You are on page 1of 18

The Development of Intentionality and the Role of Consciousness

Author(s): Michael Lewis


Source: Psychological Inquiry, Vol. 1, No. 3 (1990), pp. 231-247
Published by: Taylor & Francis, Ltd.
Stable URL: http://www.jstor.org/stable/1449759 .
Accessed: 08/10/2011 05:09

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Taylor & Francis, Ltd. is collaborating with JSTOR to digitize, preserve and extend access to Psychological
Inquiry.

http://www.jstor.org
Psychological Inquiry Copyright 1990 by
1990, Vol. 1, No. 3, 231-247 LawrenceErlbaumAssociates, Inc.

TARGET ARTICLE

The Development of Intentionality and the Role of Consciousness

Michael Lewis
Institutefor the Study of Child Development
Robert WoodJohnson Medical School
Universityof Medicine and Dentistry of New Jersey

Startingwith the observationof a veryyoung child's learningof an instrumentalresponseand the


child's frustration in being unable to control what was once a controllable behavior, I seek to
explore the origins of intentionality.Threemodels are considered.In the mechanisticmodel, the
construct of intentionalityis considered irrelevantbecause action is viewed as caused by the
environmentor internalbiological dispositions. In a second worldview, intentionalityexists but
not always and notfor all social creatures; it is an emergentpropertyof organisms. The third
view, presented here, arguesfor intentionas a propertyof all goal-directedsystems. Withinthis
view I seek to describe the various levels of intentionalityand to developa systemthat will enable
us to explain intentionin such divergentactions as a plant's movementtowardthe sun, a newborn
child's movement toward a brightly colored object, and an adult's conscious goal-directed
behavior.

Two World Views These do not imply reinforcementcontrolbut structuralcon-


trol. In such external control paradigmswe need not infer
Two views of humannaturepredominatein ourtheoriesof will, intention, or plan.
development.In the first, the humanis actedon by surround- In contrastto this passive view is the constructivistpara-
ing forces, and in the second, the humanacts on these forces digm based on the world view that the organism acts on its
(Reese & Overton, 1970). The reactive view generatestwo environmentand participatesin it. The organismhas desires
major theoreticalparadigms, biological and social control. and plans. These desires and goals are constructed, as are
The active view, on the other hand, has generatedthe con- most of the actions enabling the organism to behave adap-
structivistor developmental-cognitivetheoreticalparadigm. tively. This view does not necessitate discardingeither bio-
The place of intentionwithin these two world views differs logical imperatives or social control as potential causes of
greatly.Let us considerthese views in theirextremeforms to behavior, because humans are both biological and social
show how their respective theories might treat the issue of creaturesandboth, to some degree andin some combination,
intention. must affect behavior.I preferto thinkof these biological and
In both the biological-motivationalor social-controlpara- social forces as nothing more than the raw materialsfor the
digms, the causes of behavioror action are forces which act construction of cognitive structures including goals and
on the organismcausing it to behave. These may be internal desires, plans, and action patternsthemselves (see Fodor,
biological featuresof the species, including species-specific 1981b, for a similar view). Takingthe example of sex-role
causes of behavior, or the external social control of con- behavior, I argue that hormones and social control become
specifics-these, too, may be species specific. In all cases the materialfor cognitive structures.Such structuresmight
within this world view, the organism is acted on and the take the form "I am male or female," "Males or females
causes of its action (including its development)are external behavethis way or thatway," "Toreceive the praiseof others
to it. Thus, for example, the majordeterminantof sex-role (a desired goal) I should act either this way or that." Such
behavior is thought to be biological, that is, determinedby cognitions and their accompanyinggoals and desires, along
sex, in this case, by the effects of hormones (Money & with cognitions concerninginformationaboutthe world, en-
Ehrhardt, 1972) or lateralization(Buffery & Gray, 1972). able the child to intend, that is, to will to act in a particular
Alternatively,sex-role behaviorcan be determinedexternally fashion.
by the shaping effect of the social environment, either the These two world views are present in all psychological
differentialrewardsof conspecifics (Fagot, 1973) or the dif- inquiry.The mechanisticmodel receives supportin the case
ferential constructionof the social world. Examples of the of the biological study of action (e.g., T-cells tracingforeign
former are already well-known (e.g., parentalpraising or proteinsthathave enteredthe body). Constructivistviews are
punishing of specific sex-role-appropriateactions, such as supportedby theories of the mind (Neisser, 1967). It should
playing with particulartoys; see Goldberg & Lewis, 1969; not go unnoticed that with the growth of cognitive science,
Rheingold & Cook, 1975). Examples of constructionof the the idea of constructingmental representations(that do not
social world include giving the child a male or female name. correspondin any one-to-one fashion with the "real"world)
232 LEWIS

and with them, plans and intentions, has become more ac- children, and adults, we may have more informationto aid us
ceptable to psychology proper (see Gardner's, 1985, in understandingintention. The problem of generalizing
review). from animals or machines to humansor from one cultureto
anotheris avoided. Forthese reasons, the study of the devel-
The Problem of Intention opment of intention is of some general interest.
One way to pursuethis topic is to ask how childrencome to
Central to this article, however, is the problem of the understandthe concept of intention. We might, for example,
developmentof intention.In so statingthe problem,I beg the ask at what age children come to understandthe differences
question of whether there is such a thing as intention. Intu- between accidentalor intentionalbehavior(see, e.g., Berndt
itively, most of us are comfortablein believing thatintention & Berdt, 1975; Dunn, 1988). The problemhere is thatsuch
exists. There is no difficultyfor any of us in using termssuch studies tend to learnmore aboutthe child's understandingof
as "I intendto go to the markettomorrow,"or in understand- the termsaccidental versus intentionalthanthey do aboutthe
ing thatan intentionalact of violence is a more serioustrans- child's knowledge of intention. Nevertheless, this approach
gression thanan unintentionalone. Nor do we have difficulty has been followed by many (e.g., Dodge, Murphy, &
in explaining our action as intentional: "I went to the re- Buchsbaum, 1984; Keasey, 1977). Anotherdifficulty is that
frigeratorbecause I intendedto get the butter." studies of this type requirethatchildrenbe able to speakor at
Even so, acceptingthe notion of intentionalityraises diffi- least understandthe language. Because the language has
culties. Some forms of action are more difficultto explain as terms such as intentionand accidental, we may be inquiring
intentional;unawareaction, for example, "I did not realize I into the child's socializationratherthaninto its logical struc-
was angry and did not intend to push you away." Freud tures. There is clear evidence that children show behaviors,
suggested that action we are unaware of is unconsciously prior to language, that would lead the observer to believe
intentional.Otheractions appearso rote and mechanicalwe they have an understandingof intention and of causality, a
hardlybelieve thatthey were plannedor were intentional,for closely related problem (see, e.g., Leslie & Keeble, 1987;
example, walking actions, or even talking or listening. Michotte, 1963).
These, too, are intentionalif we considerthatintentionneed
not be always pure intention, but can be "intentionsin ac- The Development of Intentionality
tion" (Searle, 1984, p. 65). Nevertheless, I claim that adult
humansare intentional,or if not so, then at least we believe It is Piaget (1936/1952) who has offeredus a developmen-
that we and others are intentional(Dennett, 1987). tal blueprintof the developmentof causalityand intentionin
Make no mistake: The claim that intentionalityexists is the opening years of life. To summarize, Piaget employs a
made without any basis of proof, only that we have such an mixed model in his descriptionof the developmentof inten-
idea; there might be culturesand times thatwould deny such tionality. In the earliest stages of development, children's
a concept. The same, however, might be said for any mental actions are simply biologically given action patterns.After a
structureor operation. Here I am willing to considerRorty's time, these action patternsproduce (still without intention)
(1989) analysis of truth. If we follow his analysis correctly, outcomes. It is these outcomes which in turn produce the
we cannotmake the claim thatintentionalityexists out there, action patterns.ThusA accidentallycauses B (an effect) and
only our understandingof it exists and is "real."Thereis no B in turnproducesA. Piagetcharacterizesthis chainof events
Truthout there to which some language (read here world as a simple circular action pattern.Notice thatthe controlof
view or model) is betterthan another. the action is associated with the simplestof mentalrepresen-
tations. The representationis the associationbetweenA and
Truthcannotbe outthere-cannotexistindependent- B as mutually connected. Nevertheless, it is B, an environ-
ly of the humanmind-because sentencescannotso mental event, (an effect in the world) which causes A (the
exist, or be out there.The worldis out there,but
of theworldarenot,onlydescriptionsof action) to occur. I thinkit is safe to conclude that, for Piaget,
descriptions the child startsthe developmentalprocess withoutintentions.
theworldcanbe trueorfalse.Theworldon itsown-
unaided by the describing activities of human However, by the end of the first year of life, children "set
beings-cannot. (Rorty,1989,p. 5) out to obtain a certain result" (Piaget & Inhelder, 1969, p.
10). By now, the means-formerly A (both action andrepre-
Adopting this conceptualizationpermits us to choose be- sentation)-have become independentof outcome, B. The
tween descriptionsof reality and allows us to test our choice mentalrepresentationassociatingA andB, which appearedat
for truthfulnessvis-a-vis another description. In this case, the beginning of life, has now been separated.In a sense,
our choice is between a model thatdoes not requireintention Piaget describesthe child as changingfromreactiveto active
and one that does. Because I wish to focus on the topic of and from rote associative reproductionto the pursuit of a
ontogenetic change in intentional behavior, I choose the goal. Here, then, he startsto speak of intention. However, it
model asserting that it exists. Our problem then becomes is still a limited intentionality;it is only the separationof
how to study the development of this concept. I choose to means and end in the utilizationof an available means for a
focus on the topic of development,for it seems to capturethe new end. It is not until the end of the fifth stage (somewhere
problemsinherentin any discussion of intention-for exam- around15 months)thatintentionalityis assumed. ForPiaget,
ple, the question of intention in animals (Griffin, 1984), in intentionalitymakes possible the creationof goals and plans
machines (Newell, 1982), and in differentcultures (D'An- in the absence of externalevents and in the establishmentof
drade, 1981). The ontogenesis of humans allows us to con- new schemata-multiple means associated with multiple
sider the issue of intention from a broad perspective in the ends.
same organismat differentpoints in its life. Because we have Piaget's model denies the existence of intentionalityat the
some idea of the similaritiesand differencesamong infants, beginningof life, but allows for its developmentover the first
INTENTIONALITYAND CONSCIOUSNESS 233
2 years. The difficulty with such a mixed model of develop- position not unlike that held by those who would see in
ment is in the question of how is intention created? If we representationsan action potential, that is, a desire for or a
accept, for example, the mechanisticmodel, we can dismiss desire to do something (e.g., Searle, 1984).
intentionas an unnecessarymentalisticconstructhavinglittle
utility at any point in development. For the constructivist, Cognition Before Emotion?
however, intentionalityis a propertyof the mind, as is believ-
From the point of view of cognition leading to emotion,
ing, desiring, hoping, fearing, and so on. Thereis no need to
create it, because it is a partof our mental states. Forexam- appraisaltheories regardemotion as the productof informa-
tion processing. Arnold's (1960, 1970) theory of emotion
ple, consider the position of Searle (1984). For him, actions
consist of two components-one mental, one physical; the has as its central construct the cognitive act of appraisal,
mental component is an intention. That is, a mental compo- whereasLazarus(1982) held to a view thatappraisalleads to
nent is an intentionbecause it is aboutsomething. These two certain kinds of activities as a way to adjustto the environ-
ment. It is the appraisal itself that leads to the emotion.
componentsgo together: "The mental componentas partof
its condition of satisfactionhas to both representand cause Discrepancy theory, a part of general appraisaltheories, ar-
the physical component" (p. 64). In this sense, Searle turns gues for an even more direct connection between cognition
the problemof how to go from thoughtto action into another and emotion (Hebb, 1946, 1949). Berlyne (1960), for exam-
problem, which is that action, by its nature, is part thought ple, suggestedthatunfamiliarevents evoke fear, a view taken
(mental) and part physical. Whatever the argument one by others (Kagan, 1974; Lewis & Goldberg, 1969). Siminov
adopts, the claim that intentionexists as a propertyof mind (1970) defined emotion in terms of informationprocessing.
does not requirethat it be created. Piaget's explanationfor In his model, emotion is the consequence of the organism's
the developmentof intentionutilizing a mixed model seems need for informationwith respect to reaching a goal, multi-
to me to be the most difficult model to justify. The problem plied by the differencebetween "necessary"and "available"
remainsof how to go from the absence of the mental state- information.Notice that here emotion is defined at the cog-
intention-to its presence?A mechanisticstance might ask, nitive-process level ratherthan at the goal level. If, on the
"If in the beginning of developmentyou do not claim inten- other hand, we defined emotion as the goal or the desire to
tion, even though one might describe the behaviorthat way, achieve the goal, we would consider emotion as a precursor
then why do you claim intentionlaterfor similarbehaviors?" to cognition. This point is important,because if emotion is to
In other words, if we do not need it in one case, what allows be defined at this point, then emotions arenot only causedby
us to claim it for another.I returnto the basis of this claim cognitions, but, in turn, produce cognitions.
shortly.
The mixed model requires that we address the question Emotion Before Cognition?
"Where does intention come from?" How can the infant
discover intentionsif it does not alreadypossess them? Like We can approachthis problem from anotherperspective.
other ideas, intention needs to be discovered. However, the Emotions have been viewed as preceding cognitions from
three perspectives:as motive, marker,and instigator.
problem of discovering something that one does know be-
comes an issue. This problemof how to know of something
one does not know is too complex to deal with here. Fodor Emotions as motives. Since Darwin (1872), the notion
(1975) in his critique of Piaget's theory of concept acquisi- of emotions as sourcesof actionhas been useful as a theoryof
tion raises the same problemof how childrencan learna new motivation. Theories that consider emotions as motives can
concept unless they already have the ability to hypothesize be divided into two classes: (a) those viewing emotion as a
the concept. If they already have this ability, then they al- consequence of thoughtand therebyreinforcingthoughtand
ready possess the concept. This is a particularproblem for (b) those viewing emotion as causing thought based on the
Piaget, because the idea of the intention is part of the log- evolutionaryhistory of the species. The centralissue of the
icomathematical structureexisting in the child's head. As hedonic traditionis the belief that people think in such ways
such, it does no good to arguethatthe child does not createit, as to reproducepleasureand avoid pain. The emotionalcon-
but ratherborrows it from the language of the adults around sequence of a thoughtis regardedas the primarycause of that
him. In some sense, then, Piaget's mixed model is of some action.
concern. However, thereis not anotherwell-articulatedtheo- This view of emotions as motivating action throughthe
ry of development to take its place. One of the purposesof emotional consequences of that action appearsreasonable.
this article, then, is to suggest a starting point for such a Forexample, studentsmay study for examinationsbecause it
theory.One way to explore the relationbetween thoughtand feels good to pass and it feels bad to fail; childrenengage in
action is to consider the literatureon cognition and emotion symbolic play to experience the pleasuresin solving a prob-
(or desire) because desire may be synonymous with lem. It should be noted that this view of emotion considers
intention. thought to be motivated by the possibility of its emotional
consequence. Even though the emotional experienceoccurs
Relationship Between Cognition and Emotion after the thought, it is believed that the reinforcementvalue
of this experienceserves to producethe same set of behaviors
To explore the relationbetween thoughtand action, I turn to reexperiencethe particularemotion. So althoughemotion
to a discussion of the association between cognition (repre- is initially a consequence of thought, the expectation or
sentations)and emotion (action or motives). The association memory of the emotional state may precede and influence
is often discussed as one leading to the other,eithercognition subsequent thought.
leading to emotion or emotion leading to cognition. Thus, in If one thinks of emotion in this way, then emotions, es-
either case we have assumed a connection between them, a pecially feeling good and feeling bad, act as rewardsto par-
234 LEWIS

ticular patterns of thought. In many cases, these hedonic & Rumelhart,1975). It is reasonableto assumethatinforma-
events seem to be unlearned.Forexample, it is unlikely that tion may enter memory not only as a functionof the content
the good feeling producedby eating when hungryis learned. or sequence of the material,but also as a functionof the type
Rather, eating feels good because of an innate biological of emotionaltag; clearly,the schemaof a man in a white coat
connectionbetween food in the digestive tractandrelief from is more likely to be rememberedif it is associated with high
hunger. On the other hand, some emotional reinforcers fear than if it is markedwith low interestand fear. Markers
seem, at least at first glance, to be learned. There is no may also be associated with the emotional content of events
intrinsic reason that it should feel good to get an "A" on a as they relate to the emotional state of the organism. The
Frenchexamination. researchon state-dependentlearningindicatesthatemotions
Emotion may be not only the rewarding outcome of may have a powerful influence on cognitive processes, in-
thought, but also its antecedent. This view of emotion is cluding free recall, imaginativefantasies, and social percep-
usually associated with biological explanationsof emotion. tion. For example, Bower (1981) found that people recall
Darwin(1872), the sourceof the evolutionarytraditionin the more events that are affectively congruentwith their mood
studyof emotion, arguedthatthe processof evolutionapplies duringrecall. Here, emotions as markersrefernot only to the
not only to anatomicalstructures,but to intellectualand ex- emotional tag attachedto the cognitive event but also to the
pressivebehaviorsas well. Emotionsareby theirnatureasso- emotional state of the subject as the subject interactswith
ciated with patternsthatthe organismneeds for survival.For the cognitive event.
example, the sight of a predatorwill elicit fear in the orga-
nism, the action patternof which is to flee. Or, a baby's cry Emotions as instigators. The third role of emotion in
will elicit nurturancein the motherwith a concomitantbehav- cognition addressesthe following question:Do certainfeel-
ioral repertoireof nursing, holding, or retrievingthe infant. ings necessarily lead people to thinkin particularways? One
Viewed in this way, emotion is both a state of the organism way to approachthis issue is to considerpatternsas relatedto
and a responsethatis basic to life and survival.In all cases of specific emotions. Emotionsmay not only lead in some bio-
positive and negative emotions, the emotional elicitor pro- logical fashion to action patterns(Plutchik, 1980), but emo-
duces specific actionpatterns(includingcognitions)as a part tions may in fact produce specific thinkingpatterns.
of the emotion. One aspect of this issue is related to the nature or the
Plutchik(1980) enumeratedeight basic functionalpatterns content of the thought. For instance, someone may tell you
of behaviorthathave adaptivesignificance for all organisms thatyourcousin was hit by a car, or someonemay tell you that
in their strugglefor survival. The prototypicpatternsinclude your cousin won the lottery. The emotions producedby the
incorporation,rejection, destruction, protection, reproduc- informationaboutyourcousin in these two cases mightinflu-
tion, reintegration,orientation,and exploration.These basic ence your subsequent thoughts. Isen (1984), for example,
adaptivepatternsarethoughtto be the functionalbases for all demonstratedthat happy moods produce more associations
emotions recognized in humansand animals. Eightemotions than unhappymoods.
accompanythe functionalpatterns:acceptance, disgust, an- This discussion suggests thatit is unreasonableto consider
ger, fear, joy, sadness, surprise, and expectancy. Although cognition and emotion as unrelated. Within the organism,
the specific behaviors that accompany these patterns may these processes coexist and are interdependent.Thatwe sep-
vary across differentspecies, their survivalfunction is com- aratethem reflects an Aristotelianview, not necessarilycor-
mon to all species. rect. In fact, it is difficult to think without action, because
Zajonc (1980) offered a view similar to the evolutionary thinking is always associated with emotions, and emotions,
position. For Zajonc, some of the behaviorsassociated with for the most part, either (a) have action patternsassociated
an emotional state may have "hard-wired"cognitive repre- with them or (b) arethemselves the goals of thought.Such an
sentations; that is, they may be independent of cognitive analysis again leads us to the connection and interdepen-
systems and, in fact, may precede perceptualand cognitive dence between thoughtand desire;that is, intentionsconsist
operations. Zajonc discussed the primacy of emotion with of actions andthoughts, althoughas we shall see, the level of
regard to preferences and attitudes, but his argumentwas thought may vary considerably.
essentially thatemotion "accompaniesall cognitions, that it
arises early in the process of registration and re- Behaviors as Measures of Intention
trieval . . . and it derives from a parallel, separate,andpart-
ly independentsystem in the organism" (p. 154). In short, Leavingthe problemof the mixed model aside, it mightbe
emotions may be associated with basic adaptive functions the case thatPiaget's observations,andthose of othersbefore
and have as their biological consequence a set of disposi- him (e.g., Baldwin, 1894/1903), might allow us to come to
tions, including actions as well as thoughts. understandhow he claims intentionalityfrom the observa-
tion of behavior.In some way, it may be useful to use Piaget's
Emotions as markers. Much attention has been observations to construct a model of the development of
focused on the roles of "hot" versus "cold" cognitions. The intentionfrom a single stance. Piaget, in thinkingaboutthe
general assumption underlying this belief is that cognitive developmentof intention, introducestwo centralfeaturesof
processes have differentlevels of efficiency or outcome de- mental life: (a) means and end and (b) schema development.
pending on how these cognitive processes are tagged with Let us consider each in turn to see whether these mental
specific emotional tones (see Zajonc, 1980). One might ar- operations and the behaviors associated with them aid our
gue that certain cognitive processes marked with emotion understandingof the developmentalprocess.
might be more efficient thanthose not marked.Forexample,
the retrievalof past events, both in short-termand long-term Means and end. The 3-day-old infant is attachedto an
memory,is facilitatedby specific affective markers(Norman apparatusthat delivers a sweet liquor if it sucks at a certain
INTENTIONALITYAND CONSCIOUSNESS 235

