imitation.pdf

Anim Cogn (2006) 9:335–353

DOI 10.1007/s10071-006-0039-2

REVIEW

Imitation: deﬁnitions, evidence, and mechanisms

Thomas R. Zentall

Received: 16 August 2005 / Revised: 3 July 2006 / Accepted: 5 July 2006 / Published online: 26 September 2006

Abstract Imitation can be deﬁned as the copying of be-

havior. To a biologist, interest in imitation is focused on

its adaptive value for the survival of the organism, but to

a psychologist, the mechanisms responsible for imitation

are the most interesting. For psychologists, the most impor-

tant cases of imitation are those that involve demonstrated

behavior that the imitator cannot see when it performs the

behavior (e.g., scratching one’s head). Such examples of imi-

tation are sometimes referred to as opaque imitation because

they are difﬁcult to account for without positing cognitive

mechanisms, such as perspective taking, that most animals

have not been acknowledged to have. The present review

ﬁrst identiﬁes various forms of social inﬂuence and social

learning that do not qualify as opaque imitation, including

species-typical mechanisms (e.g., mimicry and contagion),

motivational mechanisms (e.g., social facilitation, incentive

motivation, transfer of fear), attentional mechanisms (e.g.,

local enhancement, stimulus enhancement), imprinting, fol-

lowing, observational conditioning, and learning how the

environment works (affordance learning). It then presents

evidence for different forms of opaque imitation in animals,

and identiﬁes characteristics of human imitation that have

been proposed to distinguish it from animal imitation. Fi-

nally, it examines the role played in opaque imitation by

demonstrator reinforcement and observer motivation. Al-

though accounts of imitation have been proposed that vary

in their level of analysis from neural to cognitive, at present

no theory of imitation appears to be adequate to account for

the varied results that have been found.

Keywords Imitation . Social facilitation . Stimulus

enhancement . Affordance learning . Observational

conditioning . Two-action procedure . Bidirectional

control . Control procedure . Sequence imitation

Introduction

This special issue of Animal Cognition (Watanabe and Huber

2006 ) devoted to research on the intelligent behavior of ani-

mals illustrates not only some of the remarkable abilities of

animals but also serves to blur the line between humans and

other animals. The present paper focuses on social learning,

an ability that has received less attention than what its adap-

tive importance deserves. Perhaps one of the most underrated

human learning abilities is learning from the observation of

the behavior of another person. Children appear to be so pre-

disposed to learn from the observation of others that Meltzoff

( 1988 ) has suggested that a more appropriate name for the

human species would be homo imitans , man who imitates.

The extent to which children learn by observation can be seen

in the adage that pits adults’ instructions to a child about how

to behave against their own adult behavior, “Do as I say, not

as I do.”

This contribution is part of the special issue “Animal Logics” (Watanabe

and Huber 2006 ).

Biological approach to social learning

T. R. Zentall ( )

Department of Psychology, University of Kentucky,

Lexington, KY, 40506-0044 USA

e-mail: Zentall@uky.edu

Tel.:

+ 606-257-4076

Biologists, who have long been interested in the adaptive

value of behavior for the survival and reproductive success of

the animal, have suggested that imitation may ﬁll an impor-

tant niche between species-typical, genetically predisposed

behavior and individual (or trial and error) learning (Boyd

Fax:

606-323-1979

Springer

Springer-Verlag 2006

336

Anim Cogn (2006) 9:335–353

and Richerson 1988 ). The advantage of species-typical

behavior is that it is reliable and does not depend on the

vagaries of environmental reinforcement. Imagine how dif-

ﬁcult they would be for a bird to have to learn to build

a nest. It would ﬁrst have to be motivated by the need to

have some place to lay its eggs, it would probably have to

lose many eggs before the appropriate shape and size were

achieved.

However, species-typical behavior may not always be

functional. Changes in the environment may force animals

to learn by experiencing the consequences of their behav-

ior. Hungry animals may have to learn to try new foods if

their familiar foods are no longer available because of cli-

mate change or increased competition from other animals.

But not all new foods are edible and some of them actually

may be poisonous. Some animals have developed ways to re-

ducing the potentially negative consequences of eating novel

foods by testing small amounts that may not be sufﬁcient to

be fatal and by the very rapid association (one trial) of a

novel taste with feeling ill (Garcia and Kolling 1966 ). Thus,

individual learning has disadvantages as well.

Social learning may provide a functional compromise.

Learning by observing others may provide more ﬂexibility

than is possible with species-typical behavior, but it may also

avoid many negative consequences of making errors that

often accompany individual learning (Boyd and Richerson

1988 ). In most cases, the behavior of others has already been

shaped by consequences; therefore, doing what others do

generally leads to favorable outcomes.

side may be sufﬁcient to elicit door opening by the observer

(Klein and Zentall 2003 ).

Social learning of this kind can be thought of as trans-

parent (Heyes and Ray 2000 ) in the sense that the observer

can see the similarity of the effect on the environment of

the model’s behavior and of its own behavior. More theoret-

ically interesting is the ability to imitate opaque or invisible

behavior; that is, a behavior that does not involve manipu-

lation of objects in the environment and furthermore, that

cannot be seen directly by the observer when the observer

is performing it. Consider a child imitating an adult who is

scratching his own head. How does the child know when she

has imitated appropriately?

According to Piaget ( 1962 ), imitation can be a product of

the sensory–motor stage of development. During this period,

children learn to coordinate visual inputs with motor outputs

such that they can efﬁciently reach for objects that they can

see. One can perhaps account for cases of transparent im-

itation (e.g., imitating a model scratching his arm) by way

of stimulus and response generalization (the observer would

engage in motor behavior until her response matches that of

the demonstrator), but opaque imitation requires additional

processes. To invoke such a process to account for the imita-

tion of head scratching, one must posit that one can visualize

the appearance of one’s own head and it is not clear how one

would show that such visualization can take place.

Alternatively, one could account for opaque imitation by

positing the ability to take the perspective of a third person.

To accomplish this, in a sense, the child must ask herself,

what I should do such that a third person would say that

my behavior matches the behavior of the model. In many

cultures, one could perhaps learn such correspondence from

extended exposure to mirrors because mirrors would allow

the individual to learn the correlation between proprioceptive

sensations and visual appearance, but the ability to recognize

oneself in a mirror appears to be unique to humans and other

great apes (Gallup 1970 ). Thus, if perspective taking were

necessary for opaque imitation, one should not be able to

ﬁnd evidence for opaque imitation in other animals.

Although there is some evidence from imitation research

with 3-year-old children (Gleissner et al. 2000 ) that they

make a similar number of errors when they cannot visually

monitor their own response (e.g., ear-lob tug) than when they

can (e.g., knee touch), the absence of an effect of invisible-

versus opaque-response in these children may reﬂect either

their experience with imitation games or their higher cog-

nitive functioning rather than an inherent absence of dif-

ference between these two classes of behavior. It would be

of interest to know whether younger or less experienced

children would show the same ability to imitate opaque

behavior.

Several reviews of social learning in animals have

attempted to categorize the different classes of social

Psychological approach to social learning

While biologists have tended to focus on the adaptive value

of social learning, psychologists have been more interested

in the mechanisms by which organisms learn from others.

Rather than asking why animals learn from observing others,

psychologists tend to be concerned with how they learn from

others. If consequences are not experienced by the observer

prior to performance, how do the observers know what to

do? 1

For certain forms of social learning (e.g., the repetition of

a bird’s song), one can propose a process similar to stimulus

matching (Zentall and Hogan 1976 ) in which the listener

attempts to match the song that it hears. A similar process

may take place when the model’s behavior has a clear effect

on the environment (such as pushing open a door to obtain

food). In this case, the social aspects of the observation may

be incidental. Seeing the door open with food on the other

1 There is a growing trend toward an integration of the biological

and psychological approaches (see, e.g., Bjorklund and Bering 2003;

Caldwell and Whiten 2002 ; Hare and Tomasello 2005 ).

Springer

Anim Cogn (2006) 9:335–353

337

learning and distinguish them from imitation (Whiten and

Ham 1992 ; Heyes 1994 ; Whiten et al. 2004 ). Imitation,

like many processes for which cognitive mechanisms are

implied, is typically deﬁned very generally and then qual-

iﬁed by specifying processes that are to be excluded. The

view that imitation, especially imitation that would be con-

sidered opaque, requires mechanisms that are as cognitive

as perspective taking means that if one studies imitation

in animals, one must develop designs that control for the

possibility that other, perhaps simpler, mechanisms are in-

volved. For example, animals may be predisposed to engage

in certain behaviors (e.g., eating) when others are seen en-

gaging in those behaviors (species-typical behaviors). Alter-

natively, being in the presence of conspeciﬁcs (other an-

imals of the same species) may result in an increase in

general arousal, which may make certain behaviors more

probable (motivational effects). Also, the behavior of oth-

ers may draw attention to a place or object independently

of the behavior itself, and that attention may facilitate learn-

ing (perceptual enhancement effects). Finally, observers may

learn the relation between the effect of the observed behav-

ior on the environment, and that may facilitate performance

by the observer (observational conditioning or affordance

learning).

To better deﬁne imitation, I will ﬁrst describe these alter-

native processes and explain why they must be experimen-

tally separated from imitation if one wants to claim that a

particular species is capable of imitation. I will then describe

several designs that have been used to assess imitation. I will

also examine several variables that have been thought to be

important in imitation by humans to determine if they have

a similar effect on animals. Finally, I will present several

hypothesis that have been proposed to account for imitation

and show that they may not be adequate to account for the

effects found.

A special case of mimicry-involving behavior is the

broken-wing display of certain ground-nesting birds, such

as the killdeer or the avocet (Sordahl 1981 ). When the fe-

male bird is near the nest and a predator approaches, the

bird ﬂies away from the nest while mimicking the erratic

ﬂight pattern that might be shown by a bird with a bro-

ken wing. Although one can speculate about the origins

of this behavior, it appears to be genetically based (i.e.,

the predator serves as a releaser ) and it does not require

learning.

Contagion

When two or more animals engage in similar behavior and

that behavior is species-typical, the coordinated behavior

is often attributed to contagion (Thorpe 1963 ; also called

mimesis, Armstrong 1951 , or response facilitation, Byrne

1994 ). Contagion can be used to describe certain courtship

displays when they involve coordinated movements between

the male and female that can sometimes appear to be virtual

mirror images (Tinbergen 1960 ). Also, antipredatory behav-

ior can be considered contagious when it involves the coor-

dinated movement of a group of animals for defensive pur-

poses. Such behavior occurs in certain mammalian species

(e.g., herding) and avian species (e.g., ﬂocking). When this

coordinated behavior is aggressive (i.e., is directed toward

rather than away from danger), it is known as mobbing

(Hoogland and Sherman 1976 ). Contagion can also be shown

in an appetitive context. For example, a satiated animal in

the presence of food will often resume eating upon the in-

troduction of a hungry animal that begins eating (Tolman

1964 ). In the case of contagion, the behavior of one animal

appears to serve as a releaser for the unlearned behavior of

others (Thorpe 1963 ).

Motivational factors

Species-typical factors

Mimicry

The typical procedure for assessing social learning is to as-

sess the probability that an observer will perform a demon-

strated behavior following observation. But performance of

a behavior may occur spontaneously and if it does, the con-

sequences of the behavior will inﬂuence the likelihood that

the behavior will be performed again. Thus, it is important to

compare the probability that the behavior will occur relative

to an appropriate control. As we will see, determination of

the appropriate control is an evolving process determined by

what alternative processes one thinks might facilitate perfor-

mance of the behavior. For example, the most obvious con-

trol would be a group of animals that acquires the response

on its own (i.e., individual-learning control); however, this is

clearly not sufﬁcient. It is possible, for example, that the mere

presence of another animal may inﬂuence the probability of

certain behaviors.

Perhaps the simplest case of copying is when one species

physically appears like another. This form of copying is

often referred to as mimicry. When a relatively defenseless

animal takes on the appearance of or acts like an animal that

has better defenses, it is known as Batesian (or Mertensian )

mimicry. A well-known case of Batesian mimicry is that of

the palatable viceroy moth mimicking the appearance of the

unpalatable monarch butterﬂy (Turner 1984 ). Such mimicry

results from natural selection of the increased ﬁtness

incurred by those moths that cannot easily be discriminated

by predators from the bad-tasting butterﬂy, and it is an

example of convergent evolution that does not involve

behavior.

Springer

338

Anim Cogn (2006) 9:335–353

Social facilitation

Transfer of fear

It has been hypothesized that the mere presence of a conspe-

ciﬁc can affect the motivational level (or level of arousal) of

an animal (Zajonc 1965 ). The effect of such an increase in

arousal, which Zajonc called social facilitation, may depend

on the context in which the animal ﬁnds itself. Consider an

experiment involving the acquisition of bar pressing by a

rat. Increased arousal produced by the mere presence of a

conspeciﬁc may increase the rat’s general level of activity,

and an increase in general activity is likely to bring the rat

into contact with the bar that is to be pressed (Gardner and

Engle 1971 ).

Alternatively, an isolated animal in a novel environment

may be fearful, and fear in an enclosed environment may

reduce exploratory behavior. If the presence of a conspeciﬁc

reduces fear and increases exploratory behavior, it may lead

to a higher probability (by chance) that the target behav-

ior will be performed (Davitz and Mason 1955 ;Morrison

and Hill 1967 ). Thus, experiments concerned with imitation

must include a control for the possibility that the presence of

another animal might result in an increase (or decrease) in

motivation that could lead to facilitated performance of the

target behavior (Zentall and Levine 1972 ; Levine and Zentall

1974 ).

One further potential source of demonstrator-provided

motivation should be mentioned. Although the mere pres-

ence of a conspeciﬁc may contribute to the motivational

state of an observer, the general (nonspeciﬁc) activity of the

demonstrator may make an additional contribution. Being in

the presence of an active conspeciﬁc (e.g., one that is work-

ing for food but is not responding in a way that is relevant to

the target response) might generate even more activity in an

observer.

Observation of a novel response being acquired (or being

performed) by a demonstrator that is motivated by the avoid-

ance of painful stimulation (e.g., electric shock) may be a

particularly good procedure to use in searching for evidence

of social learning because of the great evolutionary value

that such social learning should have (i.e., in the natural

world aversive events are often fatal). However, if one is

interested in imitation, the observation of a demonstrator in

distress presents the need for a special kind of control. Emo-

tional cues provided by a conspeciﬁc either escaping from or

avoiding shock may provide emotional cues of pain or fear of

pain that could instill fear in an observer. For example, John

et al. ( 1968 ) found that cats that had observed a demonstra-

tor being trained to jump over a hurdle to avoid foot shock

learned the hurdle-jumping response faster than controls that

did not observe the demonstrators. It may be, however, that

being in the presence of a cat being shocked was sufﬁcient

to increase the observers’ fear (motivation) associated with

the conditioning context, and the difference in initial fear ex-

perienced by cats that observed the demonstrator and those

that did not observe the fearful demonstrator may have been

sufﬁcient to facilitate acquisition.

To make interpretation of the results of experiments in-

volving observation of aversive conditioning even more dif-

ﬁcult, under different conditions, induced fear of this kind

can actually impair learning by an observer. For example,

Sanavio and Savardi ( 1980 ) found that rats that observed a

trained demonstrator that had acquired a discriminated shut-

tle avoidance response acquired that response faster than rats

that observed a merely present demonstrator; however, rats

merely exposed to the empty shuttle box acquired the shuttle

response fastest. Thus, trying to identify mere presence and

emotional/motivational effects and isolate them from the ef-

fects of social learning may be quite difﬁcult. One approach

is to use well-trained demonstrators to reduce the likeli-

hood that pain-produced cues might be transmitted to the

observers (Del Russo 1975 ); however, it may not be possible

to avoid the effects of demonstrator-provided, fear-produced

cues.

One way to reduce problems associated with differen-

tial motivational cues encountered with observation of aver-

sively motivated conditioning is to include a control group

that is exposed to performing demonstrators but with the

observer’s view of a critical component of the demonstra-

tor’s response blocked. Such a control was included in

an experiment by Bunch and Zentall ( 1980 ) who used a

candle-ﬂame-avoidance task originally developed by Lore

et al. ( 1971 ). Laboratory rats that have had no prior ex-

perience with a ﬂame show a natural curiosity when pre-

sented with a candle ﬂame. They cautiously approach the

ﬂame and withdraw quickly on contact (when the ﬂame

Incentive motivation

Reinforcement provided to the observer during the demon-

stration of a target response may also play a role in the rate

at which the response is acquired (i.e., it may provide incen-

tive motivation – the knowledge that food can be acquired

in this context). Del Russo ( 1971 ) found signiﬁcant facili-

tation of bar pressing by a group of observers that got fed

whenever their bar-pressing demonstrators got fed (relative

to an individual-learning control). This facilitation may have

involved a general increase in arousal on the part of the

reinforced observer or a more speciﬁc association of the ap-

paratus context with reinforcement. In either case, observers

that receive reinforcement when demonstrators make the ap-

propriate response would likely be more active following

observation than nonreinforced comparison groups, and a

more active animal would be more likely to learn on its own

through incidental contact with the bar.

Springer

Anim Cogn (2006) 9:335–353

339

singes their nose or their whiskers). Even so, repeated con-

tacts are made until they learn to avoid contact with the

ﬂame.

Bunch and Zentall ( 1980 ) found that rats learned the

candle-ﬂame-avoidance task faster after having seen a

demonstrator acquire the task, as compared with (1) a group

for which a small barrier was placed in front of the candle

such that the observer’s view of the rat’s contact with the can-

dle was blocked and (2) a social facilitation control group that

made no contacts with the ﬂame. Thus, although a variety of

auditory cues (a potential by-product of the demonstrator’s

pain), olfactory cues (e.g., potentially produced by singed

whiskers, defecation, and urination), and visual cues (e.g.,

seeing the demonstrator approach and then rapidly withdraw

from something directly behind the barrier) associated with

the task should have provided comparable motivational cues

for these control observers, task acquisition was not facili-

tated as much as for observers that could also observe the

demonstrator’s contact with the candle.

What is interesting about this ﬂame-contact-blocked con-

trol group is that it differs from the typical control group,

which would attempt to remove the fear-inducing cues pro-

vided by the demonstrator and ask if there were still fa-

cilitated learning. Instead, this control group should have

provided all of the demonstrator’s fear-inducing cues but

eliminated the important cause of those cues, the contact

with the ﬂame.

Another means of controlling for potential motivational

cues provided by the demonstrator performing a pain- or fear-

motivated task is to expose the observers to demonstrators

performing a discrimination (Kohn 1976 ; Kohn and Dennis

1972 ). In this research, rats that observed a demonstrator per-

forming a relevant shock-avoidance discrimination acquired

that task faster than controls for which the demonstrator’s

discrimination was the reverse of the observer’s (i.e., the cue

that signaled shock for the demonstrator signaled safety for

the observer and vice versa).

Local enhancement

Local enhancement refers to drawing attention to a locale or

place (Roberts 1941 ). This attentional response may lead to

observation that would not otherwise have been as readily

seen (Thorpe 1963 ). For example, Lorenz ( 1935 ) noted that

ducks enclosed in a pen may not pass through a hole, large

enough for them to escape, unless they happen to be near

another duck as it is escaping from the pen. The sight of a

duck passing through the hole in the pen may draw attention

to the hole and allow the observer to notice it.

Local enhancement has also been implicated in the ﬁnding

that puncturing the top of milk bottles by great tits in Great

Britain spread in a systematic way from one neighborhood to

another (Fisher and Hinde 1949 ). Although the technique of

pecking through the top of the bottle may be learned through

observation, it is also likely that attention was drawn to the

bottles by the presence of the feeding birds, and once at

the bottles, the observers found the reward and consumed it.

Then, learning to identify milk bottles as a source of food can

readily generalize to other open bottles, and drinking from

opened bottles can readily generalize to an attempt to drink

from a sealed bottle, which in turn can lead to puncturing of

the top.

As Denny et al. ( 1988 ) have shown, local enhancement

can be studied in its own right. Exposing rats to the movement

and sound of a bar being activated (by the experimenter from

outside the chamber) followed by the presentation of food,

can facilitate the acquisition of the bar-pressing response by

the observers, relative to various control procedures.

Local enhancement may also account for John et al.’s

( 1968 , Experiment 2) ﬁnding of socially facilitated acquisi-

tion of lever pressing by cats. Cats in an experimental group

that observed another cat lever pressing for food, learned

to press that lever faster than cats in a control group that

observed another cat that was fed periodically without lever

pressing. But observation of lever pressing may draw atten-

tion to the lever rather than to the cat pressing the lever. Lo-

cal enhancement is especially likely in this context, in which

observation of the moving lever might encourage lever ap-

proach upon removal of the demonstrator (especially by cats,

a species known for its motivation to explore).

Similarly, local enhancement may play a role in the faster

acquisition of lever pressing by kittens that observed their

mothers as demonstrators, than by kittens that observed an

unfamiliar female demonstrator (Chesler 1969 ), because ori-

entation toward the mother may be more likely than orienta-

tion toward an unfamiliar cat.

Local enhancement may also be involved in John et al.’s

( 1968 , Experiment 1) ﬁnding of facilitated acquisition of an

aversively motivated hurdle-jump response. The distinction

between imitation and local enhancement may be a subtle

one in this case, but observation of the jumping demonstrator

Perceptual factors

When the observation of a demonstrator draws attention to

the location of an action or the object of an action (e.g., a

lever), it may alter the salience of the lever (stimulus en-

hancement) or the place in the environment where the lever

is located (local enhancement). Thus, whenever the behav-

ior being demonstrated is directed toward an object, that

object may attract the observer’s attention independently of

what the demonstrator does to that object (i.e., the behavior

to be imitated). In fact, animals may learn much about the

functional signiﬁcance of objects in nature by having their

attention drawn to those objects when manipulated by con-

speciﬁcs. But learning facilitated by such attention-getting

would not qualify as imitation.

Springer

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: