Evidence of deontic reasoning in 3- and 4-year-old children

Transcription

1 Memory & Cogitio 1996, 24 (6), Evidece of deotic reasoig i 3- ad 4-year-old childre DENISE DELL<\ROSA CUMMINS Uiversity ofarizoa, Tucso, Arizoa Deotic reasoig is reasoig about what oe may, ought, or ought ot do i a give set of circumstaces. Virtually all of our social istitutios ad child-rearig practices presume the capacity to reaso about deotic cocepts, such as what is permitted, obligated, or prohibited. Despite this, very little is kow about the developmet of deotic reasoig. Two experimets were coducted thatcotrastedchildre'sreasoigperformaceo deotic ad idicative reasoigtasks (i.e., the reduced array selectio versio of the Waso card selectio task). Like adults, childre as youg as 3 years of age were foud to adopt a violatio-detectig strategy more ofte whe reasoig about the deotic case tha whe reasoig about the idicative case. These results idicate thatviolatio detectio emerges as a effective deotic reasoig very early i huma developmet. Deotic reasoig is reasoig about what oe may, ought, or must ot do i a give set ofcircumstaces (Hilpie, 1971, 1981; Maktelow & Over, 1991). Virtually all of our social istitutios-from schools ad movie theaters to the justice system-presuppose a capacity to uderstad ad reaso about what is permitted, obligated, prohibited, or advised. Failure to reaso effectively about deotic cocepts ca have disastrous cosequeces, icludig scoldig, expulsio, legal actio, ad eve icarceratio. So fudametal is the presumed capacity to reaso about deotic cocepts that it uderlies most ofour childrearig practices. Adults typically try to guide ad structure childre's behavior by requirig them to adhere to coditioal deotic rules, such as "If you wat to play outside, you must wear a coat," or "All kidergarters must stay i the playgroud." I order to behave appropriately, childre must determie which actios are permitted, obligated, or prohibited, ad uder which coditios. Moreover, referece to social rules appear i childre's justificatios of their behavior as early as 24 moths of age (Du, 1988). Reasoig about deotic situatios, therefore, costitutes a major issue i childre's lives. Yet despite this, very little is kow about the developmetofdeotic reasoigad its relatio to other types ofreasoig. I recet years, deotic reasoig has loomed large i the adult psychological reasoig literature. I cotrast to their performace o statistical reasoig (see, e.g., Kahema, Slovic, & Tversky, 1982), idicative reasoig (e.g., Waso & Johso-Laird, 1972), ad mathematical or scietific problem-solvig tasks (e.g., Chi, Feltovich, & Glaser, 1981), adults typically perform cosistetly ad well o tasks requirig deotic reasoig (e.g., Cheg Correspodece should be addressed to D. D. Cummis, Psychology Departmet, Califoria State Uiversity, 6000 J Street, Sacrameto, CA ( dcummis@saclik.csus.edu). & Holyoak, 1985, 1989; Griggs & Cox, 1983; Maktelow & Over, 1991, 1995). I fact, so robust ad reliable is performace o deotic reasoig tasks that umerous proposals have bee put forth to explai it. The first explaatio is pragmatic reasoig schema theory (Cheg & Holyoak, 1985, 1989; Cheg, Holyoak, Nisbett, & Oliver, 1986). Accordigto this theory, adults excel at deotic reasoig because deotic cocepts costitute classes of frequetly ecoutered situatios for which collectios ofdomai-specific, goal-orieted rules are iduced. Oe such schema, the permissio schema, captures the relatioship betwee actios ad precoditios i a collectio ofrules, such as "Ifthe precoditio is satisfied, the the actio may be take." A secod theoretical explaatio is social exchage theory, which aalyzes deotic reasoig i terms ofcost/beefit aalysis ad cheater detectio (Cosmides, 1989; Cosmides & Tooby, 1994). These strategies are proposed to be iate, havig bee selected durig the evolutio ofour species to facilitate reasoig about social exchage (cooperative actio for mutual beefit). A third theory aalyzes deotic reasoig i terms ofviolatio-detectio strategies that are iate ad domai-specific for reasoig about deotic situatios (Cummis, i press). Fially, a fourth theory explais the deotic effect i terms ofthecostructio ad maipulatio ofmetal models o the basis ofsubjective utility (Maktelow & Over, 1990, 1991, 1995). Despite the very cosiderable differeces amog these theoretical explaatios, they all have oe thig i commo, ad that is the cetral role afforded violatio detectio i dea tic reasoig. As Gigerezer ad Hug (1992) poit out, a crucial part of reasoig deotically is appreciatig the ecessity of detectig violatios of deotic rules. For example, i the case of permissios, oe must esurethat o oe has take a specifiedactio without havig also satisfied some coditios (e.g., "Ifyou wat to take a book out of the library <permitted actio>, you must have a valid library card <coditio>"). 823 Copyright 1996 Psychoomic Society, Ic.

2 824 CUMMINS I the case of obligatios, oe must esure that o oe has avoided doig what is obligated (e.g., "If you lost a library book <coditio>, you must pay $25 i fies <obligatory actio>"). I the case of prohibitios, oe must esure that o oe has doe somethig forbidde (e.g., "No oe may slide dow the slide backward"). I cotrast, what costitutes optimal performace o idicative reasoig tasks (where reasoers are asked to test the truth of a rule) is the subject of some cotroversy. Traditioally, violatio detectio played a cetral role i ormative theories of idicative reasoig, because observig a violatio ofa idicative rule (hypothesis) disproves the rule (e.g., fidig a white rave disproves the rule "all raves are black") ad hece provides icotrovertible proofofits truth cotet (Popper, 1959). Seekig cofirmig evidece, which is the typical strategy employed by adults o idicative reasoig tasks, was cosidered at best a bias i the reasoig process ad at worst a error (Evas, 1989; Waso, 1968). Recetly, however, this assumptio has bee challeged by the argumet that a "seek cofirmatio" strategy ca be optimal uder certai circumstaces (Oaksford & Chater, 1994). However oe measures the ormative value ofthe strategies that adults typically employ o idicative ad deotic reasoig tasks, the fact remais that whe reasoig about deotic rules, adults spotaeously adopt a violatio-detectio strategy, ad whe reasoig about idicative rules, they spotaeously adopt cofirmatioseekig strategies (see Cummis, i press, for a review of this literature). Very little is kow, however, about whe these distict strategies emerge i childhood, owig primarily to the difficulty i developig tasks that work with populatios youger tha 6 or 7 years ofage (see, e.g., Girotto, Light, & Colboure, 1988; Komatsu & Galotti, 1986; Light, Blaye, Gilly, & Girotto, 1989; Light, Girotto, & Legrezi, 1990; Overto, Ward, Noveck, Black, & O'Brie, 1987; Sodia, Zaitchik, & Carey, 1991). I this paper, I describe the results oftwo experimets i which this issue was ivestigated with the use ofa task that allows childre's reasoig to be directly compared with that of adults. The task is based o the Waso reducedarray card selectio task, which requires the reasoer to choose whether to ispect potetially cofirmig or violatig cases whe testig a rule. The rules used here are "All squeaky mice are i the house" ad "All squeaky mice must stay i the house." The first is a idicative rule cocerig a factual state ofaffairs, whereas the secod is a deotic rule that specifies expected behavior. Discoverig squeaky mice who are outside the house falsifies the idicative rule; that is, it proves that the rule is false. But such discovery does ot disprove the deotic rule. Rather, it idetifies violators who have disobeyed the rule ad, presumably, will face appropriate cosequeces. The procedure employed is a modificatio ofoe used by Girotto et al. (1988) with 9- to 1O-year-old childre. Girotto et al. reported that childre i this age group were more likely to adopt a violatio-detectio strategy whe testig the rule "All buzzig bees must stay i the house" tha whe testig the idicative rule "All buzzig bees are i the house." Experimet 1 is a replicatio of Girotto et al. with a much youger age group, 3- ad 4-year-old childre. I Experimet 2, the tasks were modified i order to rule out competig explaatios ofthe deotic effect. Ifviolatiodetectio does emerge early as a deotic reasoig strategy, we would expect the same patter ofresults as is typically reported i the adult literature; that is, the childre should show a greater propesity to choose to ispect potetially violatig cases whe reasoig about a deotic rule tha they do whe reasoig about a idicative rule. EXPERIMENT 1 The methodology i this experimet replicates the crucial compoets ofgirotto et al. (1988), which is itself a modificatio of Waso's reduced array selectio task (Johso-Laird & Waso, 1970; Waso, 1968). I this task, reasoers are preseted with a coditioal rule (If <p>, the <q» i uiversally quatified form (All <p> are <q», ad are asked to test the rule by choosig whether to examie potetially cofirmig istaces (q) or potetially violatig istaces (). I order to iduce a deotic iterpretatio of the rule i the deotic task, the deotic reasoig task icluded a story i which a authority (Quee Mouse) uttered a prescriptive rule cocerig what the mice should do, ad the rule cotaied the modal "must" ("All squeaky mice MUST stay i the house"). I cotrast, the story i the idicative task described a hypothesis-testig situatio i which the experimeter uttered a descriptive rule (o modal) whose epistemic status eeded to be tested ("All squeaky mice ARE i the house"). It was predictedthat childre would be more likely to adopt a violatio-detectio strategy whe reasoig about a deotic rule tha whe reasoig about a idicative rule. Method Subjects. The subjects were 3- ad 4-year-old childre ( = 44 ad 40, respectively) recruited from day-care ceters ad preschools i Tucso. The age data for the 3-year-olds were as follows: rage = 3,1-3,11, Md = 3,6 i the deotic coditio, ad rage = 2,11-3, II, Md = 3,5 i the idicative coditio. For the 4-year-olds, rage = 4,0-4, II, Md = 4,4 i the deotic coditio, ad rage = 4,0-4, II, Md = 4,3 i the idicative coditio. Paretal coset was obtaied for each participat prior to participatio i the study. Materials. Materials icluded (I) 10 yellow rubber mice, 5 of which squeaked whe squeezed ad 5 of which did ot, (2) a "mouse house" 19.5 X 28 X 25 em, costructed from a large cardboard box ad spray-paited white, with a detachable 26 X 24 em door, (3) backyard play props, icludiga toy slide, picic table, ad tree. The backyard equipmet was arraged ext to the house so that mice iside ad outside the house were equally visible ad reachable. Oe squeaky ad oe osqueaky mouse (the latter will hereafter be referred to as quiet mice) were reserved for demostratio purposes. The remaiig eight mice were divided evely betwee the house ad the backyard, so that two quiet mice ad two squeaky mice appeared i each locatio. The mice were idetical i appearace. Detectio oftheir differece required squeezig them.

3 REASONING SCHEMAS 825 Procedure. Childre were tested idividually i their day-care ceter or preschool. The locatio of the house (o the child's right or left) was couterbalaced betwee subjects. Sessios were videotaped. All sessios bega with the followig verbatim protocol: <Name of child>, I'd like you to playa preted game with me. I'd like you to preted that this is a house, ad these mice [poitig to mice] live there. That's their home. Some of the mice have goe out ito the backyard to play, see? They're havig a good time, slidig dow the slide, ad playig o the picic table, ad aroud the tree. So these mice are i the house [poitig] ad these mice are i the backyard playig [poitig]. At ight everybody goes i the house, ad they close the door [door is affixed] so that they're ice ad safe ad othig ca get them. [Door removed.] Now <ame of child>, would you do me a quick favor? Would you poit to the mice that are i the house? Ad would you poit to the mice that are i the backyard? I have somethig else very iterestig to tell you about these mice. See how they all look exactly alike? But they're really differet because some of them squeak [demo squeaky mouse squeezed] ad some of the them are quiet [demo quiet mouse squeezed]. This is a squeaky mouse [squeezed] ad this is a quiet mouse [squeezed]. Squeeze this mouse ad you'll see it makes a lot of oise [child was allowed to squeeze squeaky mouse]. Now squeeze this oe [child was allowed to squeeze quiet mouse]. See? It's really quiet. It does't make ay oise at all. The deotic protocol cotiued as follows: I have somethig else iterestig to tell you. Sometimes i the eveig, the mice like to go out i the backyard to play. But whe the squeaky mice play, they get really excited ad they start to squeak, like this [squeezig mouse repeatedly]. Ad the you kow what happes? The eighborhood cat hears that squeakig, ad he comes ruig, ad pouces, ad chases the mice all aroud. [Stuffed toy cat appeared, chasig the mouse.] So it's ot safe for the squeaky mice outside. It's oly safe for the quiet mice. Now, the Quee Mouse heard about this. [Miie Mouse doll appeared.] The Quee Mouse is their mama, ad she makes importat rules that everyoe has to follow. So let's liste carefully because she's goig to make a rule ad we're goig to have to make sure obody disobeys. Quee Mouse: "Oh, I'm so worried about the squeaky mice because of that cat. It's ot safe outside for the squeaky mice because of that cat, so I'm goig to make a rule, ad the rule is ALL SQUEAKY MICE MUST STAY IN THE HOUSE. YES, ALL SQUEAKY MICE HAVE TO STAY IN THE HOUSE." She said all the squeaky mice must stay i the house, so let's make sure obody is breakig that rule, Ok? Let's make sure o oe is disobeyig the Quee. Which mice should we check, those that are i the house, or those that are i the backyard playig? The order of the query (iside first or outside first) was couterbalaced betwee subjects. The idicative coditio protocol cotiued after the geeral itroductio with the followig: Now, I'm goig to tell you somethig, but I might be trickig you, so listecarefully. I kow somethig. I kow that ALL THE SQUEAKY MICEARE IN THE HOUSE. YES, ALL THE SQUEAKY MICE ARE IN THE HOUSE." Was I trickig you? Which mice should we squeeze to fid out whether I was trickig you, the oes that are i the house, or the oes that are i the backyard playig? The order of the query (iside first or outside first) was couterbalaced betwee subjects. The child was required first to poit ad state verbally which mice had to be squeezed to test the rule, ad was the allowed to actually squeeze them. Results ad Discussio All childre chose to squeeze oly the mice that they had i fact poited to i orderto aswer the iitial query. The umber ofchildre i each age group who idicated that the mice i the backyard had to be checked (i.e., the violatig case) is show i Table I. Table 1 Number Correct (- q) Selectios Amog 3- ad 4-Year-Olds i Deotic ad Idicative Reasoig Cotexts (Experimet 1) Age 3 4 Deotic Reasoig Cotext 22 EXPERIMENT 2 22 Although the results of Experimet I are compellig, they do ot provide uequivocal evidece for the deotic effect due to the structure ofgirotto et al.'s (1988) embeddig stories. It might have bee the case that childre i the deotic coditio were more likely to check the mice who were outside because of the dager posed by the lurkig cat, ad ot because ofay particular propesity to adopt a violatio-detectio strategy oly whe reasoig about deotic situatios. To rule out this possibility, the procedure was modified to iclude the cat ad all of its ramificatios i the idicative story coditio. A secod modificatio was the removal of the referece to "trickig" o the part ofthe experimeter. Although childre as youg as 2 years of age are capable of deliberately lyig about their actios, reliable ad cosistet reasoig about acts ofdeceptio ad false beliefs i others are difficult to demostrate i populatios youger tha 4 (see Leekam, 1992; Leslie, 1994; ad Leslie & Roth, 1993, for reviews ofthis literature). Istead, Miie Mouse was itroduced as a potetially ureliable character who stated that all the squeaky mice were i the house, ad 7 7 Idicative Amog 3-year-olds, 68% selected the case i the deotic coditio, whereas oly 32% did so i the idicative coditio, a doublig i performace that was statistically reliable [G2(1) = 5.95,p <.02]. Amog 4-yearolds, 85% selected the case i the deotic case, whereas oly 35% did so i the idicative coditio, a differece that was also statistically reliable [G2(1) = 11.03, p <.0 I]. These results replicate those ofgirotto et al. (1988) with a much youger age group, thereby idicatig that childre as youg as 3 years of age adopt a violatiodetectio strategy whe reasoig about deotic rules. I cotrast, they do ot adopt this type ofstrategy whe reasoig about idicative rules; they adopt istead a cofirmatio-seekig strategy. I other words, by the age of3, the deotic-idicative distictio is already apparet i childre's reasoig strategies. The magitude of this performace differece is approximately the same as that observed i adult performace (see, e.g., Cheg & Holyoak, 1985, 1989; Griggs & Cox, 1983; Maktelow & Over, 1991). This suggests that a distictio betwee reasoig about what oe should do (deotic) ad determiig the epistemic status ofa rule (idicative) emerges quite early i developmet ad persists ito adulthood.

4 826 CUMMINS the childre were asked to fid out whether or ot she was wrog. Although this seems to require reasoig about false beliefs, this procedure has bee used successfully with 3-year-olds o false belieftasks (Wellma& Bartsch, 1988; Zaitchik, 1991). The major differece betwee this ad the stadard false belief task is that the childre do ot kow where the squeaky mice really are, ad so they are ot required to choose betwee the curret reality (where the mice really are) ad a couterfactual belief (where someoe falsely believes them to be). Ifthe results of Experimet 1were i fact due to greater propesity to seek violatios whe reasoig about deotic rules, the the results ofexperimet 1 should be replicated. Ifthe differece i reasoig strategies observed i Experimet 1 was istead due to cocer about keepig the mice safe from the cat, the childre shouldshow a preferece for checkig the outside () mice i both the deotic ad the idicative coditios. Method Subjects. Subjects were 3- ad4-year-old childre ( = 70 ad 40,respectively) recruited from day-care ceters ad preschools i Tucso. The age data forthe 3-year-olds were as follows: rage = 2,6-3,11, Md = 3,6 ithedeotic coditio, ad rage = 2,11-3,11, Md = 3,5 itheidicative coditio. Forthe4-year-olds, rage = 4,0-4,II, Md = 4,6 ithedeotic coditio, ad rage = 4,0-4,II, Md = 4,6itheidicative coditio. Paretal coset was obtaied foreach participat prior to participatio i thestudy. Materials. The same materials asthose iexperimet I were used here. Couterbalacig was otcompletely eve amog 3-year-olds, owig toueve class sizes. Procedure. Allsessios were audiotaped. The deotic storywas idetical totheoeused iexperimet I. The idicative story coditio was modified sothatthe sectio followig theitroductio cotiued thus: I have somethig else iterestig to tell you. Sometimes i the eveig, the mice like to go out ithe backyard to play. But whe the squeaky mice play, they get really excited ad they start to squeak, like this [squeezig mouse repeatedly]. Ad the you kow what happes? The eighborhood cat hears that squeakig, ad he comes ruig, ad pouces, ad chases the mice all aroud. [Stuffed toy cat appeared, chasig the mouse.] So it's ot safe for the squeaky mice outside. It's oly safe for the quiet mice. Now, I'd like you tomeet someoe. This is Miie Mouse. [Miie Mouse stuffed aimal appeared.] Miie likes totell kids thigs, so let's liste carefully because we're goig to have to figure out whether what she says isright orwrog. Miie Mouse: "Hi, <ame ofchild>! Ikow somethig about the mice. Uh, huh! I kow that IT'S NOT SAFE OUTSIDE FOR THE SQUEAKY MICE BECAUSE OF THAT CAT, SO ALL THE SQUEAKY MICE ARE IN THE HOUSE! YES, ALL THE SQUEAKY MICE ARE IN THE HOUSEl" Now, I woder ifshe could be wrog about that. Let's fid out. Which mice should we squeeze to fid out ifshe's wrog, the mice that are iside the house, or the mice that are i the backyard playig? Results ad Discussio Oe 3-year-old i the deotic coditio chose to squeeze all of the mice despite havig idicated "outside" durig the iitial query, so his data were excluded from aalysis. The remaiig umber of childre correctly selectig the case (backyard) is show i Table 2. I the deotic coditio, 62% of the 3-year-olds selected the mice playig i the backyard i the deotic coditio, whereas oly 37% did so i the idicative co- Table 2 Number Correct() Selectios Amog 3- ad 4-Year-Olds i Deotic ad Idicative Reasoig Cotexts (Experimet 2) Age Reasoig Cotext Deotic Idicative GENERAL DISCUSSION 35 ditio, a relatioship that was sigificat [G2(l) = 4.23, p <.05]. Similarly, 80% ofthe 4-year-olds correctly idicated that the mice playig i the backyard eeded to be checked, whereas oly 30% did so i the idicative coditio, a statistically reliable effect [G2(l) = 10,60, p <.01]. Compariso ofthese percetages with the oes obtaied i Experimet 1 suggest that the modificatio i the idicative story coditio had little effect o the childre's performace. I Experimet 1, 32% of the 3-year-olds selected ; 37% did so here. Amog 4-year-olds, 35% selectedthe case, whereas 30% did so here. Give the similar levels of performace o the idicative-cotet task across the two experimets, it seems reasoable to coclude that the crucialfactor ifluecigchildre sreasoig was the deotic cotext, ad ot the dager posed by the cat i the stories. Moreover, the differece i performace levels betwee the coditios is particularly strikig whe oe cosiders how similar the reasoig rules were-that is, "ALL SQUEAKY MICE MUST STAY IN THE HOUSE" versus "ALL SQUEAKY MICE ARE IN THE HOUSE." The oly differece is to be foud i the deotic iterpretatio afforded by the deotic story cotext ad the presece of the modal "must." The data reported here clearly show that childre as youg as 3 years ofage adopt a violatio-detectio strategy whe reasoig about deotic situatios ad a cofirmatio-seekig strategy whe reasoig about idicative tasks. These results strogly suggest that these domai-specific reasoig strategies emerge early i developmet, beig i place by at least the 3rd year oflife. These results are cosistet with the explosio of research showig evidece of early emergig domaispecific reasoig capacities, icludig reasoig about the permaece ad rigidity of objects as well as costraits o their motio (Baillargeo, 1987, 1994; Spelke, 1994), physical causality (Leslie, 1984; Leslie & Keeble, 1987), biomechaical movemet (Bertethal, 1984, 1985), the abstract cocept ofumber ad arithmetic operatios (Starkey, Spelke, & Gelma, 1990; Wy, 1992), the meaig ofemotioal facial expressios (Campos & Steberg, 1981), the reciprocal ature of certai social iteractios (Vadell & Wilso, 1987), otological category (Carey, 1985; Keil, 1986), ad certai aspects of moral reasoig (Miller, 1986; Miller & Bersoff, 1988; Turiel, 1983, 1989).

5 REASONING SCHEMAS 827 These results do ot provide uequivocal support for the stroger view, take by may developmetalists i explaiig their results, that early emergece occurs because deotic reasoig strategies are iate. The possibility exists that by the age of3, youg childre have experieced a sufficiet umber of deotic situatios to eable them to iduce a suitably abstract schema. This is a explaatio that is cosistet with pragmatic reasoig schema theory i that it attributes early emergece to the frequecy ad urgecy with which deotic reasoig situatios preset themselves durig early childhood. I other words, there is ample opportuity to iduce a schema for reasoig about deotic situatios. Alteratively, the childre i this study might have bee reasoig aalogically, detectig sufficiet similarity betwee the mice's deotic cotext ad oes that they themselves have experieced to allow them to retrieve relevat kowledge cocerig the outcomes ofobediet ad disobediet behaviors from memory. There are other reasos, however, to believe that deotic reasoig strategies might i fact costitute a iate part ofour cogitive architecture. Cummis (i press) argues that the deotic effect provides a uique glimpse ito the iate structure of the huma reasoig architecture. At the heart of this positio lies a evolutioary argumet. Evolutioary theory is based o the assumptio that there is a causal relatioship betwee the adaptive problems that a species repeatedly ecouters durig its evolutio ad the desig ofits pheotypic structures. My positio is that oe ofthe largest selective pressures faced durig the evolutio of our reasoig architecture was the eed to reaso effectively about domiace hierarchies, ad that survival withi the domiace hierarchy requires effective deotic reasoig strategies. I fuctioal terms, a domiace hierarchy is simply the statistical observatio that "particular idividuals i social groups have regular priority of access to resources... i competitive situatios" (Clutto-Brock & Harvey, 1976). It is a characteristic of early all mammalia ad avia societies. The role ofdomiace is most proouced i situatios characterized by high levels of competitio for resources, such as high populatio desity or the oset of breedig seaso. This suggests that strog evolutioary pressure has favored the evolutio of reasoig capacities aimed specifically at recogizig ad exploitig social domiace relatios, a coclusio that is supported by empirical observatio. Cosider first the cocept of permissio. Those who curretly domiate resources determie who may egage i which activities whe, ad they puish trasgressors. For example, domiat male primates moopolize reproductio opportuities by aggressio agaist females ad subordiate males who are caught cosortig (de Waal, 1982). To avoid agoistic ecouters, it is therefore crucial to reaso effectively about what is permitted ad what isforbidde. It is i the iterest ofsubordiates, o the other had, to broade their access to available resources. I other words, it is i their iterest to move up i rak. Amog primates, domiace rakig is ot cor- related with size. Istead, oe's rak i the hierarchy depeds crucially o the ability to form ad maitai strog alliaces, ad alliaces are formed ad maitaied o the basis ofreciprocal obligatios, aother deotic cocept. For example, Cheey ad Seyfarth (1990, pp ) have reported that vervet mokeys are more likely to respod to calls from o-ki durig agoistic ecouters if the caller has groomedthem recetly; they also form the strogest alliaces with idividuals who groom them most ofte. This patter ofbehavior suggests a appreciatio ofa obligatio structurewherei groomigobligates the groomee to itervee o the part ofthe groomer i future agoistic ecouters, while givig the groomee grouds to expect such assistace, ad hece permissio to egage i ecouters he or she would otherwise avoid. Primate field studies, therefore, suggest that sesitivity to deotic cotet plays a crucial role i determiig a idividual's fate withi primate social groups. The eed to reaso effectively about permissio/prohibitio structures ad reciprocal obligatios looms exceedigly large i primate commuities. Failure to adhere to permissio ad obligatio structures leads to baishmet from the social group, a situatio that ca have disastrous cosequeces for survival. Clearly, ifour reasoig architecture evolved i respose to the eed to reaso effectively about adaptively crucial problems, ad ifsurvival depeds crucially o stayig withi the social group, the few problems carry greater survival cosequeces amog social species tha those ivolvig deotic cotets. This strogly suggests that deotic reasoig strategies-or their precursors-are part ofour primate geetic heritage. For this reaso, we would expect to observe the early emergece ofviolatio-detectio as a domai-specific deotic reasoig strategy, just as we have observed here with 3-year-old childre. There are two importat cotrasts with regard to the evolutioary explaatio offered by Cummis (i press) ad the social exchage theory proposed by Cosmides ad Tooby. The first is whether the deotic effect is species specific-that is, particular to humas. Social exchage theory explais the deotic effect i terms ofa cheater-detectio reasoig strategy shaped ad selected by evolutioary forces specifically to facilitate reasoig about social exchage, a activity that is most developed amog humas ad that appeared most strikigly amog humas durig the Pleistocece (Cosmides, 1989; Cosmides & Tooby, 1989, 1992, 1994). The evidece cited from the primatology literature suggests that violatio-detectio is ot specific to our species. The secod is whether the deotic effect reflects a strategy that is specialized for reasoig about social exchage or for deotic situatios broadly defied. Deotic situatios are ay situatios that require reasoig about what is socially obliged, permitted, prohibited, cautioed, or advised (what oe ought to do). Social exchage is just oe type of deotic situatio. It is difficult to frame the curret tasks i terms of social exchage, sice there is o cooperatio betwee idividuals for mutual beefit; istead, there is oly obediece to a authority's pre-

6 828 CUMMINS scriptive rule. For this reaso, perhaps the most parsimoious explaatio of the results reported here is that the early emergece of a domai-specific violatiodetectio strategy reflects a iate propesity for recogizig deotic rules ad the ecessity of seekig violatios of them. REFERENCES BAILLARGEON, R. (1987). Object permaece i 3Yz- ad 4 Yz-mothold ifats. Developmetal Psychology, 23, BAILLARGEON, R. (1994). How do ifats lear about the physical world? Curret Directios i Psychological Sciece, 3, BERTENTHAL, B. I. (1984). Ifat sesitivity to figural coherece i biomechaical motios. Joural ofexperimetal Child Psychology, 37, BERTENTHAL, B. I. (1985). The developmet of sesitivity to biomechaical motios. Child Developmet, 56, CAMPOS, J. J., & STENBERG, C. (1981). Perceptio, appraisal, ad emotio: The oset of social referecig. I M. Lewis & L. Roseblum (Eds.), Ifat social cogitio: Empirical ad theoretical cosideratios (pp ). Hillsdale, NJ: Erlbaum. CAREY, S. (1985). Coceptual chage i childhood. Cambridge, MA: MIT Press. CHENEY, D. L., & SEYFARTH, R. M. (1990). How mokeys see the world. Chicago: Uiversity ofchicago Press. CHENG, P.w., & HOLYOAK, K. J. (1985). Pragmatic reasoig schemas. Cogitive Psychology, 17, CHENG, P. W., & HOLYOAK, K. J. (1989). O the atural selectio of reasoig theories. Cogitio, 33, CHENG, P.w., HOLYOAK, K. J., NISBETT, R. E., & OLIVER, L. M. (1986). Pragmatic vs sytactic approaches to traiig deductive reasoig. Cogitive Psychology, 18, CHI, M. T. H., FELTOVICH, P. J., & GLASER, R. (1981). Categorizatio ad represetatio ofphysics problems by experts ad ovices. Cogitive Sciece, 6, CLUTTON-BROCK, T. H., & HARVEY, P. H. (1976). Evolutioary rules ad primate societies. I P. P. G. Bateso & R. A. Hide (Eds.), Growig poits i ethology (pp ). Cambridge: Cambridge Uiversity Press. COSMIDES, L. (1989). The logic of social exchage: Has atural selectio shaped how humas reaso? Studies with the Waso selectio task. Cogitio, 31, COSMIDES, L., & TOOBY, J. (1989). Evolutioary psychology ad the geeratio culture: Part I!. Case study: A computatioal theory of social exchage. Ethology & Sociobiology, 10, COSMIDES, L., & TOOBY, J. (1992). Cogitive adaptatios for social exchage. I 1. Barkow, L. Cosmides, & 1. Tooby (Eds.), The adapted mid: Evolutioary psychology ad the geeratio of culture (pp ). New York: Oxford Uiversity Press. COSMIDES, L., & TOOBY, J. (1994). Beyod ituitio ad istict blidess: Toward a evolutioarily rigorous cogitive sciece. Cogitio, 50, CUMMINS, D. D. (i press). Evidece for the iateess ofdeotic reasoig. Mid & Laguage. DEWAAL, F. B. M. (1982). Chimpazee politics. New York: Harper & Row. DUNN, J. (1988). The begiigs ofsocial uderstadig. Oxford: Basil Blackwell. EVANS, J. ST. B. (1989). Bias i huma reasoig. Hillsdale, NJ: Erlbaum. GIGERENZER, G., & HUG, K. (1992). Domai-specific reasoig: Social cotracts, cheatig, ad perspective chage. Cogitio, 43, GIROTTO, v., LIGHT, P.,& COLBOURN, C. J. (1988). Pragmatic schemas ad coditioal reasoig i childre. Quarterly Joural ofexperimetal Psychology, 40A, GRIGGS, R. A., & Cox, J. R. (1983). The effects ofproblem cotet ad egatio o Waso's selectio task. Quarterly Joural ofexperimetal Psychology, 35A, HILPINEN, R. (1971). Deotic logic: Itroductory ad systematic readigs. Bosto: ReidellKluwer. HILPINEN, R. (1981). New studies i deotic logic: Norms, actios, ad the foudatios ofethics. Bosto: ReidellKluwer. JOHNSON-LAIRD, P. N., & WASON, P.C. (1970). Isight ito a logical relatio. Quarterly Joural ofexperimetal Psychology, 22, KAHNEMAN, D., SLOVIC, P., & TVERSKY, A. (1982). Judgemet uder ucertaity: Heuristics ad biases. Cambridge: Cambridge Uiversity Press. KEIL, F. C. (1986). The acquisitio ofatural kid ad artifact terms. I W. Demopoulos & A. Marras (Eds.), Laguage learig ad cocept acquisitio: Foudatioal issues (pp ). Norwood, NJ: Ablex. KOMATSU, L. K., & GALOTTI, K. M. (1986). Childre's reasoig about social, physical, ad logical regularities: A look at two worlds. Child Developmet, 57, LEEKAM, S. R. (1992). Believig ad deceivig: Steps to becomig a good liar. I S. 1. Ceci, M. DeSimoe Leichtma, & M. E. Putick (Eds.), Cogitive ad social factors i early deceptio (pp ). Hillsdale, NJ: Erlbaum. LESLIE, A. M. (1984). Spatiotemporal cotiuity ad the perceptio of causality i ifats. Perceptio, 13, LESLIE, A. M. (1994). ToMM, ToBY, ad agecy: Core architecture ad domai specificity. I L. A. Hirshfel & S. A. Gelma (Eds.), Mappig the mid: Domai specificity i cogitio ad culture. Cambridge: Cambridge Uiversity Press. LESLIE, A. M., & KEEBLE, S. (1987). Do six-moth-old ifats perceive causality? Cogitio, 25, LESLIE, A. M., & ROTH,D. (1993). What autism teaches us about misrepresetatio. I S. Baro-Cohe, H. Tager-Flusberg, & D. 1.Cohe (Eds.), Uderstadig other mids: Perspectivesfrom autism (pp. 83 III). Oxford: Oxford Uiversity Press. LIGHT, P., BLAYE, A., GILLY, M., & GIROTTO, V. (1989). Pragmatic schemas ad logical relatios i 6- to 8-year-old childre. Cogitive Developmet, 4, LIGHT, P., GIROTTO, v., & LEGRENZI, P. (1990). Childre's reasoig o coditioal promises ad permissios. Cogitive Developmet, 5, MANKTELOW, K. I., & OVER, D. E. (1990). Deotic thought ad the selectio task. I K. 1.Gilhooly, M. T. Keae, R. H. Logie, ad G. Erdos (Eds.), Liesofthikig (Vol. I, pp ). Chichester, U.K.: Wiley. MANKTELOW, K. I., & OVER,D. E. (1991). Social roles ad utilities i reasoig with deotic coditioals. Cogitio, 39, MANKTELOW, K. I., & OVER, D. E. (1995). Deotic reasoig. I S. E. Newstead & 1. St. B. Evas (Eds.), Perspectives o thikig ad reasoig(pp ). Hillsdale, NJ: Erlbaum. MILLER, J. (1986). Early cross-cultural commoalities i social explaatio. Developmetal Psychology, 22, MILLER, J., & BERSOFF, D. M. (1988). Whe do America childre ad adults reaso i social covetioal terms? Developmetal Psychology, 24, OAKSFORD, M., & CHATER, N. (1994). A ratioal aalysis ofthe selectio task as optimal data selectio. Psychological Review, 101, OVERTON, W., WARD, S., NOVECK, I., BLACK, J., & O'BRIEN, D. (1987). Form ad cotet i the developmet of deductive reasoig. Developmetal Psychology, 23, POPPER, K. (1959). The logic ofscietific discovery. Lodo: Hutchiso. SODIAN, B., ZAITCHIK, D., & CAREY, S. (1991). Youg childre's differetiatio ofhypothetical beliefs from evidece. Child Developmet, 62, SPELKE, E. (1994). Iitial kowledge: Six suggestios. Cogitio, 50, STARKEY, P.,SPELKE, E. S., & GELMAN, R. (1990). Numerical abstractio by huma ifats. Cogitio, 36, TuRIEL,E. (1983). The developmet ofsocial kowledge: Morality ad covetio. Cambridge: Cambridge Uiversity Press. TuRIEL,E. (1989). Domai-specific social judgmets ad domai ambiguities. Merrill-Palmer Quarterly, 35, 89-1 J4.

7 REASONING SCHEMAS 829 YANDELL, L., & WILSON, K. S. (1987). Ifat's iteractios with mother, siblig, ad peer: Cotrasts ad relatios betwee iteractio systems. Child Developmet, 58, WASON, P. C. (1968). Reasoig about a rule. Quarterly Joural ofexperimetal Psychology,, WASON, P. c.,& JOHNSON-LAIRD, J. L. (1972). Psychology ofreasoig: Structure ad cotet. Lodo: Batsford. WELLMAN, H. M., & BARTSCH, K. (1988). Youg childre's reasoig about beliefs. Cogitio, 30, WYNN, K. (1992). Additio ad subtractio by huma ifats. Nature, 358, ZAITCHIK, D. (1991). Is oly seeig really believig? Sources of true beliefs i the false belieftask. Cogitive Developmet, 6, 91- I03. (Mauscript received October 13,1995; revisio accepted for publicatio February I I, 1996.)