Social Learning: Theoretical and Methodological Issues
Bennett G. Galef, Jr.
Since the latter part of the 19th century, scientists have discussed the possibility that animals are capable of learning by imitation. Darwin (1871) explained difficulties in poisoning or trapping wild animals as the result of their ability "to learn caution by seeing their brethren trapped or poisoned" (p. 49). Wallace (1870) interpreted consistency from generation to generation in the structure of the nests of birds of the same species as the result of young observing and imitating the nest of their parents. Romanes (1884) treated imitation learning and subsequent biological inheritance of imitated behaviors as responsible for both continuity across generations in species-typical patterns of behavior and the perfection of instincts. During the early part of the present century, many of the major figures in the early history of experimental psychology (Hobhouse, 1901; Kohler, 1925; Lashley, 1913; McDougall, 1924; Morgan, 1900; Thorndike, 1911; Watson, 1908), as well as any number of less well remembered behavioral scientists, studied and speculated about the process of imitation learning (Berry, 1906, 1908; Cole, 1907; Davis, 1903; Haggarty, 1909; Kempf, 1916; Kinnaman, 1902; Porter, 1910; Sheperd, 1910, 1911, 1923; Small, 1900, 1901; Witmer, 1910).
In consequence, in discussing imitative learning in animals, one has to
consider a long and venerable history that provides sources of both comfort and confusion: Comfort, in that study of learning by imitation in animals for more than a century suggests the topic of animal imitation is of intrinsic interest; confusion, in that historical diversity in approaches to study of imitative behavior has produced incompatible conceptual frameworks for analysis of imitative phenomena. One man's example of true learning by imitation is another's paradigmatic case of "pseudo-imitation" and each can cite historical precedent for treating phenomena of interest as he does.
Early work on imitation learning is not only of historical interest. The latter half of the 19th century saw the formulation of alternative approaches to the study of imitative phenomena that, even today, shape research in the area. The views of major figures in the behavioral biology and psychology of the last century, provide an important foundation for understanding the origins of much contemporary disagreement and confusion as well as a benchmark from which to measure a century's progress in the study of imitative behavior.
Early Perspectives on Learning by Imitation
The major impetus for 19th-century discussion of imitation arose out of disagreement among leading scientific figures of the period concerning the origins of the higher mental faculties of man. Darwin and Wallace, co-formulators of evolutionary theory, differed profoundly over the possibility of employing the principle of evolution, of descent with modification, to understand the development of the human mind. As a contemporary, George Romanes (1884), stated the issue:
. . . the great school of evolutionists is divided in two sects; according to one the mind of man has been slowly evolved from the lower types of psychical existence, and according to the other the mind of man, not having been thus evolved, stands apart, sui generis from all other types of existence, (p. 9)
The dispute was similar to modern debate over whether animals, like men, are capable of conscious thought, "for them to know, or think consciously about the eventual results of what they are doing" (Griffin, 1985, p. 480); the issue today, as in 1884, is the continuity of human and animal mind. In one way, the controversy at the end of the last century was more respectable than its modern counterpart; during the former debate, there was some consensus as to evidence that would decide the issue: indication that animals had humanlike emotions such as shame, remorse, jealousy, and benevolence, that they could use tools or act deceitfully, that they were able to solve complex problems or imitate complex acts.
For both George Romanes, a staunch advocate of the Darwinian view, and for his opponents, demonstrations of imitative learning in animals were seen as providing important evidence of an evolutionary origin of the higher mental faculties of man. The capacity for imitation in animals was viewed as ancestral to the unique human faculty for culture.
Because of the view of phylogeny held by Romanes and many of his contemporaries, failure to find evidence of gradually increasing complexity in imitative behavior as one ascended the great chain of being would have disconfirmed the continuity position. Romanes did not share Darwin's conception of phylogeny as a branching process (Galef, 1986). Rather, Romanes's discussions of evolution have implicit within them the older Spencerian (1855) view (now discredited; Hodos & Campbell, 1969) that it is possible to trace a historically meaningful, linear development of mind across extant species. In consequence, Romanes believed the Darwinian notion of continuity required the presence in living animals of a graded series of primitive precursors of human mental and moral faculties.
