Beyond Anecdotes: An Empirical Study of "Anthropomorphism"

Paul Morris, Margaret Fidler, and Alan Costall 1

The status of "anthropomorphic" descriptions of animals in terms of intentions and emotions has been generally regarded as a prescriptive methodological concern. In contrast, in the study of human social psychology the nature of psychological descriptions of other people has been approached as a substantive empirical issue. Following this lead, the present study investigated the nature of people's descriptions of short videotaped episodes of animal behavior. The descriptions obtained were predominantly anthropomorphic and structured according to a limited set of "event units" whose psychological meaning was highly consistent across the observers. In common with many social psychologists we conclude that consistency of anthropomorphic description suggests that meaning is specified within the structure of behavior.

This paper is concerned with the anthropomorphic or psychological description of animals behavior. Strictly speaking, however, "anthropomorphism" refers to a specific and highly influential theory (and associated method) of how psychological descriptions of other animals are possible. This theory, developed by Clifford, Romanes, and Morgan, maintained that the sole way in which a person can attribute intentions to another animal (human or nonhuman) is by an inference from his or her individual case. We most definitely do not subscribe to this theory, at least as a general explanation of the understanding of others. For example, it could hardly explain how a vegetarian prey animal (such as a zebra) could ever catch on to the quite alien intentions of a carnivorous predator (such as a lion).

The place of anthropomorphic description in the scientific study of animal behavior has been the subject of controversy for more than a century and seems destined to remain so (Breland & Breland, 1966 p. 12; Kennedy, 1992; Mitchell, Thompson, & Miles, 1997; Skinner, 1974). Over the years, however, there has been a growing reaction to the prohibition—wrongly attributed to Morgan (Costall, 1995)—against the psychological description of animal behavior. Psychological description is now widely accepted (Crist, 1999; Goodall ,1990; Griffin, 1983; Rollin, 1989; & Sanders ,1999). Yet the justification for this change has remained largely at the level of theory, such as the good argument that this prohibition runs counter to the Darwinian commitment to psychological continuity across species. But there is a lack of relevant empirical research. Indeed, the available evidence, supposedly demonstrating the unreliability of the psychological description of animal behavior, is itself suspect and often anecdotal, including Morgan's famous example of his terrier, Tony, opening the garden gate with his nose with apparent insight and intelligence. Some may object that Morgan made repeated and careful observations of his dog and should not be deemed anecdotal; however, the same may be said of many observations of animal trainers, farmers, and pet owners in general.

The question of making sense of the behavior of "others" is not specific to non-human animals. The scientific investigator encounters the same problem of difference—and hence enters into the realm of comparative psychology—when dealing with children, the insane, "primitives," or indeed anyone who is not "an Englishman, of the same social grade as ourselves, of like tastes and habits of thought, educated under the same school system ... though even here there must be slight individual differences" (Morgan, 1903, p. 41).

So what have psychologists had to say on the topic of being anthropomorphic about people? In the standard textbooks, the topic arises in two separate places. The first is in the introductory chapter setting out the scientific credentials of the discipline, and here the topic is dealt with in the familiar way, as a prescriptive methodological issue: Psychological accounts of other people, the textbooks claim, are inevitably indirect and hence prone to error (Brehm & Kassin, 1996, p. 94; Hogg & Vaughan, 1995, p. 68; Lippa, 1994, p. 117). The same issue also arises in the chapters on social psychology, yet here it is treated as a substantive subject for empirical investigation: Just how accurate and consistent are people in describing other people? A long line of experimental research seemed to indicate that such descriptions are indeed largely inaccurate, but these studies, as is now widely acknowledged, were flawed, having used static and highly simplified displays. There is however, an alternative approach, largely stemming from Gestalt psychology, which, while often using simplified displays, has attempted to capture or abstract the relevant spatio-temporal structures upon which person perception might be based (Berry & Springer, 1993; Heider & Simmel, 1944; Kassin, 1982; Massad, Hubbard, & Newtson, 1979; Springer, Meier & Berry, 1996; Runeson & Frykholm, 1983; Waseda, 1966). These studies have demonstrated impressively high consistency among observers in the perception of other people's actions and intentions—including deceptive intentions—and, on this basis, have argued that such consistency must have a basis in the structure or form of actions and posture (Asch, 1952; Michotte, 1950; Newtson, 1990; Runeson, 1985).

About This Study

Taking the lead from this empirical approach within human social psychology, the following study is an attempt to determine the incidence and consistency of people's psychological descriptions of the behavior of animals. Curiously, there seems to be just one previous study in the literature on animal behavior that relates closely to our own. Hebb (1946) gave an account of a study he conducted at the Yerkes laboratories, in which he compared anthropomorphic and non-anthropomorphic descriptions of the behavior of chimpanzees. He claimed that "objective" descriptions failed to capture aspects of behavior captured in anthropomorphic descriptions critical to effective interaction with the chimpanzees. However, "with the use of frankly anthropomorphic concepts of emotion and attitude one could quickly and easily describe the peculiarities of the individual animals, and with this information a newcomer to the staff could handle the animals as he could not safely otherwise" (p. 88). Unfortunately, Hebb's own article provides just a few illustrative examples, but no systematic data about the anthropomorphic descriptions nor their consistency. Although there is a growing body of research on lay people's judgments of the plausibility of pre-prepared texts framed in anthropomorphic language describing animal behavior (Eddy, Gallup, & Povenilli, 1993; Mitchell & Hamm, 1997), we know of no recent studies of how people describe actual examples of animal behavior they have just observed.

Our study is modeled on a classic study by Heider & Simmel (1944) demonstrating the immediacy and consistency of people's descriptions of social behavior. In their study, they constructed a simple cartoon film of geometrical shapes (triangles and squares) in which they attempted to convey actions such as chasing, and emotions such as anger, in the way these shapes moved about. Despite the simplicity of the display, and the open-ended instructions merely to "write down what happened in the picture," observers not only described the movements in psychological terms, but also showed remarkable agreement in the meanings of these movements.

In our own study, we have used short video-recordings of the actual behavior of dogs in everyday settings but with a similar aim: to determine the extent to which human observers use psychological descriptions of animal behavior and use such descriptions in a consistent way. Like other authors both for and against anthropomorphism, we define psychological/anthropomorphic descriptions largely by contrast, namely descriptions that go beyond mere movements. In our view, the actual nature of such descriptions should be a matter for empirical enquiry, as in the following study.

Procedure

Twenty-eight first-year undergraduate students (modal age 18 years; 22 female and 6 male of whom 11 had lived at some time with a "pet" dog and 17 had not) were asked to view four videotaped interactions of dogs and their "owners," or caregivers. After viewing each episode twice, they were asked to write down "what you think is going on, focusing particularly on the dog." No instructions were given as to form or content of the requested descriptions. Two minutes were allowed for writing each account. Participants viewed the video episodes in groups of up to eight, but no conferring was allowed.

Video Episodes

The selected video episodes lasted from 15 to 50 seconds and were selected from lengthier videotapes. The dogs and owners who were videotaped were selected on the basis that they were known to one of the investigators. Owners were asked to make tapes of typical interactions between themselves and their pets. The major criterion for selection of the video episodes was that the owner and dog were interacting for a major portion of the episode. Four video episodes were included in the study: The first episode showed a seated owner stroking his dog (referred to as "stroking episode" in the remainder of the paper); the second showed an owner eating with her dog at her feet (eating episode); the third showed a dog being teased and talked to by its owner (teasing episode); the fourth showed a dog with owner and the reaction of the dog to the owner leaving (leaving episode)

The purpose of the initial stage of analysis was to discover whether there was consistency in content and/or structure of the participants' accounts of the behavior they had observed. We used a parsing technique that was developed for textual analysis by Mandler & Johnson (1977). Parsing in this context means the identification and classification of salient and recurrent elements in the descriptions provided by participants. The recurrent elements, which are termed event units, are thus identifiable sub-episodes of behavior derived from the descriptions of an episode. An exhaustive list of event units within episodes was compiled for each video episode. For example, we identified the following set of event units from all the descriptions of the stroking episode: owner stroking dog #1; dog is relaxed #2; owner stops stroking the dog #3; dog raises paw to encourage further stroking #4; owner starts again #5; owner stops again #6; dog raises paw and head to encourage more attention #7 (see Figure l); looks at owner to encourage the owner to stroke more #8; owner strokes again, gives in #9; dog enjoys being stroked #10.


