Modeling Theory

Modeling Theory
Monkey See, Monkey Do

Here’s an interesting study that was done many years ago to make a point.  I’ll set it up, then you guess what happened.

BoboDollA teacher made a special video to show her class of third graders.  The video was shot in a school playroom with lots of toys that 5 year olds really like.  One of the toys was an inflated Bobo doll that stood about as tall as a first grader.  Near Mr. Bobo was a large plastic baseball bat.

What the teacher did is this.  She filmed one little boy in the playroom having fun with Mr. Bobo.  She specifically instructed the little boy to pick up the large plastic bat and to knock the beejeebers out of Mr. Bobo.  And, being a good little boy, the kid whacked Mr. Bobo like it was two out in the bottom of the ninth with the home team trailing by one.  Whack, wham, and bash.

Now, here’s the interesting part of the study.  The teacher took this video and brought it to another first grade class one day.  Just before the children went to their playroom for a little recreation, the teacher played the video for them.  It showed many kids playing the playroom, but it also featured our home run hitter knocking Mr. Bobo into orbit.

Okay, class, the $64,000 question.  What happened when this audience of first grade kids went to the playroom after watching the video?

Of course.  They went hunting for Mr. Bobo and the Louisville slugger.  And when they found them, well, it wasn’t a pretty sight.

This study seems so obvious that one wonders why it was ever done.  Of course those kids observed the videotape, then when they got the chance, they applied what they had seen.  Every parent knows all about Monkey See, Monkey Do.  So what’s the big deal with Modeling Theory?

Three points.  First, it is surprising that people can be influenced so easily.  Just by watching what other people do, we can acquire new ideas and behaviors.  Second, modeling seems to be a dominant way that people get new behaviors.  Whenever we are in a new situation, we almost always look around to see what others are doing.  Third, the whole process requires very little thinking on the part of the observer.  Indeed, modeling is faster if you simply copy the model rather than try to figure out everything that is going on.

Process Of The Theory

Modeling Theory operates in three simple steps.  Here they are in overview.

1.  You observe a model.

2.  You imitate the model’s actions.

3.  You get a consequence.

The marvel of this theory is that people are influenced simply as a result of observing other people (monkey see, monkey do).  From the observation of others, we learn what to do, what not to do, when to do it, and what to expect when we do it.  Very simple, very direct, and very easy.

After we observe the model, we then imitate.  That is, when we get in a similar situation that we had observed earlier, we now produce the same behaviors we saw the model produce.  We observe someone put a plastic card in a machine, press some buttons, then get money.  So, we walk over to the machine, look for a place to put our card, look for some directions about those buttons, press a few, and voila, money.

Now, our imitation should lead to the desired consequence.  We saw the model get the money, right?  If our imitation produces money for us, too, we got the desired consequence and now we have truly been influenced.  (I watch you do it, and when I do it, I get what I want.)  If our imitation fails, then we will drop the model.

An Interesting Historical Footnote

The catch phrase, “Monkey See, Monkey Do” has more than a common sense basis.  Just before the start of World War I (the real old one, 1914-1918), a German graduate student named Wolfgang Kohler was on Canary Island conducting learning experiments with a colony of chimps.  The island was a British possession, so when the War started, Mr. Kohler became a guest of the British government on the island.  (And to hear some graduate students complain nowadays about their research assignments?!?)

Kohler made the best of the internment period and conducted some of the best animal learning research ever done.  Kohler was primarily interested in testing cognitive function in apes, particularly “insight learning.”  While all of that work made Kohler’s research important, some of his tests tangentially demonstrated the monkey, see; monkey, do effect.

See, Kohler arranged an experimental cage with several different objects.  He would hang bananas high in the cage so that they were inaccessible to chimps.  The chimps would holler and jump for the bananas without success, but some of the chimps looked around the cage, saw the various objects, and figured out how to build a scaffold they could climb to reach the bananas.  Kohler had film on this that clearly showed various chimps having that “AH-HA!” experience of insight learning where they stood there stupidly surveying the scene, then “getting it” and putting together the various objects.

Kohler Modeling Insight Apes

Now, what’s insight learning go to do with modeling?  Nothing.  The experiment, however, does have a lot to do with modeling.  Other chimps would observe the chimp in the cage, see the failure, and then see the solution.  When these chimps got in the cage, bang-zoom, they got to the solution a lot faster, arguably due to modeling effects.

Media Modeling

Beavis ButtheadIf you’ve got kids and TV, you have observed Monkey See, Monkey Do at work repeatedly.  Melanie’s nephews grew up with MTV and Beavis and Butthead.  Melanie and I laughed like crazy at the boys’ copycat antics, while the parents Phil, Melanie’s brother, and Cherie, his wife, were less than pleased.  Anytime that show came on, the boys would start modeling the various stupid tricks, vocalizations, and catch phrases.  These kids nowadays, right?

Three StoogesWell, when grandparents Clif and Claire saw the boys acting like Beavis and Butthead, they too were amused, but in a different way.  They recalled watching Melanie and Phil as little kid sister and brother watching the Three Stooges on TV and modeling the various stupid tricks, vocalizations, and catch phrases from that era’s Beavis and Butthead.

Think now about pornography.  Researchers have correlated changes in pornographic content and changes in self reported sexual activity.  Remember “Deep Throat?”  That XXX movie made oral sex cool well before Mr. Clinton and Ms. Lewinsky.  More recently, pornography has featured heterosexual anal sex with more frequency and guess what?  People are reporting more anal sex in their own bedrooms.

Sure, this is obvious, but let’s consider other implications.  A fair amount of good research demonstrates modeling effects with other media messages, like violence and aggression.  Consider the interesting work by David Phillips on the impact of mediated violence on suicide and murder rates.  Phillips employed a simple database analysis where he would correlate the amount of media coverage (number of days on the front page) for a violent event, then run that against a national murder or suicide database to check for variations.  He also ran a wide series of statistical controls to adjust for base rates, time of year, etc.

Phillips correlated national suicide rates with the amount of media coverage of “famous” suicides (Marilyn Monroe; Kurt Cobain).  Phillips found that the longer the media covered the story (measured at the number of days on the front page), national suicide rates increased at a small effect (43/57 windowpane).  Phillips also looked at media coverage of heavyweight boxing matches and then correlated that coverage with murder rates.  Phillips found a moderate effect (34/66 windowpane) increase such that longer media coverage produced higher murder rates.

And, just to round out the effect, consider this:  If antisocial media can produce antisocial modeling, what happens with prosocial media?  Shouldn’t you get the same impact, just in a positive direction?  Give yourself a gold star if this occurred to you because the prosocial effect does hold.  Hundreds of experiments have exposed people (usually children) to a prosocial model (Mister Rogers!) then compared following behavior to people not exposed or exposed to a neutral model.

Modeling Persuasion Play

Modeling can also be applied as a kind of persuasion play that influences the thoughts, feelings, and actions of another person.  Therapists use modeling when they match or mirror the verbal and nonverbal behaviors of a clients in a psychotherapy context.  This modeling can function in several different ways.

Sometimes modeling is a potent form of feedback for the client where the therapist behaviorally demonstrates how the client looks while discussing a particular topic.  Merely seeing how we look in posture or facial expression on another person can be arresting feedback that gives us insight into our own behavior (Geez, that looks just like Dad did when he was drinking!).

Sometimes modeling demonstrates empathy from the therapist.  If you want to feel the way another person feels, try moving your body and face the way they do.  Mirror them gesture for gesture.  Sit in the same position.  Move with them.  You may begin to experience emotions that seem like what the client is reporting.  More importantly, that client will begin to sense that you do feel as they do, because they cue off your behavior to infer your feelings.  If clients “see” that you “move” like they do it looks like you “feel” like they do and you’ve established empathy.

