perceptual mechanisms that have been studied requires some kind of experience. In some cases the experience that is necessary is tightly constrained, and the animal is predisposed to be affected by very specific classes of stimuli, while in other cases the experience can be quite general. We shall discuss examples of both kinds.
Most songbird species need to learn their song from a tutor male (Thorpe, 1961; Marler, 1976; Nelson, 1997). Under certain circumstances, young males of some species can learn their songs, or at least part of their songs, from tape recordings of tutor songs. When fledgling male song sparrows (Melospiza melodia) and swamp sparrows (Melospiza georgiana) were exposed to taped songs that consisted of equal numbers of songs of both species, they preferentially learned the songs of their own species. Males of both species are able to sing the songs of the other species. Thus, it appears that they are predisposed to perceive songs of their own species; Marler (1991) called this ‘the sensitization of young sparrows to conspecific song’ (p. 200). It is noteworthy that many aspects of birdsong learning have been found to be relevant to the development of human language (Marler, 1976; Doupe & Kuhl, 1999). For example, Kuhl and her colleagues have shown that infants less than 6 months of age learn to perceive phonemes unique to their linguistic environment, but that they do not learn to utter these sounds until several months later.
In an extensive series of experiments published between 1975 and 1987, Gottlieb (1997) investigated the mechanisms underlying the preferences that young ducklings of a number of species show for the maternal call of their own species over that of other species. He found that differential behavior toward the species-specific call could already be observed at an early embryonic stage, before the animal itself started to vocalize. However, a post-hatching preference for the conspecific maternal call was only found when the animals received exposure to embryonic contact-contentment calls, played back at the right speed and with a natural variation, within a certain period in development. Thus, the expression of the species-specific preference in ducklings is dependent on particular experience early in development.
The development of filial behavior in the chick involves two perceptual systems that are neurally and behaviorally dissociable (Bolhuis, 1996; Bolhuis & Honey, 1998; Horn, 1985, 1998). On the one hand, there is an effect of experience with particular stimuli (i.e., filial imprinting). On the other hand, there is an emerging predisposition to approach stimuli resembling conspecifics (see Fig. 1). Training with a particular stimulus is not necessary for the predisposition to emerge: the predisposition can emerge in dark-reared chicks, provided that they receive a certain amount of non-specific stimulation within a certain period in development (Johnson etal., 1989).
Figure 1. Mean preference scores, expressed as a preference for the stuffed fowl, of chicks previously trained by exposure to a rotating stuffed junglefowl (gray), a rotating red box (white), or exposure to white light (black). Preference scores are defined as: activity when attempting to approach the stuffed jungle fowl divided by total approach activity during the test. Preferences were measured in a simultaneous test either 2 h (Test 1) or 24 h (Test 2) after the end of training. kl-k4 represent the differences between the preferences of the trained chicks and the controls; Ay represents the difference in preference between the control chicks at Test 2 and at Test 1.
The stimulus characteristics of visual stimuli that allow the filial predisposition to be expressed were investigated in tests involving an intact stuffed junglefowl versus a series of increasingly degraded versions of a stuffed junglefowl (Johnson & Horn, 1988). The degraded versions ranged from one where different parts of the model (wings, head, torso, legs) were re-assembled in an unnatural way, to one in which the pelt of a junglefowl had been cut into small pieces that were stuck onto a rotating box. The intact model was preferred only when the degraded object possessed no distinguishable junglefowl features. Further studies showed that the necessary stimuli are not species- or even class-specific: eye-like stimuli are normally important, but other aspects of the stimulus are also sufficient for the expression of the predisposition (Bolhuis, 1996).
There are interesting similarities between the development of face recognition in human infants, and the development of filial preferences in chicks. Newborn infants have been shown to track a moving face-like stimulus more than a stimulus that lacks these features, or in which these features have been jumbled up. Similarly, in both human infants and young precocial birds, the features of individual objects need to be learned. Once learned, both infants and birds react to unfamiliar objects with species-specific behavior patterns that tend to bring them back to the familiar object or caregiver (Blass, 1999; Johnson & Bolhuis, 2000).
Figure 2. Conception of behavior systems. Stimuli from the external world are analyzed by perceptual mechanisms. Output from the perceptual mechanisms can be integrated by central mechanisms and/or channeled directly to motor mechanisms. The output of the motor mechanisms results in behavior. In this diagram, central mechanism I, perceptual mechanisms 1, 2, and 3, and motor mechanisms A, B, and C form one behavior system; central mechanism II, perceptual mechanisms 3, 4, and 5, and motor mechanisms C, D, and E form a second behavior system. Systems 1-A, 2-B, and so on can also be considered less complex behavior systems.
Development of behavior systems
Kruijt (1964), in his classic monograph on the development of social behavior in the junglefowl (Gallus gallus spadiceus), suggested that in young chicks – and obviously, in the young of other species as well – many of the motor components of behavior appear as independent units prior to any opportunity for practice. Only later, often after specific experience, do these motor components become integrated into more complex systems such as hunger, aggression, or sex. Hogan (1988) has generalized this proposal by Kruijt and suggested a framework for the analysis of behavioral development using the concept of behavior system (see Fig. 2). A behavior system consists of different elements: a central mechanism, perceptual mechanisms, and motor mechanisms. These mechanisms are considered to be structures in the central nervous system, and one could also call them cognitive structures. The definition of a behavior system is “… any organization of perceptual, central, and motor mechanisms that acts as a unit in some situations” (Hogan 1988, p. 66). According to Hogan, behavioral development is essentially the development of these mechanisms and the changes in the connections among them. Often, these mechanisms and their connections only develop after functional experiences (i.e., experience with the particular stimuli involved, or with the consequences of performing specific motor patterns).
An example of a developing behavior system is the hunger system in the junglefowl chick (Hogan 1971, 1988). This system includes perceptual mechanisms for the recognition of features (e.g., color, shape, size), objects (e.g., grains, mealworms), and functions (food versus non-food). There are also motor mechanisms underlying behavior patterns such as ground scratching and pecking, and there is a central hunger mechanism. Importantly, several of the connections between the mechanisms (shown by dashed lines in Fig. 3) only develop as a result of specific functional experience. For instance, only after a substantial meal will the chick differentiate between food items and non-food items (Hogan-Warburg & Hogan, 1981). On the motor side of the system, a young chick’s pecking behavior is not dependent on the level of food deprivation before 3 days of age. Only after the experience of pecking and swallowing some solid object do the two mechanisms become connected, and only then is the level of pecking dependent on the level of food deprivation (Hogan 1984).
A similar phenomenon occurs with respect to suckling in rat pups, kittens, puppies, and human infants (Hinde, 1970, p. 551). For instance, human newborns sucked more when satiated and experimentally aroused than when food-deprived. In the case of rat pups, suckling does not become deprivation-dependent until about two weeks after birth (Hall & Williams, 1983). Unlike with chicks, we do not yet know what experience is needed to connect the suckling motor mechanism with the central hunger mechanism in the rat pup or human newborn.
The development of behavioral structure is not uniform, but may proceed along different pathways for different behavior systems. For example, dustbathing is a behavior that adult birds of many species frequently engage in. It consists of a sequence of coordinated movements of the wings, feet, head, and body that serve to spread dust through the feathers. The function of this behavior in adult fowl is to remove excess lipids from the feathers and to maintain good feather condition (van Liere & Bokma, 1987). Unlike the development of feeding behavior in rats or chicks, dustbathing is deprivation-dependent as soon as it appears in the animal’s behavioral repertoire (Hogan etal., 1991). Thus, in this case, chicks do not require functional experience to connect the motor mechanisms with the central dustbathing mechanism. On the perceptual side, other experiments have shown that initially the chick will perform dustbathing on virtually any kind of surface, including wire mesh, suggesting that the perceptual mechanism and the central mechanism are not yet connected (Vestergaard, Hogan and Krisijt, 1990; Petherick etal., 1995).
Figure 3. The hunger system of a young chick. Perceptual mechanisms include various feature-recognition mechanisms (such as of color, shape, size, and movement), object-recognition mechanisms (such as grain-like objects and worm-like objects), and a function-recognition mechanism (food). Motor mechanisms include those underlying specific behavior patterns (such as pecking, ground scratching, and walking) and an integrative motor mechanism that could be called foraging. There is also a central hunger mechanism (H). Solid lines indicate mechanisms and connections that develop prefunctionally; dashed lines indicate mechanisms and connections that develop as the result of specific functional experience.
