,.. Journal of Personality and Social Psychology 1984,Vol. 46, No.2, 269-279

Copyright 1984 by the sychol no; cal A . 11. I

American P

-".

ssocla on, nc.

""

~, ,

i

The Emergence of Action Daniel M. Wegner

Robin R. Vallacher

Trinity University

minois Institute of Technology

Gary Macomber, Russell Wood, and Kevin Arps Trinity University Peoplesometimesfind themselvesdoing things that they did not set out to do. The theory of action identification suggests that peoplewill make suchdiscoveriesunder certain circumstancesand then will continue to perfonn the action as newly understood;new action, then, will be the result. This action emergencephenomenon wasinvestigatedin two experiments.Each wasdesignedto test the idea that people would embracea new understandingof action-an emergentact identity-to the degree that this identity provided a more comprehensiveunderstandingof the actionthan did a previousact identity. In Experiment 1,somesubjectswereinduced to think aboutthe details of the act of "going to college" (e.g.,"studying"), whereas otherswere led to focus on more comprehensivemeanings(e.g., "preparing for a career").Thosewho concentratedon detailswere more susceptibleto an emergent understandingof the act. They came to agree with an article that suggestedthat "going to college" results in "improving one's sex life" or "impairing one's sex life." Experiment 2 revealed that emergentidentification can be translated into emergentaction. Subjectsin this studywho wereinduced to think about the details of "drinking coffee"-by drinking their coffee in unwieldy cups-were more susceptiblethan thosewho drank from normal cupsto a suggestedactionidentification. They came to believe that "drinking coffee" amounts to "making myself seek stimulation" or to "making myself avoid stimulation," and subsequentlyfollowed the suggestedaction identification by turning up or down the volume of music they were hearing.

it'

Do people know what they are doing? Traditionally, the thinking of psychological theorists on this question has been strongly divided. Some have been impressed with the human capacity to make plans, to verbalize intentions, and to consider possible courses of action, and so have argued that people know what they are doing in advance of each action. This was the way James (1890) framed the role of mind in action, and such ideas have been echoed in many theories since (e.g., Harre & Secord, 1973; Luria, 1961; Miller, Galanter, & Pribram, 1960). Other theorists, however,

have been more taken with the human tendency to err, to stumble blindly into courses of action, and to imagine in retrospect what the action might be. Freud (1914/1960) saw knowledge of action largely as a matter of such "self-discovery," and theorists since have argued that people can only begin to understand an action once it is complete (e.g., Bem, 1972; Mead, 1938; Ryle, 1949). In this article, we hope to show how both of these seemingly contradictory views can be correct. For this purpose, we draw on the theory of action identification (Vallacher & Wegner, in press), a theory that suggeststhat people always have available some conception of what This h . b N onalSci they are doing. The theory indicates that in researc was supported In part Y an ence ... Foundation GrantBNS78-26380. Wewishto thankKim some cases, an unders~ding of actIon arIses Allio andGeorgeKierstedfor helpin conducting the re- before the act and contInues to be relevant searchand ToniGiuliano,William B. Swann,Jr., and during and after the action. In other cases, RobertA. Wicklundfor valuablecomments on anearlier however the conception that was held prior D th '. h Th draft ofests thisfiarticle.. ts h ld be t t Dani I M Wi to e actIon may c ange. may di th .d .fte person . or repnn s ou sen 0 University, e" egner,equ Department of Psychology, Trinity 715 ~ove~ at some new i enti catIon 0 f th e "

Stadium Drive, San Antonio, Texas78284.

actIon is more apt and then may set out to 269

",~:

. 270

WEGNER, VALLACHER, MACOMBER, WOOD, ARPS

continue the action as newly conceptualized. In this analysis,knowledgeof what oneis doing may thus contribute to the maintenanceof old action or to the emergenceof new action. The interesting issue for this theory, then, is the determination of the conditions under which each of these processeswill occur: When is Jamesright, and when is Freud right? To provide a perspective on this issue, we concern ourselvesfirst with the theory of action identification and then with its implications for the nature of action emergence. The Theory Action identification theory holds that peopIe do what they think they are doing. The person who is asked "What are you doing?" will typically report one of a variety of potential identifications of the action, and we believe that this report is descriptive of the nature of the ongoing action-in a way that other identificationsare not. This immediately availableact identity will be the person'sprincipal cognitive representation of the action, and despite the person's knowledge of other identities, will serve as the template for engaging in the action. A person who is questioned while eating, for example, might identify the action as "chewing," as "gaining weight," as "overcoming boredom," or in yet other ways. The identity that is prepotent for the person, however, will guide the action. Thus, the person who thinks of the act as "gaining weight" is in fact likely to be putting on pounds (Wegner,Vallacher,Ewert, Dizadji, & Reno, 1983). The theory holds, then, just as James would have it, that the mind is stronglyimplicated in the performance of action. To the degreethat people identify an action in a certain way, in advance, they will maintain the action with reference to that identification and often wind up doing the action exactlyasidentified. This idea constitutes the first principle of the theory: An action is maintained in terms of its prepotent identity. Act identities may differ from one another, however,in an important way.The theory goes on to propose that the different identities a person might have in mind for an action can vary in their level in a cognitive hierarchy (cf. Goldman, 1970). Some identities convey the details or mechanicsof the action; such lower level identities indicate howan action is done.

