Review - Perspectives on Imitation
From Neuroscience to Social Science
by Susan Hurley and Nick Chater (Editors)
MIT Press, 2005
Review by Shaun Gallagher, Ph.D.
Jul 4th 2006 (Volume 10, Issue 27)

Imitation is a hot topic in developmental psychology, cognitive neuroscience, animal studies, philosophical considerations of theory of mind and intersubjectivity, as well as educational studies.  Hurley and Chater have collected a large number of first-rate essays and commentaries, originating in a conference at Royaumont Abbey, and focused on a variety of themes, gathered together under two headings: Mechanisms of Imitation and Imitation in Animals (Vol 1) and Imitation, Human Development, and Culture (Vol 2).  The two volumes are composed of 29 papers and 38 commentaries.  The Introduction by Hurley and Chater is printed in both volumes.

Rather than try to rehearse all of the papers, or touch on all of the topics covered in these volumes (something that Hurley and Chater do nicely in the Introduction, and something I won't try to imitate), I will be selective and discuss three issues that are raised in this collection.  First, can animals imitate?  Second, how does imitation facilitate social interaction in humans?  And third, how is imitation related to moral development?

Can animals imitate?

As you may expect, this depends on how one defines imitation.  There is a diversity of definitions offered in these volumes.  Rizzolatti, for example, follows Thorndike and defines imitation as "learning to do an act from seeing it done" (I, 55).  Meltzoff offers a more complex characterization.  Imitation must meet three conditions: "(1) the observer produces behavior similar to the model, (2) the perception of an act causes the observer's response, and (3) the equivalence between the acts of self and other plays a role in generating a response" (II, 55).  One person's concept of imitation, however, is another person's concept of perceptual priming, simulation, mirroring, contagion or emulation.  Thus, Jesse Prinz defines imitation as "a process by which one organism comes to exhibit a state or behavior exhibited by another organism through perceiving the other organism exhibit that state or behavior" (II, 276).  But he immediately adds that it requires mentally mediated replication.  This is still, as he indicates, a broad definition that includes crying contagion, although it is not clear why one should consider crying contagion a form of mentally mediated replication.  Others, like Tomasello (cited by Prinz), offer a narrow definition, in which imitation requires a duplication of both the means and the end of an action.

On a wide definition it seems clear that animals imitate; on a narrow definition, perhaps they can't.  Thomas Zentall nicely captures this thought in his commentary on Anisfeld.  Zentall cites his study of Japanese quail who, after observing another quail step on a treadle or peck at a treadle, will respectively step or peck when given access to the treadle.  Stepping and pecking turn out to accomplish the same result -- a specific movement of the treadle.  Quail are also capable of deferred imitation: imitating the behavior after a certain amount of time.   Zentall focuses on the "two-action method" of showing imitation.   Using this methodology, two groups of individuals watch a model behavior that accomplishes the same task (moving the treadle), but each model uses a different action (stepping vs. pecking).  If each group of individuals tends to match the behavior that is demonstrated, this is considered to be imitation.  On this operational definition, as Richard Byrne points out, "a growing list of species are now claimed to show imitation" (I, 226). The list includes Japanese quails, budgerigars, and human neonates.  Byrne points out important complications involved in some of the experiments on learning by imitation conducted with animals.  In some cases, there are no attempts to find out about preexisting repertoire.  In a study of four chimpanzees and six gorillas (conducted by Stoinski and colleagues), the chimps were likely acculturated in a human environment, whereas the gorillas were zoo animals.  The chimps learned a sequence of actions whereas the gorillas did not.  Is there a firm conclusion to be drawn from this?  Other studies have shown that great apes who are brought up by human caregivers can acquire "human" behaviors (I, 227).  Sorting out imitation from perceptual priming, efficiency of behavior, social facilitation, stimulus enhancement, or emulation is difficult, however.

Irene Pepperberg offers what she considers to be a clear case of auditory imitation that cannot be confounded with priming, social facilitation, stimulus enhancement, or anything else.  Namely, the replication of human speech in the African Grey Parrot.  How precise is their imitation?  Pepperberg offers a detailed phonological account to show that Alex the parrot shows good fidelity of vowel and stop imitation, limited only by differences in mechanisms of vocal tract, supporting her contention that Alex physically imitates her speech.  The fact that Alex's vocalizations are not merely phonetic reproductions, but are also referential is important to distinguish "mere" mimicry from true imitation.

"If an act is performed because the imitator understands its purpose -- to reach a goal, be it an object or intentional communication, that is otherwise impossible to obtain -- then the imitation is intentional and complex, most likely indicating cognitive processing" (I, 248).

Pepperberg offers a helpful discussion of the neurological correlates of imitation (mirror neurons), distinguishing between simple mimicry (relatively meaningless copying), low-level imitation (involving some social interaction), and high-level imitation (involving reference to goals or creation of improbable acts).

Throughout the discussions of animal imitation reference is constantly made to human neonate imitation, as demonstrated in experiments by Meltzoff and Moore.  This raises questions about the importance of the link between cultural settings and imitation, and concepts of animal culture.  Whiten et al. are thus motivated to point out that "even though a substantial cultural repertoire may be acquired by imitative copying, neither children nor chimpanzees copy all they see others around them doing" (I, 264), and thus the question arises: What determines what is imitated or not imitated?   They note that in studies of chimp imitation only parts of what is modeled gets copied, but that this is less the case for infant imitation.  In both cases, however, there is selectivity, and this suggests that there is something that we might call "smart" imitation.  For example, a subject might smartly ignore irrelevant details of particular movements and imitate only those aspects that are goal related; or a subject may be able to represent, for meaning, specific aspects of meaningful movement in alternative (non-motor) fashion, and therefore not need to imitate that aspect.  This kind of "adaptive flexibility" is explored experimentally by Whiten et al. as a way to provide substance to the distinction between emulation (which has a high degree of selectivity) and imitation (which is less selective).  The interesting question that they pose is whether animals who partially imitate are poor imitators, or are actually smartly emulating behavior.  Their experiments suggest that both animals and children engage in degrees of emulation, and that in some contexts chimps engage in a higher degree of emulation than 3-year old humans.  They conclude that apes "appraise the 'meaning' of components of an act they see associated with desirable outcomes" (I, 279), and thus imitate intelligently.

Intersubjectivity

Is the problem of intersubjectivity equivalent to the problem of "other minds," in regard to which we must make inferences because other minds are otherwise not accessible?  Or is it equivalent to the problem of understanding the perceived embodied actions of others?  Many of the authors writing in Perspectives on Imitation start from embodied actions and work their way towards the minds of others.  Pursuing this strategy, the concept of simulation constitutes not only an important tool (in the context of experiments, for example, where subjects may be asked "to mentally simulate an action" [Decety and Chaminade, I, 127]), but also the solution to the puzzle (as one finds explicated in simulation theory [ST] approaches to theory of mind).  This is a frequent strategy found in neuroscientific accounts of understanding and empathizing with others.  Decety and Chaminade, for example, introduce the notion of explicit simulation as found in ST (as in Goldman, who compares explicit simulation and imitation, II, 92; see Hurley and Chater II, 26-27), and describe how this concept comes to be used in the social neuroscience of mirror neurons, shared (neural) representations, and perception-action common coding mechanisms.  Vittorio Gallese equates these processes with "automatic, implicit, and nonreflexive simulation mechanisms" and, following Adolphs, calls them "as if body loops" (I, 117).  Robert Gordon equates this "constitutive mirroring" with simulation (II, 100ff; see Hurley I, 189).  Authors of two other chapters in Vol 1, Marco Iacoboni and Giacomo Rizzolatti respectively, have elsewhere employed the same terminology.

"Mirror neurons allow us to grasp the minds of others not through conceptual reasoning but by direct simulation.  By feeling, not by thinking" (Rizzolatti, quoted by Blakeslee 2006).

"When you see me perform an action - such as picking up a baseball - you automatically simulate the action in your own brain" (Icoboni, quoted in Blakeslee 2006). 

There are two questions to ask in regard to equating sub-personal neural processes with simulation.  Both questions relate to the very concept of simulation as it is developed in ST, that is, in discussions that take ST to be a solution to the problem of intersubjectivity or mind-reading.  Simulation, as it is developed in those discussions, is consistently characterized as having two aspects.  (1) The instrumental aspect: the simulation is a model that we purposively use to understand something that we do not understand, namely, the other person's mind.  And (2) the pretense aspect: the simulation involves off-line pretend states, and has the status of the subjunctive "as if."  I pretend to believe or act "as if" I were you.  These two aspects are obvious enough to anyone familiar with this literature, and I won't try to make the case here (but see Gallagher, in press). 

The two questions should now be obvious, however.  First, in what sense can we regard sub-personal automatic neuronal processes to be something that I (or the brain) use(s) instrumentally as a model of something else?  At the very least we can note, as Gordon does (II, 104-106), that "the mirroring phenomena ... are not "my own" in the requisite sense.  If I am aware of them at all, I am aware of them as underlying the other's behavior, not my own" (105).  Indeed, we should make this a stronger claim.  We do not activate or control activation of the neuronal processes, nor does the brain, in any proactive way; rather, in instances relevant to intersubjectivity, the actions of other people elicit that neuronal activity.  There is no instrumental control or use of these processes by the subject, or even the subject's brain.  A claim such as the following, for example, can be made only by reading personal level vocabulary into subpersonal processes:

"Using our own motor capacities to understand the actions performed by others is at the core of the simulation theory. ... the neural motor system involved in the preparation and execution of action, is also part of a simulation network which is used to interpret the perceived actions performed by others" (Chaminade, Meary, Orliaguet, Decety 2001; emphasis added). 

The second question is this: In what sense can we regard subpersonal automatic neuronal processes to be "as-if" processes?  Is there something like neuronal pretense or a neuronal subjunctive?  Notwithstanding Gordon's suggestion that neurons can respond "as if I were carrying out the behavior" (II, 96), there really is not an "as if it were I" or an "as if I were you" at the neuronal level.  As vehicles, neurons simply fire; they don't pretend to fire.  And in terms of what they "represent", there is now general consensus that mirror neurons and shared representations are neutral, that is, they represent neither first-person (my action) nor third-person (your action), but simply action (for which a "who" is not yet determined; see, e.g., Gallese I, 110-111).  If there is no "I" or "you" represented, it would be difficult to claim that there is a representation of "as if I were you". 

In both of these regards, then, subpersonal neuronal processes of the mirroring kind fail to meet the requirements for what makes a simulation a simulation as specified by ST.

How is imitation related to moral development?

The relation between imitation and moral development has been an important topic of philosophical discussion, at least since the time of Plato.  What is fascinating in these two volumes of essays is the recent empirical research and the various contexts in which this relationship comes to be explicated.  If we think of moral contexts as involving action and relations to others, then imitation seems central.  Consider the variety of relevant contexts.  Imitation has been shown to be positively related to whether two people like each other, with empirical evidence that shows that when others imitate us, our liking of them increases (Dijksterhuis II, 210).  Likewise, when we desire to affiliate with another person, our imitation of them increases.  Dijksterhuis cites a study by van Baaren et al (2003) that shows the positive effects of verbal imitation on the amount of tips received by waiters in a restaurant.  Research also shows the link between imitation and non-conscious trait inferences and stereotyping.  The famous experiment by Bargh et al. (1996) involved priming stereotypical "old people" vocabulary (e.g., gray, bingo, Florida), resulting in a slower pace of walking in the primed subjects.  Dijksterhuis and van Knippenberg (1998) showed that after subjects thought about typical behaviors and attributes of college professors (or in contrast soccer hooligans), the subjects were able to do better (or worse if they thought of soccer hooligans) on general knowledge games.  As a professor in Florida, these studies have some practical interest for me.  I now understand why, when I meet people, they seem very smart, but walk very slow.

Other topics related to questions of moral context concern deceptive imitation (Gambetta II, 221ff), the importance of emotion in the role played by imitation in moral development (J. Prinz II, 267ff), and the effect of violence portrayed in the mass media.  In regard to the latter, although it is very difficult to measure or show causality, Eldridge draws a reasonable conclusion based on existing evidence: "childhood exposure to violence in the media has lasting effects on behavior through a high-level process of imitation in which cognitions that control aggressive behavior are acquired" (II, 264). Several other essays (Donald, Sugden, Gil-White, Greenberg, Chater, and many other commentators on these essays) tell the evolutionary story of how we move from mimesis and imitation to adopting the values of our tribe, and thence to the effects of broader cultural memes for shaping our rationality and our cultural practices.

These various discussions all point to the importance of imitation in moral development, a view that would certainly support the Aristotelian conception of practical wisdom.  For Aristotle, to become a good person one must mix with good people and do what they do.  That is, one must imitate and practice good actions while understanding and appreciating the worth of the ends accomplished by these actions.  Aristotle also pointed out that despite our general sense that this is the way to attain virtue, there is no science of ethics.  In part, this means that there are no rules that will necessarily lead us to the good life.   This also seems to be confirmed in the various empirical studies of imitation.  There are no guaranteed, clear, or strong predictions that we can make about any individual concerning whether they will imitate what they see, or whether imitation, if it occurs, will shape their behavior in the right way. 

References

Blakeslee, S.  2006. Cells that read minds. New York Times (online) January 10, 2006.

Chaminade, T., Meary, D., Orliaguet, JP, and Decety, J. 2001. Is visual anticipation a motor simulation? A PET study. NeuroReport. 12: 3669-3674

Gallagher, S. (in press). Logical and phenomenological arguments against simulation theory.  In D. Hutto and M. Ratcliffe (eds.), Minding our Practice: Folk Psychology Re-assessed.  Springer Publishers.

© 2006 Shaun Gallagher

Shaun Gallagher, Ph.D., Professor of Philosophy and Cognitive Science at The University of Central Florida.