Figure 1: Dog raises paw and head to encourage more attention (Stroking episode, Event Unit 7)


An independent rater (using the event unit descriptions agreed by the investigators) then examined each participant's description of an episode to establish whether a particular event unit was present or absent. The range of agreement in the compiled tables of event units between the investigators and the independent rater ranged from 87% for the leaving episode to 91% for the stroking episode. Disagreements about the presence or absence of a unit in a particular account were resolved through an iterative process, although, in fact, most of the disagreements involved obvious oversights. Almost all accounts could be accommodated within the event units identified. A very small percentage, 3.8%, of the otherwise typical accounts occasionally contained unique and unclassifiable units.

The descriptions corresponding to each event unit were classified as anthropomorphic, "as if" or mechanistic. A unit was classified as anthropomorphic if psychological meaning was an integral part of the description ("the dog watched the person eating and moved to a position to get eye contact and sat and tried without break to get the person's attention"); "as if" descriptions employed anthropomorphic language but with the qualification of "as if" or equivalent ("dog appears to get excited; curious, he turns around as if looking for what the owner is talking about"); units classified as mechanistic simply referred to movements of the animal devoid of meaning and purpose ("when the owner stops stroking, dog moves a paw and turns stomach toward the owner"). Classification into these categories by two independent raters was highly consistent; agreement across all four episodes was above 94%.

A further classification of the event units was then undertaken to establish whether participants were indeed attaching the same psychological meanings for the same event unit in either their anthropomorphic or "as if" descriptions. Again the consistency between the two independent raters in this classification was high (98%).

For the purposes of this paper the most important aspect of the results is the degree of consistency of the anthropomorphic descriptions of particular event units. Consider for example the descriptions used for event unit 4 from the stroking episode. Our own gloss for this event was "dog raises paw to encourage further stroking." Here we list all the accounts obtained from the participants who referred to this event unit:

Dog prompted for more;
Dog raised paw to encourage further stroking;
Dog moves paw and moves slightly so the man will carry on stroking;
Dog trys to get stroked more;
Dog seeks contact with owner, shown by paw movement, indicating that dog did not want owner to stop petting;
Dog rests and moves paw up to gain attention and stroking from the person in the chair;
Dog signals to continue the stroking;
Dog entices the adult to continue rubbing chest by begging with paw;
Dog raised paw up and looked at owner signalling owner to carry on stroking;
Dog lifted leg a couple of times to encourage the person to continue;
Dog wants more and indicated this by raising paw;
Dog waves paw and turns stomach toward owner, gets more tickles;
Dog asks for more by raising paw;
Dog raised paw to be stroked more;
Dog twitches ear with paw to indicate owner should continue;
Dog lies in such a way to encourage the stroking;
Dog shakes paw and looks pleadingly.

Dog waves leg as if hinting for more;
Dog lifted a paw as if wanting more;
The dog raised paw as if in protest;
The dog waved paw again as if wanting the person to continue;

When the owner stops stroking, dog moves paw and turns stomach toward the owner.

Not only were anthropomorphic descriptions predominant but also identical meanings were attached to nearly all the anthropomorphic and "as if" descriptions. In fact, there were few occasions where there was any inconsistency in content. A rare example of an anthropomorphic description of an event unit that deviates from the rest can be found in event unit 1 of the "person leaving" episode. We list typical examples of the description of the event unit followed by the anomalous description in italics. Our gloss for this event unit is "dog enjoying attention from owner."

Obviously enjoying the attention
Dog enjoys being rewarded and patted
Dog enjoys petting from owner
Dog enjoys the fuss
Dog delighted in being petted
Dog was very happy to be petted
Dog slightly reluctant to the petting

Consistency of description was not restricted to the most widely included units. For example, only seven of the participants referred to event unit 3 for the teasing episode, but all seven descriptions were consistent on the dog's confusion:

But a little confused as to what was being asked to do; then acts confused and tries to look around the room;
Dog was being asked to find something, though I suspect had no idea what for;
Dog is confused by owner's tone of voice;
Confusion is evident;
Dog acts confused;

Dog seemed very alert but a little confused.

In fact, event units for all the episodes display a similar high level of consistency of description. Limitations of space preclude the inclusion of all the descriptions obtained, but Tables 1-4 present the percentage of descriptions that were classified as having the same content for a particular type of description; the number next to the percentage is the number of participants who actually used the event unit for the particular description type.
 

From Tables 1-4, it is clear that the number of non-anthropomorphic event units is small (7% of total). Furthermore there is no pattern to the distribution of non-anthropomorphic event units across or within episodes.

There was just a small number of accounts (five of the 112 descriptions) that differed qualitatively from the other accounts. These atypical accounts tended to provide an overall description without attention to detail such as, "Submission of the dog. Forming a loving relationship between handler and dog" from the stroking episode; such descriptions could not be classified into event units.

Statistical analysis (F=1.21, p=.33) revealed no difference in the number of words used in units that were classed as anthropomorphic (mean 9.45), mechanist (mean 10.16) or "as if" (mean 9.00). Thus, at least in relation to the present paradigm anthropomorphic description was not being used as a convenient shorthand.

The analyses we have presented so far bear most directly on the purpose of our study. It was not the case however that observers included the full set of event units in their descriptions

Participants tended to be restricted in their use of units (the mean number of units used in stroking episode was 4.25 out of a possible 10; the mean for eating episode was 2 out of possible 4; the mean for teasing episode was 1.78 out of a possible 5; the mean of leaving episode was 3 out of a possible 5) which accounts for the low column totals in tables 1-4. There was no pattern to the inclusion or exclusion of event units by the participants. Despite variation in the number of event units used by participants, the essence of the episode was conserved, as in the following examples of the stroking episode, which we give verbatim:

The owned leant down and stroked his dog, when the owner stopped stroking, the dog raised its paw to encourage further stroking, which the owner did. Once the owner stopped again, the dog again raised its paw and lifted its head, looking directly at the owner to encourage further stroking, which it was obviously enjoying.
The dog moves his paw up to gain attention and stroking from the gentleman in the chair. The man strokes the dog.
In summary, participants varied greatly in the number and selection of event units in relation to any particular episode. Nevertheless, their reports drew upon a very definite and limited set of event units.

Details were recorded of whether participants had owned or lived with a dog. There was no significant effect of experience with pet dogs on the number of event units used (mean number of event units used by participants with experience of pet dogs = 2.99 and mean without experience =2.97; F=.32, p=.575). Similarly there was no impact on usage of non-anthropomorphic language between the two groups ((2 = .67, p=.88). All the inconsistent event unit descriptions were from participants with no experience with pet dogs, however, the overall number of such units (three) is so small it is difficult to know how to interpret this finding.

Conclusions

Whereas "anthropomorphism" has largely been regarded as a methodological and prescriptive issue in the study of animal behavior, the present study has approached the status of how people describe animal behavior as substantive, empirical questions. There were four main findings:

1. The descriptions provided by the observers of the video-recorded activities were predominantly psychological

2. The descriptions drew upon a definite and limited set of "event units."

3. There was remarkable agreement amongst participants in the meanings of the anthropomorphic descriptions they applied to any particular event unit.

4. No participant ever included all of the event units in their descriptions, nor was there any consistent pattern of which units were included or excluded depending on the extent of elaboration of the description, yet the "gist" of the episodes conveyed in the descriptions was nevertheless similar.

In short, the so-called "anthropomorphic" descriptions we obtained in our study were far from inconsistent between the observers. But what is the basis of such consistency? In the literature on human social psychology, the consistency of the perception of other people's behavior has been presented as prima facie evidence that there must be something in the structure of human actions and posture specific to different intentions and emotions. On the basis of our findings, we wish to make a similar claim about the structure of animal behavior: The consistency of the anthropomorphic descriptions we have obtained must be grounded in the structure of the animals' behavior. If this claim is true, however, it raises in turn yet another empirical question for further research: What precisely is it in the structure of human and animal activities that affords such remarkable consistency in the anthropomorphic description of those activities.

References

Asch, S. E. (1952). Social Psychology. Englewood Cliffs, NJ: Prentice-Hall.

Berry, D. S. & Springer, K. (1993). Structure, motion and preschoolers' perceptions of social causality. Ecological Psychology, 5 , 273-283.

Brehm, S. S. & Kassin, S. M. (1996). Social Psychology. Boston: Houghton Mifflin Company.

Breland, K. & Breland, M. (1996). Animal Behavior. Toronto: Macmillan.

Costall, A. (1993). How Lloyd Morgan's canon backfired. Journal of the History of the Behavioural Sciences, 29 , 113-122.

Crist, E. (1999). Images of Animals, Anthropomorphism and Animal Minds. Philadelphia: Temple University Press.

Eddy, T. J., Gallup, G. G., Jr., & Povenilli, D. J. (1993). Attribution of cognitive states to animals: Anthropomorphism in comparative psychology. Journal of Social Issues, 49 , 87-101.

Goodall, J. (1990). Through a Window. London: Weidenfeld & Nicolson.

Griffin, D. (1983). Animal Thinking. Cambridge, MA: Harvard University Press.

Hebb, D. O. (1946). Emotion in man and animal: An analysis of the intuitive processes of recognition. Psychological Review, 53, 88-106.

Heider, F. & Simmel, M. (1944). An experimental study of apparent behavior. Journal of American Psychology, 57, 243-259.

Hogg, M. A. & Vaughan, G. M. (1995). Social Psychology: An Introduction. London: Prentice Hall, Harvester Wheatsheaf.

Kassin, S. M. (1982). Heider and Simmel 1944 revisited: Causal attribution and the animated film technique. Review of Personality and Social Psychology, 3, 145-170.

Kennedy, J. S. (1992). The New Anthropomorphism . Cambridge: Cambridge University Press.

Lippa, R. A. (1994). Introduction to Social Psychology. 2nd ed. Pacific Grove, California: Brooks/Cole Publishing Company.

Mandler, J. M. & Johnson, N. S. (1977). Remembrance of things parsed: Story structure and recall. Cognitive Psychology, 9, 111-151.

Massad, C., Hubbard, M. & Newtson, D. (1979). Selective perception of events. Journal of Experimental Social Psychology, 15, 513-532.

Michotte, A. (1950). The emotions regarded as functional connections. In M. L. Reymert (Ed.), Feelings and emotions; The Moosehearts symposium ( pp. 114-126). New York: McGraw-Hill.

Mitchell, R. W. & Hamm, M. (1997). The interpretation of animal psychology: Anthropomorphism or behavior reading? Behaviour, 134, 173-204.

Mitchell, R. W., Thompson, N. S., & Miles, H. L. (1997). Anthropomorphism, anecdotes, and animals. Albany: State University of New York Press.

Morgan, C. L. (1903). Introduction to comparative psychology (2nd ed. rev., p. 41.) London: Walter Scott.

Newtson, D. (1990). Alternatives to representation or alternative representations: Comments on the ecological approach. Contemporary Social Psychology, 14, 163-174

Rollin, B. E. (1989). The unheeded cry: Animal consciousness, animal pain and science. Oxford: Oxford University Press.

Runeson, S. (1985). Perceiving people through their movements. In B. D. Kirkcaldy (Ed.), Individual differences in movement. Lancaster, UK: MTP Press Ltd.

Runeson, S. & Frykholm, G. (1983). Kinematic specification of dynamics as an informational basis for person-and-action perception: Expectation, gender recognition, and deceptive intention. Journal of Experimental Psychology: General, 112, 585-615.

Rollin, B. E. (1989). The unheeded cry: Animal consciousness, animal pain and science. Oxford: Oxford University Press.

Runeson, S. (1985). Perceiving people through their movements. In B. D. Kirkcaldy (Ed.), Individual differences in movement. Lancaster, UK: MTP Press Ltd.

Runeson, S. & Frykholm, G. (1983). Kinematic specification of dynamics as an informational basis for person-and-action perception: Expectation, gender recognition, and deceptive intention. Journal of Experimental Psychology: General, 112, 585-615.

Sanders, C. R. (1999). Understanding Dogs: Living and working with canine companions. Philadelphia: Temple University Press.

Skinner, B. F. (1974). About behaviorism. London: Jonathan Cape.

Spivey, N. N. (1997). The Constructivist metaphor: Reading, writing, and the making of meaning. San Diego: Academic Press.

Springer, K., Meier, J. A., & Berry, D. S. (1996). Nonverbal bases of social perception: Developmental change in sensitivity to patterns of motion that reveal interpersonal events. Journal of Nonverbal Behavior, 20, 199-211.

Waseda, H. (1966). A phenomenal analysis of social perception. Journal of Child Language, 2, 1-26.

Note

1 Correspondence should be sent to Paul Morris, Department of Psychology, University of Portsmouth, King Henry 1 Street, King Henry Building, Portsmouth PO1 2DY. E-mail: paul.morris@port.ac.uk
 

For Abstracts of all issues, including the most current, click Article Abstracts

To order Society & Animals Journal, go to our secure online ordering page

You can Search the online issues of Society & Animals, as well as the entire Society & Animals Forum (formerly PSYETA) website,
for topics and keywords of your interest: