Outline for a functional analysis of imitation in animals.
It is generally acknowledged that there are three interrelated means by which individuals initially emit specific behavioral repertoires; through phylogenetic predispositions to behave in certain ways, through trial-and-error learning in the physical environment, and through observational learning of behavioral repertoires emitted by conspecifics. From an ethological perspective the major determinants of behavior are phylogenetic in origin. That is, it is assumed that conspecifics behave in similar ways because they share a common phylogenetic history. In general, phylogenetically determined behaviors are typified by both their topographical complexity and their resistance to change. Behaviors which primarily originate in this manner can be emitted without requiring prior experience, either directly, in terms of emitting similar behavior, or indirectly, through observation of the same behavioral patterns being emitted by a conspecific. Strong phylogenetic influences on the emission of specific behavioral repertoires have been shown in a number of species (e.g., species-characteristic tail flicking in dense foliage by passerine birds, Andrew, 1956; elaborate courtship displays by male fruitflies, Bastock, 1956; and inheritance of dominance in deer mice Peromyscus maniculatus, Dewsbury, 1990).
At the phylogenetic level, new behaviors arise through the evolutionary process of spontaneous variation and natural selection of genetic characteristics which control the emission of behavioral patterns. According to Skinner (1984) a process akin to this also operates at the ontogenetic level. From findings in operant research, it is known that behaviors are strengthened or weakened in particular contexts in response to their environmental consequences. With repeated exposure to similar environmental stimuli, complex behavioral patterns may develop. This process may lead to behavioral homogeneity within a group because group members are likely to have similar ontogenetic histories. In other words, as conspecifics tend to occupy the same types of habitats, and are consequently exposed to similar contingency requirements during their lifetime, environmental contingencies may individually shape their behaviors in related ways.
At the ontogenetic level it has been suggested that new behaviors may also be learned without requiring direct contingency shaping. Instead, individuals may emit unusual behaviors in novel contexts following the observation of others engaged in those activities (e.g., tool-use by captive baboons and by Californian sea otters, Beck, 1972; Hall & Schaller, 1964; lever pressing by naive rats, Corson, 1967). The processes by which organisms acquire new behavioral repertoires in this way come under a number of broad labels, such as social learning, observational learning, and imitation (Box, 1984). The term imitation shall be used in this paper because it is most commonly associated with instances of socially acquired behavior which are similar in appearance to the behavior exhibited by the observed organism.(1) By means of the processes involved in imitation, similar behavioral activities tend to be exhibited by individual members of a social species. Group activities which arise in this manner are often termed cultural or traditional activities (e.g., Aisner & Terkel, 1985; Lefebvre, 1986). Many cultural practices have been observed in field studies of nonhuman social species (see Lefebvre & Palameta, 1988, for an extensive bibliography). These behaviors are generally beneficial to the individuals concerned. For instance, previously unobtainable food resources are often made available by this means (e.g., diving and fishing for molluscs by Norway rats, Gandolfi & Parisi, 1973; termite digging in chimpanzees, Goodall, 1970; and the "intentional" stranding by killer whales when hunting seals lying on the seashore, Lopez & Lopez, 1985).
Insofar as social learning can account for increases in the variation and number of adaptive behaviors emitted within a group of conspecifics, it compares favorably in a number of ways with phylogenetic selection and operant conditioning. For example, in terms of speed of acquisition, the initial performance of behaviors acquired through observational learning occur more rapidly than by individual trial-and-error experience (Box, 1984). Behaviors acquired socially are also generally advantageous, or at least harmless, when emitted (e.g., socially acquired food preferences in rats and gerbils, Forkman, 1991; Galef, Kenneft, & Wigmore, 1984; Laland & Plotkin, 1991). For instance, it is less risky to copy others than to be innovative in an environment where inappropriate behavior may cost an individual its life. If certain behaviors are emitted by others in the group then it is likely that they will be safe to emit by an observing conspecific and have beneficial consequences. In addition, the maintenance of group activities by means of imitative behavioral processes is a more adaptable strategy than behaviors which are largely phylogenetic in origin because the former can both emerge and be replaced more rapidly and easily than behaviors which rely on natural selection for change. It can be seen, then, that imitated behavior possesses some of the advantages of genetically elicited behavioral repertoires and individually learned behaviors while avoiding their disadvantages. In particular, the advantages of imitation appear to be greatest for social species where opportunistic behavior can present new benefits to individuals within the group (Palameta & Lefebvre, 1985).
Although the functional advantages of social learning are evident, the study of imitative behavior in animals has presented difficulties in both the delineation and understanding of the processes involved (Galef, 1976). This is understandable given that different behavioral occurrences have been shown to be facilitated by social interaction to varying degrees, depending on the influence exerted upon them by phylogenetic factors and other ontogenetic factors. The purpose of the remainder of this paper is to outline the major explanatory classes into which imitation in animals has been categorized. We also present a detailed analysis of a category defined as "true imitation" in which separate social processes are believed to predominate. By then examining typical experiments performed to demonstrate true imitation, we highlight the difficulties encountered when employing some of the particular theoretical assumptions and experimental techniques used by ethologists. The conclusions drawn from this analysis provide the rationale for the subsequent presentation of a behavioral approach to the phenomenon of imitation in animals.
The study of imitation in animals has been largely left for ethologists to examine, and subsequently its development has been heavily influenced by approaches common to this perspective. When reviewing the literature, it is not long before one realizes that contradictory labels, definitions, and explanations abound. Galef (1988), for example, lists over 20 terms for imitative behavior such as local enhancement, observational conditioning, modeling, contagious behavior, and matched dependent behavior. Definitions for such terms are often narrowly stated and allow only a limited number of behavioral mechanisms, or else they may be so broadly based as to encompass a large number of possible processes.
Fortunately, most of these variously defined types of imitative behavior in animals can be classed into a few major groups according to the features they hold in common. In this paper, the categories listed by Fragaszy and Visalberghi (1989) will be described. Table 1 shows their classification system for types of imitation in animals, with some proposed explanations and examples of behaviors thought to occur through these means. Four major descriptive categories can be seen. In each category, the demonstrator refers to the conspecific who initially performs the target behavior, and the observer refers to the conspecific who, upon observing the demonstrator, emits similar behavior. The first category of general activity (GA) is characterized as an increase in behavioral movement by the observer following exposure to the demonstrator's behavior. In this case, an increase in "general" bodily movement by the observer may lead to the observer inadvertently interacting with similar stimuli to those being manipulated by the demonstrator. The observer may then behave in a similar, fashion to the demonstrator as similar contingencies come to operate on both organisms. Local enhancement (LE) is similar to general activity (GA). However, it differs in that by observing the demonstrator, the observer's behavior then becomes directed toward the location or contextual stimuli with which the demonstrator is interacting. The common context shared by the observer and the demonstrator then facilitates the emission of behavioral repertoires which happen to bear similarity to the demonstrator's behavior. Social facilitation (SF) describes imitative occurrences where specific behaviors or behavioral sequences emitted by the demonstrator lead to the emission of similar behaviors by the observer. The topographies of the behaviors in this category are not unusual to the observer in that they are either phylogenetically common to the species or ontogenetically familiar to the observer. Usually some aspect of the demonstrator's behavior acts as a stimulus for the observer to emit specific behaviors. The control exerted by the demonstrator over the observer may be phylogenetic or ontogenetic in nature. The final category, true imitation (TI), is defined as occurring when the topography of an unusual or novel activity exhibited by the demonstrator is emitted for the first time by the observer. Evidence for the existence of this category usually rests upon the rejection of LE or SF as causative factors of the imitative activity.
[TABULAR DATA 1 OMITTED]
It can be seen from Table 1 that the term imitation is a rather crude abstraction for a heterogenous group of social-environmental interactions classed together solely on their topographically similar outcome to one another. In other words, very different behavioral processes have given rise to behavioral patterns in which conspecifics come to behave in similar ways following interaction with one another. The categories of GA, LE, and SF have been labeled as "pseudoimitative" by some researchers because in each case the specific topography of the "imitated" behavior emitted by the observer does not arise through attention to the behavioral topography of the demonstrator (see Galef, 1988, for a review). Instead, the behavior is either already part of the observer's ontogenetic or phylogenetic repertoire prior to observing the demonstrator, or else it has arisen independently by trial-and-error learning subsequent to observing the demonstrator.
Explanations for TI in animals are distinctive in that they must account for the behavior emitted by the observer when it is dependent upon attention to topographical aspects of the behavior of the demonstrator. The precise nature of this control presents many difficulties for researchers (Kymissis & Poulson, 1990). Some believe it to be a separate learning phenomenon in itself, distinct from operant and classical conditioning theories (e.g., Bandura, 1977). In general it is proposed that "cognitive" variables control the organism's behavior. Although explanations for TI in animals are not clearly outlined, most researchers in the field use mentalistic terms, such as "intentional," "goal orientated," and "purposive," thinking to explain the observer's imitative behavior (e.g., Galef, 1988; Palameta & Lefebvre, 1985; Thorpe, 1963).
Explanations of this nature place TI within the category of higher cognitive functions (e.g., self-awareness, creativity, insight) more characteristic of our own species. However, increasingly complex behaviors thought to require higher abilities have now been observed in animals, (e.g., counting, Davis & Memmott, 1982; perceptual concept learning, Herrnstein, 1984). Reflecting this orientation, some ethologists are becoming more cognitive in orientation (e.g., Griffin, 1984). In addition, those concerned with animal welfare (e.g., Dawkins, 1985) and animal rights (e.g., Regan, 1987; Singer, 1985) have been keen to highlight the continuity between humans and animals by drawing attention to similarities in our cognitive functioning.
In view of the above considerations, it is not surprising that a number of attempts have been made to demonstrate TI in various nonhuman species (e.g., in budgerigars, Dawson & Foss, 1965; in capuchin monkeys, Fragaszy & Visalberghi, 1989, in rats, Heyes & Dawson, 1990; in pigeons, Palameta & Lefebvre, 1985). However, demonstrations of TI in animals under controlled laboratory conditions have been inconclusive (Box, 1984; Davis, 1981; Galef, 1988). It is argued here that the various methodological and theoretical approaches that have been used are responsible for this state of affairs. By way of highlighting the difficulties involved in this area we describe below some of the studies that have been conducted.
Studies on True Imitation in Animals
Most of the early literature on imitation in animals consists of anecdotal field accounts of individuals who were found to be emitting adaptive but novel behavior which then spread throughout the rest of the group. Many of these studies claimed to be showing TI. For example, Romanes (1884) described how one day he observed a few honeybees from one hive exhibiting a behavior usually only performed by bumblebees. This activity consisted of cutting holes in the base of flowers and then sucking the nectar from them. The next day, bees from all of the other hives were involved in this novel activity, the transmission of which, Romanes interpreted as having originated by a process similar to TI. Later, more comprehensive field studies observed what were also heralded at the time as examples of TI in animals. These examples include the spread of milk bottle top opening in blue tits (Fisher & Hinde, 1949) and potato washing in Japanese macaques (Kawai, 1965). However, it has been explained earlier that imitative behavior may result from a number of related processes such as individual learning, GA, LE, and SF. Indeed, following on from the conclusions by Fisher and Hinde (1949), studies by Sherry and Galef (1984; 1990) redefined bottle top opening in tits in terms of the effects of local enhancement and trial-and-error learning.
Primarily noninvasive studies limit the study of imitation to a purely descriptive level of analysis. Such studies validate the existence in the wild of some form of imitative behavior in its most general sense, but they do little to elucidate the variables controlling the emergence of socially transmitted behavior. What is needed is an experimental approach which systematically manipulates aspects of the behavior emitted by the demonstrator and the observer so that the relative contributions of basic processes can be determined. This approach requires that TI be investigated under controlled laboratory conditions in animals with similar phylogenetic and ontogenetic histories.
An appreciation of the limitations of field studies has led to a steady increase in the number of laboratory studies attempting to demonstrate true imitative behavior in animals (e.g., Dawson & Foss, 1965; Denny, Clos, & Bell, 1988; Galef, Manzig, & Field, 1986; Palameta & Lefebvre, 1985). Typically, such experiments use the duplicate-cage method (Warden & Jackson, 1935) consisting of two adjoining identical cages seperated by a clear partition. A conspecific is placed in either side; one animal, the demonstrator, has previously been trained to perform a task, while the other one, the observer, is naive to the task. The experimenter arranges the contingencies within the apparatus such that the demonstrator performs the target behavior in the presence of the observer. The observer can see the demonstrator and generally the environmental stimuli governing the latter's target behavior, but the observer cannot physically interact with these stimuli or receive any reinforcement while watching the demonstrator. The demonstrator is then removed and the observer is placed in the same conditions under which the demonstrator emitted the target behavior. Imitation is considered to have occurred if the observer performs the required behavior correctly in a single trial or in fewer trials than it would take without exposure to the demonstrator's behavior (Davis, 1981). A group design is primarily used in order to rule out possibilities of other processes governing the imitative behavior. It is generally assumed that TI must be operating if no other processes can be said to explain the observed imitative behavior. This approach is only possible if novel behaviors are chosen which are unlikely to be subject to phylogenetic influences.
One way to highlight some of the drawbacks of the above approach is to briefly describe one carefully implemented but typical study. Palameta and Lefebvre (1985) performed such a study in which they claimed to show true imitation in pigeons using a variation of the design described above. One experiment in their study consisted of exposing groups of experimentally naive pigeons to the sight of a demonstrator pigeon performing different aspects of a food-finding problem. The groups were each made up of five pigeons and were defined as follows:
1. True imitation (TI) group: observers were exposed to a demonstrator piercing the red half of a paper top covering a food box and eating the food underneath.
2. Social facilitation (SF) group: observers were exposed only to a demonstrator piercing the red half of a paper top covering a food box.
3. Local enhancement (LE) group: observers were exposed only to a demonstrator eating from a hole cut in the red half of a paper top covering a food box.
4. No model (NM) group: observers were not exposed to a demonstrator.
Each observer was given ten 10-minute trials twice a day in which they could observe the demonstrator through a clear partition. At the same time they also had access to a covered food box identical to the demonstrator's food box. All the TI pigeons, and 4 out of 5 of the LE pigeons, pierced the paper and ate within the 10 trials, although the former group performed this task more quickly. None of the NM group and only 1 of the SF pigeons pierced and ate within the 10 trials. From these results, Palameta and Lefebvre concluded that it was very likely that the TI group of pigeons had used socially transmitted information because acquisition of the food-finding technique had taken less time than in the LE group. Although the LE group had also succeeded in performing the task, this was considered to be caused by local enhancement and trial-and-error effects, thus accounting for the greater number of sessions generally taken to complete the task. As the SF group had failed to pierce and eat, the authors considered it probable that "observers copied the model only if they could anticipate a rewarding outcome of their action and that "TI" . . . pigeons learned at least part of the required piercing technique by observational learning [i.e., true imitation]" (p. 895).
Although much care went into the design of this experiment, it has since been rejected as a definitive demonstration of true imitation in animals (e.g., Galef, 1988; Heyes & Dawson, 1990). The topography of the behavioral sequence under analysis (i.e., pecking followed by eating) is probably under strong phylogenetic influence in pigeons. For instance, the classic study demonstrating that pigeons will peck discs without shaping when food is presented at frequent intervals (Brown & Jenkins, 1968) and the finding that chicks tend to automatically peck when a tapping sound is made, without prior behavioral observation or practice Cronhelm, 1970), both suggest that pecking and eating alone are not sufficiently novel behavioral topographies to be useful in imitation studies. Thus social facilitatory effects may also have accounted for the shorter latency to task completion by the TI birds in comparison with the LE birds. Local enhancement effects may be sufficient to explain each observer's tendency to pierce the red half of the covering if exposed to demonstrators pecking the red half. Certainly, the actual topography of the demonstrator's behavior would have been similar if the demonstrator had pecked the uncolored half of the container top. It is by no means conclusive, therefore, that the observers were paying attention to the topography of the demonstrator's movements.
Aware of the above drawbacks of Palameta and Lefebvre's study, Heyes and Dawson (1990) addressed TI by manipulating the topography of their demonstrator's behavior while holding constant the functional consequence of obtaining food. Using rats, they trained their demonstrators to push a joystick either to the right or to the left. The observers watched this from a different orientation and afterwards were allowed access to the joystick from the same orientation as the demonstrator. The observers subsequently received reinforcement for pushing the manipulandum in either direction. It was found that the observers pushed the manipulandum in the same direction as the demonstrator significantly more often than in the opposite direction. From their findings, Heyes and Dawson concluded that the rats had learned a "response-reinforcer relation" through observation alone; in other words, TI had occurred. The authors noted, however, that subtle stimulus cues such as the different material on the front wall (to the right of the joystick) and the back wall (to the left of the joystick) could have allowed them to learn stimulus-response relations through observation of the joystick's position in the box when access to food was made available to the demonstrator. In other words, it is possible that the functional consequences of the demonstrator's behavior (i.e., joystick near to "correct" wall) rather than the demonstrator's topographical movements could have been the salient factor controlling the imitative behavior emitted by the observers. Replications of this study should incorporate another control group to elucidate the functional control held by the movement of the joystick with relation to the walls. For example, the authors suggest that this group could be presented with the observation of the joystick being automatically moved toward either wall followed by the presentation of food. In this way, the function of the activity would vary without requiring the behavioral movements of the demonstrator. If this group of observers tended to push the joystick in the direction they had observed it being moved in, then it would not be possible to state that true imitation was solely responsible for any behavior-behavior concordance found.
To summarize, both Palameta and Lefebvre (1985) and Heyes and Dawson (1990) failed to empirically discriminate between control exerted by the topography of the demonstrators' target behaviors and control exerted by functional consequences of the demonstrators' behaviors. In the case of Palameta and Lefebvre's study, each observer's imitative behavior may have been either governed by the pecking movements of each demonstrator (i.e., similar topography) or by their observation of the demonstrator's manipulation of stimuli to obtain food (i.e., similar function). It is impossible to determine which of these alternatives is correct because neither were manipulated in their study. In the case of Heyes and Dawson's study, both the topography and the function of the demonstrator's behavior were altered simultaneously, so again the source of behavioral control was not apparent. In other words, when the joystick was orientated left instead of right, both the topography changed (i.e., movement to organism's left) and the function changed (i.e., joystick nearer left wall) and vice versa when the joystick was moved to the right.
According to Galef (1988) it is essential to devise studies which directly address the topography-function distinction if TI is to be demonstrated. He suggests that future studies follow the methodology employed by Dawson and Foss (1965) in which observers were exposed to demonstrators exhibiting different topographical behaviors with the same functional outcome. Dawson and Foss found that budgerigars used the same technique as their demonstrator to remove square lids from pots to obtain food, either removing the square by edging it off with their beak, lifting it off with their beak, or using a foot to dislodge it. Galef has proposed that each observer must have shown a sensitivity to the movements by the demonstrator in addition to the completion of the task in hand, thereby indicating that imitation of behavioral topography was taking place. In other words, true imitation appeared to have occurred. Unfortunately this study did not have any control groups to assess which topographically different behavioral techniques would have been used without exposure to a demonstrator. In addition, a careful replication study by Galef, Manzig, and Field (1986) which did incorporate controls found rather insignificant results and they could not conclude that imitation had occurred. Although not yet yielding positive results of TI in animals, this approach is empirically sound and warrants further attention, perhaps with other species, or observers who have been previously trained to be more "attentive" to the behavior of demonstrators (Hogan, 1988; Zentall, 1988).
Topographical and Functional Definitions of Behavior
An important conclusion to be drawn from the discussion above is that some of the disagreements among researchers stem from their failure to clarify the difference between imitative behavior which is topographically similar to the behavior of the demonstrator (e.g., Dawson & Foss, 1965) and imitative behavior which is functionally similar to the behavior of the demonstrator (e.g., Palameta & Lefebvre, 1985). In order to clarify the importance of making this distinction it is necessary to focus on the differences between topographically and functionally defined behavioral categories, and the consequences of adopting each.
A topographical analysis of behavior primarily consists of a description of the organism's postural activities and bodily movements without regard to any antecedents or consequences of the behavior (Lee, 1988). Take, for example, the act of a child brushing his/her teeth. A topographical description of such an act would involve a detailed description of the bodily movements made by the child. For instance, the child's elbows, wrists, and fingers may move toward each other, followed by clenching of the right fist while maintaining the left hand in a stationary position. This may be followed by bending the right elbow and moving the right hand toward the mouth. Upward and downward motions at slightly varying angles may be made for a number of minutes, followed by the moving of the right hand away from the mouth, an increased curving of the back and finally the opening of the jaw and lips with concurrent contracting of the muscles in the chest and mouth.
Behavior is functionally defined in terms of the contextual changes which precede, accompany, and follow it in a predictable manner.(2) A functional definition of teeth cleaning would therefore describe the contingencies operating on this act, without reference to behavioral movements. For instance, in the context of the bathroom at bedtime and the presence of the toothbrush and the toothpaste, the child may deposit toothpaste on the toothbrush. This consequence can then act as an antecedent in the sense that it sets the occasion for the child to remove tartar from its teeth. Various antecedent factors such as the physical sensation of clean teeth and the termination of an established brushing pattern occur prior to the child finally depositing its saliva in the wash-hand basin and rinsing the brush under the faucet.
The functional relationships between behavioral events and environmental stimuli outlined in the teeth-cleaning example concern contextual events which are relatively close in time and space to the behavior under observation. Behavior analysts, however, also acknowledge spatially and temporally separated functional relations in their definitional accounts of behavior (Lee, 1988). In theory, a full functional account of a behavioral event would include all contingently related contextual events extending backward and forward in time covering any spatial distance (Hayes & Brownstein, 1986).
From this perspective phylogenic contingencies lead to the selection of an organism's biological structure and function, to the form and function of behaviors it may perform, and to the salience required by particular stimuli in eliciting certain responses and shaping others. Ontogenic contingencies between the biological propensities of the organism and its environment determine the probability that it will respond in particular ways to certain stimuli. In turn, the probability of certain behavioral events occurring in the future will depend upon the relationship between the environmental consequences of the organism's current behavior and its phylogenic and ontogenic history. The nature of this relationship is summarized in the words of Keenan and Toal (1991):
At any one instance, the characteristics of the behavioral system
are dependent upon the interplay between the "plasticity" or
dynamic limitations inherent in the adaptiveness of the biological
system, and the dynamics imposed across time by the structure
of the prevailing contingencies. (p. 113)
As can be seen, then, both behaviors that are subject to strong phylogenic influences (e.g., sexual imprinting, courtship ritualization) and behaviors that are primarily shaped within the organism's lifetime are adequately dealt with in a full functional definition of behavior which is not spatially or temporally limited.
With respect to the descriptions of topographically and functionally defined behavior, it is clear that Dawson and Foss (1965) regarded TI as the copying of response topography. In their experiment, the outcome or functions of the demonstrator's activities were kept constant while their behavioral topographies varied across the groups. They reported that all the observer budgerigars removed the lids of the pots using the same gross behavioral movements as the demonstrator suggesting topographically imitative behavior.
The combined act of pecking downwards and then eating describes the topography of the target behavior to be imitated in the study undertaken by Palameta and Lefebvre (1985). This is a topographically common behavioral pattern for adult pigeons which terms, however, it novel in that respect to the observers. In functional terms, however, it was novel because each observer was required to peck the red half of a food coverting, a task which had never previously performed.
One way to reduce the likelihood of miscategorizing studies investigating imitation in animals is to define any target behaviors (i.e., those behaviors that are to be imitated by the observer) in terms of both their topography and their function. Studies can then be designed so that functional and topographical aspects of each demonstrator's behavior are manipulated independently of one another in order to determine where the stimulus control lies between the demonstrator behaving in context and the subsequent imitative behavior of the observer in the context defined by the demonstrator's behavior.
The distinction between functionally defined behavior and topographically defined behavior is also important when individual occurrences of imitative behavior are being classed together for scientific analysis. This classification is done on the basis of common defining features extracted by the experimenter and it will produce very different behavioral classes for topographically defined behaviors than for functionally defined behaviors (see Johnston & Pennypacker, 1980, for a detailed discussion). Topographically defined behavioral classes contain responses with common topographical features but which may occur under different settings, with different consequences, controlling stimuli and most probably different underlying neural mechanisms. For example, the topographical class of closing and opening an eye may include eye movement when winking, blinking, and sleeping which are very different in terms of the behavioral mechanisms and environmental stimuli that control their emergence, maintenance, and extinction.
In contrast, the classing together of functionally defined behaviors with common antecedents and/or consequences will include individual behaviors within the class which are very different in appearance. For example, the behavioral class of obtaining food by pigeons may include behaviors such as pecking a key in a controlled experimental setting, joining other feeding birds in a market square, and visiting a rubbish tip.
Classifying behaviors according to common topography without regard to functional antecedents and consequences has led to difficulties in observational learning research concerning the precise nature of their subject matter. As stated earlier, TI is distinguished from GA, LE, and SF by means of experimental manipulation. Thus when little attention is given to the functional consequences of behavioral patterns, then the information required for such analysis will probably be incomplete or inaccurate. If the processes controlling TI are to be known, then some degree of functional analysis must eventually take place in order for some of the unsuitable members of a topographically defined group to be omitted and for controlling variables to be chosen. However, by this stage, research time will have been wasted and behaviors that do not belong to the same topographical class rejected, even though they may be functionally similar. In contrast, the decision at the outset to define a behavioral class functionally will have already helped in the analysis of the data by outlining the variables that are likely to predict when certain behaviors will occur and how they may be controlled (Johnston & Pennypacker, 1980). Unlike topographically defined behavioral classes, then, those which are functionally defined include a fuller range of different behaviors controlled by functionally similar stimuli and followed by functionally similar consequences. Conclusions about the data should then be more widely applicable.
On a conceptual level, classifying behaviors in terms of topography rather than function may lead to the subsequent tendency to reify the label given to the behavioral class. This is a relatively common error in psychology (Lee, 1988) which can lead to the reified label being used as an explanation for the occurrences of its members. For instance, a researcher making this kind of conclusion about animal imitation would typically explain that when an organism emits imitative behavior it does so because it has an imitative ability. Explanations such as these add nothing to our understanding of the phenomenon and may inhibit further analysis of the environmental variables actually in control. In the absence of known environmental variables functionally related to it, a particular behavioral pattern such as imitative behavior may appear to arise solely from within the organism rather than as a consequence of its interaction with its environment.
Although behavior analysis is employed to examine behavioral patterns occurring relatively frequently in the laboratory, behavior synthesis offers an alternative approach ideally suited to the study of complex behavioral phenomena which are typically uncommon under controlled conditions. According to Catania (1984),
[The procedures involved in behavior synthesis] ... construct
performances that are related in some of their properties to
behavior outside the laboratory. Their laboratory study may
then reveal features that were not accessible in the
nonlaboratory situations from which they were derived. (p. 187)
If researchers suspect how TI may develop and if it is feasible to reproduce these or similar conditions in the laboratory, then the employment of behavior synthesis is possible. In other words, the observers can be "trained" to perform imitative behavior that is dependent upon observation of the topography of the demonstrator's behavior.
The procedures involved in behavior synthesis are increasingly being adopted to examine complex behavioral processes in animals. The considerable gains of this approach are evident from the recent series of studies carried out by Epstein and Skinner on symbolic communication, insight, self-awareness, and the spontaneous use of memoranda (see Epstein, 1986, for review and discussion).
A study by Baer, Peterson, and Sherman (1967), concerning the application of behavior synthesis techniques to produce imitative behavior in humans, offers conceptual and procedural ideas for the emergence of TI in animals. They trained retarded children, who, according to their care givers, were previously considered completely nonimitative, to imitate. Their procedure consisted of the training of many specific stimulus-response chains in which the children were shown how to emit particular behaviors which were then shaped to occur in the presence of the demonstrator performing similar behavior.
In this case, the demonstrator's behavior was a simple discriminative stimulus for the timing of the emission of a behavior which happened to bear topographical similarity to the properties of the stimulus (previously termed "matched dependent behavior" by Miller & Dollard, 1941). This type of behavioral interaction may look like true imitation to the casual onlooker who observes its occurrence but does not experimentally manipulate the prevailing contingencies to see what is governing it. However, such stimulus-response patterns, or as Catania (1984) calls them, cases of "specific imitation," are not examples of TI because the organism cannot imitate conspecific behavior that has not been specifically trained. However, Baer et al. (1967) found that after teaching many specific imitative behaviors, untrained or probe demonstrator behavioral topographies were then imitated immediately by the subjects without shaping. The term "generalized imitation" is used by radical behaviorists to describe this phenomenon because it appears that the stimulus control generated by the demonstrator's behavior over the observer's behavior has generalized along a stimulus dimension where the changing behavior of the demonstrator functionally controls a similar change in the behavior of the observer. In other words, the children emitted true imitative behavior by reproducing the topographical features of the demonstrators' behaviors rather than functional aspects of their behaviors.
It is not understood what the specific nature of this control is, or how it appears to develop. In the case of the above study, generalized imitative behavior was produced by exposure to many examples of specific imitative behavior. One cannot conclude, however, that this is the only path to the production of such behavior.
The use of the behavior-synthesis approach to imitation in animals will probably be seen as "cheating" by the majority of more ethologically orientated researchers in the field. However, as Epstein (1986) points out, some ways of behaving (such as TI) may not develop through appropriate natural contingencies for some species. In nature, the congtingencies generated through typical social interaction may not bring individuals under their control becuase they have no negative repercussions, and/or alternative ways of behaving may obtain similar rewards more immediately and easily. From this perspective, therefore, attempting to train TI to occur in the laboratory increases the chances of demonstrating it in various animal species. It is possible that future studies employing this technique could deduce whether or not a species' failure to exhibit TI in its natural environment is either a product of it being in an unsuitable context for TI to emerge or else evidence of a biological limitation of the species under observation. In particular, primate research in this area using the behavior synthesis approach may increase our understanding about the phylogenetic and ontogenetic changes leading to the development and prevalence of TI in our own species.
At present the search for true imitation in animals is in a state of deadlock which may lead to its eventual abandonment if a different approach is not taken and found to be more fruitful. The failure of the majority of studies to distinguish possible cases of TI from other types of imitative behavior has been explained in terms of their failure to experimentally distinguish between the control exerted by topographical features of demonstrators' behaviors and functional features of their behaviors. The current paper has proposed that a radically different approach exists which will allow a detailed understanding of the processes involved in the emission of imitative behavioral patterns contingent upon a predefined series of social interactions and which might facilitate the emergence of TI in various species under controlled conditions. Behavior synthesis has already been successfully applied to the analysis of other behaviors in animals which were supposedly a function of mental abilities.
With a high degree of experimental rigor then, behavior analysts are now tackling complex phenomena which have primarily been the domain of cognitivists, and to a lesser degree, of ethologists. Unlike them, however, behavior-analysts' explanations for the observed phenomenon are found in the organism's historical and current context and not inside the organism's head. It has been argued here that questions directed at events which are unobservable, or moreover, totally hypothetical, can only misdirect the researcher who is intent on understanding his or her subject matter to a degree where it can be predicted and controlled. By predicting and controlling their subject matter, researchers can leave the rather speculative realm of inner causes. Instead they can seek for explanations in the functional relationships existing between the behavioral patterns emitted by conspecifics, defining them in terms of phylogenetic predispositions, ontogenetic histories, and currently operating contingencies. (1) Most researchers in the field of observational learning focus only on behavioral patterns that have initially been obtained by one or more individuals through trial-and-error learning (Davis, 1981). It should not be forgotten, however, that observational learning lies within a broader dynamic social context which initiates, maintains, and shapes much of the everyday behavior of group members. This is evident when animals are deprived of social contact during their development. Social deprivation studies have shown that many animals, and in particular mammals, require lengthy periods of social contact with their parents and peers in order for "typically normal" behaviors to emerge which are essential to the species survival. For example, juvenile rhesus monkeys which received little contact from other monkeys failed to acquire normal social behaviors as adults and displayed inadequate parenting behavior (Bowlby, 1969; 1973; 1980). (2) The predictability of an organism's activity is empirically determined by systematically manipulating the presentation of stimuli which are suspected to be functionally related and then observing the behavioral changes.
AISNER, R., & TERKEL, J. (1985). Habitat exploitation through cultural transmission: Pine cone feeding behavior in black rats. Paper presented at the 19th International Ethological Conference, Toulouse, France. ANDREW, R. J. (1956). Intention movements of flight in certain passerines, and their use in systematics. Behavior, 10, 179-204. BAER, D. M., PETERSON, R. F., & SHERMAN, J. A. (1967). The development of imitation by reinforcing behavioral similarity to the model. Journal of the Experimental Analysis of Behavior, 10, 405-416. BANDURA, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice-Hall. BASTOCK, M. A. (1956). A gene mutation which changes a behavior pattern. Evolution, 10, 421-439. BECK, B. B. (1972). Tool use in captive hamadrys baboons. Primates, 13, 276-296. BOWLBY, J. (1969). Attachment and loss: Vol. 1: Attachment. London: Hogarth Press. BOWLBY, J. (1973). Attachment and loss: Vol. 2: Separation, anxiety and anger London: Hogarth Press. BOWLBY, J. (1980). Attachment and loss: Vol. 3: Loss, sadness and depression. London: Hogarth Press. BOX, H. O. (1984). Primate behavior and social ecology. London: Chapman and Hall. BROWN, P. L., & JENKINS, H. M. (1968). Auto-shaping of the pigeons key-peck. Journal of the Experimental Analysis of Behavior, 11, 1-8. CATANIA, A. C. (1984). Learning (2nd ed.). London: Prentice-Hall. CLAYTON, D. A. (1978). Socially facilitated behavior. Quarterly Review of Biology, 53, 373-391. COOK, M., MINEKA, S., WOLKENSTEIN, B., & LAITSCH, K. (1985). Observational conditioning of snake fear in unrelated rhesus monkeys. Journal of Abnormal Psychology, 93, 355-372. CORSON, J. (1967). Observational learning of a lever pressing response. Psychonomic Science, 7, 197-198. CRONHELM, E. (1970). Perceptual factors and observational learning in the behavioral development of young chicks. In J. H. Crook (Ed.), Social behavior in birds and mammals: Essays on the social ethology of animals and man (pp. 393-439). London: Academic Press. CURIO, E. (1988). Cultural transmission of enemy recognition. In T.R. Zentall & B. G. Galef, Jr. (Eds.), Social learning: Psychological and biological perspectives (pp. 75-97). Hillsdale, NJ: Lawrence Eribaum. DAVIS, M. (1981). Imitation. In D. McFarlands (Ed.), The Oxford companion to animal behavior (pp. 298-303). Oxford University Press. DAVIS, H., & MEMMOTT, J. (1982). Counting behavior in animals: A critical review. Psychological Bulletin, 92, 547-571. DAWKINS, M. S. (1985). The scientific basis for assessing suffering in animals. In P. Singer (Ed.), In defence of animals (pp. 27-40). Oxford: Basil Blackwell. DAWSON, B., & FOSS, B. M. (1965). Observational learning in budgerigars. Animal Behavior, 13, 470-474. DENNY, M. R., CLOS, C. F., & BELL, R. C. (1988). Learning in the rat of a choice response by observation of S-S contingencies. In T R. Zentall & B. G. Galef, Jr. (Eds.), Social learning: Psychological and biological perspectives (pp. 207-223). Hillsdale, NJ: Lawrence Erlbaum. DEWSBURY, D. A. (1990). Fathers and sons: Genetic factors and social dominance in deermice, Peromyscus maniculatus. Animal Behavior, 39, 284-289. EPSTEIN, R. (1986). Bringing cognition and creativity into the behavioral laboratory. In T. J. Knapp & L. C. Robertson (Eds.), Applications to cognition: Contrasts and controversies (pp. 91-110). Hillsdale, NJ: Lawrence Erlbaum. FISHER, J., & HINDE, R. A. (1949). The opening of milk bottles by birds. British Birds, 42, 347-357. FORKMAN, B. (1991). Social facilitation is shown by gerbils when presented with novel but not with familiar food. Animal Behavior, 42, 860-861. FRAGASZY, D., & VISALBERGHI, E. (1989). Social influences in the acquisition of tool-using behaviors in tufted capuchin monkeys (Cebus apelia). Journal of Comparative Psychology, 103, 159-170. GALEF, B. G., Jr. (1976). Social transmission of acquired behavior: A discussion of tradition and social learning in vertebrates. In J. S. Rosenblatt, R. A. Hinde, E. Shaw, & C. Beer (Eds.), Advances in the study of behavior (Vol. 6, pp. 77-100). New York: Academic Press. GALEF, B. G., Jr. (1988). Imitation in animals: History, definition, and interpretation of data from the psychological laboratory. In T. R. Zentall & B. G. Galef, Jr. (Eds.), Social learning: Psychological and biological perspectives (pp. 3-28). Hillsdale, NJ: Lawrence Erlbaum. GALEF, B. G., Jr., KENNETT, D. J., & WIGMORE, S. W. (1984). Transfer of information concerning distant foods in rats: A robust phenomenon. Animal Learning & Behavior, 12, 292-6. GALEF, B.G., Jr., MANZIG, L. A., & FIELD, R. M. (1986). Imitation learning in budgerigars: Dawson and Foss (1965) revisited. Behavioral Processes, 13, 191-202. GANDOLFI, G., & PARISI, V. (1973). Ethological aspects on predation by rats, Rattus norvegicus (Berkenhout), on bivalves Unio pictorum, L. and Cerastoderma lamarki (Reeve). Bollettino di Zoologia, 40, 69-74. GARDNER, R. A., & GARDNER, B. T. (1969). Teaching sign language to a chimpanzee. Science, 165, 664-672. GOODALL, J. (1970). Tool-using in primates and other vertebrates. Advances in the Study of Behavior, 3, 195-249. GRIFFIN, D. R. (1984). Animal thinking. London: Harvard University Press. HALL, K. R. L., & SCHALLER, G. B. (1964). Tool-using behaviour of the California sea offer. Journal of Mammalogy, 45, 287-298. HAYES, S. C., & BROWNSTEIN, A. J. (1986). Mentalism, behavior-behavior relations and a behavior-analytic view of the purposes of science. The Behavior Analyst, 9, 175-190. HERRNSTEIN, R. J. (1984). Objects, categories, and discriminative stimuli. In H. L. Roitblat, T. G. Bever, & H. S. Terrace (Eds.), Animal cognition (pp. 233-261). Hillsdale, NJ: Erlbaum. HEYES, C. M., & DAWSON, G. R. (1990). A demonstration of observational learning in rats using bidirectional control. The Quarterly Journal of Experimental Psychology, 42B, 9-71. HOGAN, D. E. (1988). Learned imitation by pigeons. In T. R. Zentall & B. G. Galef, Jr. (Eds.), Social learning: Psychological and biological perspectives (pp. 225-238). Hillsdale, NJ: Lawrence Erlbaum. HUMPHREY, G. (1921). Imitation and the conditioned reflex. Pedagogical Seminary, 28, 1-21. JOHNSTON, J. M., & PENNYPACKER, H. S. (1980). Strategies and tactics of human behavioral research. Hillsdale, NJ: Lawrence Erlbaum. KAWAI, M. (1965). Newly acquired pre-cultural behavior of the natural troop of Japanese monkeys on Koshima Inlet. Primates, 6, 1-30. KEENAN, M., & TOAL, L. (1991). Periodic reinforcement and second-order schedules. The Psychological Record, 41, 87-115. KREBS, J. R. (1974). Colonial nesting and social feeding as strategies for exploiting food resources in the great blue heron (Ardea herodias). Behavior, 51, 99-131. KYMISSIS, E., & POULSON, C. L. (1990). The history of imitation in learning theory: The language acquisition process. Journal of the Experimental Analysis of Behavior, 54, 113-127. LALAND, K. N., & PLOTKIN, H. C. (1991). Excretory deposits surrounding food sites facilitate social learning of food preferences in Norway rats. Animal Behavior, 41, 997-1005. LEE, V. (1988). Beyond behaviorism. Hillsdale, NJ: Lawrence Erlbaum. LEFEBVRE, L. (1986). Cultural diffusion of a novel food-finding behavior in urban pigeons: An experimental field test. Ethology, 71, 295-304. LEFEBVRE, L., & PALAMETA, B. (1988). Mechanisms, ecology, and population diffusion of socially learned, food-finding behavior in feral pigeons. In T R. Zentall & B. G. Galef, Jr. (Eds.), Social learning: Psychological and biological perspectives (pp. 141-164). Hillsdale, NJ: Lawrence Erlbaum. LOPEZ, J. C., & LOPEZ, D. (1985). Killer whales (Orcinus orca) of Patagonia, and their behavior of intentional stranding while hunting nearshore. Journal of Mammology, 66,181-183. MILLER, N. E., & DOLLARD, J. (1941). Social learning and imitation. New Haven: Yale University Press. PALAMETA, B., & LEFEBVRE, L. (1985). The social transmission of a food-finding technique in pigeons: What is learned? Animal Behavior, 33, 892-896. REGAN, T. (1987). The struggle for animal rights. Clarks Summit, PA: The International Society for Animal Rights. ROMAN ES, G. J. (1884). Mental evolution in animals. New York: AMS Press. SHERRY, D. F., & GALEF, B. G., Jr. (1984). Cultural transmission without imitation: Milk bottle opening by birds. Animal Behavior, 32, 937-938. SHERRY, D. F., & GALEF, B. G., Jr. (1990). Social learning without imitation: More about milk bottle opening by birds. Animal Behavior, 40, 987-988. SINGER, P. (1985) (Ed). In defense of animals. Oxford: Blackwell. SKINNER, B. F. (1984). The phylogeny and ontogeny of behavior. The Behavioral and Brain Sciences, 7, 669-711. STEPHENSON, G. (1967). Cultural acquisition of a specific learned response among rhesus monkeys. In D. Starck, R. Schneider, & H. Kuhn (Eds.), Progress in primatology. Stuttgart: Gustav Fisher Verlag. THORNDIKE, E. L. (1911). Animal intelligence. New York: Macmillan. THORPE, W. H. (1963). Learning and instinct in animals (2nd ed.). London: Methuen. WAITE, R. K. (1981). Local enhancement for food finding by rooks (Curvus frugilegus) foraging on grassland. Zeitschrift fur Tierpsychologie, 57, 15-36. WARDEN, C. J., & JACKSON, T. J. (1935). imitative behavior in the rhesus monkey. Journal of Genetic Psychology, 46, 103-125. ZENTALL, T. R. (1988). Experimentally manipulative behavior in pigeons and rats. In T. R. Zentall & B. G. Galef, Jr. (Eds.), Social learning: Psychological and biological perspectives. Hillsdale, NJ: Lawrence Eribaum.
This article was written while Maureen Howard was the holder of a Research Grant from the Department of Education for Northern Ireland. Requests for reprints should be addressed to Maureen Howard, or Michael Keenan, Department of Psychology, University of Ulster, Coleraine, Northern Ireland, BT52 1SA.
|Printer friendly Cite/link Email Feedback|
|Author:||Howard, Maureen L.; Keenan, Michael|
|Publication:||The Psychological Record|
|Date:||Mar 22, 1993|
|Previous Article:||On the context of yawning: when, where, and why?|
|Next Article:||The effects of chronic lead exposure on taste-aversion conditioning in rats.|