The perceptual mechanism itself develops more quickly with some substrates (peat or sand) than with others (wood shavings or wire mesh), which is similar to the development of perceptual mechanisms in song learning (Marler, 1991) and filial predispositions (Bolhuis, 1991). Furthermore, preferences for functionally unlikely surfaces (in this case a skin of junglefowl feathers) can be acquired as a result of experience with them (Vestergaard & Hogan, 1992). This is another example of the development of a perceptual mechanism, and one that is not dissimilar to filial imprinting.
Finally, Hogan (2001, pp. 254-257) has discussed how it is possible to consider human language to be a behavior system that is similar in many respects to those we have just discussed. Learning to perceive and produce phonemes has been mentioned above. It is also possible to identify two major central components of the language system: the semantic system (Shelton & Caramazza, 1999) and the syntax system (Chomsky, 1965; Pinker, 1994). The basic lower-order units of the language system are morphemes (words). Numerous studies have shown that the same morphemes can be expressed equally well with auditory-vocal units (normal spoken language) or visual-manual units (sign language). Development of the organization of both the semantic system and the syntax system proceeds in the same way regardless of how the morphemes are expressed.
Ethological theories and methods have played an important role in the formulation and development of John Bowlby’s (1969, 1991) theory of attachment in humans. This theory was originally developed to explain the behavior of children who had been separated from their mother and raised in a nursery during the Second World War, and was greatly influenced by Lorenz’s ideas about imprinting. This is not the place to discuss attachment theory in detail, but we can point out that, in many ways, the attachment system postulated by Bowlby is analogous to the filial behavior system in young birds. In both cases, the newborn infant or chick possesses a number of behavior patterns that keep it in contact with the parent (or other caregiver) and that attract the attention of the parent in the parent’s absence. Furthermore, both infant and chick must learn the characteristics of the parent, which is considered to be the formation of a bond between the two. Factors influencing the formation of the bond are also similar, including all the factors we have discussed above such as length of exposure, sensitive periods, irreversibility, and predispositions. Studying the importance of these factors in the human situation has resulted in a large body of literature, some of which has supported the theory, and some not (Rutter, 1991, 2002). The theory itself has been modified to take these results into account, and has also been expanded to include development of attachments throughout life.
Gathering data to test hypotheses about human behavior always presents special challenges because of the ethical issues involved. To study the effects of maternal separation on infant behavior, Harlow (1958), for example, raised infant rhesus monkeys in complete social isolation, which led to horrific effects on the infant’s subsequent behavior. Less intrusive methods such as raising infants with other infants (Harlow & Harlow, 1962), or separating infants from their mothers for brief periods of time (Hinde & Spencer-Booth, 1971; Hinde, 1977) led to less dramatic results, but these methods are still unacceptable for human research. Bowlby felt that the best method for studying human development was to observe infants in real-life situations, in much the same way as many ethologists study the behavior of other animals in natural or semi-natural settings.
Much of his theorizing about human attachment was based upon such research carried out by Mary Ainsworth (1913-1999). She and her colleagues (1978) developed a standardized ‘Strange Situation’ test in which a stranger approaches an infant with and without the parent being present, and various aspects of the infant’s behavior are measured. This method is now widely used, and has allowed researchers to characterize specific patterns of attachment and their determinants. Use of basically similar methods has allowed results from both human and other animal studies to be more easily compared, and has led to mutual benefits with respect to both theory (Kraemer, 1992) and methods (Weaver & de Waal, 2002).
Ethology, as a set of theories distinct from those in other disciplines, no longer exists. Nonetheless, many workers trained in the framework of ethology have made important contributions to the study of behavioral development. The debate over the conceptualization of the roles of nature and nurture, ubiquitous in the older literature, has led to a modern synthesis that is generally accepted by all developmental biologists. This can be seen in the studies of filial imprinting and behavior system development that we have reviewed, as well as in studies of sexual imprinting and birdsong development. These studies, in turn, have had an important influence on such topics in child development as early attachment, face recognition, and language development. Students of behavioral biology interested in development are now devoting much of their energy toward investigating aspects of cognition in humans and other animal species using techniques from neuroscience and psychology, as well as the observational and experimental techniques used by the early ethologists. The search for grand theories is likely to continue, but in a much wider context.