Read zelazo.pdf text version

Intention, Act, and Outcome in Behavioral Prediction and Moral Judgment

Philip David Zelazo, Charles C. Helwig, and Anna Lau

University of Toronto

ZELAZO, PHILIP DAVID; HELWIG, CHARLES C ; and LAU, ANNA. Intention, Act, and Outcome in

Behavioral Prediction and Moral Judgment. CHILD DEVELOPMENT, 1996,67,2478-2492. 72 children at 3,4, and 5 years of age and 24 undergraduates were required to use information about intention under a normal causal system or a noncanonical one (e.g., hitting causes pleasure) to predict an agent's behavior. Additionally, they were asked to integrate intentions, acts, and outcomes to judge an act's acceptability and assign punishment. 3-year-olds performed poorly on behavioral prediction in the noncanonical condition. Most participants at all ages made categorical judgments of act acceptability based solely on outcome, although quantitative ratings reflected an age-related increase in sensitivity to intention information. When assigning punishment, many 3-year-olds used a simple intention or outcome rule, whereas older participants were more likely to use a conjunction rule (if outcome is negative and intention is negative then punish). Together, the results reveal both an early understanding of harm and changes in the complexity of the rules that children use to predict behavior and integrate information.

A large body of researcb on social reasoning has demonstrated that, by the preschool years, children differentiate moral acts involving harm and unfairness from social conventions, judging the former to be independent of external authority, punishment, and explicit sanctions (Nucci & Turiel, 1978; Smetana, 1981, 1985; Smetana & Braeges, 1990; Smetana, Kelly, & Twentyman, 1984). For example, in a study by Smetana and Braeges (1990), children judged the permissibility, seriousness, generalizability, and rule and authority contingency of both moral and conventional transgressions. In addition, these children rated the amount of punishment that the transgressors deserved. Older children (42 months) distinguished moral and conventional acts on all criteria, whereas the youngest children (24 months) did not distinguish these acts on any of the criteria. The intermediate age group (34 months) judged moral transgressions to be more generalizably wrong (i.e., wrong across social contexts) tban social conventional transgressions, although they did not distinguish these acts on any other dimension.

Although these findings suggest that children have begun to formulate a distinct domain of moral judgment by the end of the third year of life, the basis for children's judgments that moral acts are wrong across social contexts remains largely unexplored. Justifications are commonly used to address this question in research with older participants (Turiel, 1983), but heavily languagebased techniques are not feasible witb young prescboolers. Researchers working from what has come to be known as a social domain perspective (Smetana & Braeges, 1990; Turiel, Killen, & Helwig, 1987) have proposed that young children abstract out properties of events to form distinctions between moral, personal, and socialconventional domains of social judgment. In the case of the moral domain, this view leads to certain expectations about the relations between what have been called the substantive features of moral concepts (e.g., harm) and the formal criteria (e.g., generalizability) proposed jointly to comprise the moral domain (Helwig, Tisak, & Turiel, 1990; Turiel, 1989). One expectation is that children should use the concept of harm to evaluate

The research reported herein was supported by a grant from NSERC of Canada to P. D. Zelazo and a grant from SSHRC of Canada to C. Helwig. A. Lau is now at the Department of Psychology, UCLA. The authors would like to thank Doug Frye, Karen Li, David Moshman, Steve Reznick, Elliot Turiel, and two anonymous reviewers for very helpful comments on an earlier draft of the article. Please send correspondence to P. D. Zelazo or C. C. Helwig, Department of Psychology, University of Toronto, Toronto, Ontario, Canada M5S 3G3 (e-mail: [email protected] or [email protected]). [Child Development, 1996,67,2478-2492. © 1996 by the Society for Research in Child Development, Inc. All rights reserved. 0009-3920/96/6705-0036$01.00]

Zelazo, Helwig, and Lau

acts either before, or at the same time that, they treat moral acts as generalizable across contexts. If judgments of generalizability occurred before children used harm to guide the evaluation of acts, this would point to other origins of early domain distinctions. An alternative account, for example, might propose that the existence of an invariant association between specific acts, like hitting, and punishment or adult sanctions leads to the judgment that these acts exclusively are wrong across social contexts, unlike other acts. This alternative account would allow for the finding that children distinguish certain acts (labeled as moral) according to a set of formal criteria prior to learning, tbrough modeling or some other social communication mechanism, that harm is the basis of adult sanctions regarding moral acts (see Edwards, 1987). Because moral acts have intrinsic consequences (Turiel, 1983), it is extremely difficult to test these alternative explanations of early moral judgments. Indeed, it is impossible, under normal circumstances, to dissociate acts (e.g., hitting) from their outcomes (e.g., harm) in order to test the hypothesis (Turiel et al., 1987) that concepts of harm underlie moral judgments in the preschool years. However, it should be possible to detennine the basis of young children's moral judgments by creating hypothetical events in which the normal relations between acts and outcomes are altered or reversed. If cbildren's judgments about moral acts such as hitting stem from an appreciation of the harm that these acts cause, and are not simply due to a negative association between external sanctions and the act itself, then young children should judge it to be acceptable to bit under circumstances wbere hitting leads to a positive consequence (pleasure). Likewise, they should judge it to be unacceptable to engage in a normally acceptable act (e.g., petting an animal) when this act leads to harm. Exploration of young children's reasoning in these situations of noncanonical causality may therefore provide important information about the features of acts to which young children are responsive. The current study used a normal and a noncanonical causal system (e.g., hitting causes pleasure) to examine preschoolers' and adults' ability to use information about intentions, acts, and outcomes to make behavioral predictions and moral judgments. Reasoning about two acts (hitting vs. petting hypothetical animals), committed by an

2479

agent under different circumstances, was examined. The circumstances differed in the agent's intention (benevolent vs. malevolent), the nature of the causal relation between the act and its associated outcome (normal vs. noncanonical), and the outcome itself (harmful vs. beneficial). Invented animals were used in order to provide a plausible context for the noncanonical causal relations. In the normal condition, the animal "feels good and smiles" when petted and "gets hurt and cries" when hit. In the noncanonical condition, however, the animal "gets hurt and cries" when petted and "feels good and smiles" when hit. The use of a noncanonical condition allowed us first to determine whether children were able to consider botb act-outcome relations and intentions in order to predict an agent's act, and second, to determine the extent to which children's moral evaluations are influenced by outcomes independent of acts. In accordance with social domain perspectives on the construction of moral judgments (Turiel et al., 1987), it was expected that even the youngest children (3-year-olds) would use outcome information to evaluate acts in the noncanonical condition, indicating that early moral judgments are oriented toward the moral features of acts (e.g., harm) rather than the acts themselves. Recent research on moral judgment has shown that, contrary to traditional developmental accounts (e.g., Piaget, 1932), young children take intentions into account in judging moral agents, especially when levels of outcome and intention are not confounded and when information about intentions is made explicit and salient (Farnill, 1974; Keasey, 1978; Leon, 1980; Nelson, 1980; Nelson-LeCall, 1985; Yuill, 1984). The earliest age at which intentions-based moral judgment bas been found is 3 years; in a study by Nelson (1980), 3-year-olds judged well-intentioned actors more favorably tban ill-intentioned actors regardless of the consequences. Intentions-based moral judgment has also been found in 3y2-year-olds by Yuill (1984) and in somewhat older preschoolers in many other studies (e.g., Berndt & Berndt, 1975; Feldman, Klosson, Parsons, Rholes, & Ruble, 1976; Moran & O'Brien, 1983; Shultz, Wright, & Schleifer, 1986; Surber, 1977; Wellman, Larkey, & Somerville, 1979). Although these studies demonstrate that young children can use information about intentions in moral judgments, tbere is ample evidence that intentions become more important and are weighted

2480

Child Development

rule systems (where complexity is measured by the degree of embedding in the systems). Further increases in rule use occur within age-related constraints on complexity. For example, it is not until 5 years of age that children first acquire the ability to combine two pairs of rules and use them simultaneously or in rapid succession (as in a matrix classification task; see Frye, Zelazo, & Palfai, 1995, Exp. 3; Inhelder & Piaget, 1964; Roberts & Fischer, 1979). The Cognitive Complexity and Control Theory accounts for age-related changes in reasoning in a variety of domains (e.g.. Das Gupta & Bryant, 1989; Diamond, Towle, & Boyer, 1994; Frye et al., 1995; Frye, Zelazo, Brooks, & Samuels, 1996; Kendler, Kendler, & Wells, 1960; Kessen & Kessen, 1961; Rudy, Keith, & Georgen, 1993; Russell, Mauthner, Sharpe, & Tidswell, 1991; Zelazo, Frye, & Rapus, 1996; Zelazo & Reznick, 1991; Zelazo, Reznick, & Pifion, 1995). The Cognitive Complexity and Control Theory predicts that children's use of information about intentions, acts, and outcomes will be constrained by more general developments in rule use. Specifically, children's ability to make behavioral predictions in the noncanonical causality condition will be constrained by their inability to use higherorder rules. In this condition, children must switch from their usual representation of the canonical relation between hitting and harm, and reason from a confiicting perspective where hitting leads to pleasure. Because they are required first to consider the system of causality and then to take into account the intentions of the actor (either benevolent or malevolent) before making the behavioral prediction, this condition involves the use of a higher-order rule: If X, then if Y, then Z. Thus, it was expected that 3-year-olds, but not 4- and 5-year-olds, would have difficulty coordinating information about intentions and act-outcome relations in order to make correct behavioral predictions. The actoutcome relation can be ignored in the normal condition, so a single pair of rules (if intention is positive then pet; if intention is negative then hit) will suffice. Thus, the theory makes the prediction that 3-year-olds, who are restricted to the use of a simple pair of rules, will make accurate behavioral predictions in the normal, but not the noncanonical, causality condition. Similarly, the theory predicts that 3year-olds will differ from 5-year-olds and from adults in act evaluations where older children and adults sometimes or typically

more heavily with age (Buchanon & Thompson, 1973; Costanzo, Coie, Grumet, & Farnill, 1973; Gutkin, 1972; Hebble, 1971; Helwig, Hildebrandt, & Turiel, 1995; Leon, 1980; Surber, 1977). In the current study, it was therefore expected that older children and adults would make greater use of intention information in their moral judgments than would younger children. Age-related changes in moral reasoning are likely to be constrained by more general changes in cognitive functioning. Accordingly, the present study draws upon research on rule use and information integration to make predictions about the characteristics of children's reasoning at different ages. Although there is evidence that young children can use information about intentions and outcomes, it was expected that the simultaneous presentation of multiple types of information would increase the complexity of the rules required to predict behavior and to make moral judgments. When various types of information must be coordinated, young preschoolers appear to focus exclusively on one dimension (intentions or outcomes), whereas older preschoolers often combine this information (Nelson, 1980; Surber, 1977; Yuill, 1984). Similar changes have been found in other domains, and it has been suggested that increased information integration instantiates a general developmental trend (e.g., Anderson & Cuneo, 1978; Kun, Parsons, & Ruble, 1974; Zelazo & Shultz, 1989). Integration rules are equivalent to the simultaneous use of two pairs of explicit rules, which has been postulated to emerge at about 5 years of age as part of tbe development of deliberate reasoning structures (Zelazo & Frye, in press). According to this approach, called the Cognitive Complexity and Control Theory, 3-year-olds can use a pair of rules, but they cannot use a higher-order rule to determine which of two incompatible pairs of rules to use. For example, they can sort cards by color (blue here, red there) or by shape (rabbits here, boats there), but tbey cannot switch fiexibly from sorting by color to sorting by shape. To switch, children must use a higher-order rule that permits them first to determine which dimension to use and then to determine which card is presented. By about 4 years of age, children can use a higher-order rule to select between rule pairs (inhibiting color while using shape). Acquisition of the ability to use a higher-order rule corresponds to an important increase in the complexity of children's

Zelazo, Helwig, and Lau

use an integration rule, such as in the assignment of punishment. Three-year-olds will be restricted to the use of simple rules such as an outcome rule (if outcome is negative then punish; if outcome is positive do not punish) or an intention rule (if intention is negative then punish; if intention is positive do not punish). They will not be able to use an integration rule such as a conjunction rule (if outcome is negative and intention is negative, then punish; otherwise do not punish), which requires the simultaneous consideration of two dimensions and ought flrst to emerge at about 5 years of age or soon sifter (e.g., Frye et al., 1995, Exp. 3; Roberts & Fischer, 1979). Order of presentation of intentions and consequence information has been found to affect moral judgments in other research (see BCamiol, 1978; Keasey, 1978). Thus, order of presentation of all relevant information in the present study (e.g., positive intentions vs. negative intentions) was systematically counterbalanced. Moreover, due to discrepancies in flndings based on different measures, Keasey (1978) has recommended that studies use more than one index of moral judgment. This recommendation appears to have gone largely unheeded in subsequent research, but the present study included assessments of qualitative and quantitative judgments of both act acceptability and punishment. Qualitative judgments (e.g.. Was it okay for X to Y?) are useful for providing general information about the features that are most important in determining evaluations, while quantitative measures provide direct information about the relative weight of different pieces of information in judgments (Buchanon & Thompson, 1973; Surber, 1977) and may be more sensitive to the influence of multiple types of information. Two types of moral judgment (act acceptability and punishment) were included because the rules that are used to make these judgments may differ. Finally, in order to ensure comprehension of all relevant information, including agents' intentions and the nature of the noncanonical causal relations, pictures similar to those employed by other researcbers (Nelson, 1980) were used, in conjunction with multiple comprehension probes and criterial comprehension questions for each piece of essential information.

2481

months; range = 35-48 months), 4 years (M = 54.0 months; range = 48--60 months), and 5 years (M = 65.9 months; range = 61--71 months), and 24 University of Toronto undergraduates (M = 23 years; range = 18-32 years). The children were recruited through local day-care centers, and the students received course credit for their participation. Participants were predominantly EuropeanAmerican and tended to come from middleclass backgrounds. Roughly half (49 out of 96; 51%) ofthe participants were male. Two 3-year-old girls refused to complete the study and were replaced. Design and Overview Half of the participants at each age were randomly assigned to the normal causality condition, and half were assigned to the noncanonical condition. A between-subjects manipulation was employed in order to minimize potential confusion about the causal relations between particular acts and outcomes. For each causality condition, there were four different scenarios that presented different combinations of intention (nice vs. mean), act (pets vs. hits), and outcome (smiles vs. cries; see Table 1). Tbe four scenarios for the normal causality condition included the same agents and acts as the corresponding scenarios for the noncanonical condition. Each agent was always associated with a particular intention and act. The outcome depended on the relation between act and the causality condition. For example, Sally (who is nice) hits the animal in both cases. In tbe normal causality condition, the dax cries when she hits it, whereas in the noncanonical condition, the mugwump smiles when she hits it. Acts and outcomes were crossed within each causality condition by designing scenarios in which the animal was responsible for effecting a particular outcome. For example, when Sally (who is nice) tries to pet the dax (in the normal causality condition), the dax jumps up, and she ends up hitting it by mistake. By attributing the cause ofthe outcome to the patient (i.e., the animal) rather than to the agent, we have endeavored to keep the agent's intention as clear and as unambiguous as possible. All scenarios were closely matched in narrative structure and wording (see Appendix for examples of a normal and noncanonical scenario), and the order in which the scenarios, and thus information about intentions, acts, and outcomes, were presented was counterbalanced. For each scenario, participants were asked several comprehension probe ques-

Method

Participants Participants included 24 children at each of three ages, 3 years (M = 41.4

2482

Child Development

TABLE 1

FOUR SCENARIOS PER CAUSALITY CONDITION INVOLVING DIFFERENT COMBINATIONS OF INTENTION, ACT, AND OUTCOME CAUSALITY

AGENT (Intention, Act) Tommy (nice, pets) Anne (mean, pets) Sally (nice, hits) Peter (mean, hits)

Normal (Outcome) Scenario Scenario Scenario Scenario 1 (smiles) 2 (smiles) 3 (cries) 4 (cries)

Noncanonical (Outcome) Scenario Scenario Scenario Scenario 1 (cries) 2 (cries) 3 (smiles) 4 (smiles)

tions designed to assess whether they understood the important details and could remember them. They were then asked three criterial confirmation questions immediately prior to being asked three types of test question: a behavioral prediction question, two act acceptability questions (qualitative and quantitative), and two punishment questions (qualitative and quantitative). In total, for each of the four scenarios, participants were asked four preliminary comprehension questions, three criterial confirmation questions, and five test questions, all of which were presented in a fixed order. The use of a fixed order within scenarios was necessitated by tbe logical relations among the questions and the narrative structure of the scenarios (e.g., prediction of the act necessarily preceded retrospective evaluations of the act). Procedure Pcirticipants were interviewed individually in sessions that lasted about 15 min. The procedure was the same for children and for adults, except that at the beginning of the session, adults were told that they would be asked a series of questions in exactly the same way that preschool children would be asked the questions, and so answers to some of the questions would seem obvious. Prior to testing, all participants were familiarized with the act acceptability rating scale, which consisted of an array of five colored faces on a 21.5 X 28 cm card. From left to right, the faces ranged from very sad (deep red with a severe frown) to very happy (bright yellow with a marked smile). Participants were told that the face on the far left meant "really, really bad," the next face was "just a little bad," and tbe middle face was "just okay." Tbe faces to the right of center were then described as "just a little good" and "really, really good." T^his 5-point scale includes two levels of discrimination (a little vs. a lot) in two directions (positive and negative), with

a neutral midpoint. It can be thought of, in effect, as two 3-point scales combined. Thus, affer making a judgment (positive or negative), children only needed to discriminate among two levels of degree. After being shown the act acceptability rating scale, participants were asked a few questions to ensure that they understood the scale and could use it appropriately. The scale was then set aside and the experimenter proceeded to tell participants the first of four stories that were each illustrated using six 21.5 x 28 cm laminated colored drawings. In each story, a child character was introduced and participants were told that the child's parents returned from their vacation with a special kind of animal. In the normal causality condition, the animal "gets hurt and cries" when hit and "feels good and smiles" when petted. In the noncanonical causality condition, however, the animal "gets hurt and cries" when petted and "feels good and smiles" when hit Immediately after being told the act-outcome relations, participants were asked to state the outcome of hitting and the outcome of petting the animal. If participants answered either of these questions incorrectly, they were told the correct answer. After the first two comprehension questions, the character's intention was stated explicitly. In two of the stories that participants received, the character was described as mean, and it was stated explicitly tbat the character wanted to hurt the animal. In the other two stories, the character was described as nice, and it was stated tbat the character wanted to make the animal feel good. Participants were then asked whether the character was mean or nice, and whether the character wanted to hurt anyone. If participants answered either of these final comprehension questions incorrectly, they were corrected as above.

Zelazo, Helwig, and Lau

At this point, participants were asked three criterial confirmation questions in succession: (1) "Now, what does a (e.g., a miugwump) do when you hit it?" (2) "And what does it do when you pet it?" (3) "And is [e.g., Sally] mean or nice?" These questions were criterial in the sense that children were required to answer all three of the questions correctly in order for their responses to the subsequent test questions to be included in the analysis (see Results). As with the previous comprehension questions, participants were told whether or not they answered the three criterial confirmation questions correctly. Immediately after the confirmation questions, participants were asked the behavioral prediction question. This question required participants to predict the character's act based on the character's intention and the stated causal system. For example, for the story involving Sally, who is nice, in the noncanonical causal condition, the question was phrased as follows: "Now, Sally knows that mugwumps are weird. She knows that a mugwump cries when you pet it and smiles when you hit it. What is Sally going to do?" After participants answered this question, they were told the predetermined outcome (see Table 1 and Appendix). Tben, they were asked the act acceptability questions. First, they were asked the qualitative question, "Is it okay for [e.g., her] to [e.g., hit] the [e.g., mugwump]?" If participants responded that it was okay, then they were asked, "How good is it to [e.g., hit] the [e.g., mugwump] ? Is it really, really good or just a little good or just okay?" and they were then presented with five colored faces that ranged from very sad to very happy (the act acceptability rating scale). They were required to point to one of the five faces (the quantitative act acceptability question). Likewise, if paridcipants answered tbe qualitative question by saying tbat it was not okay, tben they were asked, "How bad is it to [e.g., hit] the [e.g., mugwump]? Is it really, really bad or just a little bad?" and they were then given the act acceptability rating scale. Responses to the quantitative act acceptability question were scored as integers from 1 (really, really bad) to 5 (really, really good). The punishment questions followed the act acceptability questions. First, participants were asked, "Sbould [e.g., Sally] get in trouble?" If they answered this

2483

qualitative question affirmatively, then they were asked the quantitative question: "A little trouble or a lot of trouble?" Responses to this questions were scored as 0 (no trouble), 1 (a little trouble), or 2 (a lot of trouble). After the punishment questions, there was a short break; tben the experimenter administered the remaining three sceneirios in succession. All sessions were audiotaped for subsequent analysis. Results Preliminary analyses revealed no significant effects of sex, so sex was dropped from subsequent analyses. Confirmation Questions For eacb scenario, participants were asked four preliminary comprehension questions, two pertaining to the causal relations between acts and outcomes and two pertaining to the character's intention. In 97% of the cases (743 cases out of 768), participants answered the causal relation comprehension questions correctly. Similarly, the intention questions were nearly always (94%) answered correctly. (In the large majority [83%] of the cases where children erred, they erred by stating that the nasty character was nice.) After the preliminary comprehension questions and prior to the three types of test question, participants were asked three criterial confirmation questions designed to ensure that they understood and remembered the relations between acts and outcomes and the intention of the story character. In 97% (374) of the 384 cases (3 questions x 4 scenarios X 96 participants), participants answered all three of the questions correctly. There were 10 cases, involving eight different children, where participants failed to do so. These included three 3-year-olds in the normal causality condition (two girls and one boy), four 3-year-olds in the noncanonical condition (two girls and two boys), and one 4-year-old girl in the normal condition. Behavioral prediction scores were converted to proportion correct based only on the items for which the participants answered all three criterial confirmation questions correctly. For the remaining two types of questions, the eight participants who failed to answer all three confirmation questions correctly for one or more of the four scenarios were eliminated from the analyses. These participants were eliminated because determination of tbe basis of tbeir judgments

2484

Child Development

cell passed the questions than would be expected by chance based on an alpha of .05. It was determined that the probability of seven or more children answering three or four out of four questions correctly by chance is .04. Thus, this analysis yielded a criterion of seven children out of 12. More children were classified as passing than would be expected by chance for all cells except 3-year-olds who received the noncanonical condition (p < .05). Act Acceptability--Qualitative Rules were fitted to each participant's pattern of responses across the four qualitative act acceptability questions. In the large majority of cases (94%), responses could be fitted to one of the six rules listed in Table 3. Table 3 also lists the percentage of participants at each age and in each condition whose responses corresponded perfectly to each rule. As is clear from this table, the outcome rule fit the majority of participants' responses at all ages in both conditions. In contrast, neither the act rule nor the intention rule was able to account for participants' responses (there was only one case where responses corresponded to the intention rule). Across both causal conditions, nearly all 3-year-olds used the outcome rule. Inspection of Table 3 indicates that the percentage of 3-year-olds using the outcome rule was comparable to the percentages of 5-year-olds and adults who did so. However, because all of the 3-year-olds in the normal causal condition used the outcome rule, data from this age group could not be included in analyses on tbe use of this rule. For the remaining three age groups, the effects of age and condition on the use of the outcome rule were examined using a categorical data analysis based on a linear models method (see Grizzle, Starmer, & Koch, 1969). The only significant effect to emerge from this analysis was a main effect of age, x^(2, N = 71) = 7.27, p < .05. Separate chi-squared tests were conducted to compare the numbers of participants in each age group (collapsing across causal condition) who did or did not use the outcome rule. These tests revealed that 4-year-oIds were less likely than 5-year-olds to use the outcome rule, X^(l, N = 47) = 5.63, p < .05, but that neither 4-year-olds nor 5-year-olds differed from adults, x^(l, N = 47) = 1.04, N.S., and X^ih N = 48) = 2.02, N.S., respectively. Act Acceptability--Quantitative Preliminary examination of act acceptability ratings confirmed that participants used the full range of the 5-point scale (rang-

on these questions required consideration of their pattern of responding across all four items. Behavioral Prediction Preliminary examination of the data revealed that adults performed almost perfectly on the behavioral prediction questions (M = 0.99, SD = .05). Because of this ceiling effect, data from adults were omitted from an analysis of variance (ANOVA) on the proportion of behavioral prediction questions answered correctly. A 3 (age) x 2 (causality) ANOVA on arcsin transformed scores revealed a main effect of age, F(2, 66) = 4.65, p < .05, and condition, F(l, 66) = 7.78, p < .01, but no significant age x condition interaction. Performance was better in the normal condition (M = .91, SD = .15, reported in raw scores) than in the noncanonical condition (M = .76, SD = .29). Tukey's honestly significant difference (HSD) tests revealed that 5-year-olds (M = .94, SD = .15) performed better than 3-year-oIds (M = .75, SD = .29; p < .05), but that neither of these groups differed from 4-year-olds (M = .82, SD = .24). (Random responding would produce a score of .50.) Comparison of individual participants' performance to chance corroborated the results of the ANOVA. Each participant was classified as passing the behavioral prediction questions according to a criterion of three or four correct responses. The numbers of participants reaching this criterion are listed in Table 2. The probability of answering three or four questions correctly (assuming that four questions were considered) by chance was .31. A second-order application of the binomial theorem was used to determine whether more children in each

TABLE 2

NUMBERS OF PARTICIPANTS CLASSIFIED AS PASSING THE BEHAVIORAL PHEDICTION QUESTIONS ACCORDING TO A CRITERION OF THREE OR FOUR CORRECT RESPONSES CAUSALITY

AGE 3 years 4 years 5 years Adult

Normal 9* 11* 12* 12*

Noncanonical 5 8* 10* 12*

NOTE.--Asterisk indicates that the number of participants exceeds the number expected by chance based on a second-order application of the binomial theorem (p < .05).

Zelazo, Helwig, and Lau

TABLE 3

PERCENTAGE OF PARTICIPANTS USING EACH AGT ACCEPTABILITY RULE BY AGE AND CAUSALITY CONDITION AGE

2485

3 Years

RULE N NC

4 Years

N NC

5 Years

N NC N

Adult

NC

Intention Act Outcome Conjunction Disjunction Positive bias Other

100 0 0 0 0

0 0 0 0 75 50 0 8 13 25 13 0 8

0 0 64 9 0 8 27

0 0 92 8 0 0 0

0 0 83 8 0 0 0

8 0

0

0 0 75 17 0 0

0 0 67 17 17 0 8 0

NOTE.--N = normal, NC = noncanonical. Due to rounding to integers, percentages may not sum to 100. Rules were as follows: Intention: If intention is positive, then act is okay; if intention is negative then act is not okay. Act: If act is canonically positive, then act is okay; if act is canonically negative, then act is not okay. Outcome: If outcome is positive, then act is okay; if outcome is negative, then act is not okay. Conjunction: If outcome is positive and intention is positive then act is okay; else act is not okay. Disjunction: If outcome is positive or intention is positive then act is okay; else act is not okay. Positive bias: Act is always okay.