rate. Within a few minutes, the infant is able to alter its intention,theschematahithertorelatedto othersitua-
behavior so as to get the liquor. It is clear that, for Piaget, tions.(Piaget,1936/1952,p. 211)
conditioningin a very young child does not representinten-
tionality. The child at very early ages may be taughtto suck The child moves from a trial-and-errorbehaviorpattern,in
on a nipple "in orderto produce" an effect (Lipsitt, 1976); which a solution is eventuallyfound, to an insightfulperiod.
however, the behavioralconnection between the sucking re- In this latterperiod, possible solutions arepresumedto occur
sponse and the outcome are not intended. These he saw as as mental representations,which are created and combined
habits, imposed externally. There is no mental separation and, only after, displayed as action.
between the means and end and thereforeno intentionality. Here, too, it is difficultto see how such behaviorcan help
Moreover,and perhapsmore important,we might arguethat us assume intention. Piaget's language employs an inten-
the simple circularreactionis as much causedby the outward tional stance to show it is intentional.It is not clear-unless
reactionas the child's desire to act and thus fails to be inten- we believe that by Stage 4 the child has gained, at least,
tional because it is imposed. primitive intention-that the actions and combinations
It is believed that intentions can be inferred only when themselves satisfy our need. Considerreachingfor an object
thereis variationin behavior.Piaget acknowledgedthis when no longer in sight, at about 8 months. We know thatpriorto
he spoke of multiple means to the same end or, alternatively, this time, the child's reachingis relatedto a complex schema
the same means to multiple ends. Although Piaget inferred which combines reaching with seeing. If seeing is blocked,
the developmentof mental structuresthat logically must be reaching ceases. The developmentof active memory at this
associated with such actions, to infer intentiononly requires point may allow the child to continue to see the object, but
that multiple means and ends be available. When they are, now in memory,thusreachingcontinues. Alternatively,chil-
we need to infer some mechanismwithin the organismena- drenmightreachfor somethingthey see becausetheywantit.
bling the choice. In doing so, we assume that there are no They cease to reach when the wanted object is out of sight
simple habits that can account for our observation. because they cannot rememberit, but with the advent of a
However, we arenot helpedas muchby this observationas maturingmemory system, they can remember,and so they
we might at first assume. In the case of a repeated similar continue to reach for it. Notice that in the formerexample,
action, we cannotassumethatthe same meansto an end does the belief in the lack of intention associated with the reach
not reflect intention. First, no action is ever the same, thus does not easily allow us to associate intentionwith the reach
even a child's simple kick is different each time (Thelen, at a latter point-the problem of the mixed model. This
Kelso, & Fogel, 1987). Moreover, certain means may be problem does not occur when we adopt a model including
preferredand are repeated not because they are controlled intentionfrom the start. Piaget's argumentfor intentionality
externally, but because they are valued. I may put my left at a particularpoint in development, but not before, strikes
shoe on first each day because I prefer it, not because the me as assuming a selective ontogenetic intentionalstance,
perceptionof shoes compels me to put the left one on first. something akin to a limited intentional stance (Dennett,
Forthe case of multiplemeansto an end, therealso may be 1987).
logical problems. Forexample, it seems possible to construct
a machine (or instructa child) to produceany one of several How Infants Learn and What Happens
actions to produce a given result. The training of such a When What Works Doesn't
complex habitor rule (or its programming,using the comput-
er metaphor)only requiresthat a particularresponse be se- Given these concerns, especially with the problem of a
lected and its effect vis-a-vis the end be evaluated.If a partic- mixed model, we need to returnto the phenomenathatneed
ularmeans succeeds, alternativemeans arenot needed. Such explaining. Rather than choose real-life examples, I use a
an analysis poses some difficultyfor any theoryof the devel- particularstudy of learning that my colleagues and I have
opmentof intentionthatassumeswe can move logically from already reported(Alessandri, Sullivan, & Lewis, in press;
an action to an inferredmental state. I do not think that any Lewis, Alessandri, & Sullivan, in press). Because this study
specific action on objects as in means-end or its develop- is carefully described elsewhere, I address only its main
ment will satisfy this demand alone, nor will multiple featureshere. We studied how childrenlearn and what hap-
observations. pens when the rules are changed. Our experimentinvolved
intentionality,and, because we studied infantsover the first
Schema development. On this topic, Piaget suggested year of life, startingwith the 8-week-old, we can explorethe
thatthe infantshows us thatit is not passive to events around topic almostfromthe child's beginningof life. Afterexplain-
it, but takes an active stance. The child in Stage 5 startsto ing the study and its findings, I use it to considera theoryof
coordinatedifferentcombinationsinto new and meaningful the developmentof intention. Fromthe outset, it should be
schemata. So, for example, the child knows how to pull on a clear that the position I adopt attributesintention to any
rug to bring the rug toward itself. The child sees an object system that is goal directed. My basic premise is that all
that it wants to possess, but is out of reach. By pulling the goal-directed systems are intentional, but that differentlev-
rug, the Qbjectcomes into reachand is possessed. This is for els or types of intention may be useful in understanding
the child the coordination of previously independent animate, inanimate, phylogenetic, and ontogenetic differ-
schemata. ences.
The experimental paradigm was intended to examine
Now,in orderthattwoschemata,untilthendetached, whether young infants could learn a simple task and, once
maybe coordinated withone anotherin a singleact, they had learnedthe task, what would happenwhen the rules
the subjectmustaim to attainan end whichis not changed. Because the learningconsistedof pulling a stringto
directlywithinreachand to put to work, with this obtain a reward,their motor actions were observedto assess
236 LEWIS

learning. In addition, their faces were continuously lated for each minute of the session. We describe only two,
monitoredto measuretheir emotional expression. A simple joy and anger face, which could be coded with over 90%
operant-conditioningtask was used (see Lewis, Sullivan, & agreement between judges.
Brooks-Gunn, 1985, for details). A string connected to a At each age (2, 4, 6, and 8 months), infantswere assigned
velcro wristcuff activated a microswitch. A pulling move- to the experimentaland yoked-control conditions. The ex-
ment of the string triggered a brief presentationof a color perimentalsubjects'armpulls resultedin the event occurring
slide showing an infant's smiling face, accompaniedby a whereasthe controlsubjectsreceived the same amountof the
recordingof children's voices singing the "Sesame Street" event as did the experimentalsubjects, but it was not related
theme song. Arm-pullingresponses were recordedand each to theirarm-pullbehavior.Forthem, therewas no possibility
child was videotaped. of associating a cause and effect.
Each experimentalsession included a 2-min baseline dur- Look first at the arm-pull data for each age group (see
ing which we were able to determinethe baseline or ongoing Figure 1). Notice that control subjects showed no change
rateof armmovement. Infantsthen received a learningphase from the base periodto the learning, extinction, and second-
of contingent stimulationin which the audio-visual stimuli learning phases. Not so for the experimental subjects: To
were activatedby each armpull. All infantslearnedthe task begin with, the infants who could cause the event to go on
within the first 3 min of the learningperiod. When learning significantly increasedtheirarm-pullbehavior.Of particular
was achieved, a 2-min extinction phase occurred, followed interest are the subjects' responses once the association be-
by a second 3-min learning phase. During extinction, no tween armpull and event ceased to work (extinction).Notice
event was presentedafter an arm pull. that when the arm pull no longer caused the event, arm-
Rates of arm pulling throughoutthe session were com- pulling behaviorsignificantly increasedratherthandeclined
puted as the total number of arm pulls per minute. Facial over the period of disassociation. In fact, during the dis-
movements were coded from videotapesof the infantsusing association phase, there was a 154%increasein armpulling
the Maximally Discriminative Facial Movement Coding over the learning phase and a 376% increase over the base
System (MAX; Izard, 1979). Coders sampledthe videotape phase! Once the extinction phase was over, the infants re-
segments of each subjectusing a frame-by-frameanalysis of turnedto the rate of armpulling they showed duringthe first
the videotape for each of three facial regions: brows, eyes, learningphase. These differenceswere all highly significant.
and mouth. After coding each component, facial expressions Now let us turnto the emotional behavior.
were identified by MAX formulasand their frequencytabu- Joy expression follows what we have reported before

2 months
Mean Arm Pulls
8
6 EXP

4
2 -C...-CON
0
B L1 EX L2
Phase of Learning

6 months 8 months
Mean Arm Puills Mean Arm Pulls
13 13
9.75 9.75 EXP

6.5
^^_
6.5
3.25 CON 3.25 _^ ---^
0 -------------- 0
B L1 EX L2 B L1 EX L2
Phase of Learning Phase of Learning

Figure 1. Arm pullingby conditionas a functionof age.


INTENTIONALITYAND CONSCIOUSNESS 237

(Lewis, Sullivan, & Michalson, 1984). There was little joy seven parts. The 40% that did, showed an increase in arm
duringthe base phase and no change for the control subjects pull and anger and a decrease in joy duringthe second dis-
(see Figure2). The subjectswho learnedshowed increasesin association, and an increasein joy and decreasein angerand
joy during the initial learning phase, a total decline during arm pull once the association was restored.
extinction, and renewedjoy once the second learningphase It is also importantto note thattherewas a high correlation
began. Angry expressions follow a reverse pattern(see Fig- between arm-pull rate during the disassociation and angry
ure 3). There is little anger during the base or during the faces. One more finding is necessarybefore we ask whatthis
initial learning phase. Anger increased markedly once the all means. We measuredthe activity in each of the child's
association between action and outcome was broken and arms and found that althoughmovement in both was present
declined as rapidly once the second learning phase began. to begin with, duringthe learningof the associationonly the
Age effects reveal that 2-month-olds behave in the same armor hand pulling the string increasedin activity;the hand
manner as do 8-month-olds. Notice that arm-pullbehavior not pulling, decreased. More important,when arm-pullrate
patternsdo not vary by age. Althougharm-pullrates, over all increasedas the angry face appeared,it was only in the arm
phases, are greaterthe older the infant, there are no interac- associated with the response. The response to the dissocia-
tions between age and phase. In other words, 2-month-olds tion, even in the 8-week-olds, was not a generalizedactiva-
show proportionallythe same increases as the oldest infants tion, but a highly specific response to a learnedassociation.
(344% for 2-month-olds and 393% for 8-month-olds from Let me try to summarizethe results of these observations.
base rate to extinction). To observe how children in the first year of life learn, and
The emotional datareveal the same findings. Again, over- what they do when what they have learned changes, we
all amountof expressiondemonstratedincreaseswith age but created a situation where an arm pull resulted in some un-
there are no Age x Phase interactions. Thus, even for 2- usual event. I say unusualbecause from our perspectivewe
month-olds, the constructionof an association between an have no reason to believe that the child, priorto our manip-
action and outcome is accompaniedby increases in positive ulation, has ever experiencedan associationbetween an arm
affect, and the disassociation between them results in the pull andthe appearanceof picturesand sounds. Certainlyit is
appearanceof anger which declines once the association is possible that the children, in theircribs, learnedthatmoving
restored.I mightpoint out thatthe originaldesign called for a their arms produced some effect such as the shaking of the
second disassociation (extinction) and a third association mobile above them, but pictures and sounds, are unlikely.
(learning)phase, but too few subjects were able to finish all Regardless of whether or not it was an unusual associa-

2 months 4 months
Mean Joy Mean Joy
2 2
1.5 EXP
1.5
EXP
1 .....
T~:~:':c-~?;$':'?,~-.~
. .
......
. . . . ~.~i~::~..
1
..N..-
'
0.5 i' . 0.5
.' . CON w---- CON
.
0 0
B L1 EX L2 B L1 EX L2
Phase of Learning Phase of Learning

6 months 8 months
Mean Joy Mean Joy
2 2
EXP
1.5 1.5
1 ... EXP 1 Al.

0.5 _ 0.5
?r "? CON
O C 0
B L1 EX L2 B L1 EX L2
Phase of Learning Phase of Learning

Figure2. Facialexpressionof joy by conditionas a functionof age.


238 LEWIS

2 months 4 months
Mean of Anger Expressions Mean of Anger Expressions
2 3
1.5 2
I
1
0.5
0 0
B L1 EX L2 B L1 EX L2
Phases of Learning Phases of Learning

6 months 8 months
Mean of Anger Expressions Mean of Anger Expressions
5
4 4
3' 3
2 2
1 1
0 * 0
B L1 EX L2 B L1 EX L2
Phases of Learning Phases of Learning

Figure3. Facialexpressionof anger by conditionas a functionof age.

tion, the children demonstratedthat they could learn, and know the increasedeffortto producethe effect and the angry
learn quickly, that the arm movement would increase the face are related to the disassociation between action and
occurrenceof this event. Moreover, as they learnedthis re- outcome because as soon as the association is restored, the
sponse they showed an interested face as well as joy and anger disappears, arm-pullfrequency declines, and the joy
surprise. All infants regardless of age learned this associa- response returns.These children, even the 8-week-olds, ap-
tion. We cannot tell if they wished to learn this response, pearto be angrywhen they do not get whatthey expect to get.
because we made the connection or association for them by Moreover, like the angry adultswho bang the pay telephone
our electronic/mechanical contraption. However, once at- when they lose their quarter,the increasedaction directedat
tached to it, they did appearinterestedand happywhen they causing the event disappearsonce the event returns.1
made it work. They could stop their arm pulling if they so
desired, and so we can arguethatits continuationreflecteda Explanation: Associative-Learning Model
desire to do so. Alternatively, Piaget (1936/1952) would
argue that they neither desired to nor were interestedin en- Let me try to apply a simple associational or learning
gaging in this task. Once in the situation,the infantcontinued perspective. Fromsuch a mechanisticview, the positive rein-
in it because the event (ouroutcome of face and voice) led to forcementof A (the armpull) with B (the pictureand sound)
its continuation. It was not the child who desired but the should result in an increase in the rate of A. So far, so good.
outcome that controlled. This is how Piaget avoidedimpart- Now, the reinforcementstops. One might expect A to stop if
ing to the 8-week-old the mental propertyof intentionality.I it was controlled by B. A continues and we explain this
returnto this point shortly,but let me reportthe remainderof continuanceas a habit, or something learned. Exactly what
the study. this habitis, is unclear;nevertheless,we can assumethatit is
Having learned this association between arm pull and some altered structurein the nervous system of the infant.
event, the event suddenly stopped. The arm pull no longer Thus A should continue, and we assume that if there was no
resulted in this event. The infants' joy disappeared, they
became angry (some infants showed fear and sadness as
1 We also measuredthe activity of the armnot attachedto the string. This
well). Notice now whatoccurred.The responsethatled to the arm did not show an increase in arm pulling as a result of the frustration.
event did not work and so infants increased their response Because of this, it is difficultto arguethatthe increasein armpull was partof
level and at the same time appearedangry. Moreover, we a general response.
INTENTIONALITY
ANDCONSCIOUSNESS 239
furtherreward,A would eventuallydecreaseits frequency,at any developmental sequence to a brief period-the first 2
least back to the level we saw at the beginning of the study. months of life-or to promotinga nativistic view.
But A does not decrease in the absence of B. Quite to the
contrary,A increases in the absence of B. Now this becomes Intention, Desire, and Consciousness
more difficult to explain by simple associative learningbe-
cause a habit learnedmay continuebut not increase. In fact, It seems obvious thatourvery descriptionof whathappens
it is necessaryto now introducea new construct,namely,the in this experiment assumes an interpretiveand therefore a
increasein A is due to the absenceof B. This increasewe call particularstance (see Hirsch, 1967, for an analysis of this
frustration/angerand claim thatonce the associationis made problem.) Given that I assume a particularstance, let me
betweenA andB, the eliminationof B producesin the infanta restateit explicitly:All goal-directedsystems areintentional.
new structurewhich calls fortha new response. But now the How these goal-directed systems differ depends on the
notion of a simple habithas brokendown andwe need fartoo process underlyingthe goals. Forall goals, intentionis deter-
many additional structuresto support a simple associative mined by the affective states comprising part of the goal.
model. Even here in this simple model, we run into trouble However, some goals contain both affective and knowledge
because even associative models of this kind involve cogni- states (or knowledge systems; see Newell, 1982), and some
tion and, possibly, intention (Rescorla, 1987). even the additionof the cognitive state of consciousness or
objective self-awareness (see Duval & Wicklund, 1972).
Explanation: Sensorimotor Intelligence Such a propositionalsystem, if developed, enables us to
deal with such diverse questions as "Does a T-cell have
As I have indicated previously, I find no ready answer intention when it moves after a foreign protein?" "Does a
within a sensorimotor-intelligenceframework.Circularreac- leaf have intentionwhen it moves towardthe sun?" "Does an
tions, as we would expect at this age, are also like habits. 8-week-old intendto learn?"or "intendto try to get the lost
There is no mentalism here. Piaget and Inhelder (1969) objectsback?" "Do adultsintendto go to workeach day?"In
stated that, at Stage 2, each case, the answerto the questionis "yes," andwe need to
see how it might be the case.
anelementary"habit"is basedon a generalsensori-
motorschemewithinwhichthereis notyet, fromthe Levels of Meaning
subjects'pointof view, any differentiationbetween
meansandends.Theendin questionis attainedonly What is meant when we say that an infant, child, or adult
bya necessarysuccessionof movements whichleadto has intention?I believe that to understandthis problem we
it, withoutone'sbeingableto distinguish
eitheranend need to resortto a levels-of-meaning analysis. Humanorga-
pursuedfromthe startor meanschosenfromamong nisms may achieve a level differentfrom animalsand within
variouspossibleschemes.(p. 9) the human organism levels may differ as a function of on-
togeny. A levels analysis has been explored by Fischer
Nor is there yet any notion in the 2-month-oldof causality (1980) and Mounoud(1976), and I borrowfrom their analy-
because the infant is unaware of the spatial and physical sis. It may be the case that differentlevels of ability require
connection between actions. differentdegrees of experientialinteraction.The lowest lev-
We recognize that in Stage 3 or even Stage 4, a child who els of an ability may requirealmost no experience for their
has learned an association between a particularmeans and emergence or, in fact, may exist at or priorto birth. Within
end will continue that means in the absence of the ends. So, the organism, higherlevels of skill may requiremore experi-
for example, the 10-month-oldchild returnsto searchfor the ence or may be totally dependent on learning and culture.
object once found in location X even when the object is no Such a view of levels allows for both a nativisticandculture-
longer to be found in that location but is seen moving to influenced world view. It may be the case that, across orga-
location Y. By Stage 5, this is no longer the case; however, nisms, the same level of an ability may be achieved through
our 2-month-olds are not in Stage 4 or 5, so we remain differentmeans. As such, one should be wary of concluding
perplexedas to why by 2 monthssuch associationcontinues. that similar abilities across species have similar histories.
Even more puzzling is why in the absence of the end (B Werner(1961) consideredthis problemin his analysisof the
event), the response increases. equivalence of behaviors and called this the "constancy
Perhapsthe movement from reflexes to intentionsof the fallacy."
kind we observe here begins earlier or occurs more rapidly This level-of-ability approach touches on an issue that
than has been thought. Perhapsthey can be observed pro- remains a considerableproblemin development. This is the
vided we constructthe appropriateexperimentalconditions. notion of equivalenceof behaviorsacross age. One can often
Such a view allows for the maintenanceof the sensorimotor observe that a very young infant can perform some action
sequence, as specified by Piaget, but requiresthatwe change that, when performedat an older age, would be consideredto
the timing of these emerging skills. By changing the timing representsome underlyingcomplex structure.Take, for ex-
in this fashion, we give to the very young infant intelligent ample, the problem of imitation. Imitation is particularly
behavioralmost from birth, or at least soon after. Thus, the importantbecause the establishmentof a true imitative re-
behaviorof these 2-month-olds is not only centered around sponse heralds the developmentof an understandingof self
their own bodies, but involves attemptsto produceenviron- (Baldwin, 1894/1903).
mental actions, and so are secondarycircularreactions. The The newborninfantwill imitatecertainbody movements.
effect of this interpretationis to reduce the mixed model Forexample, a tongue protrusionby an adultwill producea
Piaget offers to a single one in which intentionsappearfrom tongue protrusionin the infant (Meltzoff & Moore, 1977).
the beginning. This also has the effect of either restricting Otherforms of imitativebehaviorhave been reported(Field,
240 LEWIS

Woodson, Greenberg,& Cohen, 1982). Althoughtheremay actionpatterns(see Darwin, 1872; Searle, 1984). The second
be some question as to the reliability of this behavior process is the knowledge-boundconnection to goals. Con-
(Anisfeld, 1988), such actions have been called "imitation." sider Newell's (1982) construct of a knowledge level. In
Imitation has a particularmeaning, usually inferring some every system, there is a hierarchyof knowledge. Even at the
intentionalityon the partof the imitator(Piaget, 1937/1954). lowest level of knowledge, the system has a body of knowl-
The finding that matching behavior exists in the newborn edge which includes knowledge of that system's goals. Each
constitutes a challenge for developmentaltheory. We could system acts using the knowledge at that particularlevel to
claim intentionalbehavior(like that in the adult)in the new- attainits particulargoals.
born. However, this is a nativistic explanationimplying that Newell would not need a thirdprocess, such as self-con-
there is no development in imitation nor in the process of sciousness (or knowledgeof knowledge of goals) because he
intention. Alternatively, the same behavior can be said to would include this as anotherlevel of the hierarchysystem.
have different meanings. We can say that the behavior at To the extent that I claim this thirdprocess has no different
Time 1 is called "X," whereas at Time 2 it is called "Y." status-it is the same substanceas the second-I have little
This solution has the effect of saying that behavior "Y," the disagreement. Nevertheless, I see this process as uniquely
more maturebehavior,does not exist until Time 2. Thus, for different from the other two. On what basis could such a
example, imitation in the newborn is called "matchingbe- claim rest? Perhapson no more thanthe generalclaim thata
havior,"whereasin the 8-month-old,it is called "imitation" metacognitionis not in the same knowledge level as another
(Jacobson, 1979). It is much like a stage-theorynotion be- cognition. Consider the case of memory. A memory of a
cause behavior "X" at Time 1 is not "Y," and it is not "Y"- memory is not of the same class as anothermemory,because
like. the systems (or levels) that supportit and the materialfrom
Another way of handling this problem of the meaning of which it is made may not be the same. We can think of this
behavior is to consider that both "X" and "Y" are func- difference in knowledge and in knowledge of knowledge
tionally similar but that they representdifferentlevels of the (what we wish to call consciousness) as emanating from
same meaning. In this case, "X" and "Y" behaviorcould be differentpartsof the brain(Jaynes, 1977). Duval and Wick-
called the same, recognizing that "X" and "Y" are at differ- lund (1972) made use of objective self-awareness (knowl-
ent levels. In the imitation example, both behaviors are edge of knowledge of goals), first by differentiatingit from
called "imitation,"but we recognize thatnewbornimitation subjective self-awareness (knowledge of goals) and then by
is at a level differentfrom 8-month-oldimitation. This posi- showing how it affects conformity.
tion requires that we consider that a particularability may Considerthe metaphor,often mentioned,thatthe property
have multiple levels. These levels are ordered and may be of wetness cannot be derived from the propertiesof oxygen
controlledby differentprocesses. Moreover,the level of the and hydrogen.In the same fashion, "consciousnessemerged
ability may be found both as a phylogenetic as well as an at some point in evolutionand in ontogeny, in a way under-
ontogenetic function. Thus, from a phylogenetic perspec- ivable from its constituentparts"(Jaynes, 1977, p. 12; italics
tive, a nonhumananimal, a rat, may imitate, but this imita- in original). Searle (1984) made a similarclaim in regardto
tion is at a level differentfrom that achieved by a 1-year-old propertiesof the mind. In discussing artificialintelligence,
child. Likewise, from an ontogenetic perspective, newborn he asserted that different types of knowledge come from
humansmay imitate, but newbornimitationis not at the same differenttypes of "machines."A humanbrainis not a com-
level as that achieved by a 2-year-old. Whetherthe levels puter and because "understandingis a propertythat comes
found phylogeneticallymatchthose foundontogeneticallyis from a certain kind of machine only, a machine like the
unknown, althoughthere is every reasonto assume thatthey human brain" (Gardner,1985, p. 174), Searle applied such
do. an analysis to intentionalityas well. He saw intentions as
Such a view of levels allows for the development of an caused by the specific propertiesof the humanbrain.
ability while at the same time allowing for its existence For this analysis, a claim is made that knowledge pro-
across the entire developmentalspan. Moreover, this devel- cesses or levels can be divided into two levels. In the first, a
opmental process may be seen both within and across spe- nonconscious knowledge level of high order exists, what
cies. The problemof equivalencyis especially relevantto the Duval and Wicklundhave called subjectiveself-awareness.
understandingof intention. Indeed, most human thinking occurs at this level. I am un-
Because the theory of intention across age requires the usually unaware(what I mean here by nonconscious) of the
assumptionof various levels of intention, I describethese in processes of my thought, althoughI may become aware of
some detail, with particularfocus on the last level to emerge. them through special effort. For example, there are times
The last level is based on the acquisition of self-con- when I desire to trace a set of associations. We all have
sciousness or objective self-awareness, and thereforetouch- experiencedthis ability to retraceour mentalsteps and arrive
es on the development of the self system. I cannot pursue at a satisfactoryrecognitionof what occurredwithoutat first
here the developmentof self per se so the interestedreaderis using our objective awareness. This subjective self-
referred to Fischer and Pipp (1984), Lewis (1990b), and awarenesshas knowledge andgoals andthereforeintentions.
Lewis and Brooks-Gunn(1979). For example, incomplete actions towardgoals set up inten-
tions (perhaps we might call these tensions if we used an-
Levels of Intention other language-see Freud, 1915/1959), which are then
satisfied.
I proposethreetypes of intentionalprocesses, all relatedto There is another knowledge level, objective self-
goals. The firstprocess is affectively boundand is connected awareness, when we areawareof ourthoughts,feelings, and
to goals as actions of desire. The emotionalpropertyof goals actions. Phenomenonologically,we appearto be watching
gives them action because emotions, in part, are themselves ourselves. This knowledge level often involves our emo-
INTENTIONALITYAND CONSCIOUSNESS 241

tional life, as when we become embarrassedby some action, 4. The levels view allows for the considerationof con-
thought, or feeling of the self (Lewis, 1990b; Lewis, Sul- sciousness as the highest level of intentionality.
livan, Stanger,& Weiss, 1989). Blushing, as Darwin (1872)
first noted, is the most humanof all emotional expressions. Table 1 presentsin some abbreviatedform the five levels
He stated (p. 325) that "it is not the simple act of reflecting we consider. It includes the names assigned to the levels,
on our own appearance,but the thinkingwhat othersthinkof their relation to sensorimotorintelligence, and the level of
us, which excites a blush." It seems clear thatthis knowledge emotion (desire) presentat each level. A full explanationof
level (objective self-awareness)is in some way uniqueto the this table is presentedwithin the text.
otherform of knowing (subjectiveself-awareness)andthatit
is reasonableto assume thata metaknowledge(knowledgeof Level I
knowledge) has differential effects vis-a-vis the levels of
intentionwe seek to explain. Intentionsassociated with this The first level of intentionality,which correspondsto the
level of knowledge phenomenonologicallyappearquite dif- reflex period in sensorimotordevelopment, we call, necessi-
ferentfrom intentionsat anotherlevel. Forexample, I intend ty. The action, although intentional, is both predicatedand
to study a particularproblem. This is a direct operationon a prescribedby survival. This intentionalitycan apply to cells
plan which is quite different from the intentions following and infants and even in some aspects to maturehumans. For
this, which proceed from the plan of which I have no example, breathingor heartaction are examples of this first
awareness (although I might), and which act, as it seems, level. Thatautomaticor involuntaryprocesses areintentional
independentof my generationof otherplans or intentions.It appearsto violate our common-sense understanding.At this
is a familiarexperience to most of us to be thinkingabout a level, intentions are all derived through adaptive functions
problem-a direct intention-when another intention relatedto survival. Commercewith the externalenvironment
arises, for example, a desire for a glass of water. Intentional is all in the service of the internaladaptivefunctionwithinthe
action on this desire does not prevent intentions at other infant. These adaptive functions necessitate no response
levels from proceeding. This possibility appearsto occur. In from the environment. They act on the environmentbut, in
fact, it is one way of conceiving of intrapsychic conflict general, are not acted on. Considerthe early capacitiesof the
which was given so much attentionin Freud'stripartiteno- infant. These response systems operate so that the infant
tion of the psyche (Freud, 1923/1961). intendsto suck when an object is placed in its mouth, to blink
In this overview, three processes have been identifiedthat its eyes when an object expands in its visual space, and to
supportthe idea of levels of intention. These are (a) adaptive grasp an object when it is placed in its palm. They grow out
intention, (b) knowledge intention (or subjective self- of the child (are part of its biological heritage) and may be
awareness), and (c) conscious intention (or objective self- accommodatedto by the environmentbut are not createdby
awareness).These levels have an ontogeneticcourseandmay them. These behaviors are goal-directed, internal, self-sus-
have a phylogenetic one as well. Most importantfrom our taining, and adaptive.
point of view, these levels, once developed, do not become The organismacts from goals built into its systems. Recall
transformed as new levels are reached. For each adult that for us, as for others, goals contain desires. Survival or
human, therefore,these threelevels coexist and at times may adaptationis the intendedgoal (or cause of action). Desires at
even conflict. Moreover, a particularlevel may be involved this level are global emotions, they are simply positive or
for one set of goals and anotherfor anotherset. The regula- negative (approach vs. avoid) states which are associated
tion of our immunocompetencesystem is usually left to the with action to satisfy a goal. At this level, necessity, inten-
first level or perhapseven to the second. Yet even here the tion, and emotion are equivalent. Organismsbehave so as to
third or conscious level may be employed. There are those have positive emotions and to avoid negative ones. The ac-
who subscribe to the view that deliberate attempts (con- tions are built into the emotions (see, e.g., Darwin, 1872,
scious) to make yourself happy, or to reduce stress, will and Plutchik, 1980, for analyses of emotions as motives and
affect the immune system. goals). Intentionality is a part of the system because the
Applying these three processes in the searchfor levels has knowledge system is controlled by the emotions. For the
proved useful, as have Dennett's (1987) work on intentional infant, this means that its action is intentionalbecause it is
stance, Fodor's (1981a, 1981b) work on propositional at- built into a system. It is controlledby desire. Behaviorsoccur
titudes, and Mitchell's (1987) analysis of deception. I have which cause pleasure and which avoid pain (as defined by
borrowedfrom each of these to outline a levels analysis. In biological information).
doing so, I have strayed from the theory of sensorimotor Necessity is synonymouswith adaptation.Puredesires are
intelligence in several respects. goals; there is no mental representationexcept as desires can
be represented(see Zajonc, 1980). Perhapsit is describedby
the statement "doing what feels good and avoiding what
1. Intentionalityas a propertyof all goal-directedsystems feels bad." Feeling good and bad are the only two emotional
is assumed, thus avoiding the problem of a mixed states attachedto desire at this level. The desire system itself
model. remainsundifferentiated.The differentiationof this system is
2. Development is not transformational,at least in this importantfor our understandingof the intentional system
regard. Such a view allows for the operation of all because as desires (goals) differentiate,the intentionalsys-
achieved levels throughoutdevelopment. This avoids tem expands. A more complete discussion of the develop-
the inherentproblemof regression, which is not readily ment of the desire system can be found in Bridges (1932),
handled by transformationaltheories. Sroufe (1979) for the earliest emotions, and Lewis et al.
3. Emotionalproperties(the desire/goal featureto all rep- (1989) for the self-conscious emotions.
resentations)underlie all intentions. Do we wish to claim that this lowest level is intentional
242 LEWIS

Table 1. Levels of Intention

Knowledge
Level Nature of Intention (Causality) Desire

I Necessity Reflexes 1. Global emotions:


(Innate) approach(+),
avoid (-).
2. Internaladaptive
functions.

II InteractiveNecessity Circular 1. Simple emotional


Reactions differentiation.
2. Internaland external
adaptive functions.

III Action Intent Complex 1. Differentiatedemotions.


Circular 2. Internaland external
Reactions adaptive functions.
3. Simple representations
and adaptive
anticipation.

IV Divergent Intent Insight 1. Differentiatedprimary


emotions.
2. Internal,external, and
anticipatedadaptive
functions.
3. Representations,
categorization, and
manipulation.

V Conscious Intent Self- 1. Complete primary


Awareness emotions.
2. Secondary emotions.
3. Internal,external, or
anticipatedadaptive
functions.
4. Representations,
categorization, and
manipulation.
5. Metarepresentationsand
emotions.

and what benefit for a theory of intentionis derivedby doing similarto the first level in the sense that intentionis directly
so? Some mightbe uncomfortablewith includingsuch action tied to emotion (desire), thus the term, interactivenecessity.
within this realm, and might argue that it is not useful to The adaptivefunctionsnow requireenvironmentalinputand
claim any form of mentalityfor such noncognitive (or intel- as such the intentionsare interactive.It could be arguedthat
ligent) action. We make the claim for several reasons. To all an organism's adaptivefunctions are interactive,that is,
begin with, we assume intentionin all goal-directedsystems. are related to the environment. Although this may be so,
Moreover, we have no difficulty in ascribing intention to there appear to be differences that warrantthe distinction.
systems operatingon the level of adaptationand containing The smile of the infantexemplifies this. Until 3 monthsor so
emotions (or desires.) This propertyis not given to reflexive of life, the infant's smile appearsrelatedto internaladaptive
behaviorby Piaget, althoughhere too, simple schemataare functions. The infant smiles often when in REM sleep
proposed which have some action component. Second, al- (Wolff, 1963). The infant may or may not smile to a human
though we might agree with a mechanistic(behavioral)view face. At this first level, the infant's smile does not appearto
that our belief in intentionalitydoes not affect the observa- be related to exogenous factors but rather to endogenous
tion, we are still troubledby any mixed-model view which adaptiveactivities. By Level II, the infant'ssmile is no long-
first does not, but then does, requireintentionality. er partof the internaladaptivegoals. The infantnow smiles
to environmentalevents.
Level II At Level II, infantintentionsarerelatedto externalevents,
their presence or absence. Forthe smile, infantssmile in the
The second level of intention (and all levels thereafter) presence of a human-face-likeevent and do not smile in the
involve the organism'sinteractionwith the environment.It is absence of this event. The intentionalitycan be describedin
INTENTIONALITYAND CONSCIOUSNESS 243

the general form, as given an environmentalevent (E1), the environment,thatnow becomes the cause of the goal. Antic-
child's adaptive goal is to smile (action a). Here, unlike ipatedemotion requiresrepresentationalabilityandan active
Level I, the adaptivegoal of smiling is tied to the presenceor memory capacity which infants of this age now possess
absence of an event. Notice that smiling is still tied to an (Schaeffer, 1974). For example, if a face produces a goal
adaptivefunction, but the smile only occurs underspecified (desire), the intentionto smile, then it is possible for a goal
conditions. (desire) to intendto producea face. This might be analogous
Because of its interactive nature, the observer often as- to superstitiousbehavior because there is no disassociation
sumes the infant's smile at this age is intentional.We do so between desire and the intended action.
because the action is specific, that is, based on select en- There are severaldifficultieshere, not the least of which is
vironmentalevents. Parentshave been known to report"She how to understandthe association between environmental
likes you because she smiles at you." Thatparents(or others) context, goal (desire) activation, and action. The latter as-
assume a more adultlikeintentionalsystem may have impor- pect gives little troublesince desires (emotions)have built-in
tant implications for its development(Kaye, 1982). action patterns. The difficulty becomes how to reverse the
Desires are specific goals which are activated by events directionof effect. In Level II, the environmentactivatesthe
outside the organism. The desires, however, are located goal. By Level III, goals have become independentof en-
within the infant. As emotional differentiationoccurs (the vironmentalcontext for their activation. Moreover,because
process for this is unclear,but may be relatedto maturational of past connections, goals (desires) have associated with
factors), more goals are produced than the initial ones of them action patternsrelative to specific environmentalcon-
approachand avoidance. At this level, it is the environment texts. The only possible explanationfor the independenceof
that still activatesthe organism'sdesire. Because of the con- goals is (a) their past association with numerous specific
textualdifferencesin the environment,andbecause thereare environmentalcontexts and (b) the greaterrepresentational
more emotions, more goals (more intentions) are differ- capacity of the child. Because goals (desires) have action
entiated. associated with them and have been repeatedly elicited
The conditional natureof the response vis-a-vis its goal across a variety of contexts, and because they can now be
allows for simple conditioning and learning. This occurs remembered,the infantis now capableof manipulatingthese
because the conditions for goal-directedbehaviorare estab- representationsand, as such, reversing the order. There is
lished by the environment;thatis, "I have a desireto smile to much in Piaget's thinking (also Baldwin, 1894/1903) that
a facelike event." If the event occurs, then my goal is acti- informsour discussion aboutLevels II and III. Forexample,
vated. Notice these goals still belong to the originaladaptive our Level II is much like the descriptionof accommodation.
functions. The change has been in the power of the environ- To paraphrase,in accommodationthe movementcan be con-
ment to activatethem. Such a simple level of interactioncan ceived, in a sense, as going from the object towardthe sub-
also be seen in adults;for example, I am likely to yawn when ject and compelling the subjectto adaptto the object, where-
anotherperson does so more than when anotherpersondoes as in assimilation (Level III), the movement proceeds from
not do so. My yawning is intentionalalthoughthe activation the subject toward the object. "In the beginning, an infant
of the goal is determinedby the conditions of the environ- only accommodatesitself to thingswhen it is forcedby them,
ment (as well as the conditions of the person or organism). whereas at the outset the infant tries to assimilate the real,
Associations or representationsare createdthroughthe con- impelled by an invincible and vital tendency" (Piaget,
tinued activation of goals (actions) with particularenviron- 1936/1952, p. 275). We view this tendency as necessity,
mental events. For example, faces elicit the desire to smile. arising from adaptivefunctions.
These representationslead to the next level. Clearly,Level III is the point where intentionalityassumes
its more maturefunction: In part, this is due to the further
Level III differentiationof desires (emotions) andthe increasedcapac-
ity to represent.Forthe first time, then, goals have represen-
This level of intentionality,called action intent, remains tationsthat can be manipulated.The ability to activategoals
the most problematic.Up until now, goals are locatedwithin (desires) with associated action-in-the-worldallows for ac-
the organismas partof the infant'sadaptivefunctions.These tion on instead of only action when. Because of its close
functions specify goals containing action patternsand these association with Level II, it is not clear that Level III repre-
goals are internal;that is, they require only the minimum sents a new level. Rather,it may be only the elaborationof
environmentalimpact. The second level is also characterized what occurs in Level II. We chose to differentiateit, pri-
by goals within the organism (again a part of its adaptive marilybecause of the increasedmemorycapacityassociated
function);however, these goals areactivatedby environmen- with this period.
tal factors as well as internalones. Moreover, the goals re-
main relatively undifferentiated, in part because desires
Level IV
(emotions) remain global. As desires differentiate, goals,
too, increase.
At some point, representationsof the goals and desires, Until this point, intentionswere producedby environmen-
and the environmentalcontexts associated with them, are tal contexts through the activation of goals. Moreover,
established.This is likely a functionof the infant'sincreased throughrepresentationalcapacity, goals and their intentions
memory capacity. These representationsare not necessarily became the activatorsof environmentalcontexts. These rep-
directional. That is, the association itself can give rise to resentationalskills enable the anticipationof goals (desires).
desires and goals which produce environmental contexts Thus, actual and anticipatedgoals (desires) supportactions.
with which they have been associated. In fact, it may be the It must be rememberedthat intentionis built into goals and
case thatit is the anticipateddesire, which was causedby the thataction is associatedwith desire. At first, it is desirealone
244 LEWIS

thatdrives goals; however, with the developmentof memory, Conscious intent, like all levels, is supportedby emotion.
representationsalong with desire now drive goals. Nevertheless, cognitive capacity (present at Level IV) and
Now, at the level of diverse intent, the goals have become the new emergent structure,objective self-awareness, now
more differentiated.In fact, by this level all primarydesires become the materialof desire. Herewe meanthatto be aware
(emotions) and their goals have become differentiated(see of desires, as opposed to having them but not being aware,
Bridges, 1932; Lewis & Michalson, 1983). Moreover, ac- becomes a new kind of desire. This supportsa new kind of
tions associated with goals themselves have differentially intention. Now, and for the first time, intentionscan be dis-
developed. The infanthas many more motorpatternsandthe associatedfrom the directeffects of emotion. Thinkingabout
ability to allow for a greater diversity of action. Finally, the self, independentof direct emotion, can sustain inten-
representationshave become complex and abstract(Fischer, tions. In a sense, my awareness that "I wish for .. ."-a
1980). Representationsand desires with their actions repre- cognitive act-creates its own emotion. The degreeto which
sent goals and, increasingly,anticipatedrepresentationsand I am not self-aware (not conscious) that I want somethingis
desires serve the role formerly held by actual desires. In the degree to which emotions control my action ratherthan
some sense, desires, and thereforegoals, have become less the other levels of knowledge. Such a view has been ad-
importantthan representations,in part, because representa- dressed in consideringthe topic of primaryversus secondary
tion of desire is possible. thinking (Freud, 1915/1959) and is consistent with the as-
All these developments give rise to the ability to utilize sumption that unconscious thought processes are not the
alternatives. Because divergence is now easy, representa- equivalent of conscious thoughtprocesses. For me, the rea-
tions are more readily manipulatedand particulargoals can son has more to do with the level of cognition and the role of
be driven by a variety of desires. That is why we call this emotion associated with unawareintentions.
fourthlevel, diverse intent. By the time this level is reached, The difference between Levels IV and V has to do with
the possibility of diverse goals (intentions),ease of represen- objective awarenessof intentions.Forexample, I believe my
tation, and differentiateddesires all permitvariationin how cat has Level IV but not Level V intentions. I see my cat
particulargoals are reached. At this level, we find inten- coming into my study. He wants to jump on my lap which is
tionality attributedby others because the definitionof inten- covered by books. To landon my lap underthis condition, he
tion is satisfied by the multiplegoals andthe multipleactions needs to jump up on my desk and walk carefully from there
available to achieve each goal. Organisms choose pos- onto my lap. This is not his usual way because he usually
sibilities from a wide arrayby anticipating(a) which goal is jumps directlyfrom the floor onto my lap. Thus, he altershis
more desirableand (b) which action is most likely to lead to means to get to the goal. This appearsto satisfy at least some
this desire. At this level, organisms choose both the goals aspects of Levels III and IV. However, to the best of our
and have the flexibility, through trial and erroror even in- knowledge, the cat does not have objective awarenessof his
sight, to select those actions most likely to lead successfully intentions. Thus, the action can be distinguished from ac-
to the goal. Actions at this level include thoughts about ac- tions by human adults who possess, at least at times, this
tions (plans) as well as actions themselves-thus, the pos- awareness.
sibility of solution to problems without observable action The differencebetween Levels IV andV can be seen in the
exists. behaviorof animals (or even infants). If intentionalsystems
Once the control featureof emotion emerges as in Level are organized into a series of increasingly more complex
III, a new desire is created, that of control itself. This new structures,then so, too, must be desires. Elsewherewe have
desire serves as a type of masteremotion for all otherdesires shown thatthereis a significantchange in emotionallife with
because it naturallyseeks variationand diversityof action to the introductionof objective awareness or consciousness.
goals. The developmentof memoryhas enabledthe develop- When consciousness emerges, two classes of emotion are
ment of categorization.Categorizationitself empowers new possible: (a) those we have called self-conscious emotions,
combinations, both of goals and their actions/desires. which include embarrassment,empathy, and envy; and (b)
those we have called self-conscious evaluative emotions,
Level V which include pride, shame, guilt, satisfaction/competency
(Lewis, 1990a, in press;Lewis et al., 1989). These emotions
The final level of intentionalityinvolves fourth-levelstatus become the basis of a new set of intentionsbecause humans
plus one furtherelaboration-consciousness (or objective act so as to avoid shame andguilt andto achieveotherprefer-
awareness of intention)-and is called conscious intent. able emotions, for example, pride.
Here the organism not only has the flexibility of abstract
representationof actionsandgoals, but is now aware(the self I have describedthe five levels of intentionand suggested
turned toward the self) of these goals and actions. This that they have both ontogenetic as well as phylogenetic
awarenessor consciousness allows the child to considerthat usefulness. However, I have not explicated how the child
it has divergentintents. Consciousness is a metaprocess;for moves from level to level. I have made suggestions about
example, the mature human, unlike other organisms, has whatprocesses areinvolved in moving to each next level, but
memory of its memory. This is capturedby the recursive these have not been as explicit as is possible. Two major
statement "I rememberthat I wished to do something." At developmental processes appearto underlie the movement
this point, intentions,which were flexible vis-a-vis goals and throughlevels: cognitive and emotional. The first has to do
actions, are now viewed by the child itself. By this act of with the change in cognitive structure, in particular,the
consciousness, intentionalityitself becomes availableto con- change in representationalability and the growthof memory
sider. As such, intentionalityitself is changed. I can now say, capacity. Both these capacities exhibit qualitativeas well as
"I am awarethatI intendedto do X, butthatis reallynot what quantitativechanges which are influenced by the genetic
I wish to do." The manipulationof intentionsthemselves is code of the species as well as the child's commerce with its
one propertyof objective awarenessor consciousness. environment.
INTENTIONALITYAND CONSCIOUSNESS 245

The second process thatundergoeschange and thatunder- The problem, afteraccepting such a world view, is how to
lies the movement throughlevels is the emotional growthof apply it to the developmentalissues of growth, change, and
the child. As I have tried to make clear, I see emotion and transformation.A possible solution might be found in the
desire as similarandthe basis of intentions.It is clear thatthe way others have treated machines or animals vis-a-vis hu-
emotions differentiate themselves. At first they appear as mans in termsof these structures.Thatis, if we arewilling to
undifferentiatedpositive and negative affects, and, within give these same structuresto organisms other than adult
the first 2 years of life, they differentiateinto the complex humans, we could give them to childrenand even infants. A
arrayseen in the adultof the species (Bridges, 1932;Lewis & reading of the literaturesuggests both a "yes" and "no"
Michalson, 1983). This differentiation, like that for cog- answer to this question. Dennett (1987) and Newell (1982),
nitive capacities, also involves both qualitative and quan- for example, seemed willing to assign such structuresto
titative changes. Here, too, both the genetic code of the others, whereas Searle (1984) would restrict them to hu-
species and the child's social-interactivelevel contributeto mans. If we distributethem to others than adulthumansand
the development and change of this capacity. give them similar form, then we run into other problems
We might ask, "Whythen shouldthese developingcapaci- concerning development. That is, by giving infants and
ties lead to changes in the level of intention?"The answer young children these structureswe cease to be developmen-
appearsobvious at first;on reflection, it may not be so clear. talists. The position results in no developmentto study and
That differentlevels of intentionrequiredifferentcapacities, no model of organism-environment interactionto explain
coupled with the fact that capacities change over time, does the basis of growthandtransformation.If, on the otherhand,
not necessarily require that levels of intention change. If, we give these structuresto others but assign them different
however, we returnto ouroriginalassumption-namely, that levels of the material,thenwe still have a problemto explore.
goal-directedsystems are intentional-our problemis made Another solution to the developmental problem, one
easier. Different capacities lead to different goals. These which Piaget seems to have adoptedis to allow the infantto
goals differ both in quantity and quality. Thus, changing create these structuresthrough the developmental process
capacities lead to changing goals and therefore changing itself. Here, at least, at the very beginning of the process of
intentions. growth, the structuresdo not exist-they areacquiredvia the
developmental process itself. The problem here, as some
How the Model Informs have suggested, arises in the difficulty of going from no
structureto structure.Discardingthis problem, the theoryof
From this discussion, it appears possible to distinguish sensorimotorintelligence is an attemptto deal with the devel-
between differenttypes of intentionalactions in an organized opmental problem and one to which we owe a considerable
fashion. Notice that such an organizationfits not only inten- debt.
tionality but causality as well. Moreover,otheractions, such In attemptingto understandthis problem, I have resorted
as deception, arereadilyincorporated,allowing us to consid- to a position thatconnects thoughtto actionthroughemotion.
er a range of such possible actions-from simple deceptions This attemptto give thought action is not new; more than a
such as an adaptive change in coloration of an animal, an century ago, Darwin (1872) saw emotion as having such a
action by which an organism hides from its predator(Level property. More contemporarytheories likewise solve this
I), to the complex deceptions of human adults (Level V), in problemoften by introducingemotion throughthe use of the
which the personis awareof his or her action andthe effect of term "desire." It seems a reasonablesolution. In this article,
that action on another (Mitchell, 1987). Because levels of I have arguedthat all goal-directed systems are intentional.
intentionalityare so closely linked to causality, the discus- Moreover,our theory suggests, along with others, thatgoals
sion also becomes relevantfor the acquisitionof knowledge. themselves possess emotional features which include other
Indeed, for Piaget, this was an importantstructuralfeatureof forms of knowing such as how to achieve the goal. Specifi-
his epistemological system. cally, for us, goals are ideas with emotion. They can be
The intentionalityquestion has been raised here to deter- generatedby the biology (as in Level I of our model) as well
mine whether levels of intention (or the knowledge of as by the thought of the organism (Level IV), or by self-
causality) can be used to explain the infant's action. Our awareness (Level V). The developmental process is this
problem, one shared with others, is to understand how change.
knowledge, at any level, can lead to action. I use here the Recall that in our study, very young infants, long before
term "desire" as the mediatinglink, giving to this construct they should be able, appearto act intentionally.First, they
the properties usually assigned to such terms as motives, persist in pulling a string to make something happen, and
drives, and innate releasing mechanisms. I do this to con- then they become angry when what works once, does not
form to the addedpropositionthatall intentionshave desires. work. Firstconsiderthe initial learningof the armpull. From
This definition allows us to move from intentionsto actions ouranalysis, this behaviorinvolves both interactivenecessity
by the expedient of attributingto intentions a motivating (Level II) and action intent (Level III). To begin with, the
power, a quality not unreasonable(Searle, 1984). I arguefor accidentalappearanceof the event (pictures and sound)pro-
a hierarchicalorganizationof intentions and thus for a hier- duces the desire. That is, the infant does not startoff with a
archical orderingof desire. schema of tryingto do something, but if it does, the things it
As I have tried to demonstrate,a mechanisticworld view, tries to do, do not include pulling a string to get a pleasant
one that does not consider mental actions such as intentions, event. Rather, the desire grows from the event's occurring
is capableof explainingthe researchfindingsI have reported. contingenton the infant'sactivity.Initially,the occurrenceof
Indeed, one strength in the mechanistic model is its sim- the event produces general activity, but within a shorttime,
plicity. Nevertheless, I reject this view in favor of one in the desire becomes focused in the arm movement which
which representations,plans, anddesires exist andconstitute results in the event. At this point, the desire, which we have
the units of human life. described as interest and some pleasure, changes into sur-
246 LEWIS

prise and joy as the infant discovers that the arm itself can Dennett, D. C. (1987). The intentional stance. Cambridge, MA: MIT
cause the event. Lewis et al. (1984) describedthis change in Press/Bradford.
desire and we reportedsimilardatato show the effect. It is at Dodge, K. A., Murphy,R. R., & Buchsbaum,K. (1984). The assessment
of intention-cuedetection skills in children:Implicationfor develop-
this point that interactive necessity becomes action intent. mental psychopathology.Child Development, 55, 163-173.
Now the infant has discovered that the desire can be pro- Dunn, J. (1988). The beginnings of social understanding. Cambridge,
ducedby itself, a new knowledge level has been reached.It is MA: HarvardUniversity Press.
the appearanceof surpriseandjoy that marksthis transition Duval, S., & Wicklund, R. A. (1972). A theory of objective self
awareness. New York:Academic.
of levels.
Fagot, B. I. (1973). Sex-relatedstereotypingof toddlers'behaviors.Devel-
Furthersupportfor the action-intentlevel comes from the opmental Psychology, 9, 429.
next phase of the study in which the cessation of the event, Field, T. M., Woodson, R., Greenberg,R., & Cohen, 0. (1982). Discrim-
ination and imitation of facial expression by neonates. Science, 218,
something imposed on the infant, results first in interactive 179-181.
necessity; that is, cessation of the event leads to a new re- Fischer, K. (1980). A theory of cognitive development:The control and
sponse (anger),somethingbuilt into the biology of the infant. constructionof hierarchiesof skills. Psychological Review, 87, 477-
This change in levels back to interactivenecessity is disrup- 531.
tive for some children and they leave the study due to fuss- Fischer, K. W., & Pipp, S. L. (1984). Development of the structuresof
iness. For the great majority of others, this level change unconscious thought. In K. Bowers & D. Meichenbaum(Eds.), The
unconscious reconsidered (pp. 88-148). New York:Wiley.
causes a reinstatement(althoughexaggerated)of Level III, Fodor, J. A. (1975). The language of thought. New York:Crowell.
the action intent-hence, the increased arm-pullbehavior. Fodor,J. A. (1981a). The mind-body problem. ScientificAmerican,244,
Clearly,this is only one of manypossible explanationsand 114-123.
our solution to the problemof the developmentof intention Fodor, J. A. (1981b). Representations:Philosophical essays on thefoun-
dation of cognitive science. Cambridge,MA: MIT Press.
raises more questions than it answers. Nonetheless, it ac-
Freud, S. (1959). Instincts and their vicissitudes. In J. Riviere (Trans.),
counts for the dataandprovidesa basis for furtherconsidera- Collected papers (Vol. IV, pp. 60-83). New York: Basic. (Original
tion, and, as such, may have heuristicvalue. Whatis clear is work published 1915)
thatthe problemof intentionalityis centralto the issue of the Freud, S. (1961). The ego and the id. In J. Strachey(Ed. & Trans.), The
standard edition of the complete psychological works of Sigmund
development of the infant's action in the world. Given our Freud(Vol. 19, pp. 3-66). London:Hogarth.(Originalworkpublished
belief in intentionality of adults and the possible inten-
1923)
tionality of animals, how we understandthe intentionalityof Gardner,H. (1985). The mind's new science. New York:Basic.
infants and its developmentremains an importanttopic. Goldberg, S., & Lewis, M. (1969). Play behaviorin the year old infant:
Early sex differences. Child Development, 40, 21-31.
Griffin, D. R. (1984). Animalthinking.Cambridge,MA: HarvardUniver-
Note sity Press.
Hebb, D. 0. (1946). On the nature of fear. Psychological Review, 53,
Michael Lewis, Institutefor the Study of Child Develop- 256-276.
Hebb, D. 0. (1949). The organization of behavior. New York: Wiley.
ment, RobertWood JohnsonMedical School, University of Hirsch, E. D. (1967). Validity in interpretation. New Haven, CT: Yale
Medicine and Dentistry of New Jersey, New Brunswick, University Press.
NJ 08903-0019. Isen, A. A. (1984). Towardunderstandingthe role of affect in cognition. In
R. Wyler & T. Srule (Eds.), Handbookof social cognition (pp. 179-
235). Hillsdale, NJ: Lawrence ErlbaumAssociates, Inc.
References Izard,C. E. (1979). TheMaximallyDiscriminativeFacialMovementCod-
ing System(MAX).Newark:Universityof Delaware, InstructionalRe-
Alessandri, S. M., Sullivan, M. W., & Lewis, M. (in press). Violation of sources Center.
expectancy and frustration in early infancy. Developmental Jacobson, S. W. (1979). Matching behavior in the young infant. Child
Psychology. Development, 50, 425-430.
Anisfeld, M. (1988). The question of early imitationoffacial gestures:An Jaynes, J. (1977). The origins of consciousness in the breakdownof the
historical-theoretical perspective and a meta-analysis. Manuscript bicameral mind. Boston: Houghton Mifflin.
submittedfor publication. Kagan, J. (1974). Discrepancy, temperament,and infant distress. In M.
Arnold, M. B. (1960). Emotionand personality (Vols. 1 & 2). New York: Lewis & L. Rosenblum(Eds.), Theorigins offear (pp. 229-248). New
Columbia University Press. York: Wiley.
Arnold, M. B. (1970). Brain function in emotion: A phenomenological Kaye, K. (1982). The mentaland social life of babies:How parents create
analysis. In P. Black (Ed.), Physiological correlates of emotion (pp. persons. Chicago: Universityof Chicago Press.
261-286). New York:Academic. Keasey,C. B. (1977). Children'sdeveloping awarenessandusage of inten-
Baldwin, J. M. (1903). Mental developmentin the child and the race (2nd tionality and motives. In C. B. Keasey (Ed.), NebraskaSymposiumon
ed.). New York:Macmillan. (Original work published 1894) Motivation (Vol. 25, pp. 219-260). Lincoln: University of Nebraska
Berlyne, D. E. (1960). Conflict, arousal and curiosity. New York: Press.
McGraw-Hill. Lazarus, R. S. (1982). Thoughts on the relations between emotion and
Berndt, T. J., & Berndt, E. G. (1975). Children's use of motives and cognition. AmericanPsychologist, 37, 1019-1024.
intentionalityin person perceptionand moraljudgment. ChildDevel- Leslie, A. M., & Keeble, S. (1987). Do six month old infants perceive
opment, 46, 904-912. causality? Cognition, 25, 265-274.
Bower, G. H. (1981). Mood and memory. American Psychologist, 36, Lewis, M. (1990a). Social knowledge and social development. Merrill-
129-148. Palmer Quarterly, 36, 93-116.
Bridges, K. M. B. (1932). Emotionaldevelopmentin early infancy.Child Lewis, M. (1990b). Thinking and feeling-The elephant'stail. In C. A.
Development, 3, 324-334. Maher, M. Schwebel, & N. S. Fagley (Eds.), Thinkingand problem
Buffery, A. W. H., & Gray, J. A. (1972). Sex differences in the develop- solving in the developmentalprocess: Internationalperspectives (pp.
ment of spatial and linguistic skills. In C. Ounsted & D. C. Taylor 89-110). Hillsdale, NJ: LawrenceErlbaumAssociates, Inc.
(Eds.), Genderdifferences:Theirontogenyand significance (pp. 123- Lewis, M. (in press). Self knowledge and social developmentin early life.
157). Baltimore:Williams & Wilkins. In L. A. Pervin (Ed.), Handbookof personality theoryand research.
D'Andrade, R. (1981). The culturalpartof cognition. CognitiveScience, New York:Guilford.
5, 179-195. Lewis, M., Alessandri, S. M., & Sullivan, M. W. (in press). Violationof
Darwin, C. (1872). The expression of the emotions in man and animals. expectancy, loss of control, and anger expressions in young infants.
London: Murray. Developmental Psychology.
ANDCONSCIOUSNESS
INTENTIONALITY 247

Lewis, M., & Brooks-Gunn,J. (1979). Social cognition and the acquisi- Piaget, J. (1954). The construction of reality in the child (M. Cook,
tion of self. New York:Plenum. Trans.). New York:Basic. (Originalwork published 1937)
Lewis, M., & Goldberg, S. (1969). Perceptual-cognitivedevelopmentin Piaget, J., & Inhelder,B. (1969). Thepsychology of the child. New York:
infancy: A generalized expectancy model as a function of mother- Basic.
infant interaction.Merrill-Palmer Quarterly, 15, 81-100. Plutchik, R. (1980). A general psychoevolutionarytheory of emotion. In
Lewis, M., & Michalson, L. (1983). Children's emotions and moods. R. Plutchik & H. Kellerman(Eds.), Emotion: Theory, research and
New York:Plenum. experience (pp. 3-33). New York:Academic.
Lewis, M., Sullivan, M., & Brooks-Gunn,J. (1985). Emotionalbehavior Reese, H., & Overton, W. (1970). Models of developmentand theoriesof
during the learning contingency in early infancy. British Journal of development.In L. R. Goulet & P. B. Baltes (Eds.), Life span develop-
DevelopmentalPsychology, 3, 307-316. ment psychology: Research and theory (pp. 116-145). New York:
Lewis, M., Sullivan, M. W., & Michalson, L. (1984). The cognitive- Academic.
emotional fugue. In C. E. Izard, J. Kagan, & R. Zajonc (Eds.), Emo- Rescorla, R. A. (1987). A Pavlovian analysis of goal-directedbehavior.
tions, cognition, and behavior (pp. 264-288). New York:Cambridge AmericanPsychologist, 42(2), 119-129.
University Press. Rheingold, H. L., & Cook, K. V. (1975). The content of boys' and girls'
Lewis, M., Sullivan, M. W., Stanger,C., & Weiss, M. (1989). Self-devel- rooms as an index of parents'behavior.Child Development,46, 459-
opment and self-conscious emotions. Child Development, 60, 146- 563.
156. Rorty, R. (1989). Contingency, irony and solidarity. Cambridge, MA:
Lipsitt, L. (Ed.) (1976). Developmentalpsychobiology: The significance CambridgeUniversity Press.
of infancy. Hillsdale, NJ: LawrenceErlbaumAssociates, Inc. Schaeffer,H. R. (1974). Cognitive componentsof the infant'sresponseto
Meltzoff, A. N., & Moore, M. K. (1977). Imitationof facial and manual strangers.In M. Lewis & L. A. Rosenblum(Eds.), The origins offear
gestures by humanneonates. Science, 198, 75-78. (pp. 11-24). New York:Wiley.
Michotte, H. (1963). Theperception of causality. London:Methuen. Searle, J. (1984). Minds, brains and science. Cambridge, MA: Harvard
Mitchell, R. W. (1987). A frameworkfor discussing deception. In R. W. University Press.
Mitchell & N. S. Thompson(Eds.), Deception:Perspectiveson human Siminov, P. V. (1970). The informationtheory of emotion. In M. B. Ar-
and non-humandeceit. Albany: State University of New YorkPress. nold (Ed.), Feelings and emotions (pp. 145-150). New York:
Money, J., & Ehrhardt,A. (1972). Man and woman, boy and girl. Bal- Academic.
timore:Johns Hopkins University Press. Sroufe, L. A. (1979). Socioemotional development. In J. D. Osofsky
Mounoud, P. (1976). Les r6volutions psychologiques de l'enfant (Ed.), Handbook of infant development (pp. 462-516). New York:
[Psychologicalrevolutionsof the child]. Archivesde Psychologie, 44, Wiley.
103-114. Thelen, E., Kelso, S. J., & Fogel, A. (1987). Self-organizingsystems and
Neisser, U. (1967). Cognitive psychology. New York: Appleton-Cen- infant motor development. DevelopmentalReview, 7(1), 39-65.
tury-Crofts. Werner,H. (1961). Comparativepsychology of mentaldevelopment.New
Newell,A. (1982). Theknowledgelevel.ArtificialIntelligence, 18, 81-132. York:Science Editions, Inc.
Norman, D. A., & Rumelhart,D. F. (1975). Explorations in cognition. Wolff, P. H. (1963). Observationson the early developmentof smiling. In
San Francisco:Freeman. B. M. Foss (Ed.), Determinantsof infant behavior (Vol. 2, pp. 113-
Piaget, J. (1952). The origins of intelligence in children (M. Cook, 138). New York:Wiley.
Trans.). New York: InternationalUniversities Press. (Original work Zajonc, R. B. (1980). Feeling and thinking: Preferences need no in-
published 1936) ferences. AmericanPsychologist, 35, 151-175.

You might also like