Seeing Is Believing: Children's Explanations of ...karlrosengren.com/documents/Seeing-Is-Believing.pdf · currences (e.g., erasing a pencil drawing). In Study 2 we presented children

Seeing Is Believing: Children's Explanations ofCommonplace, Magical, and ExtraordinaryTransformations

Karl S. RosengrenUniversity of Illinois

Anne K. HieklingUniversity of Michigan

ROSENGREN, KAKL S., and HICKLING, ANNE K. Seeing Is Believing: Children's Explanations ofCommonplace, Magical, and Extraordinary Transformations. CHILD DEVELOPMENT, 1994, 65,1605—1626. Ghildren's magical explanations and beliefs were investigated in 2 studies. In Study1, we first asked 4- and 5-year-old children to judge the possibility of certain object transforma-tions and to suggest mechanisms that might accomplish them. We then presented several com-monplace transformations (e.g., cutting a string) and impossible events (magic tricks). Prior toviewing these transformations, children suggested predominantly physical mechanisms for theevents and judged the magical ones to be impossible. After seeing the impossible events, many4-year-olds explained them as "magic," whereas 5-year-olds explained them as "tricks." In Study2, we replaced the magic tricks with "extraordinary" events brought about by physical or chemi-cal reactions (e.g., heat causing paint on a toy car to change color). Prior to viewing the "extraordi-nary" transformations, children judged them to be impossible. After viewing these events, 4-year-olds gave more magical and fewer physical explanations than did 5-year-olds. Follow-upinterviews revealed that most 4-year-olds viewed magic as possible under the control of an agent(magician) with special powers, whereas most 5-year-olds viewed magic as tricks that anyonecan learn. In a third study, we surveyed parents to assess their perceptions and conceptions ofchildren's beliefs in magic and fantasy flgures. Parents perceived their children as believing ina number of magic and fantasy flgures and reported encouraging such beliefs to some degree.Taken together, these findings suggest that many 4-year-olds view magic as a plausible mecha-nism, yet reserve magical explanations for certain real world events which violate their causalexpectations.

Recent accounts of cognitive develop- have imaginary playmates (Taylor & Cart-ment portray young children as constructing wright, 1991). Other research suggests thatsystematic and coherent theories of the it is not uncommon for children to hold be-world. These accounts contrast with older liefs in the reality of supernatural beingsaccounts (e.g., Piaget, 1929, 1930) which (Glark, 1991; Harris, Brown, Marriott, Whit-viewed young children as inclined toward tall, & Harmer, 1991; Rosengren, Kalish,illogical, even magical, thinking. However, Hickling, & Gelman, 1994), have trouble dif-the modem accounts do not completely cap- ferentiating fantasy from reality (Subbotsky,ture the nature of children's thought. Not all 1985), and label certain events as magicof children's thinking can be described as (Johnson & Harris, 1994). At issue, however,systematic, logical, or scientific. Numerous is the extent to which children truly believereports of young children have shown that that supernatural events and magic (seen asthey enjoy a rich fantasy life, engage in pre- a supernatural force or mechanism) are pos-tend play (Singer & Singer, 1990), and may sible. Ghildren may merely use "magic" to

This research was supported by a Rackham Faculty Grant from the University of Michiganto the flrst author. We gratefully acknowledge the children, teachers, and staff of the Universityof Michigan Ghildren's Genters, Children's Playspace, and the Little Farm Preschool. We wouldlike to thank Charles Kalish for helpfril discussions of magic, Allison Gordon for research assis-tance, Andrea Backscheider and Stephen Silverman for statistical assistance, and Renee Baillar-geon, Judy DeLoache, Lawrence Hirschfeld, Gharles Kalish, Sarah Mangelsdorf, Kevin Miller,and four anonymous reviewers for providing helpful comments. Address for correspondence:Karl S. Rosengren, Department of Psychology, 603 East Daniel, University of Illinois, Cham-paign, IL 61820.

IChild Development, 1994,65,1605-1626. © 1994 by the Society for Research in Child Development, Inc.All rights reserved. 0(X)9-3920/94/6506-0010$01.{)0]

1606 Child Development

label certain events that they see as extraor-dinary. It remains unknown what childrenmean when they use the term "magic." Dothey use the term to refer to "tricks," as inparlor magic, simply as a fallback mecha-nism when all else fails, or does its use re-flect an underlying belief in the possibilityof extraordinary or supernatural events? Thegoal of the present research was to examinethe types of events children label as magic,what it means for children to label events asmagic, and how children might come to usemagical explanations.

The term "magic" has been used bypsychologists in two different ways (Chan-dler & Lalonde, 1994). First, "magic" hasbeen used to label others' (usually chil-dren's) seemingly irrational thinking (Luhr-mann, 1986; Neusner, 1989; Piaget, 1930).For example, Piaget (1930) labeled as "magi-cal" children's beliefs that they could causesome event merely through their ownthoughts or gestures. Ghildren acting as ifthey could obtain a prized toy by wishingwould be said to be engaging in magical rea-soning.

Magic has also been used to refer to aset of causal beliefs involving supernaturalabilities held by particular individuals. Inthis second sense, magic is a causal mecha-nism invoked by the subjects of study them-selves to explain certain events. Few re-searchers have devoted much attention tothis sense of magic, although Fiaget (1929)suggested that these beliefs, which he re-ferred to as "social magical beliefs," aremore systematic and stable than other as-pects of children's early causal reasoning.Ghandler and Lalonde (1994) have sug-gested that such thoughts about magic mayplay an important developmental role inallowing children to bracket occurrenceswhich might otherwise threaten their"fledgling theories."

Several recent investigations have ex-amined the extent to which children believethat magic and supernatural events are pos-sible. These studies have shown that youngchildren make a shairp distinction betweenpossible and impossible events, yet leaveopen the possibility of certain extraordinaryevents. For example, Rosengren et al. (1994)found that while 4- and 5-year-old childrendid not accept the notion that animals couldbe made smaller or undergo changes inshape, they did accept the idea that a magi-cian could bring about these changes. Fewof these children suggested that magicians

would resort to tricks and deception toachieve these ends. It is not clear from thisstudy, however, whether children believedthat magicians are real people who havepowers to perform impossible events orwhether the children merely placed theseimpossible events in the context of a storyor some sort of fantasy.

Subbotsky (1985,1991, 1994) has shownthat, in verbal reports, 4- and 5-year-old chil-dren normally make a clear distinction be-tween events which are possible in the realworld and those which occur only in booksand fairy tales. However, when children areplaced in a conducive context they may actas if supernatural events are possible. Forexample, following the reading of a fairytale, adult instruction about a supernaturalevent, or being shown some sort of "magi-cal" apparatus, 4- and 5-year-old childrenacted as if they believed that pictures couldbe turned into three-dimensional objects,that inanimate objects could come alive, andthat objects could be moved by words alone.

Two other investigations have exam-ined in greater detail the extent to whichchildren actually think about magic. In oneof these investigations, Johnson and Harris(1994) presented 3-5-year-old children withvarious hypothetical object transformations,some possible and some impossible by nor-mal means. Ghildren consistently labeledthe impossible occurrences "magic." Thus,preschool children seem to make a princi-pled distinction between outcomes theyconsider magical and more ordinary out-comes. However, Johnson and Harris foundthat when 3—7-year-old children were askedto pretend that an attractive object was in anearby box, approximately half of the chil-dren in each age group opened the pretendbox and responded as if they wondered ifthe pretend entity was actually in the box.Johnson and Harris suggest that, althoughmost children distinguish magical eventsfrom other types, children under the age of7 vary in the extent that they take a credu-lous or skeptical stance toward magical oc-currences.

A recent investigation (Ghandler & La-londe, 1994) examined the extent to which3- and 4-year-old children are willing to la-bel events as "magic." Rather than beingpresented with hypothetical or pretendtransformations, in this study preschool chil-dren were presented with an apparatusadapted from one of Baillargeon's infant ha-bituation studies (Baillargeon, 1991). The

apparatus included a screen which appearedto pass completely through a solid object inblatant violation of physical laws. Children'sreactions to this impossible event were re-corded. Two thirds of the children labeledthis occurrence as "magic." However, afterrepeated viewings of the event, almost all ofthe children suggested that it involved atrick. Thus, when faced with a clear viola-tion of physical law, children initially werewilling to label the event as magic but aban-doned this explanation when given the op-portunity to consider alternative explana-tions.

Taken together these studies suggestthat "magic" is indeed an active category inpreschoolers' thinking. Children willinglylabel impossible or unexpected events as"magic." However, the robustness of chil-dren's tendency to invoke "magic" remainsunclear. Specifically, it is not known whatfactors beyond physical impossibility mayencourage children to invoke "magic" andwhether children truly believe magic to bea real, supernatural force. Finally, it is notclear how children's magical beliefs changewith age.

The primary purpose of this investiga-tion was to examine when and to what extentpreschool children resort to magical expla-nations. Specifically, we examined whedierchildren reserve "magic" for unexpected orseemingly impossible events and the extentto which older and younger preschoolers usedifferent mechanisms to account for such oc-currences. To address these issues, we pre-sented 4- and 5-year-old children with seem-ingly impossible object transformations,using a set of events typically included inmagic shows. We chose to use prototypicalmagic tricks to see whether children viewthem as truly magical or whether they viewthese events as merely involving tricks ordeception. In this first study, we contrastedchildren's explanations of these magicalevents with those of more commonplace oc-currences (e.g., erasing a pencil drawing). InStudy 2 we presented children with transfor-mations of objects, which though possiblewithout tricks or deception, involved mecha-nisms expected to be poorly understood bychildren (and many adults). In both studieswe asked children to judge the possibilityof potential mechanisms of change prior toseeing the actual transformations. T"his al-lowed us to examine children's explanationsof both hypothetical and actual transforma-tions, that is, to make a direct comparison ofchildren's spontaneous reactions and their

Rosengren and Hiekling 1607

explanations upon seeing the transforma-tions. Finally, in order to examine the extentto which children actually believe thatmagic is a real, supernatural force we inter-viewed the children about their magical be-liefs. In sum, by interviewing the childrenspecifically about their magical beliefs, pro-viding a wider range of magical items thanin previous studies, and including actual aswell as hypothetical transformations, wehoped to gain a clearer understanding ofchildren's reasoning about magic.