More strategically, modeling can be a control tactic.  Milton Erickson famously employed modeling as a hypnotic induction with difficult clients.  He would mirror the resistant client until he synchronized with them, then he would change the client’s nonverbal actions by making the client begin to mirror Erickson’s behaviors.  The client then began to move when and how Erickson wanted and at the proper moment of control, Erickson would start the hypnotic induction.

Perils and Precautions about This Modeling

I am hesitant mention Dr. Erickson here because his work has been employed by people motivated by profit particularly in some sales and personal growth groups.  You’ve probably seen web come-ons from these groups . . . a lot of percentages and exclamation points.  Erickson was legit as a scholar, researcher, and practitioner whose work is still cited and studied many years after his death.  The sales and growth gurus have never published anything remotely scientific and indeed some of their claims have been scientifically studied and found wanting.  You need to know that some folks use words like “mirroring,” or “matching,” and tactics based on nonverbal behavior.  Talking the talk is not walking the Erickson walk, however.

You don’t need to conduct your own randomized controlled experiment on these claims.  The therapeutic setting is wildly different from any sales situation.  Clients make appointments with a health care professional in an office.  The client pays the professional who provides services.  The client has a strong motivation to follow the professional.  In a sales setting the psychology is obviously different.  The client is now a customer and the professional is selling something.  It’s in a marketplace setting with other customers milling around.  Trying to use modeling as I’ve just described while sitting in a department store or district manager’s office is transparently silly.  Furthermore, a therapy relationship implies cooperation and openness while a sales relationship implies competition and advantage-seeking.  So, consider yourself warned.  Don’t buy what those gypsy girls sell you.

However, therapeutic modeling does work.  If you are particularly unscrupulous, you can try model-mirroring as a pick up tactic in bars.  Once you establish a conversation, begin nonverbal matching.  Move in sequence with your partner.  Get the flow and follow it.  Then when you are in sync, begin to change your nonverbals to a more forward, intimate, or sexual style.  For example, if your partner has a particular way of gesturing, mirror that gesture until you are in sync, then one time add to it an ending where you finish the gesture with a light touch on your partner’s hair or face, then go back to mirroring.  Continue the process of extending your partner’s moves with an addition that increases familiarity.  You’ll probably find your partner adding these “new finishes” and when you do, you’re in.

Caveats, Warnings, and Admonitions!  If you try this, you are alone on the perimeter and you will deserve what you get.  Persuasion can manipulate intimacy quite easily, but you get what you play for.  Deceptions like this produce explosive discoveries and surprising reversals.  Do you want to get caught playing with another like this?  Do you want to risk getting trapped by someone who models better than you?

What Kind of Difference Does Modeling Make?

Give somebody a behavioral task whether simple like an arm movement or complex like a series of assembly actions to put something together and provide two conditions.  The first is the control condition or Just Do It.  You tell the participants what you want and let ‘em rip.  The second is the treatment condition or Modeling.  You provide the task, plus a live or mediated example of a model doing the task.  Let everyone do some practice at the learning, then run a trial to criterion or This One Counts.

A meta-analysis of these kind of studies provides two interesting pieces of knowledge.  First, people who get the model do better, but only a small effect (45/55).  Second, people who get the model are a lot better (large effect or nearly 25/75) at the behavioral “fluency” or the smoothness and coordination of the behavior.  Thus, modeling helps people be more accurate or correct or error-free (small effect, though) and modeling helps people perform a lot more smoothly.

The effect sizes are similar when looking at media modeling effects.  Susan Hearold’s meta analysis of both antisocial and prosocial messages found a moderate size (35/65 windowpane) with more media, more behavior whether pro or antisocial.  (Paik and Comstock conducted a more recent, 1994, meta analysis and replicated the finding.)  Modeling, whether through direct observation of real models or indirect through observation of mediated models, produces predictable changes in behavior.

Real World Applications

Among the many uses of modeling, I want you to consider four practical implications.

1.  You have to know what is being modeled.

Do you remember Mrs. Reinforcer and her student, Bad Bill?  Bad Bill broke a Rule and Mrs. Reinforcer used punishment to influence Bill’s behavior.  (Except Bad Bill really wanted the punishment to escape the classroom and so he kept doing the bad thing, which confused Mrs. Reinforcer.)  Something else was also going on in Mrs. Reinforcer’s classroom.  Every other kid was watching the event and because of the principles of modeling, every kid was being influenced.  Each one of them learned, simply through observation, several important lessons.

Many students learned that bad kids do get punished.  That’s good.  When you enforce a Rule as a teacher, everybody in the room, not just the target, is influenced because of modeling.  But bad things are learned, too.  Some of the kids learned that if they act like Bad Bill they can escape Mrs. Reinforcer’s room.  Others learned (by seeing what happened before Bill got thrown out) all the things they can do and still not get in trouble.  Finally, some learned how to pull Mrs. Reinforcer’s chain.

The point of this example is direct.  When things happen, people may be modeling.  And we aren’t just talking about little kids in elementary school either.

2.  Use modeling to change behavior.

Modeling Theory is designed primarily to explain behavioral influence.  It is less useful in creating or understanding changes in thinking or feeling.  Therefore, whenever you want to influence student behaviors, consider modeling.  For other types of changes, use other persuasion tools.  However, note that modeling can lead to the creation of new attitudes and beliefs that are consistent with the behavior.  For example, a child observes racist behavior from a parent and models it.  Even though the parent may never express any belief or attitudes about the prejudice, the child will acquire their own behavior-consistent cognitions to support the modeled behavior.

3.  Show modeling. (Don’t tell.)

As noted at point 2, Modeling Theory works well at influencing behavior.  The best way to implement modeling is to do it rather than to say it.

4.  The Say-Do Gap.

People also use Modeling Theory to help make Attributions about you.  Observers may look to you for the “correct” behavior from you to guide their own actions.  If, however, they note a “say-do” gap where your actions and your words are inconsistent, the modeling process will certainly breakdown and may lead to other persuasion problems for you.  This is a classic problem for actors in leadership, management, or supervisory positions.  Not only should you use Modeling Theory to influence subordinate behavior, but to demonstrate consistency between what you say (key values, policies, regulations, goals, etc.) and what you do (you model “exceptions” or “Mulligans” or just plain inconsistencies).


We are changed and we can change others merely through observation, imitation, and consequence.  We can nonverbally persuade observers to perform desired behaviors, but we need to insure not only imitation, but also insure that the right consequences follow.

And, we’ll revisit Monkey See, Monkey Do in the CLARCCS section on Cues.

References And Recommended Readings

Bandura, A. (1962). Social learning through imitation. In M. Jones (Ed.), Nebraska symposium on motivation (pp. 211-269). Lincoln, NE: University of Nebraska Press.

Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice-Hall.

Ashford, D., Bennett, S., & Davids, K.  (2006).  Observational Modeling Effects for Movement Dynamics and Movement Outcome Measures Across Differing Task Constraints: A Meta-Analysis. Journal of Motor Behavior. Vol 38, 185-205.

Kohler, W. (1925). The mentality of apes.  New York, NY: Harcourt Brace.

Haley, J. (1986).  Uncommon Therapy: The Psychiatric Techniques of Milton H. Erickson, M.D.  New York:  W.W. Norton.

Hearold, S. (1986).  A synthesis of 1043 effects of television on social behavior.  In G. Comstock (Ed.), Public Communication and Behavior, vol. 1, (pp. 66-135).  Orlando, FL:  Academic Press.

Paik, H., & Comstock, G. (1994). The effects of television violence on antisocial behavior: A meta-analysis. Communication Research. Vol 21, 516-546.

Phillips, D. (1986). The found experiment:  A new technique for assessing the impact of mass media violence on real-world aggressive behavior.  In G. Comstock (Ed.), Public Communication and Behavior, vol. 1, (pp. 260-308).  Orlando, FL:  Academic Press.