Imitation learning was a particularly important test case for Romanes (1884, 1889) because he believed that the imitative faculty reached its highest levels of perfection, not in rational, adult, European man, but in slightly inferior forms: monkeys, children, savages, and idiots (Romanes, 1884, p. 225). Hence, imitation was a faculty one would expect to find, in at least rudimentary form, in species standing yet lower on the psychological scale. Seeking evidences of primitive imitative capacities in animals, Romanes was quick to find them. Romanes's (1884, 1889) classic texts provide many examples.
The first instance of imitation, and the one described by Romanes (1884) at greatest length, is an example of imitation by honeybees of a behavior exhibited by bumblebees.
One morning for the first time, I1
saw several humble-bees . . . visiting flowers [of the kidney bean}, and I saw them in the act of cutting with their mandibles holes through the under side of the calyx, and thus sucking the nectar: all the flowers in the course of the day became perforated, and the humble-bees in their repeated visits of the flowers were thus saved much trouble in suckling. The very next day I found all the hive-bees, without exception, sucking through the holes which had been made by the humble-bees. How did the hive-bees find out that all the flowers were bored, and how did they so suddenly acquire the habit of using the holes? . . . I must think that the hive-bees either saw the humble-bees cutting the holes, and
understood what they were doing and immediately profited by their labour; or that they merely imitated the humble-bees after they had cut the holes, and when sucking at them. (p. 220—221)
Romanes then briefly mentions a number of additional cases of imitative learning reported by other correspondents: (1) dogs in the Falkland islands that learned from one another the best way of attacking cattle, (2) chickens learning to respond to "the danger cries and signals employed by other species," (3) birds imitating the songs of different species, (4) birds of some species that "articulate words" or "songs having a proper musical notation," (5) dogs foster-reared by cats acquiring feline patterns of behavior such as face-washing, avoidance of water, and stalking mice, (6) juvenile birds taught by their elders to fly, (7) hawks taught by their parents "to more perfectly swoop upon their prey," and (8) newly hatched chicks learning to drink water by imitating their fellows.
Romanes (1884) justified treating this diverse collection of observations as exemplifying a single underlying process, imitation, by inferring that in each case "there must first be intelligent perception of the desirability of the modification on the part of certain individuals, who modify their actions accordingly" (p. 229).
In Romanes's view, modification of behavior as the result of interaction with others implied both intelligence and intentionality in the imitator. These inferences of intelligence and intentionality from evidences of imitation were both crucial to Romanes's main line of argument and a recurring problem in succeeding decades.
If imitation in animals results from psychological processes qualitatively different from those underlying imitation in man (presumed to be intentional and intelligent), then instances of apparent imitative learning in animals are not true precursors of the human faculty for culture; such examples of animal imitation would be, in modern terms, analogues rather than homologues of human imitation. In consequence, Romanes's use of evidence of imitation in animals to provide a bridge between the minds of animals and the minds of men required interpretation of instances of animal imitation as examples of the exercise of rudimentary versions of humanlike capacities for intelligent, intentional action. J. T. Bonner's (1980) recent tracing of the evolution of culture has a similar underlying philosophy.
C. L. Morgan
The need to determine whether a given instance of animal imitation depended on faculties of mind similar to those assumed to be employed in imitation learning by humans was recognized early in the history of behavioral biology. C. L. Morgan (1900) proposed that imitation may be of two basic types, either
"instinctive," or "reflective"2
and that it is only the latter type, "deliberate and intentional imitation . . . directed to a special end more or less clearly perceived as such" (p. 193), that should properly be considered imitation in the sense the term is used in describing the behavior of humans after infancy.
A chick sounds the danger note; this is the stimulus under which another chick sounds a similar note. . . . Such a procedure may be described as imitative in its effects, but not imitative in its purpose. Only from the observer's standpoint does such instinctive behaviour differ from other modes of congenital procedure. It may be termed biological but not psychological imitation. And if it be held [as Romanes asserted] that the essence of imitation l...