Study 1METHOD

SubjectsForty-eight children participated in the

study. There were 24 children in a youngerage group (M = 4-4, range = 3-11 to 4-10;9 males, 15 females) and 24 children in anolder age group (M = 5-5, range = 5-0 to5-10; 12 males, 12 females). In addition, 22adults (9 males, 13 females) performed simi-larity, familiarity, and/or possibility ratingson the items used in the transformations.The children were from two university pre-schools and another local preschool, and theadults were recruited from undergraduatepsychology classes. The sample consisted ofprimarily white, middle-class children andadults. All participants were native Englishspeakers.

MaterialsThe items for Study 1 included eight ob-

jects whose appearance could undergo atransformation of one of four types: color,shape/size, part/whole, or appearance/dis-appearance. Four "magic" items were pro-fessional quality magic tricks which couldundergo seemingly "impossible" transfor-mations involving nonobvious causal mech-anisms. The four "commonplace" itemswere everyday objects which could undergo"possible" transformations involving moreobvious and familiar mechanisms. The pos-sibility/impossibility and familiarity/unfa-miliarity of the transformations were con-firmed by ratings of two groups of 11 adults."Magic" items were rated as significantlyless familiar than the commonplace ones,t(9) = 10.5, p < .001, and magical transfor-mations were rated as less possible than"commonplace" ones, t(9) - 24.6, p < .001.The items, transformations, and mean adultratings appear in Table 1.

ProcedureGhildren were tested individually by

one of two female experimenters in a private


TABLE 1

EXPERIMENTAL ITEMS USED IN STUDIES 1 AND 2, TRANSFORMATIONS, AND MEAN RATINGS OFFAMILIARITY AND POSSIBILITY BY ADULTS

Items Transformation

Mean Familiarity Rating(N = 10)

(Range = 0-10,Maximum = 10)

Mean Possibility Rating(N = 10)

(Range = 0-10,Maximum = 10)

Magic:Dots Color changeColoring book Color appearanceNickel Size changeRope Parts-whole

Commonplace:Play-Doh ball Shape changeString Whole-partsBalloon Size/shape changePencil lead dot .... Disappearance

Experiment 2:Glowsdck Appearance of lightFickle foam Golor changeColor car Color changeInvisible ink Appearance of picture

4.65.23.44.94.8

9.08.99.79.7

5.55.84.05.7

.81.8.9.1.4

10.010.010.010.0

7.96.51.39.4

room. To introduce the session, the experi-menter explained the task as a game inwhich she would show some toys to thechild and then do something with each toyto make it look different. In order to focusattention on the relevant aspect of each item,the experimenter presented each of theeight items (in a random order) and askedchildren to describe its initial appearance.For example, children were shown a ballof Play-Doh and asked, "What shape doesthis look like?" All but one child gave ac-curate descriptions of the items' initialstates.

To obtain initial reactions regardingwhat sorts of mechanisms could be used tochange each item's appearance, childrenwere presented with a hypothetical transfor-mation. That is, they were asked to think ofways to make the item look different (e.g.,"How could I make this look a differentshape? Gan you think of some ways?"). Chil-dren were encouraged to suggest severalmechanisms. Children's responses werecoded as belonging to one of several catego-ries: "mechanical" if they referred to work-manship (e.g., painting, hammering, cut-ting), "natural" if they invoked naturallyoccurring mechanisms of change (e.g., rust-ing, fading), "deception" if they mentionedtrickery, "religious" if they referred to Godor Jesus, or "magical" if they referred tomagic words, wands, or people. Reliabilityof mechanism coding, assessed by two inde-

pendent coders on 18 children's responses,was 97%.

The majority of children spontaneouslysuggested some sort of mechanism to per-form the hypothetical transformations. How-ever, if children were unable to suggest anypossible mechanisms for these transforma-tions or did not give the one to be used inthe actual transformation, the experimenterproposed a mechanism and asked whether itwould succeed. For the commonplace items,the proposed mechanisms were the actualones to be used in the transformations (e.g.,"cutting the string," "blowing air into theballoon," "hitting the Play-Doh with myhand," "erasing the circle with an eras-er"). For the "magic" items, the proposedmechanisms were the same "impossible"-appearing transformations to be performedby the experimenter (e.g., putting two piecesof rope together by "just pulling on them,"making two colored dots change color by"pressing them together," making colors ap-pear on pages of a blank coloring book by"flipping the pages with my fingers," mak-ing a nickel shrink in size by "by placingthis [disk] on it").

Following this initial set of questions,the experimenter performed a transforma-tion on each item. After each transformation,the experimenter asked the children to againdescribe the item's appearance (e.g., "What

shape does this look like now?"). All chil-dren correctly described the new appear-ance of the objects. The experimenter thenasked children to explain how the transfor-mation was performed ("How did I makethis look [new appearance]?"). If childrenresponded "I don't know" or gave no re-sponse, the experimenter prompted themwith three choices (given in random order):(1) the first mechanism suggested by thechild to the initial question about the hypo-thetical transformation, (2) magic, and(3) trick. The actual question was phrased:"Did I do it by (child's mechanism), or wasit magic, or was it a trick?"^ Agreement be-tween coders, assessed on 18 children'stransformation explanations, was 96%. Afterseeing each transformation, if children askedto try the item, we allowed them to play withthe item. For the magic items, however, wesubstituted an identical-appearing, non-working copy. We used nonworking copiesof the magic items for a number of reasons.First, we wanted to make our task similar tosituations where magic tricks are performed.Magicians rarely reveal the mechanisms in-volved in their tricks. Second, by providingthe children with nonworking copies, we en-sured that children who did ask to play withthe items did not have the opportunity todiscover more about the item than did chil-dren who did not ask to play with the item.Finally, when we purchased the items at thelocal magic store we agreed never to revealthe true mechanisms underlying the tricks.Nearly all children asked to try one or moreof the commonplace items. Only eight of theyounger children and five of the older chil-dren requested to try any of the magic items.

Next, children were asked whether theythemselves, a friend, each of their parents, ateacher, and a magician could perform eachtransformation. Children were always firstasked whether they could perform thechange but were asked about other agents ina random order.

Finally, children were asked a set ofquestions designed to determine whetherthey hold a set of coherent beliefs regardingmagical phenomena. Children were askedwhether magic is something real (as opposedto tricks) and whether people can find magicin the "real world" or only in books andmovies. Specifically, we asked: "What ismagic? Is it something real, or is it always

Rosengren and Hickling 1609

tricks?" Children were asked to explaintheir answers (e.g., we asked them to explainthe meaning of "trick"). Children's re-sponses were coded as: 1 = magic involvesreal supernatural events, 2 = magic involvestricks or deception, 3 = magic is a combina-tion of tricks and real supernatural events, 4= magic is fantasy or it only occurs in booksor movies, and 5 = unclear responses. Wethen asked the children: "Is magic only instories and movies, or does it happen in thereal world sometimes?" Depending on theirresponses, children were asked to describemagical events they had seen or to explainhow they knew magic was not "real." Chil-dren's responses were coded as: 1 = magicoccurs in the real world, 2 = magic occursonly in stories and movies, 3 = magic occursin the real world as well as in books andmovies, and 4 = unclear responses. Chil-dren were also asked questions about magi-cians, such as: "How does someone becomea magician? Do magicians just know how todo magic or do they have to learn how?"These answers were coded as: 1 = magi-cians are bom with special powers, 2 = mag-ical powers are passed on by another indi-vidual with magic powers, 3 = magic islearned from books, 4 = magic is learnedfrom a magician or in magic school. Finally,we asked: "Do magicians have special pow-ers?" Responses to this question were codedas: 1 = magicians have special powers, 2 =magicians have special knowledge but notspecial powers, and 3 = magicians do nothave special powers. Reliability on thesequestions, assessed on 18 children's re-sponses, was 93%.

On the basis of interview responses andexplanations of the transformations, we ratedthe degree of each child's belief in magic(i.e., as a possible causal mechanism distinctfrom tricks or deception). Two independentcoders used a seven-point scale (1 = childclearly views magic as always involvingtricks or deception to 7 = child believesmagic is possible and real). Children givena score of 1 clearly stated that magic was notreal and responded that magic involvedtricks or deception, and these childrenstated that magic did not involve any typeof special or supernatural powers. Childrengiven a score of 7 mentioned that magic wasreal, that it involved special or supernaturalpowers, and these children did not use tricks

' Eighteen of the 48 children used the term "magic" prior to any mention of this term bythe experimenter. '


or deception to explain any of the magicevents. Children given a score in the mid-range responded to our questions in a mixedmanner. These children typically used both"magic" and "tricks" to explain the eventsand were inconsistent in their responses toour questions concerning the reality statusof magicians. Reliability for the belief rat-ings, assessed on 50% of the children's inter-views, was 94% within one scale point (84%exact agreement).

RESULTS

Ghildren made a sharp distinction be-tween the commonplace and magical events.All of the subjects accepted the notion thatthe hypothetical transformations of the com-monplace items could occur (Af = 3.9 out of4 events)^ and rejected the notion that thehypothetical magical transformations couldoccur (M = .3, SD = .7 out 4; *(47) = 23, p< .001).

Children provided a variety of causalmechanisms to explain the transformations.Since most of the responses (92%) were cate-gorized as physical/natural, magic, or decep-tion, only these explanation types were in-cluded in the analyses.'^ To comparechildren's explanations prior to and afterseeing the transformations, we conductedthree separate, nonindependent 2 (age: 4, 5)X 2 (time: before, after) x 2 (item type:magic, commonplace) repeated-measuresANOVAs, each with frequency of a differentexplanation type (physical/natural, magic,deception) as the dependent measure.^ Theresults appear in Table 2. The "magic" and"trick" analyses revealed significant age dif-ferences, with 5-year-olds providing more"trick" responses (M = .76) than did 4-year-olds, M = .17, F(l, 46) = 15.3, p< .001, and4-year-olds giving more "magic" responses(M = .79) than did the older children, M =.37, F(l, 46) = 11.0, p < .005. Four- and5-year-olds were equally likely to give phys-ical/natural explanations, M = 2.60, M =2.67 respectively, F(l, 46) = 0.3, N.S. Eachanalysis revealed main effects of time and

item type. Children of both ages gave morephysical/natural explanations prior to seeingthe transformations than after seeing theevents, F(l, 46) = 36, p < .001, but gavemore trick and magic responses after seeingthe transformations than before seeing them,Fs(l, 46) > 50, p < .001. More physical/natu-ral explanations were provided for the com-monplace events than for the magic events,F(l, 46) = 445, p < .001, but more magicand trick explanations were provided for themagic events than for the commonplaceevents, Fs(l, 46) > 49, p < .001. Significanttime X item type interactions were obtainedfor each of the three types of explanations,Fs(l, 46) > 36, ps < .001, and age x time xitem type interactions, Fs(l, 46) > 14, ps <.001. Tukey HSD pairwise comparisons ofthe means (p < .05) were used to follow upthe ANOVA results. Children of both agesgave significantly more physical/natural ex-planations prior to seeing the magic eventsthan after they viewed these events (see Ta-ble 2). The younger children, however, pro-vided significantly more physical/natural ex-planations for the events involving thecommonplace items after viewing the events(M = 4.0) than before viewing the events (M= 3.3). There was no significant differencein the number of physical/natural explana-tions provided by the older children beforeand after viewing the commonplace events.Very few magic explanations were providedfor the commonplace events before or afterthe viewing of the events; however, bothgroups of children provided significantlymore magic explanations following themagic events than prior to these events.There was no difference between the twoage groups in the number of magic explana-tions given prior to seeing the magic events;however, after viewing the magic events the4-ye£ir-oIds gave significantly more magic re-sponses (M = 2.96) than the 5-year-oIds did(M = 1.09). Similar to the results for themagic explanations, few trick responseswere provided for the commonplace eventsbefore or after the viewing of the common-place events. Both groups of children pro-

^ One child did not accept one of the four transformations.^ The exceptions included one religious response and the responses of two children who

merely restated the items' appearance on the hypothetical, magical transformation.* We chose to conduct three separate analyses rather than including explanation type as

another factor in a single repeated-measures ANOVA, since the frequencies with which childrengave each of the three explanation types were not independent. The use of multiple-repeatedmeasures ANOVAs offered the advantage of allowing for comparison of each explanation typeacross age, item types, and time. The additional nonparametric tests reported below allow foranalysis of the relative frequency of the different explanation types.

Rosengren and Hiekling 1611

TABLE 2

CHILDREN'S MEAN (SD) RESPONSES IN STUDY 1 AS A FUNCTION OF EXPLANATIONTYPE, AGE, AND ITEM TYPE BEFORE AND AFTER SEEING THE TRANSFORMATION

MAGICAL COMMONPLACE

Before After Before After

Physical/natural:4-year-olds 2.76 (1.2) .36 (.7) 3.28 (.8) 4.00 (.0)5-year-olds 2.52 (.9) .70(1.1) 3.52 (.7) 3.96 (.2)

Magic:4-year-olds 16 (.4) 2.96(1.4) .04 (.2) .00 (.0)5-year-olds 26 (.4) 1.09(1.3) .13 (.5) .00 (.0)

Trick:4-year-olds 00 (.0) .68 (.6) .00 (.0) .00 (.0)5-year-olds 44 (.4) 2.17 (1.1) .44 (.2) .00 (.0)

NOTE.—Mean scores per category range from 0 to 4, IV = 24 per age group.

vided significantly more trick responses forthe magic events after viewing them thanbefore viewing them; however, after view-ing the magic events the older children pro-vided significantly more trick responses (M= 2.17) than the younger children did (M =0.68).

To compare more directly the relativefrequencies with which children used thedifferent explanation types, we examinedindividual children's dominant responsepatterns after seeing the magical trans-formations. Each child was scored as giv-ing predominantly physical/natural (pin),magic, or trick explanations to the four magi-cal transformations (i.e., at least two re-sponses in the same category, with no ties;ties were scored as "mixed"). A 2 (age) x 4(pattern type) chi-square analysis confirmedthat the modal response patterns in the twoage groups differed, x^(3, N = 48) = 16.5, p< .001, with 4-year-olds conforming to the"magic" response pattern more often than toother kinds (N^̂ gie = 19, N^/^ = 1, Nêk =3) and with 5-year-olds most often conform-ing to the "trick" pattern (Ntnck = 12, N^g^^= 5, Np/̂ = 4).

To examine children's responses re-garding whether different agents could per-form each of the actual transformations, wegave each "yes" response a score of 1 andentered these data into a 2 (age) x 2 (itemtype) X 6 (agent) ANOVA. Overall, childrenof both ages believed that the various agentswere more capable of performing common-place (M = 3.7 out of 4) than magical trans-formations, M = 2.0; F(l, 46) = 143.9, p <0001. Children believed that agents vary intheir ability to perform the transformations.

F(5, 280) = 9.6, p < .0001, with magicians(M = 3.4) being more powerful than otheragents, by Tukey HSD pairwise compari-sons of the means, p < .01. For other agents,the mean number of "yes" responses peritem type ranged from 2.5 (child) to 2.8(teacher). Children believed magicians to besignificantly more likely to cause magicevents than the other agents, Af = 3.2; F(5,230) = 21.8, p < .0001, Tukey HSD pairwisecomparisons, p < .01. On the magic itemsthe mean number of "yes" responses for theremaining agents ranged from 1.5 (child) to1.8 (teacher).

Children's responses to the structuredinterview questions suggest that childrenhave substantial knowledge about the cate-gory of magic, though their beliefs about thenature of magic vary. When asked whatmagic is, many of the 4-year-oIds (IV = 13)responded that magic was real and involvedextraordinary or supernatural powers, whilethe majority of 5-year-olds (N = 14) said thatmagic involved trickery or deception. Sev-eral children (six 4-year-olds, five 5-year-olds) suggested that the nature of magic de-pends on the situation. For these childrenmagic involved real supernatural powers insome situations and trickery in others. Twochildren (one of each age) gave "fantasy" re-sponses, indicating that magic is somethingfound only in books, movies, or stories. Onthe question asking whether magic can befound only in fantasy or also in the realworld, children of both ages gave a varietyof responses. Several children (four in eachgroup) relegated magic to the realm of fan-tasy, whereas 15 children (eight 4-year-olds,seven 5-year-olds) believed in the possibil-


ity of finding magic in the real world. Theremaining children believed magic to besometimes real and sometimes fantasy (threein each group), didn't know, or gave no re-sponse. Thus, children vary in their knowl-edge and beliefs about the boundaries ofmagic as a category.

When asked about magicians, the major-ity of both 4- and 5-year-olds agreed that ma-gicians are "magic" people. However, whileseveral 4-year-olds believed that people be-come magicians by being born one (N = 5)or by having powers passed on (N = 3), fewolder children gave such responses. The ma-jority of 5-year-olds (N = 17) and several 4-year-olds (N = 5) responded that magicianslearn magic from other magicians, or frombooks, or by going to "magic school." Over-all, data from the structured interview sug-gest important individual differences in theextent to which children believe magic toentail supernatural or impossible occur-rences, versus stagecraft and trickery,though as a group, 4-year-olds appear morewilling than 5-year-olds to believe in thepossibility of "real" magic and magicalagents.

Finally, examination of the belief rat-ings revealed considerable age and individ-ual differences in children's credulity to-ward magic. Overall, 4-year-oMs were ratedas more believing than 5-year-olds, M = 5.6and 3.3, respectively; f(36) = 4.13, p < .001.Although nearly two-thirds of 4-year-oIds(63%) believed that magic was real (score of6 or 7), only about a fifth of 5-year-olds (21%)held such strong beliefs (see Fig. 1). In con-trast, 38% of 5-year-olds believed that magicwas not real (score of 1 or 2), as compared toonly 10% of 4-year-oIds. In sum, aJthoughchildren of both ages did show some credu-lity toward the possibility of magic, a devel-opmental trend emerged, with 4-year-oldsappearing more believing than 5-year-olds.

DISCUSSION

These results suggest that 4- and 5-year-old children have a fairly clear notion aboutwhat events are possible versus impossiblein the real world. In general, children usednatural or physical mechanisms to explaincommonplace transformations. When pre-sented with events which they believed tobe impossible a few children continued toprovide natural, physical mechanisms to ex-plain the phenomena. However, many chil-dren (primarily 4-year-olds) resorted to la-beling these events as magic. Other

children, primarily 5-year-olds, suggestedthe "impossible" events were achieved bytrickery. The interview results suggest thatmany of the younger children have specificnotions about what sort of events should belabeled as magic, who can p)erfonn magic(typically magicians, and not themselves,other children, parents, or teachers) and howone becomes a magician (typically by beingborn a magician, by having the powerspassed on, or by learning magic from magi-cians, books, or school). Thus, it seems that,for many of these younger children, magicis not simply a label used to categorize somephenomena as unusual but describes certainspecial powers or mechanisms which are be-yond the ordinary. In contrast, these findingssuggest that the majority of 5-year-oIds viewmagic as little more than trickery.

Study 2In Study 1 we found that children

clearly distinguish between possible andimpossible events. Children labeled magictricks as impossible and explained their oc-currence by appealing to magic or decep-tion. However, the extent to which childrengive magical explanations for other sorts oftransformations, that is, more everydayevents not involving prototypical magictricks or deception, remains to be explored.Ghildren are often confronted with eventswhich were historically thought to be impos-sible (airplanes flying) or extraordinary(electric light, television, remote conixolledappliances). Today these events are seen asquite ordinary by adults, yet it is unlikelyihat most adults would be able to give a so-phisticated or accurate description of howthese events occur. In Study 2, we exploredhow children reason about this type of eventby presenting a series of "extraordinary"events analogous to these more common butstill dramatic events. These extraordinaryevents included chemical and physicaltransformations of objects which resembleditems typically found in the home. Theseitems included a glowstick (similar to a plas-tic tube), a "magic" car (similar to a toy car),an invisible ink pad (similar to a pen andpaper), and a foam pad that changes color(similar to a sponge). These items could allbe transformed by natural means (touchingthe foam pad, dipping the car in hot/coldwater, bending the glowstick) to yield fairlydramatic results (a change of color or theemission of light). Additionally these trans-formations involved mechanisms whichmost adults do not understand very well. Al-


S

u

o

1

Not real

FIG. 1. Percent of 4- and 5-year-old children in Study 1 categorized by degree of magic belief. N= 20 per age group.

3 4

Unclear

Rating of

5

Belief

6 7Real

though we chose these items to be relativelyunfamiliar to 4- and 5-year-old children, allcould be purchased in toy stores, and all ofthe transformations could be performed byanyone provided they had some understand-ing of the underlying mechanism. As inStudy 1, we presented both the common-place and extraordinary transformations, inorder to assess whether children rely on dif-ferent causal mechanisms to explain theiroccurrence.

Previous researchers examining youngchildren's causal reasoning have come to dif-ferent conclusions regarding the extent towhich children use nonnaturalistic or magi-cal explanations when they are presentedwith unfamiliar phenomena or events whichthey do not understand. For example, Huang(1930) reported that children rarely usedmagical or supernatural causality to explainevents. In this investigation, children be-tween 4-10 and 10-3 were presented witha set of events with unexpected outcomes.These events included a few magic tricksand several perceptual illusions (e.g., theMuller-Lyer illusion), but the majority of theevents involved physical phenomena withrelatively unfamiliar mechanisms (e.g., aglass filled with water is covered with apiece of paper and the water remains in theglass when the glass is turned upside down).Fewer than 3% of the children's explana-tions of the events made reference to magi-

cal or supernatural causality. Rather, the ma-jority of the children's explanations involveddiscussion of physical and natural mecha-nisms. Other researchers, however, have re-ported that it is common for children to usenonnaturalistic or magical explanations forunfamiliar events (Baldwin, 1955; Berzon-sky, 1970, 1971; Nass, 1956; Russell, 1956).In this research, familiarity has beenequated with knowledge of the underlyingmechanism causing the event. Thus, anevent was considered to be familiar if chil-dren could experience the process and out-come of the causal sequence (e.g., a boatsinking in a tub of water due to a hole inthe bottom). Unfamiliar events were thosewhere the underlying casual mechanismwas "remote" or not readily observable (e.g.,clouds moving, stars shining). Berzonsky(1971) found that 6- and 7-year-old childrenprovided more nonnaturalistic explanationswhen asked to explain remote events orevents where the children's expectationswere violated (e.g., a scale failing to tilt tothe heavier side when support was with-drawn) than when asked to explain familiarevents. One of the primary goals of the pres-ent study was to sort out these disparatefindings by examining more closely howyoung children reason about unexpectedevents with unfamiliar causal mechanisms.

The "extraordinary" transformations in-cluded in the present study allowed us to


examine children's reactions to less remote,yet dramatic transformations that violate ex-pectations and which are accomplished bymechanisms which are poorly understood bymost adults. As in Study 1, we examinedwhether children might label these transfor-mations as caused by "magic," "tricks," ornatural means. Thus, in this study we exam-ined whether children's magical responseswere reserved for prototypical magic tricksinvolving deception, or whether childrenwould use magic to explain the occurrenceof events which both violated their expecta-tions and for which they had little under-standing of the underlying causal mecha-nisms. In addition, to determine whetherprevious experience with the objects mightinfiuence children's causal explanations, weasked children if they had played with orowned these objects prior to participating inthe study. This last question allowed us toexamine whether familiarity with any of theitems infiuenced the manner in which thechildren explained the events. As in Study 1,we assessed children's beliefs about magicusing a structured interview following theexperimental session.

METHOD

SubjectsForty children participated in Study 2.

There were 20 children in a younger agegroup (M = 4-5, range = 3-11 to 4-11; 10males, 10 females) and 20 children in anolder age group (M = 5-5, range = 5-0 to5-11; 10 males, 10 females). The same 20adults who performed similarity, familiarity,and/or possibility ratings on the items inStudy 1 performed similar ratings for thisstudy. The children were from a universitypreschool or another local preschool, and theadults were recruited from undergraduatepsychology classes. All of the subjects werenative English speakers with predominantlywhite, middle-class backgrounds.

MaterialsAs in Study 1, the items were eight ob-

jects capable of undergoing appearancetransformations. The four "commonplace"items were the same as those used in Study1. However, the four "magic" items were re-placed by objects which could undergotransformations considered by adults more"possible" than magic tricks, though involv-ing unfamiliar mechanisms. Ratings pro-vided by the same students who made rat-ings of die items for Study 1 confirmed thatthese "extraordinary" item transformationsare more possible than those of the "magic"

items, t(6) = 3.8, p < .01, but less familiarthan those of the "commonplace" items, t{6)= 8.6, p < .001, at least as viewed by naiveadults. See Table 1 for a list of "extraordi-nary" items, transformations, and meanratings.

ProcedureThe experimental procedure was nearly

identical to that in Study 1. Again, the exper-imenter asked the children to suggest mech-anisms that might account for hypotheticaltransformations of each of the eight items. Ifthe children did not suggest an appropriatemechanism on a particular item, the experi-menter proposed a specific mechanism andasked whether it would succeed. For the ex-traordinary items, the proposed mechanismswere descriptions of ihe transformations tobe performed (e.g., making a toy car changecolor by "putting it in a glass of water," mak-ing a picture appear "by just scribbling onthe paper," making light appear "by bend-ing this [tube] in half," coloring a piece offoam by "touching it with my hand").

After the assessment of the children'sinitial reactions to the items, the childrenviewed the actual transformations. As inStudy 1, the experimenter asked the chil-dren to explain each transformation and, ifnecessary, prompted them with three spe-cific choices: (1) magic, (2) trick, (3) the firstmechanism suggested by the child to ac-count for the hypothetical transformation.These choices were provided in a randomorder. As in Study 1, if children asked to tryany of the items after seeing each transfor-mation, we allowed them to play with theitems. As in the procedure used in Study 1,we substituted nonworking copies of the ex-traordinary items. We used nonworking cop-ies in this study for the same reasons as inStudy 1 and to be consistent across the twostudies. Once again, nearly all of the chil-dren asked to play with the commonplaceitems, but only seven of the younger chil-dren and six of the older children asked toplay with the extraordinary items. Followingeach explanation, the experimenter askedthe children to judge whether various causalagents could perform the same transforma-tion. At the end of the experimental session,the experimenter asked the childrenwhether they had seen or played with eachitem prior to the study. All coding categorieswere the same as those used in Study 1. Reli-ability was assessed on 10 children's re-sponses by two independent coders. Agree-ment was 96% on coding of children'ssuggested mechanisms, 100% on explana-


TABLE 2

CHILDREN'S MEAN (SD) RESPONSES IN STUDY 1 AS A FUNCTION OF EXPLANATIONTYPE, AGE, AND ITEM TYPE BEFORE AND AFTER SEEING THE TRANSFORMATION

MAGICAL COMMONPLACE

Before After Before After

Physical/natural:4-year-oIds 2.76(1.2) .36 (.7) 3.28 (.8) 4.00 (.0)5-year-olds 2.52 (.9) .70 (1.1) 3.52 (.7) 3.96 (.2)

Magic:4-year-olds 16 (.4) 2.96(1.4) .04 (.2) .00 (.0)5-year-olds 26 (.4) 1.09(1.3) .13 (.5) .00 (.0)

Trick:4-year-olds 00 (.0) .68 (.6) .00 (.0) .00 (.0)5-year-olds 44 (.4) 2.17(1.1) .44 (.2) .00 (.0)

NOTE.—Mean scores per category range from 0 to 4, N = 24 per age group.

tions of transformations, and 97% on the gen-eral questions about magic and magicians.

RESULTS

As in Study 1, children described theinitial and end states of the items accurately(100% of the time). Upon hearing about thehypothetical transformations, all children ac-cepted the commonplace transformations aspossible (M = 4.0, SD = 0.0) but were lesslikely to accept the extraordinary transforma-tions as possible (M = 1.7, SD = 1.1), t(39)= 7.4, p < .0001. The children were signifi-cantly less familiar with the extraordinarytransformations (M = 1.6, SD = 1.2), t(l, 39)= 12.5, p < .001, compared to the common-place ones (M = 3.96, SD = 0.2).

Overall, children provided fewer differ-ent causal explanations than in Study 1. Tocompare explanations of hypothetical andactual transformations of the different items,we conducted three nonindependent 2 (age)X 2 (time) x 2 (item type) ANOVAs, usingeach of the three most frequent response cat-egories (physical/natural, magic, trick) as de-pendent variables. The results appear in Ta-ble 3. Of the three analyses, only the"magic" analysis revealed a main effect ofage, F(l, 38) = 6.2, p < .05, with 4-year-oldsproviding significantly more magic re-sponses than 5-year-olds. Each of the analy-ses revealed main effects of item type andtime. Significantly more magic and trick re-sponses were given to the extraordinaryitems than to the commonplace ones, Fs(l,38) = 15.1, p < .001, but more physical/nat-ural responses were provided for the com-monplace items than the extraordinary ones,F(l, 38) = 74.3, p < .001. Children were

significantly more likely to provide physical/natural explanations before seeing theevents than after seeing them, F(l, 38) =18.3, p < .001, but were significantly morelikely to provide both trick and magic re-sponses after seeing the events than beforeseeing them, Fs(l, 38) > 13.5, p < .001. Sig-nificant item X time interactions were ob-tained for each of the explanation types,Fs(l, 38) > 17.6, p < .001. Tukey HSD pair-wise comparisons of means (p < .05) wereused to examine these effects. Childrenwere significantly more likely to providephysical/natural explanations for the extraor-dinary items prior to seeing the transforma-tion (M = 3.2) than after seeing the transfor-mation (M = 1.8). After witnessing theevents, children were significantly lesslikely to provide physical/natural explana-tions for the extraordinary items comparedto the commonplace ones (M = 3.95). Whilechildren rarely provided trick responsesprior to seeing either event type (M = .03for both types), after witnessing the eventsthey provided significantly more trick re-sponses to the extraordinary items (M = .5)than to the commonplace ones (M = .00). Asimilar result was obtained with the magicexplanations. While children rarely usedmagic explanations prior to seeing theevents for either item type (M = .1 for theextraordinary items, M = .08 for the com-monplace ones), after seeing the events chil-dren gave significantly more magic explana-tions for the extraordinary items (M = 1.7)than to the commonplace ones (M = .03).No significant three-way interactions wereobtained.

To compare the relative frequency with

karlsrosengren2013

Highlight


which children used each explanation type,we scored children's responses aHer seeingthe four actual, extraordinary transforma-tions as predominantly physical/natural,magic, or trick. A 2 (age) x 4 (pattern type)chi-square analysis indicated that the distri-bution of children's responses did not de-pend on age, x^(3, N = 40) = 3.40, p < .4.The majority of children (N = 30) gave pre-dominantly physical/natural (N = 15) ormagic (N = 15) responses. Four-year-olds'modal pattern was "magic" (N = 10), while5-year-olds' modal pattern was "physical/natural" (N = 10). In contrast to Study 1,very few children (three 5-year-oIds) gavepredominantly "trick" responses. However,several children (five 4-year-olds, two 5-year-olds) provided an equal number of"magic" and "physical/natural" responses.

Many children were not consistent intheir use of particular explanation types.Twelve children (eight 5-year-olds, four 4-year-olds) gave physical/natural responsesfor three or four of the extraordinary items,while eight (six 4-year-olds, two 5-year-olds)never did so. Eleven children (seven 4-year-olds and four 5-year-olds) gave magic re-sponses for three of the four extraordinaryitems, while nine children (three 4-year-olds, six 5-year-olds) never gave magical ex-planations. Approximately half of the chil-dren (eleven 4-year-olds, twelve 5-year-olds)never gave trick explanations.

Since many children did not respondconsistently across the extraordineiry items,we examined children's responses on eachitem in more detail. Specifically, we investi-gated the relation between children's "pos-sibility" judgments prior to seeing the trans-formations, their familiarity with each of theitems, and their "magic" explanations aflerviewing the transformations. First, to inves-tigate the relation between familiarity andpossibility judgments, we conducted 2 (fa-miliarity) X 2 (possibility) ohi-square analy-ses on children's responses for each item.The analyses conflrmed that, for the mostpart, children unfamiliar with a particulartransformation considered the change im-possible (78.3% of the time, averaged acrossitems), and, conversely, children familiarwith a particular transformation generallyconsidered it possible (59% of the time, av-eraged across items), x^s(l, N = 40) > 3.6,ps < .05. Therefore, children's responses in-dicated a reliable, though not perfect, rela-tion between familiarity ratings and possi-bility judgments.

To investigate the relation between pos-sibility and famili^ty judgments with chil-dren's explanations, we conducted possibil-ity X explanation type and familiarity xexplanation type chi-squares for each of theitems. None of the four possibility x expla-nation type chi-squares proved significant,xh(2, N = 40) < .17, N.S. On each item,responses of children who judged the trans-formation to be impossible split primarilybetween "physical/natural" and "magic" ex-planations of the actual transformations.Few trick responses were given. Of the fourfamiliarity x explanation type chi-squares,only that for the invisible ink coloring bookrevealed a reliable relation between famil-iarity and explanation type, x^(2, N = 40) =6.3, p < .05, with the majority of "familiar"children (N = 18) giving "physical/natural"explanations (N = 12) and the majority of"unfamiliar" children (N = 22) giving"magic" explanations (N = 14). For each ofthe other three items, "unfamiliar" childrengave "magic" and "physical/natural" expla-nations (44.6% and 39.2% of die time, re-spectively, averaged across the items) moreoften than "trick" responses (17.6%, on theaverage). Interestingly, on these three items,"familiar" children ^ v e "magic" responsesnearly as frequently (40.1% of the time, onthe average) as did "unfiuniliar" children.Apparently, children are sometimes willingto label as magic transformations which theyhave seen before.

To investigate children's beliefs aboutwhich agents are capable of performing theextraordinary transformations versus morecommonplace ones, we conducted a 2 (age)X 2 (item type) x 6 (agent) ANOVA on thenumber of "yes" responses children ^ v ewhen asked whether a particular agent couldperform each transformation. Overall, chil-dren believed the various agents to be capa-ble of performing both commonplace and ex-traordinary transformations. However,children of both ages believed the agents tobe somewhat more capable of perfoimingcommonplace (M = 3.7, SD = 0.4) than ex-traordinary transformations, M = 3.1, SD =0.9, F(l, 38) = 20.9, p < .0001. Children alsobelieved the agents to vary in their abilitiesto perform the extraordinary transforma-tions, as indicated by a significant agent xitem type interaction, F(5, 190) = 6.22, p <.0001. Simple effects analysis revealed thatonly magicians were considered equally ca-pable of performing both the extraordinaryand commonplace transformations (Ms =3.5, SD = 1.1, SD = 1.0, respectively, F(l,


38) = N.S). In spite of item type differencesfor each of the other agents, Fs(l, 38) > 6.3,ps < .05, children generally believed otheragents to be fairly capable of performing the"extraordinary" transformations. The meannumber of "yes" responses on extraordinaryitems ranged from 2.8 (mom, SD = 1.3) to3.2 (dad, SD = 1.2). Thus, whereas childrenin Study 1 believed magicians but not otheragents to be capable of performing magicaltransformations, children in Study 2 be-lieved all agents to be fairly capable of per-forming transformations of the extraordinaryitems.

As in Study 1, children's responses tothe structured interview suggested varyingknowledge and beliefs about magic. Again,4-year-olds were likely to say that magic wassomething "real" (N = 8) or sometimes realand sometimes trick (N = 8). Five-year-olds(N = 6) were more likely than younger chil-dren (N = 3) to say that magic always in-volved tricks but were equally likely to re-spond that magic can be either real ortrickery (N = 8), depending on the situation.In addition, 5-year-olds said magic was"real" nearly as often (N = 6) as did 4-year-olds. Only one child (a 4-year-old) placedmagic in the realm of fantasy. When askedwhether magic occurs in the real world or instories only, children gave a variety of re-sponses. Several children (seven in eachgroup) believed that magic exists only in afantasy context, while others (five 4-year-olds, seven 5-year-olds) believed it could oc-cur in the real world. The remaining chil-dren believed magic to be sometimesfantastic and sometimes real (five 4-year-olds, four 5-year-olds), didn't know, or gaveno response. These children, like those inStudy 1, varied in their beliefs about the na-ture of magic.

On the questions about magicians, chil-dren of both ages tended to agree that magi-cians were real magic people. The majorityof children in this study (fifteen 5-year-olds,eleven 4-year-olds) believed that people be-come magicians through learning rather thanbeing born with special powers (N = 3) orhaving special powers passed on (N = 3).However, a majority of 4-year-olds agreedthat magicians have real magic powers (N =12), while the majority of 5-year-olds dis-agreed (N = 12). Overall, these data are con-sistent with those obtained in Study 1, con-firming that 4-year-olds as a group are morewilling than 5-year-olds to consider the pos-sibility of "real" magical occurrences andagents.

DISCUSSION

The overall results of Study 2 are similarto those obtained in Study 1. Both 4- and5-year-olds provided natural and physicalmechanisms to explain hypothetical and ac-tual commonplace transformations. In Study2 children of both ages gave more "magic"and "trick" and fewer "physical/natural" re-sponses after witnessing the extraordinarytransformations than the other events. Moreyounger than older children gave magicalexplanations after viewing the extraordinarytransformations. In contrast, the majority ofthe older children continued to respond thatthese events were caused by natural or phys-ical means.

Familiarity with the particular extraor-dinary items and their transformations didnot account for children's patterns of magicalexplanations. For three of the four extraordi-nary items, children who were familiar withthe item provided the same level of magicalresponses as children who said they wereunfamiliar with the items. It seems that chil-dren are willing to label some transforma-tions of items which they have seen beforeas "magical." Children generally suggestedthat the extraordinary events were not possi-ble. Thus, although some children reportedthat they had seen the items before, theymay not have seen the transformations weused before and may have been surprisedby the transformations.

The results of Study 2 also suggest that4-year-olds as a group are more willing thanolder children to consider the possibility of"real" magic events and people. However,it is clear from these results that individualchildren vary considerably with respect totheir magical knowledge and beliefs.

Study 3Studies 1 and 2 provide evidence that

many 4-year-olds believe in the existence ofmagic. They use magic to explain the occur-rence of seemingly impossible events andhold a rich set of beliefs regarding the natureof magical phenomena. How might these be-liefs develop? One possibility is that parentsmight encourage the development of magi-cal and fantasy beliefs. Rosengren et al.(1994) have found that parents report en-couraging magical and fantasy beliefs tosome extent. Belief in fantasy figures suchas Santa Claus is also sanctioned by our cul-ture, and previous research (Clark, 1991; Ro-sengren et al., 1994; Scheibe, 1989) suggeststhat children commonly believe in the real-


ity of fairies, witches, Santa Claus, and otherfantasy figures. In previous work the relationbetween parental support for these beliefsand children's magical beliefs has not beenexamined. The goal of Study 3 was to exam-ine how parental encouragement of beliefsin magic and fantasy might relate to chil-dren's magical beliefs. In addition, we exam-ined how parents' reports of their children'sbeliefs related to their children's magical re-sponses in our two studies.

METHOD

SubjectsParents of children participating in

Studies 1 and 2 were asked to complete aquestionnaire designed to gather basic infor-mation on how parents talk to preschool agechildren about magic and fantasy and howparents perceive and guide their children'sfantasy beliefs.^ Parents of 37 of the 48 chil-dren who participated in Study 1 and par-ents of 26 of the 40 children who partici-pated in Study 2 returned the questionnaire(72% response rate). Of the respondents, 32were parents of 4-year-olds (17 females, 15males), and 31 were parents of 5-year-olds(15 females, 16 males). All but four of therespondents were mothers.

ProcedureStudy 1 survey.—The first of the sur-

vey's several sections included three ques-tions about each of nine magical or fantasycharacters. These characters included threefigures associated with specific events (SantaClaus, Easter Bunny, and Tooth Fairy) andsix supernatural fantasy figures (magicians,witches, fairies, ghosts, monsters, and drag-ons). We included both kinds of characterssince we believed that responses might varydepending on whether the character is gen-erally viewed as good or evil. To assess per-ceptions of children's fantasy figure beliefs.

we asked parents to indicate (yes, no, notsure) whether their children currently be-lieved that each of the nine characters wasreal. We also asked parents to indicate (yes,no, neutral) which characters they currentlyencouraged their children to view as real,as well as which characters they themselvesonce believed were real.® Together, thesequestions allowed us to examine the rela-tions among parental encouragement of be-lief in fantasy characters, parents' percep-tions of their children's beliefs, andchildren's magical beliefs as measured inthe experimental setting.

In order to assess how parents mighttalk to their children about fantasy charac-ters such as those listed above, we asked theparents how they would respond if theirchildren asked whether certain fantasy char-acters are real. We asked these questions re-garding magicians, Santa Claus, EasterBunny, ghosts, and the Tooth Fairy. Re-sj>onses were coded as "yes" if the parentssaid they would reply that the character wasreal, "no" if the parents said they would re-ply that the character was not real, or "avoid-ant" if the parents said they would not givea direct answer (i.e., respond by asking forthe child's opinion). Since many parents in-dicated that they would not give a straight-forward yes or no reply to these questions,we also coded the degree of certainty of in-formation provided about the reality statusof each character. Responses were coded as"definite" if the parents said they wouldgive an informative response, "evasive" ifthe parents said they would not give definiteinformation, or "alternative" if the parentssaid they would give a reply implying thatthe character was real in a metaphysicalmanner (e.g., Santa is the spirit of the sea-son). Agreement between coders on thesetwo coding dimensions was 83%, assessedfor two independent coders on 15 surveys.

' This survey was adapted from that reported in Rosengren et al. (1994) but is described ingreater detail here.

* It is possible that parents may have interpreted the questions concerning magicians in amanner different from those concerning the other figures. That is, parents may have interpretedour question about magicians as concerning whether there exist actual people who perform tricksand illusions and not (as we intended) whether there exist people called magicians who haveextraordinary powers. We think that parents were unlikely to have interpreted our questions inthe former manner for three reasons. First, our questions concerning magicians were embeddedin a questionnaire asking about the reality of fantasy figures. It would seem unlikely in thiscontext that the questions about magicians would be construed to be about the reality status ofindividuals who perform tricks. Second, it is doubtful that parents would interpret our questionsabout the encouragement of beliefs in magicians to be about individuals performing bricks andillusions. Rather, it is much simpler to treat this question as one about the reality of magic as asupernatural power. Finally, in informal discussions with parents, including participants in ourresearch, they have never misinterpreted similar questions ahout magic and magicians.


their children s questions requiring compli-cated technical or scientific responses. For?S" i^ of awould explain the operation of such a com^plex device and (2) whether they would pro-vide a magical explanation in response to

a auestion. Rp«r.r,r,ses were coded as^ 1 they would try to

^.iA ,i! ^''^"âte explanation, "magic" if theysaid they would give fantastical information

ZT.tr. ""^^ "'i^' t? «."*?rity" if theyto a

simoiv adm,. . r " ^ '^ *^^ P^'^"* ôuldsimply admit not knowing an answer. Sincef/ '"*^;âted that their responses

t o f th ^ ^were f

- departure of the modified sur-vey was an expanded focus on how parentsexplain transformations involving complexor unknown mechanisms. Rather than givingthe garage door example, we described thetour extraordinary" items used in Study 2and asked parents how they would explaineach transformation. This allowed for a com-parison of parents hypothetical explanationsand dieir children's reactions to the actualtransformations. The contents of parents' re-sponses were coded according to the catego-ries used for the Study 1 survey (scientificmagic, refer to authority, don't know) Reli-ability of coding was assessed on eight sur-veys, yielding intercoder agreement of 91%AS tollow-up questions, we asked parents(yes, no) whether they would explaintransformations as magic, how often (rai. f e ^ r i f " *« ""o-̂ e than daily")

X

coders, who reached 81% agreement on thelo surveys assessed. RESULTS

guish fantasy from reality on their own) ine rang-

to parents of children in Study 1. Parentresponded to the same three questions aboutevent-related and supematumf fantesf figures, as well as to the same questions IboutAe importance of fantasy in their children"

be^fef f ^*'«° Pêntal belief, childencouragement) x 2 (age: 4

b L X ^ ^ ' ^ ' ^ ^ ^ ANOVAs on the numêvent SateTfi" '^ ' ^" 'x^ ''"P^^^tural andf d f 1/ ^ ^

based „„


TABLE 4

PERCENT OF PARENTS REPORTING BELIEF IN VARIOUS FANTASY FIGURES AS CHILDREN, IN THEIR OWNCHILDREN, AND BELIEFS THEY ENCOUHAGE IN THEIR OWN CHILDREN (N = 63)

Figure

Supernatural:DragonsWitchesGhostsMonstersFairiesMagiciansMean number of "yes" responses ....(SD)

Event-related:SantaEaster BunnyTooth FairyMean number of "yes" responses ....(SD)

Parents' ChildhoodBeliefs

Children's CurrentBeliefs

304356415754

2.8(2.2)

8979762.4

(1.0)

2232383044592.2(2.0)

7972642.2(1.3)

EncouragedBeliefs

5983

2828

.7(1.3)

7268632.0

(1.3)

dren. The analyses did reveal main effectsof question type, F(2,118) = 49.5, p < .001,for the supernatural, F(2, 118) = 5.3, p <.01, for the event-related. Neuman-Keulspost hoc compsaisons revealed that the levelof parental encouragement of belief in su-pernatural figures (M = 0.7 out of 6, SD =1.3) was significantly less than those re-ported for parents' childhood beliefs (M =2.9 out of 6, SD = 2.2) and parents' reportsof children's current beliefs (M = 2.2 out of6, SD = 2.0).

For the event-related figures, parents re-ported that as children they had believed insignificantly more figures (M = 2.5 out of 3,SD = 0.9) than they reported encouragingtheir children to believe in (M = 2.0 out of3, SD = 1.3). The number of event-relatedfigures that parents reported their childrencurrently believing in (M = 2.2 out of 3, SD= 1.3) was not significantly different fromeither the number of figures the parents' re-ported believing in as children or the num-ber of figures the parents' reported encour-aging their children to believe in.

As far as parents' general beliefs aboutthe role of fantasy in their children's livesare concerned, responses varied consider-ably. When asked in the Study 1 survey howmuch they encouraged beliefs in magic andfantasy, parents' responses were fairlyevenly distributed, ranging from activelydiscourage (N = 2) to actively encourage (JV= 5). On average, parents reported that theyencouraged fantasy beliefs somewhat (Af =3.3 on a 0—5 scale). As far as their percep-

tions of the role of parents in guiding chil-dren's fantasy understanding are concerned,parents again gave a variety of responses,though the majority (N = 22) believed thatchildren should learn the distinction mostlyon their own or with occasional parentalhelp. In general, these data suggest substan-tial individual differences in parental beliefsregarding encouragement of fantasy-realityunderstanding.

Relations between Parental Beliefs,Children's Beliefs and ParentalE ncouragement

In order to examine the relations amongparental beliefs during childhood, percep-tions of children's current beliefs, and pa-rental encouragement, we performed sev-eral sets of correlations. Specifically, weexamined the results obtained on the threequestions across all characters, as well asseparately for each of the two subtypes (i.e.,event-related and supernatural). The totalnumber of fantasy characters in which par-ents reported that they encouraged belief(range 0—9) was significantly related to thetotal number of characters parents reportedthat their children believed were real, r(63)= .71, p < .Ol. For the supernatural figures,parents' reports of their children's beliefs(range 0—6) were found to be significantlyrelated to parents' reports of their own child-hood beliefs, r(63) = .73, p < .01, and paren-tal encouragement of the child's beliefs,r(63) = .62, p < .01. For the event-relatedfigures, parents' reports of their children'sbeliefs (range 0-3) were significantly related


TABLE 5

PARENTS' RESPONSES TO QUESTIONS CONCERNING THE REALITY STATUS OF MAGICIANS, GHOSTS, SANTA,THE EASTER BUNNY, AND THE TOOTH FAIRY IN THE STUDY 1 SURVEY (N = 37)

REALITY STATUS? CERTAINTY OF STATUS? AGE DEPENDENCE?

FIGURE Yes No Avoid Definite Evasive Alternative Yes No

Magician 9 23 3 32 3 0 12 15Ghost 3 24 8 24 10 1 8 18Santa 21 7 7 24 7 4 23 5Easter Bunny .... 19 9 6 25 5 3 20 6Tooth Fairy 16 10 3 25 4 0 16 6

NOTE.—Because of missing data, rows within questions do not total 37.

to both reported parental encouragement,r(63) = .75, p < .01, and parents' reports oftheir own childhood beliefs, r(63) = .59, p< .01.

Parental Responses to Fantasy QuestionsTable 5 shows the results for the Study 1

survey questions asking how parents wouldrespond if their children asked about the re-ality of certain fantasy figures. Parents gavemany more definite responses (78% of re-sponses) than either evasive or alternativeones. The majority of parents responded thatthey would tell their children that ghostsand magicians were not real and that Santa,the Easter Bunny, and the Tooth Fairy werereal. When asked about event-related fig-ures, parents suggested that their responseswould vary with the age of the child. How-ever, parents were about evenly divided onmagicians (12 responding yes, 15 re-sponding no), and most parents (18 of 26 re-spondents) claimed that their responses con-cerning ghosts would not vary with thechild's age.

Parents' Use of Magical ExplanationsIn our Study 1 survey, parents re-

sponded that they would not provide magi-cal explanations to their children's questionsvery often. When asked how they would ex-plain the operation of an automatic garagedoor opener, nearly all parents reported thatthey would give a scientific response (N —20), refer to an authority (i.e., more knowl-edgeable adult, book, N = 2), or some com-bination thereof (A/ = 7). Only one parentreported that she might use magic as an ex-planation. Similarly, on the question askingdirectly whether they ever use "magic" as aresponse to children's questions, the major-ity of parents (73%) reported that they wouldnever use magic as an explanation. Two par-ents reported using "magic" only in teasing,four sometimes as a fallback explanation,and two often as a fallback response.

Slightly different results were obtainedin the Study 2 survey. When parents wereasked how they would explain the mecha-nisms underlying each of the four "extraor-dinary" transformations, parents gave "sci-entific" responses most (68%) of the time.Parents sometimes responded that theywould refer their children to an authoritativesource (9% of the time) or gave mixed "sci-entific/authority" responses (11% of thetime). Parents very rarely (3%) said theywould give magical explanations. However,when asked more directly whether theywould ever give magical explanations ofthese or similar "extraordinary" transforma-tions, 11 parents (42%) admitted that theymight provide a magical explanation some-times (Af = 3) or often {N = 8). Similarly,when asked if they generally use "magic" asa fallback response to their children's diffi-cult questions about causal mechanisms, 11parents admitted that they do (occasionally= 8, often = 3). Thus, although most parentsnormally give "scientific" responses to chil-dren's causal questions, some may occasion-ally resort to magical explanations of un-usual events.

Relation between Parental Reports andChildren's Magical Responses

To examine the relation between paren-tal reports and children's magical responseswe conducted a number of exploratory corre-lational and median split group analyses.Among the parent variables examined werenumber of fantasy figures in which parentsencouraged belief (total, as well as supernat-ural and event-related figures separately),number of fantasy figures in which parentsthought their children believed, reports ofchildren's belief in magicians, number offigures in which parents once believed, con-tent of information parents would give re-garding figures' reality status (across all fig-ures, as well as for magicians only), and each


TABLE 6

NUMBER OF CHILDREN BELIEVING IN MAGIC (Based on the Belief Rating) AS A FUNCTION OFPARENTAL VIEW OF CHILDREN'S BELIEF IN MAGIC AND PARENTAL ENCOURAGEMENT

OF BELIEF IN MAGIC (N = 37)

Level of child's belief:Not real (0-3)Real (5-7)

PARENTAL REPORT OF CHILD'SBELIEF IN MAGIC

Real

813

Not Real

33

Unsure

25

PARENTAL REPORT OF ENCOURAGE-MENT OF BELIEF IN MAGIC

Real

39

Not Real

56

Unsure

56

NOTE.—Three children's data are omitted from this table since they were coded at the midpoint of the beliefrating.

of the general questions regarding the im-portance of fantasy beliefs and parentalguidance in fantasy-reality understanding.We examined these parental variables in re-lation to children's belief in magic, mea-sured by the belief rating obtained in Study1, as well as by number of "magic" re-sponses given in the experimental sessionsin both Studies 1 and 2. Table 6 shows thepattern of results obtained when children'sbelief was assessed along with parental re-ports of their children's belief in magic andparental encouragement of these beliefs. Ta-ble 7 shows the pattern of results obtainedwhen children's magical responses werecompared with parents' hypothetical expla-nations of extraordinary events and parents'use of magical explanations. In the analyses,we collapsed across various categories to in-crease cell size, then performed chi-squareanalyses. These comparisons, as well asanalyses examining other combinations ofthe variables described above, failed to yieldsignificant correlations or group differences.Given these results, there appears to be nostraightforward relation between parental

reports and children's behavior in our exper-iments. Rather, the relation between par-ents' perceptions and encouragement of so-cial magical beliefs (i.e., beliefs in fantasyfigures) and children's magical beliefs andexplanations of causal events may be bothsubtle and complex.

DISCUSSION

Parents appear to view their 4- and 5-year-old children as having some troubledistinguishing fantasy from reality. In oursurvey, parents reported that their childrenbelieved that Santa Claus, the Easter Bunny,and the Tooth Fairy were real figures. Magi-cians and other fantasy figures were not re-ported as believed to be real to the sameextent. It seems that parents primarily reportencouraging belief in magical agents whomthey sometimes portray. With respect to ma-gicians, about half of the parents reportedthat their children believed magicians werereal.

How might children come to label someevents as "magic"? In this study we used a

TABLE 7

NuMRER OF CHILDREN PROVIDING MAGIC RESPONSES FOR 0 TO 4 OF THE EXTRAORDINARY EVENTS AS AFUNCTION OF PARENTAL REPORTS (N = 26)

Number of magic responses:01234

PARENTAL REPORT OF USE OFMAGIC FOR EXTRAORDINARY

EVENTS

Never

33432

Sometimes

10011

Often

23300

PARENTAL REPORT OF USE OFMAGIC IN GENERAL FORDIFFICULT QUESTIONS

Never Occasionally

3 33 06 11 32 1

Often

03000

parent survey to determine the extent towhich parents admit to encouraging beliefsin fantasy figures and magicians in their chil-dren. Only a third of the parents reportedthat they encouraged their children to be-lieve that magicians were real, and the ma-jority of the parents answered that, if explic-itly asked, they would tell their children thatmagicians were not real. On the other hand,the majority of the psirents responded that ifexplicitly asked, they would tell their chil-dren that Santa, the Easter Bunny, or theTooth Fairy were real individuals. Manyparents also reported encouraging belief inthese figures. In general, parental encour-agement of belief in fantasy figures wasfound to be positively related to perceptionsof children's beliefs in the reality of fantasyfigures.

The relation between parents' reports ofbelief in fantasy and magic and our assess-ment of children's willingness to accept"magical" changes was not as clear. Wefound no significant relations between sur-vey responses and children's performanceon either of our experimental tasks. Al-though this finding may at first seem surpris-ing, there are a number of possible explana-tions. One possibility is that parents do nothave a very good understanding of their chil-dren's beliefs. Parents do tend to underesti-mate children's beliefs (Singer & Singer,1990; Taylor & Cartwright, 1991). In our sur-veys we found that parents of 4- and 5-year-olds reported similar levels of beliefs intheir children and encouraged similar levelsof belief in fantasy figures. However, inStudies 1 and 2 we found significant differ-ences in the number of magical responsesgiven by 4- and 5-year-olds. Taken togetherthese findings suggest that parents may nothave a very good understanding of their chil-dren's beliefs. It is also likely that parents'encouragement of belief in magic is subtleand infrequent and hence not easily cap-tured in a parental report measure. Finally,while parents and the general culture do en-courage belief in fantasy figures and magicto some extent, it is clear that children do notacquire these beliefs solely by these means.Harris (1994) has suggested that it is onlywhen children have begun to understandthe distinction between those events whichare possible and those which are not thatchildren can come to appreciate certainevents as magical. At this point, childrenmay learn to label as magic those eventswhich violate normal causal principles. Wesuggest that parents may initially providethose labels.


In our surveys, we found that, althoughmost parents report normally providing "sci-entific" responses to children's causal ques-tions, some parents occasionally resort tomagical explanations for unusual events. Wesurmise that, while these magical explana-tions may not be very common, they may beused to explain events that are particularlysalient. That is, it is unlikely that parents usemagical explanations at all to explain natu-ral, commonplace events. However, parentsmay use magical explanations when pre-sented with dramatic and perhaps novelevents that violate expected causal se-quences and of which they do not have acomplete understanding.

General DiscussionEven though young children appear to

understand certain principles which under-lie various physical, biological, and mentalphenomena (Wellman & Gelman, 1992), 4-and 5-year-old children do not rule out theexistence of extraordinary, even supernatu-ral events. Four-year-olds and, to a lesser ex-tent, 5-year-olds labeled both prototypicalmagic tricks and extraordinary events asmagic. Follow-up interviews revealed thatmany of the 4-year-olds view magic as a realphenomenon, controlled by magicians.Magic for many of these children is notsomething that is learned but something thatinvolves special powers an individual is ei-ther bom with or which are bestowed uponan individual by someone who possessesthese powers. The majority of the 5-year-olds in these studies viewed magic as in-volving tricks and deception. These childrenviewed magic as a skill which could belearned through reading, other magicians, orat a special "magic" school.

The results of our first two studies con-trast with those obtained by Chandler andLalonde (1994), who found that 3- and 4-year-old children quickly shifted from magi-cal to trick responses when allowed re-peated viewing of an unusual event. Thetwo investigations, however, differ in a num-ber of important ways. First, we presentedchildren with only a single viewing of eachevent rather than multiple viewings. It maybe, as Chandler and Lalonde suggest, thatchildren use magical explanations as tempo-rary place holders while they search forsome more familiar causal mechanisms toexplain unusual events. In general, magi-cians perform a given trick only once pershow, perhaps to keep the audience mem-bers from searching too far for nonmagical


explanations. Clearly, the stability of chil-dren's magical explanations needs to be ex-amined in more depth. Second, Chandlerand Lalonde allowed their subjects to freelyexplore the apparatus involved in their un-usual event, whereas we presented childrenwith the opportunity to explore nonworkingcopies of the magic and extraordinjiry items.Few children, however, asked to play withthe magic and extraordinary items. Thus, inour studies it was highly improbable thatchildren would discover the actual mecha-nisms underlying the events. For this rea-son, children may have been less able to gobeyond their initial magical explanations.Finally, we presented children with a seriesof unusual events rather than a single one.The cumulative impact of viewing a seriesof unexpected events may have increasedchildren's willingness to use magical expla-nations. In a sense, we may have been bettermagicians by creating a more "magical" con-text than Chandler and Lalonde. Presentingunusual events in an unusual or "supernatu-ral" context may shifl children away fromusing more naturalistic explanations of phe-nomena. Likewise, had we performed ourstudies in a religious context, we might haveinduced children to provide more religiousexplanations. Subbotsky (1985,1994) has be-gun to examine certain contextual issuesleading to magical responses. The impor-tance of contextual information in triggeringvarious causal models or in causing childrento shift from using one causal framework toanother should be investigated in further re-search.

Children, however, do not use magic toexplain all types of events, regardless of thecontext. In both of our first two studies, chil-dren did not use magic to label and explaina variety of commonplace events. One possi-bility that we explored in Study 2 is that chil-dren might reserve magical explanations fortransformations of objects which were unfa-miliar. Familiarity with the objects did notseem to determine whether a child will labelit as magic. Even many of the children whoreported having previous experience withsome of our extraordinary items suggestedthat magic was involved in the transforma-tions. Since many of the children reportedthat the hypothetical transformations of theextraordinary events could not occur, it maybe that, while these objects were familiar tothe children, the mechanisms themselves vi-olated the children's expectations regardingwhat is a typical event for such items. Thisviolation of familiar causal sequences may

underlie children's fascination for and de-sire to repeat events similar to the extraordi-nary events presented in Study 2. Furtherresearch should examine more closely howfamiliarity with the items and familiaritywith the causal mechanisms relates to chil-dren's magical explanations.

Children did, however, often believethat they could perform the transformationsof the extraordinary events after witnessingthem. In this regard children's responses forthe extraordinary events were quite differ-ent from those for the magical ones, whichthey did not believe they could perform. Atpresent, the cause of this difference is un-clear. All of our magic and extraordinaryevents occurred relatively abruptly. Thus itdoes not seem likely that children used thetiming of the event to determine whetherthey could perform it. It is possible that chil-dren noticed subtle differences in the typesof actions used to produce the differenttypes of events. We tiiink that this explana-tion is unlikely, since we used events whichcontained similar types of discrete actions(e.g., pulling, pushing actions). In addition,to perform both of these events the agentneeded to have some specific knowledgeabout the conditions which were necessaryto produce the event. For the magic events,this knowledge constitutes the "trick." Forthe extraordinary events this knowledge in-volved some basic knowledge of the under-lying mechanism. For example, in order tomake the car change color in water, oneneeds to know (or realize) that the car waspainted with heat sensitive paint and thatthe water must be a certain temperature. Inthis sense, our extraordinary events involved"tricks" as well. It is unlikely that the chil-dren in these two investigations were anymore knowledgeable of the "tricks" in-volved in the extraordinary events than theywere of those involved in the magic events.

Children do seem reluctant to use magi-cal explanations unless faced with an ex-traordinary event (Subbotsky, 1994). This re-luctance to label events as magical or to citemagical mechanisms may be due to some ex-tent to perceived social costs (see WooUey& Phelps, 1994), but what seems truly toseparate the magical from the commonplaceis whether or not children believe the eventto be possible. For both hypothetical magicand extraordinary events, children re-sponded first that such transformations wereimpossible by the means we suggested andthen proceeded to suggest physical mecha-nisms (e.g., color the magic coloring book


with a crayon). Only after witnessing the ac-tual magic and extraordinary events did chil-dren resort to magic. Thus, in the first twostudies presented here, magical explana-tions were reserved for those items whichthe majority of the children believed to beimpossible. At present it is not clear whetherchildren's magical explanations are moregreatly influenced by violations of familiarcausal sequences or by the occurrence of un-expected outcomes. It may be that youngchildren view "magical" events in a princi-pled fashion, reserving magical explanationsfor those events which violate expectationsbecause the events are incompatible withknown, familiar causal mechanisms.

Recent evidence has shown that chil-dren come to appreciate what events arepossible and which are not by a relativelyearly age (Baillargeon, 1991; Rosengren,Kalish, & Gelman, 1992). This ability to clas-sify events as possible or impossible indi-cates that children have a relatively good un-derstanding of the causal principles whichapply to events in the world. The ability toseparate events into possible and impossiblecategories also means that children realizethat causal principles provide one with theknowledge to predict what is possible andalso to recognize those events that are notpossible (Harris, 1994). Harris (1994) sug-gests that this ability to recognize certain ex-traordinary events opens up the possibilityof a domain of magical events. According toHarris, classification of events as magic en-ables the child to cope emotionally and cog-nitively with cultural practices that supportbelief in the existence of supernatural be-ings (e.g., Santa Claus, God). From this per-spective, it is only after children have devel-oped an understanding of what iscommonplace that children can come to ap-preciate certain events as magical or extraor-dinary.

The results of our surveys suggest thatthe relation between parental encourage-ment of beliefs and children's actual beliefsis quite complex. Clearly, cultural input isnot the sole source of children's magical be-liefs. It is unlikely, however, that childrencome to label events as magic completely ontheir own. Parents in our study reported thatthey encouraged their children to believethat certain fantasy figures are in fact real.Our assumption is that adults may use theterm "magic" to highlight interesting eventsfor children (e.g., in describing the "magic"of nature) or may even use magical explana-tions as a shortcut when asked to explain

complex events which appear to violatecommon causal sequences. Our survey re-sults provide evidence that some parents onoccasion resort to magical explanations. Wehave also found in our work in preschoolclassrooms that teachers sometimes label re-markable or extraordinary events as"magic." For example, in demonstratinghow a magnet could make a row of paperclips hold together in a chain, one kindergar-ten teacher explained that the magnet was a"magic wand." On another occasion, ateacher showed a group of 4-year-olds a spe-cial kind of white paint which turned colorupon contact with paper, and said, "Whatjust happened? Who knows?!—maybe it'smagic!!" We believe that these magicalmarkers may be provided to a greater extentthan parents and other adults report, thoughwe expect magical explanations to be re-served for certain special, extraordinaryevents, which are not easily explained bycommon causal mechanisms.

On the basis of our survey results, itseems that parents may actively support andencourage certain supernatural beliefs inyoung preschoolers. However, many parentsresponded that children should come tolearn the distinction between fantasy and re-ality by age 5 or 6, suggesting at the sametime that children should learn to make thisdistinction on their own. This implies thatparents may shift from actively encouragingbelief in certain supernatural figures toallowing children to figure things out ontheir own. Two types of parental responsesto children's questions regarding the realitystatus of certain supernatural figures supportthis. First, some of our parents respondedthat if their children asked them if certainfantasy figures were real they would re-spond, "I don't know, what do you think?"Other parents responded to this type ofquestion with vague metaphysical responsessuch as "Santa's real in the spirit." It appearsthat this shift in parental support may occurat the same time that children enter formalschooling and are provided with both more"scientific" causal mechanisms and peerswith different belief systems. As childrencome into greater contact with peers withdifferent beliefs, as well as an educationalsystem which actively discourages magicalbeliefs, they may replace their magical ex-planations with more natural ones and beginto think that tricks and deception underliethe occurrence of seemingly impossibleevents. Likewise, as children come intogreater contact with peers who have compet-


ing interests and desires and gain a greaterunderstanding of the differences betweenappearances and reality, children may beginto understand the deception involved inmagic. By this account, parents and the cul-ture at large may at first actively supportmagical beliefs while peers and schools areleft to debunk these childhood myths.

ReferencesBaillargeon, R. (1991). Reasoning about the height

and location of a hidden object in 4.5- and6.5-month-old infants. Cognition, 38, 13-42.

Baldwin, A. L. (1955). Behavior and developmentin childhood. New York: Dryden.

Berzonsky, M. D. (1970). Efifects of probing chil-dren's phenomenisdc explanations of causeand effect. Developmental Psychology, 3,407.

Berzonsky, M. D. (1971). The role of familiarity inchildren's explanation of physical causality.Child Development, 42, 705-716.

Chandler, M. J., & Lalonde, C. E. (1994). Surpris-ing, magical, and miraculous turns of events:Children's reactions to violations of theirearly theories of mind and matter. BritishJournal of Developmental Psychology, 12,83-95.

Clark, C. D. (1991). The cultural psychology ofChristmas and Easter—with special refer-ence to children. Unpublished doctoral dis-sertation. University of Chicago.

Harris, P. L. (1994). Unexpected, impossible andmagical events: Children's reactions to causalviolations. British Journal of DevelopmentalPsychology, 12, 1-7.

Harris, P. L., Brown, E., Marriott, C , Whittall, S.,& Harmer, S. (1991). Monsters, ghosts, andwitches: Testing the limits of the fantasy-reality distinction in young children. BritishJournal of Developmental Psychology, 9,105-123.

Huang, I. (1930). Children's explanations ofstrange phenomena. Psychologische For-schung, 14, 63-182.

Johnson, C , & Harris, P. L. (1994). Magic: Specialbut not excluded. British Journal of Develop-mental Psychology, 12, 35-51.

Luhrmann, T. M. (1986). Witchcraft, morality andmagic in contemporary London. Interna-tional Journal of Moral and Social Studies, 1,77-94.

Nass, M. L. (1956). The effects of three variables

on children's concepts of physical causality.Journal of Abnormal and Social Psychology,53, 191-196.

Neusner, J. (1989). Religion, science, and magic:In concert and in conflict. New York: OxfordUniversity Press.

Piaget, J. (1929). The child's conception of theworld. London: Kegan Paul.

Piaget, J. (1930). The child's conception of physi-cal causality. London: Kegan Paul.

Rosengren, K. S., Kalish, C. W., & Gelman, S. A.(1992). No way? Children's understanding ofpossible and impossible transformation ofanimals and artifacts. Manuscript submittedfor publication.

Rosengren, K. S., Kalish, C. W., Hickling, A. G.,& Gelman, S. A. (1994). Exploring the relationbetween preschool children's magical beliefsand causal thinking. British Journal of Devel-opmental Psychology, 12, 69-82.

Russell, D. H. (1956). Children's thinking. NewYork: Ginn.

Scheihe, G. (1989). Developmental differences inchildren's reasoning about Santa Claus andother fantasy characters. Unpublished doc-toral dissertation, Gomell University.

Singer, D. G., & Singer, J. L. (1990). The house ofmake-believe: Children's play and devel-oping imagination. Gambridge, MA: HarvardUniversity Press.

Subbotsky, E. V. (1985). Preschool children's per-ception of unusual phenomena. S octet Psy-chology, 23, 91-114.

Subbotsky, E. V. (1991). A life span approach toobject permanence. Human Development, 34,125-137.

Subhotsky, E. V. (1994). Early rationality andmagical thinking in preschoolers: Space andtime. British Journal of Developmental Psy-chology, 12, 97-108.

Taylor, M., & Gartwright, B. S. (1991). The distinc-tion between fantasy and reality: A compari-son of children with and without imaginarycompanions. Paper presented at the biennialmeetings of the Society for Research in GhildDevelopment, Seattle.

Wellman, H. M., & Gelman, S. A. (1992). Gogni-tive development: Foundational theories ofcore domains. Annual Review of Psychology,43, 337-375.

Woolley, J. D., & Phelps, K. E. (1994). Young chil-dren's practical reasoning about imagination.British Journal of Developmental Psychol-ogy, 12, 53-67.

Documents

Seeing Is Believing: Children's Explanations of ...karlrosengren.com/documents/Seeing-Is-Believing.pdf · currences (e.g., erasing a pencil drawing). In Study 2 we presented children