o'neill, astington y flavell (1992) young children's understanding of the role that sensory...

7/27/2019 O'Neill, Astington y Flavell (1992) Young Children's Understanding of the Role That Sensory Experiences Play in Kn

1/17

Young Children's Understanding of the Role

That Sensory Experiences Play in

Knowledge Acquisition

Daniela K. O'NeillStanford University

Janet Wilde Astington

University of Toronto

John H. Flavell

Stanford University

O'NEILL,DANIELAK.; ASTINGTON,JANETWILDE;and FLAVELL,JOHNH. Young Children's Under-standing of the Role That Sensory Experiences Play in Knowledge Acquisition. CHILDDEVELOP-MENT,1992, 63, 474-490. 3 studies investigated whether young children understand that theacquisition of certain types of knowledge depends on the modality of the sensory experienceinvolved. 3-, 4-, and 5-year-old children were exposed to pairs of objects that either looked thesame but felt different, or that felt the same but looked different. In Study 1, 36 children wereasked to state, when one of these objects was hidden inside a toy tunnel, whether they wouldneed to see the object or feel it in order to determine its identity. In Study 2, 48 children wereasked to state which of 2 puppets knew that an object hidden inside a tunnel possessed a givenvisual or tactile property, when one puppet was looking at the object and the other was feelingit. In Study 3, 72 children were asked, in a scenario similar to Study 2, to state for each puppetwhether he could tell, just by looking or by feeling, that the hidden object possessed a certainvisual or tactile property. Children were also asked what was the best way to find out whethera given object possessed a certain visual or tactile property. Results of all 3 studies suggest thatan appreciation of the different types of knowledge our senses canprovide (i.e., modality-specificknowledge) develops between the ages of3 and 5. The results are discussed in relation to youngchildren's developing understanding of the role that informational access plays in knowledgeacquisition.

"Monika knows the ball is red." When

we utter this sentence, what mental state

of Monika's are we describing by using

the verb "to know"? According to Dretske

(1981), we qualified Monika's belief asknowledge because there was evidence to

support it and because the belief was based

on that evidence. For example, Monika saw

the ball and seeing the ball caused her to

believe that it was red. As adults, this is the

understanding that we have of knowledge

states. But is this understanding of knowl-

edge shared by children? Presumably, such

a characterization of knowledge requires

an understanding of how knowledge is ac-

quired. If children do not understand this,

then their understanding of what it means toknow something may differ from our adultunderstanding. One fundamental aspect of

knowledge acquisition that children must

leam is that what we can know about an ob-

ject or event depends on the sensory mo-

dality through which it is perceived. For

example, we know that Monika could know

This research was supported in part by a Stanford University Fellowship to the first author,a grant from the Natural Sciences and Engineering Research Council of Canada to the second

author, and an NIMH grant MH 40687 to the third author. We would like to thank the staff,parents, and children of the day-cares and nursery schools in Toronto, Palo Alto, Mountain View,and Los Altos who participated in this study. We would also like to thank Fred Dretske, DerekKoehler, Angeline Lillard, Herbert Pick, and three anonymous reviewers for helpful comments,and Steve Cole and Louis Moses for their statistical help. Portions of this paper were presentedat the annual meeting of the American Educational Research Association in Boston, MA, April16-20, 1990, and at the biennial meeting of the Society for Research in Child Development,Seattle, April 18-20, 1991. Correspondence should be sent to the first author at: Department ofPsychology, Jordan Hall, Stanford University, Stanford, CA 94305-2130.

[Child Development, 1992,63,474-490. 1992by the Society forResearch in Child Development, Inc.All rights reserved. 0009-3920/92/6302-0014$01.00]


2/17

the color of the ball from seeing it, but notfrom feeling it. The development of thisunderstanding is the focus of the presentthree studies.

The impetus for the first study camefrom a study by O'Neill and Gopnik (1991)in which children had to identify differentobjects hidden under a toy tunnel just byseeing or feeling them. Attimes,young chil-dren did not appear to understand that allthe properties of an object cannot always bederived froman isolated sensoryexperience.A number of 3-year-olds told the experi-menter, after feeling a ball hidden under-neath a toy tunnel-a ball that they had

never seen before-that they could tell itwas a blue ball. These children apparentlydid not understand that while they couldknow the ball's color by seeing it, or by be-ing told it, it was impossible to find out itscolor just by feeling it. Becausethis findingwas an incidental one, it was not exploredfurther in the O'Neill and Gopnik (1991)study.

A fewother findings suggestthat youngchildren have difficulty isolating the types

of knowledge resulting fromindividual sen-sOry experiences. Four-year-olds have atendency to attribute their own personalknowledge of a picture (e.g.,the name of adepicted animal) to a naive observer who

just sees the picture (Sodian, 1986, cited inFerner, 1991; Taylor, 1988).MossIer, Mar-vin, and Greenberg (1976)used a privilegedinformation situation to investigate youngchildren's ability to engage in conceptual

perspective taking. Children were exposedto both the audio and visual portions of avideotaped story and were askedabout theirmother's knowledge of both portions aftershe had received only the visual portion. Ofthe 3-, 4-, and 5-year-olds, 95%, 40%, and15%, respectively, overattributed knowl-edge (i.e., attributed knowledge of the audio

portion) totheir mothers. Suchresults couldbe interpreted as a failure on the part ofthese children to differentiate the types ofinformation that can and cannotbe acquiredfrom only one sensory modality.

In our first study, therefore, we explic-itly tested whether preschool children un-derstand that the acquisition ofcertain typesof knowledge depends speCificallyon themodality of the sensory experience in-volved. Sight and touch were chosen as thetwo sensory modalities to be investigated.Children's understanding was assessed bytheir. ability to state correctly whether they

O'Neill, Astington, andFlavell 475

would need to see or to feel an object inorder to determine whether it had a givenvisual or tactile property.

Study I

METHOD

Subjects

Subjects were 36 children, largely frommiddle-class families, attending urban day-care centers: 123-year-olds (mean 3-6, range3-0 to 3-11), 124-year-olds (mean 4-6, range4-0 to 4-11), and 12 5-year-olds (mean 5-5,range 5-1 to 5-11). There were approxi-mately equal numbers of boys and girls ineach age group.Information on race was not

systematically collected.

Materials

A red "tunnel" (approximately 30 x 25x 15 cm) wasconstructed out of Styrofoam.

The openings at either end were covered byfelt flaps. Four pairs of objects were used.Two pairs looked the samebut felt different:(a) two identical toy felt cats, one stuffedwith beans that felt hard and lumpy and onestuffed with cotton wool that felt soft andsmooth, and (b) two identical piggybanks,

one full of pennies and one empty. Theyconstituted thefeel condition. The other twopairs of objectsfelt the same but looked dif-ferent: (a ) twosmall toyfootballsofthe samesize, shape, and make, one green and theother red, and (b) two birthday cards of thesame size, shape, and make, one depicting astriped dragon on the front and the other a

polka-dotted dragon. They constituted thesee condition. A small, hard, red ball and asmall, soft, spongy, yellow ball were used

in the introductory task. Two small picturecards of a hand and an eye were also used.

Procedure

Introduction.-Children were testedindividually. They sat facing the tunnel, infront of which were placed the two smallpicture cards. They were told that in thisgame they would be asked about things wecan know by seeing and things we can know

by feeling. The experimenter then hid a redball inside the tunnel and asked children to

lift up the tunnel, look inside, and tell herwhat color the ball was. After children hadresponded red, the experimenter pointedout to them that they knew it was red be-cause they could see with their eyes that itwas red. Atthis point, children were shownthe card with the eye on it and were toldthat this card meant "I have to see." Theywere reminded that this was one of the an-swers they might have to give in the game.


3/17

Following this, the experimenter hid asoft, spongy, yellow ball inside the tunnelwithout the children seeing it. They wereasked to put their hands inside the tunnel

and to feel whether the ball was soft or hard.It was made clear to children that they couldonly feel the object under the tunnel and notlook at it. Care was taken here to ensure thatthey did not pull the object out of the tunnelwhile feeling it, or peek inside the tunnel.Once they had answered soft, the experi-menter pointed out that they knew it wassoft because they could feel with their handsthat it was soft. Children were then shownthe card with the hand on it and were toldthat it meant "I have to feel," and that thiswas also an answer they might have to givein the game. They were then reminded ofthe two answers they might have to giveandwere asked to state what each card meant. Ifthey would not respond verbally, they wereasked, "Which card means 'I have to feel!see?'" and were allowed to point to the an-swer. While this introduction was elaborate,

pilot studies showed that it was necessary toensure that children understood they couldonly feel or only see an object under the tun-

nel, but not do both.

Experimental task.-Following the in-troduction, children received four experi-mental trials. At the beginning of each trial,children examined one of the four pairs ofobjects. They were shown and told that theobjects were the same size and shape, andtold and shown, appropriately, whether theobjects looked the same or felt the same. Forexample, given the footballs, children were

told they felt the same and were encouragedto feel this for themselves. They were alsotold and shown, appropriately, how the ob-

jects felt or looked different, for example,that the two footballs were different colors,namely, red and green.

After children had finished examiningthe two toys, the experimenter told them shewould now take both toys away, hide thembehind her back, and put only one toy insidethe tunnel. Once this was done, children

were given the test question, which asked

them what they would haveto do to findoutfor sure the identity of the object inside thetunnel. For example, they were asked in thecase of the footballs: "Now, to find out for

sure what color the footballinside the tunnelis, what would you have to do?" Childrenwere first given a chance to respond sponta-neously (either verbally orbypointing tothecard) with one of the two alternatives, "Ihave to feel" or "I have to see." If childrendid not respond spontaneously, the two al-ternatives were presented in a forced-choicequestion (e.g., "Do you haveto see the foot-

ball or do you have to feel the football?").The order of presentation of the forced-choice alternatives was counterbalancedacross trials and subjects. This procedurewas repeated for the remaining three pairsof objects. For the other see condition trial,the test question was: "Now, to find out forsure which dragon is on the card under thetunnel, what would you haveto do?" Forthetwo feel condition trials, the test questionswere: "Now, to find out for sure which pig-gybank is under the tunnel, what would youhave to do?" and "Now, to find out for surewhat the toy cat under the tunnel is stuffedwith, what would you haveto do?"l The or-der of presentation of the four pairs of ob-

jects was counterbalanced across trials andsubjects.

RESULTS AND DISCUSSION

In the introductory task,all the childrencorrectly answered the controlquestion con-cerning the meaning of the two cards, andthe majority did so without needing the

forced-choice alternatives to be given. Oneach of the four test questions, children re-ceived a score of 1 if they answered a givensee and feel condition trial correctly, 0 ifthey answered it incorrectly (see Table 1).Children's responses in each age group didnot differ significantly as a function of theparticular object pair used within a givenmodality condition (binomial test, all p's >.38), and so the two scores in each of theseconditions were combined. There was no

main effectof gender, not didit interact with

IWhere possible, the test question included mention of the superordinate distinguishing

visual or tactile property (i.e., color, stuffing), without mention of the categorical instances of the

properties themselves (e.g., red, soft). In two cases, however, this was not possible (e.g., weight,

pattern) as the questions became too difficult for the children (e.g., What is the weight of the

piggybank under the tunnel?). In these two cases, therefore, children were asked " ... which

piggybank/dragon .... " Since, however, the two objects on each trial differed only with respect

to a visual or tactile property, asking the children what they had to do to find out "which X" or

"what the property of X" was under the tunnel was equivalent in both cases to asking them what

they had to do to find out X's identity.


4/17

PERCENTAGE OF TRIALS ANSWERED CORIl.ECTLYBY

AGE AND MODALITY CONDITION IN STUDY1

AGE

MODALITY CONDITION 3 4 5

See .............................. 38 33 88

Feel ............................ 71 83 96

any of the variables of interest, soit will not

be considered further.

Atwo-way, repeated-measures ANOVA,

with age (3 vs. 4 vs. 5) as the between-subjects variable and modality condition(see vs. feel) as the within-subjects variable,revealed significant main effects of age,F(2,33) =8.63, p


5/17

edge to be gained from individual sensoryexperiences.

Study 2

We designed Study 2 in such a way thatchildren's responses would not be con-founded by their own wishes to perform oneaction over another: Children made knowl-edge assessments of puppets who wereseeing or feeling an object. We also testedthe hypothesis that the ability to makemodality-specific knowledge assessments isacquired in the later preschool years. Previ-ous studies show that young preschool chil-dren perform well on tasks that require themto infer whether a puppet whohas or has notlooked inside a box knows what is in the box(Pillow, 1989; Pratt & Bryant, 1990). Pre-sumably, they can master such tasks becausethey need to monitor only whether a puppetdid or did not have perceptual access to theobject (O'Neill & Gopnik, 1991). However,without an understanding of how differentsensory experiences result in different typesof knowledge, these same children shouldfail an equivalent question concerning a

puppet's knowledge that requires them todifferentiate among the features of objects

perceivable by means of different sensoryexperiences. For instance, suppose that one

puppet feels a toy and another puppet seesthe same toy. When assessing which of thetwo puppets knows certain visual or tactile

properties ofthe toy, these children may errbecause they will not take into account thetype of sensory access that each puppet had.Since both puppets have sensory access to

the toy in either case, these children mayattribute knowledge to both puppets, regard-less of the property or sensory experienceinvolved, and give random answers aboutwhich puppet knows a certain property.

So, in Study 2, children were asked toassess the knowledge state of puppets whohad either seen or felt an object hidden in-side a tunnel. (We will call this modality-specific knowledge assessment.) Childrenwere also asked (in a task modeled after that

used by Pratt & Bryant [1990])to assess theknowledge state of puppets when only oneof the two puppets had visual or tactile ac-cess to an object hidden inside a tunnel (pri-mary knowledge assessment). If our hypoth-esis is correct, then young children should

perform well on the primary knowledge as-sessment task, yet perform poorly on themodality-specific knowledge assessmenttask.

METHOD

SubjectsSubjects were 48 nursery school chil-

dren, largely from upper-middle-class fami-

lies: 16 3-year-olds (mean 3-5, range 3-0 to3-11), 16 4-year-olds (mean 4-5, range 4-0 to4-11), and 165- and 6-year-olds (mean 5-11range 5-6 to 6-6; referred to hereafter as th~5Ih-year-olds). There were approximatelyequal numbers of boys and girls in each agegroup. Information on race wasnot systemat-ically collected.

MaterialsThe same tunnel was used as in Study

1. In the primary knowledge assessment

task, two pairs of objects were used. Onepair looked the same but felt different: twoidentical yellow sponges, one wet and theother dry. The other pair felt the same butlooked different: two plastic dinosaurs ofthe same size, shape, and make, one greenand the other blue. In the modality-specificknowledge assessment task, four pairs of ob-

jects were used. Two of the four pairs of ob-jects looked the same but felt different: (a)the two felt cats from Study 1, and (b) two

identical" squeaky toy" pigs, one filled withplaster of paris that felt hard and one emptythat felt squishy. These pigs were substi-tuted for the piggybanks used in Study 1

because the dimension of hardness-squish-iness was thought to be easier for childrento understand than that of heavy-light, andto be more distinguishable as a tactile prop-erty. These two pairs of objects constitutedthe feel condition. The other two pairs ofobjects felt the same but looked different:

(a) the two toy footballs from Study 1, and(b) two cardboard dragons of the same sizeand shape, one of which was spotted and theother striped. These two pairs ofobjects con-stituted the see condition. Two "SesameStreet" puppets, Ernie and Bert, were used.These puppets were familiar to all the chil-dren and were easily named by them. A toyspoon, a spongy ball, some blue and greenmarbles, and a toy car were used in the intro-duction, pretest, and control tasks.

ProcedureIntroduction.-Children sat facing the

tunnel, on either side of which stood Ernieand Bert (which side was counterbalanced).Children were given a small introduction tothe puppets, the tunnel, and the task itself.First, we checked that children knew thenames of the puppets. Then, using a toyspoon and a spongy ball, the experimenterfamiliarized children with how to look at or


6/17

- , > - , - < , ~ c - : . y ~ ~. , . ~ " - " " " , ' ; > . . \ _ tit P" "

I",~~'i.:o'

r ~~ ;'c~ ~,~ ~ ~ :,~ ,\~~.. Gfr;;ii~"~iJO'Neill, Astington, and Flavell 479

feel objects inside the tunnel. ~:~ wouldn't see which toy the children hidthis, children were told that in this game inside the tunnel. The experimenter thenthey were going to help the experimenter whispered, "Which one shall we hide?"hide some toys inside the tunnel and that Once children replied, the experimenter

Ernie and Bert were going to take turns said, "Good, so we'll hide the [blue/green]seeing or feeling what was inside. They one," and it was hidden inside the tunnel.were also told that they would have to tell As a memory check for the property bythe experimenter which puppet, Ernie or which the objects differed, children wereBert, knows what is inside the tunneL then asked, "So which one is inside the

tunnel?"Pretest.-A pretest familiarized chil-

dren with the procedure of choosing one ofthe puppets in response to the experi-menter's questions. Children were shownsome blue and green marbles and asked toname the colors of each. Then the experi-

menter said, "Now I'm going to give Erniesome blue marbles and Bert some greenmarbles." Following this, children wereasked, "Now, who has the blue/green mar-

bles?" and asked to point to this puppet.Half the children in each age group wereasked one or the other of the test questionforms.

Control task.- The control task wasused to assess that children could easily

distinguish the two actions of seeing orfeeling what was inside the tunnel whenthe puppets performed them. A toy car wasplaced inside the tunnel. The experimenterthen made one puppet look into the tunnel

by bending its head over and lifting one flapup over its head. The puppet's hands stuckclearly out in back of it. The other puppetwas made to feel inside the tunnel by liftingthe other flap slightly on one side and slid-ing the puppet's hand into the tunnel. The

contents ofthe tunnel were not made visibleto this puppet at any time. To ensure thatchildren understood that the puppet whowas feeling could not see inside the tunnel,its eyes faced forward toward the childrenand 90 degrees away fromthe tunnel open-ing. Children were then asked: (a) "Which

puppet is feeling the toy car inside the tun-nel?" and (b) "Which puppet is seeing thetoy car inside the tunnel?" Whether it wasErnie or Bert who saw or felt the car, theorder in which they did so and the order of

the test questions were counterbalanced.

Primary knowledge assessment task:See condition.- The two dinosaurs were

placed on the table in front of the children.The children, the experimenter, and the

puppets then looked at and felt these dino-saurs until all agreed that the two felt thesame but looked different. Following this,children were told that Ernie and Bert werenow going to turn around so that they

The puppets were now turned backaround and the relevant actions of the pup-

pets were carried out. For example, childrenwere told, "Now, this time, Bert's going tolook inside the tunnel. [Bert looked inside.]

Bert's lookinginside the tunnel. And Ernie'sgoing to put his hand on the side of the tun-nel. [Ernie's hand was placed resting onthefront side of the tunnel, clearly in view andin front of the child.] Ernie's feeling on theside of the tunnel." While the puppets re-mained in these positions, children wereasked the test question: "Now, one of themknows the dinosaur inside the tunnel is[blue/green] and one ofthem doesn't know.Who knows the dinosaur inside is [blue/

green]?" Children responded by naming orpointing to the puppet. If they just namedthe puppet, they were asked to point aswell.Note that, unlike in Study 1, the relevantcategorical instances (e.g., blue, wet) werenow mentioned in the test question itself.This presumably made it easier for childrento remember the property by which the ob-

jects differed.

Primary knowledge assessment task:Feel condition.- The two sponges were

placed on the table in front of the children.The children, the experimenter, and the

puppets then looked at and felt thesesponges until all agreed that the two lookedthe same but felt different. The procedurethen continued in the same manner as in thesee condition except that one puppet feltinside the tunnel while the other puppet

placed his hand on the front side of thetunnel.

Which action each puppet performed'

and the order of the two actions were coun-terbalanced within and across conditions.The order of the see and feel conditionsthemselves was also counterbalanced. Thetwo primary knowledge assessment trialswere blocked to appear either before orafterthe four modality-specific knowledge assess-ment trials.

Modality-specific knowledge assess-ment task.- The procedure of this task was


7/17

almost identical to that of the primary knowl-

edge assessment task, except that both pup-

pets now had access to the object inside

the tunnel. The task consisted of four trials.

On two trials the property in question was

known only to the puppet who was feeling

inside the tunnellfeel condition, cat and pigpairs). On two other trials it was known only

to the puppet who was looking inside (see

condition, dragon and football pairs).

On each trial, children, the experi-

menter, and the puppets all first examined

the pairs of objects as already described for

the primary knowledge assessment task.

Then, Ernie and Bert were turned around

and children hid one of the objects insidethe tunnel. Next, the puppets were turned

back around and one puppet looked inside

the tunnel while the other felt inside the

tunnel as described in the control task.

While the puppets remained in these posi-

tions, children were asked the test question,

which was identical to that used in the pri-

mary knowledge assessment task, for exam-

ple: "Now, one of them knows the pig inside

the tunnel is [hard/ squishy] and one of them

doesn't know. Who knows the pig inside is

[hard/ squishy]?"

The other three trials followed the same

format. The order of the four trials was coun-

terbalanced. The order of the see condition

trials and feel condition trials alternated so

that children never received two trials of the

same modality type one after the other. (This

included the primary knowledge assessmenttrials that came before or after.) Which action

each puppet performed and the order of the

two actions was counterbalanced within andacross all four trials.

RESULTS

All the children passed the pretest and

control task without difficulty. In both

knowledge assessment tasks, all the children

also answered the memory control check

correctly. In both knowledge assessment

tasks, children received a score of 1 if they

answered a given see or feel condition trialcorrectly, 0 if they answered incorrectly (see

Table 3). Children's responses did not differ

significantly (binomial test, all p's >.16) as

a function of the particular object pairs used

within a given modality condition of the mo-

dality-specific knowledge assessment task,

and so these two scores were combined in

each of these two conditions. Because the

number of see and feel trials differed across

the two tasks, all scores were calculated in

PERCENTAGE OF TRIALS ANSWERED CORRECTLY BY

AGE, TASK TYPE, AND MODALITY IN STUDY 2

Modality-Primary' Specificb

AGE See Feel See Feel

3 .... 81 69 62 404 . . . .100 94 88 475 11 z . 100 100 100 59

a Total number of trials is 16 per cell.

b Total number of trials is 32 per cell.

terms of proportions, which were converted

using an arcsin transformation prior to statis-

tical analysis. There was no main effect of

gender, nor did it interact with any of the

variables of interest, and therefore this factor

was not considered further.

A five-way, repeated-measures ANOVA

was carried out with age (3 vs. 4 vs. 51fz), trial

order (see/feel vs. feel/see), and task order

(primary knowledge assessment task first

vs. modality-specific knowledge assessment

task first) as the between-subject variables,

and task type (primary vs. modality-specific

knowledge assessment) and modality condi-

tion (see vs. feel) as the within-subject vari-

ables. The analysis revealed a significant

main effect of age, F(2,36) = 12.86, p

o'neill, astington y flavell (1992) young children's understanding of the role that sensory...

Documents