ing from 1 = really, really bad to 5 = really, really good). Thus, a four-way mixed ANOVA (age X causality x act x intention) was conducted on these ratings. In this analysis, the combination of a particular act (e.g., petting) and a particular causal system (e.g., noncanonical) always corresponded to a particular outcome (e.g., crying). As a result, the significant act x causality interaction, F(I, 80) = 437.75, p < .0001, reflected the influence of outcome: Acts were rated as more acceptable when they were committed in a causal condition where they produced smiling rather than crying (see Fig. 1). For example, hitting was rated as more acceptable in the noncanonical condition than in the normal condition. The ANOVA also revealed a main effect of intention, F(l, 80) = 12.89, p < .0005, that was qualified by a significant age X intention interaction, F(3, 80) = 3.6S, p < .05. Simple effects indicated that the effect of intention was only significant for 5-year-olds, F(l, 22) = 10.45, p < .005, and for adults, F(l,22) = 14.61, p < .001. Means and standard errors for the age x intention interaction are presented in Table 4. There were no other significant main effects or interactions. Punishment Questions--Qualitative As with act acceptability questions, rules were fitted to each participant's pattern of responses across the four qualitative punishment questions. Responses could be fitted to one of the rules listed in Table 5 in 92% ofthe cases. Table 5 also lists the per-

centage of participants at each age and in each condition whose responses corresponded to each rule. The majority (75%) of adults, but very few 3-year-olds, appeared to use the conjunction rule. The percentage of 4- and 5-year-olds who used this rule was intermediate. However, for the 5-year-olds, it was the dominant rule in the noncanonical condition, where it accounted for the responses of 42% of the children. Across both causal conditions, only one 3-year-old used the conjunction rule. Because none of the 3-year-olds in the noncanonical causal condition used the conjunction rule, data from the 3-year-olds could not be included in categorical analyses on the use of this rule. For the remaining three age groups, the effects of age and condition on the use of the conjunction rule were examined using a categorical data analysis based on a linear models method. This analysis revealed a strong main effect of age, x^(2, N = 71) = 20.89, p < .0001. Separate chi-squared tests revealed that, across causal conditions, both 4-yearolds and 5-year-olds were lass likely than adults to use the conjunction rule, x^(l> ^ = 47) = 13.33, p < .001, and x^(l, N = 48) = 8.39, p < .005, respectively. Punishment Questions--Quantitative As with the act acceptability ratings, participants used the full range of the 3point scale for assigning punishment (0 = none, 1 = a little, and 2 = a lot). A four-way mixed ANOVA (age x causality x act x intention) on punishment ratings revealed a

2486

Child Development

5-1

'4.5-

Nice/Pets

Mean/Pets Nice/Hits Scenario (Intention/Act)

Mean/Hits

FIG. 1.--Mean act acceptability ratings--quantitative (and SEs) as a function of causality condition, act, and intention.

significant act x causality interaction, F(l, 80) = 85.85, p < .0001, which refiected the fact that punishment ratings were higher when the outcome was negative (see Fig. 2). There was also a significant main effect of intention, F(l, 80) = 74.97, p < .0001, and an age x intention interaction, F(3, 80) = 4.28, p < .01. However, all of these effects were qualified by a three-way act X causality X intention interaction, F(l, 80) = 30.75, p < .0001, and by a four-way age x act x causality x intention interaction, F(l, 80) = 6.95, p < .0005. The conjunction rule predicts the three-way interaction (i.e., there should only be an effect of intention when the combination of act and causality leads to a negative outcome). The four-way interaction (for which cell sizes ranged from 8 to 12) appears to refiect an age-related increase in the use of the conjunction rule. Visual inspection of Figure 2 shows that the punishment ratings were highest when (a) the

outcome was negative (petting in the noncanonical causal condition and hitting in the normal causal condition) and (b) the intention was negative. To decompose the fourway interaction, we conducted separate comparisons examining the effect of intention for each type of outcome (positive and negative) at each age and in each causal condition (for a total of 16 comparisons). Comparisons were made using the Bonferroni t statistic to guard against Type I error. These comparisons revealed that there was a significant main effect of intention only for adults when the outcome was negative, *(22) = 266.2, p < .01, for hitting in the normal causal condition, and t(22) = 137.5, p < .01, for petting in the noncanonical causal condition. Discussion Preschool children and adults were required to use information about intentions under normal or noncanonical caused systems in order to predict an agent's behavior. In addition, they were asked to integrate intentions, acts, and outcomes to judge the acceptability of an act emd to assign punisbment to the agent. Behavioral prediction improved with age and was more difficult when children were required to reason about a noncanonical causal system. Most 3year-olds and many 4-year-olds performed at chance in the noncanonical condition. At all ages, most participants made categorical judgments of act acceptability based solely on outcome rather than on the canonical acceptability of the acts or on intentions. Quantitative ratings of acceptability refiected an age-related increase in sensitivity

TABLE 4

MEAN RATINGS OF ACCEPTABILITY (and Standard

Deviations) BY AGE AND INTENTION

INTENTION

AGE 3 years 4 years 5 years Adult

Positive 3.21 (.59) 2.70 (.99) 3.40 (.81)* 3.44 (.61)*

Negative 2.85 2.85 2.46 2.69 (.96) (.85) (.83) (.60)

NOTE.--Asterisk indicates that the mean for the positive intention is higher than the corresponding mean for the negative intention (p < .05). Ratings were on a 5point scale.

TABLE 5 PERCENTAGE OF PARTICIPANTS USING EACH PUNISHMENT RULE BY AGE AND CAUSALITY CONDITION AGE 3 Years RULE Intention Act Outcome Conjunction Disjunction Positive bias Negative bias Other N 22 33 11 33 0 0 NC 13 0 38 0 0 25 0 25 4 Years N 17 0 17 25 0 25 0 8 8 8 8 8 0 27 18 8 17 0 8 NC 8 0 25 25 8 0 0 17 5 Years N 17 0 33 42 0 NC 8 0 N 25 0 0 75 0 0 0 Adult NC 17 0 0 75 8 0 0 0 0

NOTE.--N = normal,NC = noncanonical. Due to rounding to integers, percentages may not sum to 100. Rules were as follows: Intention: If intention is negative, then assign punishment; if intention is positive, then do not assign punishment. Act; If act is canonicaljiy negative, then assign punishment; if act is canonically positive, then do not assign punishment Outcome: If outcome is negative, then assign punishment; if outcome is positive, then do not assign punishment. Conjunction: If outcome is negative and intention is negative, then assign punishment; else do not assign punishment Disjunction: If outcome is negative or intention is negative, then assign punishment; else do not assign punishment Positive Bias: Always assign punishment Negative Bias: Never assign punishment

Normal Noncanonical

Nicaff>els

Mean/PM*

NIcalHHs

Mean/HHs

Nice/Pets

MeenA>eto

Nice/Hits

HeenMHi

Scenario (liil«ntloiVAcl)-3-YEAR-OU)S

Scarario (intemion/AciH-YEAR-OLOS

NIcafPets

Mean/Pets

Nica/Hlts

Mean/Hits

Scenario (lntamion/Act)-5-YEAR-0LDS

NiceA>ats MeanA>ata NiceMits Meanmita Scenario (Intenlk)n/Act)-ADULTS

FIG. 2.--Mean punishment ratings--quantitative (and SEs) as a function of age, causality condition, act, and intention.

2488

Child Development

ments about act acceptability, but for 5-yearolds and adults, ratings of degree of acceptability were sensitive to additional information (i.e., intentions). In contrast to the acceptability judgments, the qualitative punishment judgments showed an age-related trend toward convergence on the conjunction rule. According to this rule, punishment is assigned only if the outcome was negative (crying) and the intention was negative (mean). This rule requires the simultaneous use of two dimensions, which has previously been found to emerge at about 5 years of age (e.g., Frye et al., 1995, Exp. 3; Roberts & Fischer, 1979), so it is perhaps not surprising to find that the conjunction rule was used by very few 3-year-olds (11% in the normal condition and none in the noncanonical condition). Collapsing across causality conditions, 6% of 3-year-olds used the conjunction rule, whereas 22% of 4-year-olds, 33.3% of 5-yearolds, and 75% of adults were able to do so. A simple outcome rule was used to assign punishment by a substantial percentage of children at all three ages (but none of the adults). Additionally, a minority of participants at all ages (ranging from 13% to 21%, collapsing across condition) used an intention rule. Both the outcome rule and the intention rule require only the use of a single pair of conditional statements. Because a majority of participants at all ages used an outcome rule in qualitative judgments of act acceptability, it may be the case that many younger children are generalizing a single rule (namely, the outcome rule) to make a number of different types of moral judgment (botb acceptability and punishment). Thus, the younger children may have been more likely than older children and adults to fail to distinguish between different types of moral judgment. A similar pattern, where moral judgments become more sensitive with age to different types of contextual applications, has been found for older cbildren and adolescents in the areas of psychological harm and civil liberties (Helwig, 1995a, 1995b, in press; Helwig et al., 1995). Further, overgeneralization of relatively simple rules by younger children has been noted outside of the domain of moral reasoning (e.g., Zelazo & Shultz, 1989) and may be a widespread characteristic of cognitive developmental change (Shultz, 1991). The age-related increase in success on the behavioral prediction questions was also consistent with predictions made from the

to intention information: Ratings were systematically infiuenced by intention only for the 5-year-olds and adults. Age differences were also found for both types of punishment question. Older participants were more likely to use a conjunction rule (if outcome is negative and intention is negative then punish) to assign punishment, whereas 3-year-olds and some 4-year-olds tended to employ simple intention or outcome rules. The results of this study lend support to the suggestions that young preschoolers are able to make moral judgments based on underlying moral constructs such as harm, and that these constructs are used to generate generalizable moral judgments (Smetana & Braeges, 1990; Turiel et al., 1987). By 3 years of age, concepts of harm would seem to be differentiated from characterizations of particular acts such that they can be employed in the context of entirely novel causal sequences. Three-year-olds' judgments of act acceptability tended to be based on outcome, and no participant in the study made judgments that were based solely on the canonical acceptability of the act performed. Within a system of noncanonical causality, most preschoolers can conceive of acceptable hitting and unacceptable petting. Children are not limited to responding on the basis of learned associations between specific acts and adult sanctions or punishment (e.g., hitting is bad). How children come to focus on harm in making moral judgments is subject to a variety of theoretical interpretations, including cognitive development^ and social learning accounts. Nonetheless, the findings of this study make clear that substantive concepts of harm and welfare, however acquired, are guiding moral judgments by tbe fourth year of life. The outcome rule, used at all ages in qualitative judgments of act acceptability, requires participants to consider only a single pair of conditional statements (if crying then punish; if smile then do not punish) and therefore should be within the ken of 3-year-olds (Zeleizo & Reznick, 1991). In contrast to the qualitative judgments, however, the quantitative act acceptability judgments were sensitive to an age-related increase in the use of intention information, consistent with numerous previous studies (e.g., Buchanon & Thompson, 1973; Costanzo et al., 1973; Gutkin, 1972; Hebble, 1971; Helwig et al., 1995; Leon, 1980; Shultz et al., 1986; Surber, 1977; Wellman et al., 1979). Participants appear to use a simple outcome rule to make categorical judg-

Zelazo, Helwig, and Lau

Cognitive Complexity and Control Theory (Zelazo & Frye, in press). Analysis of individual patterns of responses across items revealed that most 3-year-olds and many 4year-olds performed at chance in the noncanonical condition, which required children first to determine the system of causality and then to determine the agent's intention in order to predict the agent's act. Unlike the normal causality condition, in the noncanonical condition, the relation between act and outcome could not be taken for granted. In the present study, 5-year-olds and adults were well able to switch from their own (normal) causality and make predictions in the noncanonical causality condition. This finding is consistent with previous studies that have found that, adthough 3year-olds can use a single pair of rules, it is not until about 5 years of age that children can use a higher-order rule to select between two incompatible pairs of rules (see Zelazo & Frye, in press). The logic of the current design resembles in a general way that used by Das Gupta and Bryant (1989) to explore preschoolers' ability to make causal inferences. An earlier study (Gelman, Bullock, & Meek, 1980) had found that 3-year-olds were able to choose the instrument (e.g., knife) that effected the transformation from an initial state (e.g., apple) to an end state (e.g., cut apple). However, this study failed to provide convincing evidence that children could infer which instrument caused the transformation because the problem could be solved simply knowing the function of the instrument and considering the end state (e.g., the cut apple). In order to assess whether children could trul)^ make inferences. Das Gupta and Bryant (1989, Exp. 2) adapted a procedure used by Gelman et al. (1980) so that children were required to consider both the initial state and the end state in order to infer the instrument. Children were presented with causal sequences that began with a noncanonical state (e.g., broken cup) and ended vdth a canonical one (e.g., mended cup). Children then had to choose an instrument from an arrajr' that included a distracter instrument that could have produced the initial noncanonical state (e.g., a hammer). Das Gupta and Bryant (1989) found that 3-year-olds typically failed to select the appropriate instrument in this condition, instead choosing the distracter instrument. In contrast, 4-yearolds performed well. The authors concluded that genuine causal inference develops during the preschool period. In the current

2489

study, the use of a noncanonical condition allowed us to determine whether children were able to make a genuine behavioral prediction. Behavioral prediction in the noncanonical condition required children first to consider the appropriate act-outcome relation (while ignoring the usual causal system) and then to consider the agent's intention in order to predict the agent's act. Many 3-yearolds appesir to base their predictions on intention information alone. Although children at all ages use intention information when making behavioral predictions, the majority of 3-year-olds did not use intention information when responding to the act acceptability questions and the punishment questions. Rather, the use of intention information to judge act acceptability increased with age and the use of intention information in combination with outcome information increased with age in judgments of punishment. An earlier study (Nelson, 1980) using a similar methodology (e.g., stories were presented both verbally and pictorially) found that 3-year-olds relied on intention (motive) information to judge how good or bad an actor was. These differences may be due to the different dimensions of moral judgment assessed. Nelson (1980) asked children to evaluate agents, rather than to make judgments about punishment or the acceptability of acts. Use of intentions in agent evaluations, as in behavioral prediction, may be developmentally prior to use of intentions in other judgments. Our results indicate that the use of intentions in moral judgment depends in part on the particular type of judgment being assessed (act acceptability or punishment) and also on the assessment methodology (qualitative or quantitative). When making moral judgments, children do not invariably respond with either an outcome or an intention orientation. In this regard, the findings from this study are inconsistent with the notion of a global stage of heteronomy (Piaget, 1932) in which younger children display a rigid orientation to moral rules and base their judgments of moral responsibility on external features such as outcomes, whereas older children and adults focus on subjective features such as intentions. Although the majority of 3-year-olds were outcome oriented, they displayed a flexible orientation to moral acts as shown by their responses in the noncanonical conditions, where they were able to conceive of hitting as good and petting as bad. In addition, even adults rarely based judgments of act acceptability

2490

Child Development

Noncanonical causality. Positive Intention, Hitting Here's Sally [drawing 1]. Sally's parents went on a trip to Brazil, far far away. You know what they found there? They found a special kind of animal called a mugwump [drawing 2] and they brought it back to Sally. Now, a mugwump is pretty weird, it has thick skin like rubber. When you hit a mugwump, its belly jiggles and wiggles and it feels good and it smiles [drawing 3]. What does a mugwump do when you hit it? It doesn't like to be petted, though [drawing 4]. That really, really hurts a mugwump, when you pet it because of its rubbery skin. When you pet it, it hurts and it cries. What does a mugwump do when you pet it? Sally is nice. She doesn't want to hurt anyone. She's nice, isn't she? Yes, she's nice. So, when her parents gave her the mugwump she wanted to make it feel good [drawing 5]. Is Sally mean or nice? Does she want to hurt anyone? Now, what does a mugwump do when you pet it? And what does it do when you hit it? And is Sally mean or nice? Behavioral prediction question: Now, Sally knows that a mugwump cries when you pet it and that it smiles when you hit it. What is Sally going to do? That's right Sally is nice, she wanted to make the mugwump feel good and she knew it liked to be hit so she hit it [drawing 3--reprise] and the mugwump smiled [drawing 6]. Act acceptability questions: Is it okay for her to hit the mugwump? How bad/good is it to hit the mugwump? Is it really, really good or a just little good or just okay? Punishment questions: Should Sally get in trouble? A little trouble or ^ lot of trouble?

or punishment on intention alone. Different moral judgments would seem to elicit the use of different types of information. This finding has important implications for future research, in that it suggests that greater attention may need to be paid to the type of dimension assessed when interpreting agerelated norms and progressions in moral judgment research. Further, for different judgments, different rules will be required to integrate information about intentions, acts, and outcomes, and the use of these rules appears to be constrained in part by their complexity, as predicted by the Cognitive Complexity and Control Theory. Thus, in addition to addressing some fundamental questions about the basis of children' moral judgments, the study illustrates the utility of considering moral development in the context of more general cognitive developmental changes. Appendix Sample Scenarios

Normal Causality, Positive Intention, Hitting Here's Sally [drawing 1]. Sally's parents went on a trip to Brazil, far far away. You know what they found there? They found a special kind of animal called a dax [drawing 2] and they brought it back to Sally. Now, a dax is pretty normal, it has skin just like you and me. When you pet a dax, it feels good and it smiles [drawing 3]. What does a dax do when you pet it? It doesn't like to be hit, though [drawing 4]. That really, really hurts a dax, when you hit it. When you hit it, it hurts and it cries. What does a dax do when you hit it? Sally is nice. She doesn't want to hurt anyone. She's nice, isn't she? Yes, she's nice. So, when her parents gave her the dax she wanted to make it feel good [drawing 5]. Is Sally mean or nice? Does she want to hurt anyone? Now, what does a dax do when you pet it? And what does it do when you hit it? And is Sally mean or nice? Behavioral prediction question: Now, Sally knows that a dax is normal. She knows that it cries when you hit it and that it smiles when you pet it. What is Sally going to do? That's right. Sally is nice, she wanted to make the dax feel good and she knew it didn't like to be hit so she tried to pet it. But you know what? When she tried to pet it, the dax jumped up and so she ended up hittkig it [drawing 4--reprise] by mistake and the dax cried [drawing 6]. Act acceptability questions: Is it okay for her to hit the dax? How bad/good is it to hit the dax? Is it really, really good or a just little good or just okay? Punishment questions: Should Sally get in trouble? A little trouble or a lot of trouble?

References

Anderson, N. H., & Cuneo, D. O. (1978). The height -1- width rule in children's judgments of quantity./oumoi of Experimental Psychology: General, 107, 335-378. Berndt, T. J., & Berndt, E. G. (1975). Children's use of motives and intentionality in person perception and moral judgment. Child Development, 46, 904-912. Buchanon, J. P., & Thompson, S. K. (1973). A quantitative methodology to examine the development of moral judgment. Child Development, 44, 186-189. Costanzo, P. R., Coie, J. D., Crumet, J. F., & Farnill, D. (1973). A reexamination of the effects of intent and consequence on children's moral judgments. Child Development, 44, 154-171. Das Cupta, P., & Bryant, P. E. (1989). Young children's causal inferences. Child Development, 60, 1138-1146. Diamond, A., Towle, C , & Boyer, K. (1994). Young children's performance on a task sensi-

Zelazo, Helwig, and Lau

tive to the memory functions of the medial temporal lobe in adults--the delayed nonmatching to sample task--reveals problems that are due to non-memory-related tasks demands. Behavioral Neuroscience, 108, 659-680. Edvrards, C. P. (1987). Culture and the construction of moral values: A comparative ethnography of moral encounters in two cultural settings. In J. Kagan & S. Lamb (Eds.), The emergence of morality in young children (pp. 123-151). Chicago: University of Chicago Press. Farnill, D. (1974). The effects of social-judgment set on children's use of intent information. Journal of Personality, 42, 276-289. Eeldman, N. S., Klosson, E. C , Parsons, J. E., Rholes, W. S., & Ruble, D. N. (1976). Order of information presentation and children's moral judgments. Child Development, 47, 1203-1215. Frye, D., Zelazo, P. D., Brooks, P., & Samuels, M. (1996). Inference and action in early causal reasoning. Developmental Psychology, 32, 120-131. Erye, D., Zelazo, P. D., & Palfai, T. (1995). Theory of mind and rule-based reasoning. Cognitive Development, 10, 483-527. Gelman, R., Bullock, M., & Meek, M. E. (1980). Preschoolers' understanding of simple object transformations. Child Development, 51, 691-699. Crizzle, J. E., Starmer, C. F., & Koch, G. G. (1969). Analysis of categorical data by linear models. Biometrics, 25, 489-504. Gutkin, D. C. (1972). The effect of systematic story changes on intentionality in children's moral judgments. Child Development, 43, 187-195. Hebble, P. W. (1971). The development of elementary school children's judgment of intent. Child Development, 42, 1203-1215. Helwig, C. C. (1995a). Adolescents' and young adults' conceptions of civil liberties: Freedom of speech and religion. Child Development, 66, 152-166. Helwig, C. G. (1995b). Social context in social cognition: Psychological harm and civil liberties. In M. Killen & D. Hart (Eds.), Morality in everyday life: Developmental perspectives (pp. 160-200). Cambridge: Cambridge University Press. Helwig, C. C. (in press). The role of agent and social context in judgments of freedom of speech and religion. Child Development. Helwig, C. C , Hildebrandt, C , & Turiel, E. (1995). Children's judgments about psychological harm in social context. Child Development, 66, 1680-1693. Helwig, C. C , Tisak, M., & Turiel, E. (1990). Chil-

2491

dren's social reasoning in context. Child Development, 61, 2068-2078. Inhelder, B., & Piaget, J. (1964). The early growth of logic in the child. New York: Harper & Row. Karniol, R. (1978). Children's use of intention cues in evaluating behavior. Psychological Bulletin, 85, 76-85. Keasey, C. B. (1978). Children's developing awareness and usage of intentionality and motive. In C. B. Keasey (Ed.), Nebraska Symposium on Motivation (Vol. 25, pp. 219-260). Lincoln: University of Nebraska Press. Kendler, T. S., Kendler, H. H., & Wells, D. (1960). Reversal and nonreversal shifts in nursery school children. Journal of Comparative Physiology and Psychology, 53, 83-88. Kessen, W., & Kessan, M. (1961). Behavior of young children in a two-choice guessing problem. Child Development, 32, 779-788. Kun, A., Parsons, J. E., & Ruble, D. N. (1974). Development of integration processes using ability and effort information to predict outcome. Developmental Psychology, 10, 721-732. Leon, M. (1980). Goordination of intent and consequence information in children's moral judgment. In F. Wilkening, J. Becker, & T. Trabasso (Eds.), Information integration by children (pp. 71-112). Hillsdale, NJ: Erlbaum. Moran, J. D., & O'Brien, G. (1983). The development of intention based moral judgments in three and four-year-old children. Journal of Genetic Psychology, 143, 175-179. Nelson, S. A. (1980). Factors influencing young children's use of motives and outcomes as moral criteria. Child Development, 51, 823-829. Nelson-LeGall, S. (1985). Motive-outcome matching and outcome foreseeability: Effects on attribution of intentionality and moral judgments. Developmental Psychology, 21, 332-337. Nucci, L. P., & Turiel, E. (1978). Social interactions and the development of social concepts in preschool children. Child Development, 49, 400-407. Piaget, J. (1932). The moral judgment ofthe child. London: Kegan Paul. Roberts, R. J., Jr., & Fischer, K. W. (1979). A developm.ental account of classification skills. Charlottesville: University of Virginia. Rudy, J. W., Keith, J. R., & Georgen, K. (1993). The effect of age on children's learning of problems that require a conHgural solution. Developmental Psychobiology, 26, 171--184. Russell, J., Mauthner, N., Sharpe, S., & Tidswell, T. (1991). The "windows" task as a measure

2492

Child Development

Turiel, E. (1989). Domain-specific social judgments and domain ambiguities. MerrillPalmer Quarterly, 35, 89-114. Turiel, E., Killen, M., & Helwig, C. C. (1987). Morality: Its structure, functions and vagaries. In J. Kagan & S. Lamb (Eds.), The emergence of morality in young children (pp. 155-244). Chicago: University of Chicago Press. Wellman, H. M., Larkey, C , & Somerville, S. C. (1979). The early development of moral criteria. Child Development, 50, 869-873. Yuill, N. (1984). Young children's coordination of motive and outcome in judgments of satisfaction and morality. British Journal of Developmental Psychology, 2, 73-81. Zelazo, P. D., & Frye, D. (in press). Cognitive complexity and control: A theory of the development of deliberate reasoning and intentional action. In M. Stamenov (Ed.), Language structure, discourse and the access to consciousness. Amsterdam and Philadelphia: John Benjamins. Zelazo, P. D., Frye, D., & Rapus, T. (1996). An age-related dissociation between knowing rules and using them. Cognitive Development, 11, 37-63. Zelazo, P. D., & Reznick, J. S. (1991). Age-related asynchrony of knowledge and action. Child Development, 62, 719-735. Zelazo, P. D., Reznick, J. S., & Piiion, D. E. (1995). Response control and the execution of verbal rules. Developmental Psychology, 31, 508-517. Zelazo, P. D., & Shultz, T. R. (1989). Concepts of potency and resistance in causal prediction. Child Development, 60, 1307-1315.

of strategic deception in preschoolers and autistic subjects. British Journal of Developmental Psychology, 9, 331-350. Shultz, T. R. (1991). Simulating stages of human cognitive development with connectionist models. In L. Birnbaum & C. Collins (Eds.), Machine Learning: Proceedings of the Eighth International Workshop (pp. 105-109). San Mateo, CA: Morgan Kaufmann. Shultz, T. R., Wright, K., & Schleifer, M. (1986). Assignment of moral responsibility and punishment. Child Development, 57, 177-184. Smetana, J. G. (1981). Preschool children's conceptions of moral and social rules. Child Development, 52, 1333-1336. Smetana, J. G. (1985). Preschool children's conceptions of transgressions: Effects of varying moral and conventional domain-related attributes. Developmental Psychology, 21, 715-724. Smetana, J. G., & Braeges, J. L. (1990). The development of toddlers' moral and conventional judgments. Merrill-Palmer Quarierly, 36, 329-346. Smetana, j . G., Kelly, M., & Twentyman, C. T. (1984). Abused, neglected and nonmaltreated children's conceptions of moral and socialconventional transgressions. Child Development, 55, 277-287. Surber, C. F. (1977). Developmental processes in social inference: Averaging intentions and consequences in moral judgment. Developmental Psychology, 13, 654-665. Turiel, E. (1983). The development of social knowledge: Morality and convention. Cambridge: Cambridge University Press.

Information

Title

16 pages

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate

980034


You might also be interested in

BETA
Title
JPS-2007-FINALjb.indd
Moral Development Theory: A Critique of Its Kantian Presuppositions
Microsoft Word - Gino Shu Bazerman OBHDP 2010
CR 36-3