Other identities conveya more comprehensive understandingof the action; such higher level identities indicate why or with what effectthe action is done. So, for example, "washing the floors" is lower in level than "cleaning the house" becauseit indicates how "cleaning the house" is done; "preparing for guests,"in turn, is higher in level than "cleaning the house" becauseit describeswhy or with what effect "cleaning the house" is done. As a rule, an identity A is higher in level than identity B if it makessenseto saythat one does A by doing B. With this sort of analysis,all the identities of a particular action can be arranged in an identity structure that stretchesfrom the highestlevel consequences of the action to the lowestlevel details. As suggestedby the first principle of the theory, a person might set out to do something identified at anyone of these levels. The secondprinciple of the theory suggests that people would prefer to identify action at the highestlevel possible: When an action can be identified at both a higher and a lower level, there will be a tendencyfor the higher level identity to becomeprepotent. It is not too surprising, after all, that people would want to be informed of what they are doing in the most comprehensiveway.When one performs an action that could be identified as both "throwing dice" and "winning a million dollars," for example, adhering to the lower level identity of "throwing dice" would seemfoolish and shortsighted. Higher level act identities inform the person of the contingencies surrounding the act-the things one doesby doing the action-and the fullest knowledge of the act inheres in such contingencies. Given a choicebetweena lowerand higherlevel identity for an action, then, people ordinarily opt for the higher level identity (Vallacher& Wegner, in press). The third principle of the theory indicates that such high-level identities can present problems: When an action cannot be maintained in terms of its prepotent identity, there will be a tendencyfor a lower level identity to becomeprepotent. If one sets out to "cook a meal," for instance, especiallyif one is not an accomplished cook, one could very well fail for the simple reasonthat one cannot perform the action without thinking about some of its details. Frequently, the preferred higher level identities for action must be set aside in favor

I

A(;l'lUN

5~ '", ,

1~ '! ~ ii 7~

i

EMERGEN,CE

271

of lower level identities that provide representations of how the act can be completed. The unskilled cook, for example, might haveto be consciouslyconcernedwith "boiling water" or even"finding the kettle" for the act of"cooking a meal" to be completedsuccessfully.The tendencyto identify an action at high levels,then, is counteracted by a tendency to identify an action at a lower level when it cannot be main, tained in terms of a higher level identity. The three principles of the theory prescribe a systemof dynamic interaction betweenaction and identification. The person attempts to perform action as identified, but because of the attractivenessof high-level identities, the person's "plans" always tend to become more comprehensiveand high level.This press toward higher levelsis thwarted, however,each time the personsetsout to perform the action and finds it cannot be maintained with reference to a high-level identity. The person movesto lowerlevelsto discern how the action can be continued and thus againbecomessus-

This sequenceis possiblebecauseany action hasmultiple high-levelidentities, independent comprehensivemeanings that can be substituted for one another. When one performs the low-level identity of "opening a door," for example, one might be performing any of a variety of higher level identities. One might be "going to work," "checking for squeaks," or even"letting in freshair." On "going to work" one morning and finding the door stuck, one might have to think, however briefly, about "opening the door." And in this instant, one would have severaloptions for understanding the action at a higher level. One might fix the door and continue to "go to work." Or, it could be that the rush of air encounteredasthe door was opened would serve as a reminder that one was "letting in fresh air." Emerging with this higher level identification of the action could conceivably lead one to continue the action in this new direction. It might seem reasonableat this point to "open a window" as well. On moving to a lower level identifi-

ceptibleto the preferencefor higherlevel identities. Over time, then, the identification of action movesin thesecyclestoward the highest level of identification at which the personcan maintain the action.

cation of action, one entersa realm of mundane details of action that could have many high-level meanings. From this position, any suggestionof a high-level meaning, even one that is very unlike the initial meaning of the act, could lead one to emergewith a new comprehensiveunderstanding. Someinitial evidencefor this hypothesized sequencehas been observed in a study of an action that many peoplerealize can changein meaning:the act of "getting married." Wegner, Vallacher,and Kelly (1983, cited in Vallacher & Wegner,in press)askeda number of people to give descriptions of what one does in "getting married" and compiled a list of the 30 most frequently mentioned identities. In a series of telephone interviews, they then asked people who were in various stagesof the .act of "getting married" to rate the degreeto which eachof theseidentitiesdescribedthe act. Factor analysisof theseratings and tests of trends in factor scoresrevealed an interesting pattern. For respondentsquestioned a month before their wedding day, "expressing my love" was a compelli~g identity of ~e a~~;a factor composedof this and ~elatedIdentities wasstrongl"! endorsed.So, prIor to the act, people saw It in terms of a fairly generalhigh-level identity. For respondentsquestioned the day before their wedding, however,a factor composedof, low-level identities was predominant; here!

The EmergenceProcess If people generally know what they are doing, asthe theory asserts,then how can they sometimes perform behaviors that they did not know they were doing in advance?We believe that such action emergenceoccurs by way of a particular sequenceof eve~ts that can be charted through the principles of the theory. The person begins an act with some particular identity (in accord with the first principle); if this identity is too high in level for effective maintenance of the action, the person moves to thinking about the act in terms of a lower level identity (in accord with the third principle); if this lower level identity is effectively maintained, the person moves back to thinking about the act in terms of a higherlevel identity (in accord with the second principle). This higher level identity, however, is not necessarilythe one with which the act wasinitiated. Emergenceof a new high-level understanding of action occurs by way of a transitional state in which the person thinks about some detail of the action.

,...iou,"

11'"

272

WEGNER, VALLACHER, MACOMBER, WOOD, ARPS

people

noted

that

tographer," vows,"

and

the it

act,

the

line

theory,

they

details

of

the

a at

but high-level

These

of

believe

in

a

the

emergence illustrate

of to

immersed sight

in of

the

adopt

a

a

process. the

knowledge

of

what

the

details

initial

of

are

the

action,

of

that

And

The

research

this

The

reported

"getting

was

of

after

their

not

theIr

causal

two

experiments.

the

idea

an

a.

n

act.

people at

low

to The

gence

new

in

a different

addition

investigated

.' which

In

into

both

move

subjects

a to

changes

either

a

.

IdentificatIon

lowed

by

them

a

to

the

act.

.

.of

I

new

high-level

the

adoption

hi~-leve

.entIty

of

a

to

serve

this

study

we

n

that an

I

of d

d epen

to

college

college."

t en

IDlght

identification plan

be of

was

to

do

m ~e

to

possibility

of

classrooms,"

led the vary

to act

embrace of

initial

"going iden-

On

the very

(n =

the

act

the

next

extremes

to

make

a

Both as they

a

The

check

s.ubject.

It described had been

10) of undergraduate were

arranged

textb~?~," campus, 5

students' to

include

'~itti~g

in

'studymg,

high-level

on was

and

identities

("be-

my exCIting

and "getting in on. college social recorded on 7-pomt scales anby

describes

very

poorly

(1)

and

(7).

page

of

article

was attributed

cri..c

of

whereas

to college. identities

as to how well These identities

and

and

was

.."..."""""",

a list

college,

of my parents," "improving a Job ."" I want; nllssmg more

emergence that

Subjects

make

constituted

("reading

class")

well

of resulting arranged to

asked

gomg many

as

around

in

to

of to p~o-

level. to

~pulation.

of 10 Identities to college."

independent of getting

designed

we:e

booklet

~

"walking notes

asked

of

small exper-

distribution

were

.or going

four

20.

identi~es

at

) as

in

the

identification

as part to list

the

of

of

random

conditions

a sample

low-level

chore~

i.de~tity life

laboratory

one

were

do

of

the

to

as a result

of

from

to

of prior

limit

page

descriptions

"taking

the

identi-

sex

and ~ measure began WIth a page

low-level

a

next

descnbes

students

Our

5

my

for at

to

prior

emergent

booklets

things," "spen~ money;' activities"). Ratings were the

the

assigned

~e~

the

~e

conditions

to rate each act of "going

free

new

1

examined

of

students

with

lmpmrmg

through

the things they were encouraged

gathered of

this

e

asked

~~

vs.

assigned

bookl~ts.

effectiveness

-the

expose

'.

emergent

the

0

p I

claims test

randomly

came

each

they

coming chances eriment

unusual

design

l?,w)

life

Subjects

high-level

up

The

level

Interest.

Ex

emer-

undergraduate

were

vs: sex.

were

.thin~

could,

was to

as

such

Sixty-one

as a manipulation

of groups

to

identity

th

of

conditions.

process.

the manIpulations Each booklet

s~bJects

are

low

.Ii actIon

d

articles

degree

stringent

X 2 factorial

conditions

in the

behavior.

arranged

.

I

varIables

of

. d

ul ar

manIpulatIon

Measures

serve

the

actIon

or

a p~c

a potential

the

planned

high

these

variables.

and

VIde both emergence.

emer-

degree

in

in

that one's

males) participated in exchange introductory psychology classes

(high my

e::rperimental

operations

of

le:vel

imental

.list

manipulation

..' of

of

the

observable

studies,

test

sus-

the of

'... IdentificatIons

emergent

translated

set

in

the

fairly

They

a 2

Procedure.

about

IdentificatIons

of

groups

to

think

tested

an

claim,

experimental

design.

and 32 in their

mdependent

exaIDlne

especially

experiment using

to

and

(l~proVIng

conducted

designed

led

become

high-level

second

process

and

was are

level ..,.

we

read

"impairing

belief

emergence

University.

~~.ation

events,

To

detail;

first who

0

ceptIble

m

The

that

cti

Trinity

only

of

mterd~pende~ce. process

in

belief a

the

conditions

.the

emergence

of

Subjects

to

.'

the

that

potential Some

to college" has the sex life"; others read

assess

the

provide

females credit

all,

sequence

to of

does

groups

Method

emergence.

study,

hypothesized

of

used

one

both

opposite

Measures

does were

and

of

desIgned

action

the results

each

what

act.

"going one's

college"

thought

power

lose

action

sense

here

married"

the

in

.extra

conceptualization

shows

then

gence

one

when, they

their

makes

were

would

People

doing

life."

the

were

Others

far-fetched,

of

making to

what

Then,

a rather

make

Some

college.

college.

that

to

act.

identities, to

identity

article

We

Freudian

action.

they

meaning

meaning

that

to

low-level

of going to

student

the

identities,

going

claiming of , 'improving

sex

natural

of list

"going

each of

high-level

exposed

an a

problems."

.(29 cIrcumstances.

current

test

the

"discover"

de-

sequence

also

conception appear

had

endorsed

illustrate

they

were

to

article effect

act

high-low-high

typifies

sense,

the

list

a part

the wedding endorsement

often

thus the

as

asking

identities

result

high-level

"getting

observations

the

by

the

to a

respondents

the

more

factor:

occurrence we

of

now

of

asked

of with

following level. Their

identities they

as

"expressing

Finally,

month high

low-level

of

principle

level

list

asked

maintain

themselves

action.

tification a

performing

longer

third

pho-

"saying

of

no

thinking

the

a

outfit,"

course

concerned the

questioned were again

clined,

the

by

with

"hiring

special

could

merely In

new

In

people

properly

were

a

like.

then,

love."

of

they

"getting

presented.

the

booklet,

the

Subjects (falsely)

opportunity

encountered to The

""

"'

,Iii

New

for a

York

~hort TImes.

i" Ii

. ~ :;

.;;;;:;

ACTION

EMERGENCE

I~ began: ".Washington, D.C. In a recent study at the NatIonal InstItute ~f M:n~ Health, researchers found that a college educatIon SIgnIficantly affects a person's future. sex life." The article continued, then, in differing versions

prior identification of variance (ANOVA)

depending

level

was

sIgnificant,

and

other

effects

dition.

on the subject's

For those

attendance

do similar went

one's

individuals

to think

con-

that

college

said,

"Dr.

college."

of the fr~m

this versi°.n other

than

do similar

The accompanying to the same overall

It

(fictitious)

and exhave

four

as the

fi

al

e

f

n

page

of emergence series

of

college

th

bo

0

k1

e

t

0

and was the same

Likert-type

life, three

tal .

e

rating

con

th

on several

were embedded

for students my

"I think

sex life will

Results

Dr. Wolfram's

at Trinity,"

and

be about

and

the

Amidst

a

was

aspects

of

level

that were in-

identity-rating

nipulation

to

check

nificant, h ot er

are true

note cluded

high-level

identification

the

task

listing

sonably

we

began

by

of

high-level

ratings

forming

the

The

proved

index

index

(a

= .33).

e

in

other

main

So,

for

Low-level

1983).

For

this

elsewhere, index

as

the

was

effective.

the

index

to

the

best

level

By

35)

tification

have

was

in

group

the

low-level

for and

the

23.8

for

the

the

e

. artic

1 e

was

not

relation

with .05.

was

support

to led

action,

in

can

opposite

by

identity

subjects,

the

low-level

check

was

the

measure

degree to

by

subjects in

suggested

terms emer-

index,

in

turn,

with

the

belief

negative

the

sig-

ofbelief,

that

college"

the

a

between

target

expected

attenuated

cor-

unreliability

of

index.

results

People

and

level

relation

high-level

the

level

supported

correlated

perhaps

low

and

adopted

The

in were

high

entirely

the

To

they

action.

the

"going

they

high-level The

the

p