Department of Educational Psychology
University of Wisconsin-Madison
Much of our knowledge and communication about intentional and physical phenomena involve general laws. For the purposes of this dicussion, "law" will be taken to mean any (idea of) general constraint on individual outcomes. For example, "An object cannot travel faster than the speed of light." is a law relevant to physical contexts. "A car cannot travel faster than the posted speed limit." is a law relevant to intentional contexts. "Law" is purposely ambiguous (as is "cause" above), encompassing both laws of nature and social laws, as well as norms, rules, and even personal resolutions.
As adults we are familiar with two different senses in which something cannot (or can, or must) happen according to a law. Continuing the example above, it is impossible for an object to travel faster than the speed of light. However, it is possible for a car to exceed the legal speed limit. What keeps a car travelling slower than light are causes; attributes of the physical structure of the world. What keeps a car travelling the speed limit are reasons; influences on the intentional decisions of the driver (see Davidson, 1963, on reasons and causes). This distinction has considerable significance. For example, we expect laws with automatic conformity to hold more reliably than those with voluntary conformity. We assign credit or blame differently when conformity is voluntary than when there is no choice. Laws with voluntary conformity often have a prescriptive function, those with automatic conformity are descriptive. Promulgating or learning the law takes on different importance-- in cases of voluntary (but not automatic) conformity one may violate the law through ignorance. Most generally, a distinction between voluntary and automatic conformity is a distinction between intentional and natural/physical causation; a distinction between two different domains.
While it is important, the distinction between voluntary and automatic conformity may not be obvious, especially to young children. Children may not distinguish the two processes of conformity or they may recognize the difference but confuse particular instances (see voluntary as automatic or vice versa). One problem is that the language of laws is often ambiguous. For example, most English modals (e.g., can, will, may, must, should) indicate both obligation (deontic modality) and possibility (epistemic modality; Shatz & Wilcox, 1991). This is illustrated by the two senses of "can't" in the laws about speed limits mentioned above. Experience also may be a poor guide; as Hume (1963/1748) pointed out, modality is not presented objectively but is a subjective assessment. Just because a law always has been followed does not mean conformity is automatic. Conversely automatic conformity is consistent with some failures if, for example, the law is probabilistic (consider conformity to "laws" of illness causation). First person experience of choice or compulsion may lead to awareness of the difference between voluntary and automatic conformity. However, it is not clear what kind of introspective access children have to their own mental states (Gopnik, 1993). Thus it remains an empirical question whether (or how) children do distinguish different types of lawful behavior:
It has often been argued that young children conflate what adults view to be very different types of laws. For example, Piaget (1965/1932) argued that children were realists about all laws. It is unclear how widely or totally children are supposed to collapse distinctions between laws. Often the error is characterized as failing to recognize that laws have different sources. For example, children believe that all laws (morals, conventions, laws of nature) exist and are true independently of human intention: Laws are discovered not invented (see Gabennesch, 1989, for review). At least occasionally the error is characterized in different terms; as not recognizing that conformity to laws may be under intentional control. For example, Taylor (1996) writes: "Prior to age 9, children do not seem to differentiate between physical and moral laws, at least in terms of whether it is possible to violate these rules." (p. 1569). In the context of the current paper, one way of taking the claim that children conflate adult distinctions between different types of laws is that they do not distinguish between voluntary and automatic conformity.
Children views of conformity to laws have not been directly studied. There is a small body of literature exploring distinctions between physical and social laws (Komatsu & Galotti, 1986; Levy, Taylor, & Gelman, 1996; Lockhart, Abrahams, & Osherson, 1977; Nicholls & Thorkildsen, 1988) and a larger literature concerned with distinctions between morals and conventions (e.g., Turiel, 1983, 1989). However the focus of this work has been on the alterability (could it be changed) and universality (appropriate everywhere) of the laws. These issues are largely idependent of conformity. For example, laws with voluntary conformity may be either universal or relative- people (may) see morals as universal and social conventions as relative (Turiel, 1983) but agree that conformity to both these laws is under intentional control. Similarly, laws with automatic conformity may be both universal-unalterable (nothing travels faster than light) or relative-alterable (the lights go on when you push the switch up).
In the course of exploring intuitions about universality and alterability, children have often been asked what can or cannot happen. For example, Levy, et. al (1995) asked, "Can people [also] steal money?" Komastu and Galotti (1986) asked children whether the mayor of a city "can" change things (e.g., so that rocks float). What these studies demonstrate is that young children do know many laws and do see them as constraints. Children say that people "can't" violate physical, moral, and conventional laws. These studies have gone on to ask under what circumstances children might answer "can" (e.g., if everyone gets together and decides). However, for the purposes of the current focus on types of conformity, the important question is what children mean when they say "can't" (or can). That is, why can't people steal or the mayor make rocks float? We would like to know, for example, whether children cite reasons and mental states for failures to steal, but causes and physical states for failures of rocks to float.
In their investigations of morals and conventions, Turiel and his colleagues (e.g., Davidson, Turiel, & Black, 1983) have asked children to justify their "can't" responses. They argue that children give different reasons for following morals (e.g., harm avoidance) than conventions (e.g., custom). In general, children do give appropriate justifications for their judgments. However these studies were not designed to distinguish causes from reasons. Justifications are assumed to be reasons, even though some explanations may be at least ambiguous (e.g,. "you always do it when you marry." [wife changes her name] Davidson et al, 1983). These studies have not compared justifications of automatic and voluntary conformity . Nonetheless, this work suggests that one valuable strategy for assesing intuitions about conformity might be to ask children "Why not?" when they say something can't happen. This strategy is pursued in Study 1 below.
In the process of acquiring adult commonsense beliefs about the nature of, and differences between, psychological and physical phenomena children must learn about voluntary and automatic conformity to laws. The studies reported below attempt to assess how far they have come by the preschool years. To what extent do preschooler-aged children share our adult intuitions about the reasons and causes for outcomes following laws? Before presenting the studies it is useful to consider three possible differences which might be found; a lack of distinction, a tenuous distinction, and/or a mis-placed distinction.
One possibility is that children fail to make the distinction between voluntary and automatic conformity. They may conflate apects of the two types of law-following or they may view all conformity as one or the other (automatic or voluntary). This represents the strongest prediction of difference between children and adults, and is thus the least likely possibility.
A slightly weaker hypothesis is that children do distinguish between voluntary and automatic conformity but have not integrated their understanding of obedience to laws with their knowledge of other causal relationships in the domains of psychology and physics. In the physical domain we know that children have difficulty coordinating theories (laws) and evidence (events; e.g., Kuhn, Garcia-Mila, Zohar, & Andersen, 1995). In the psychological domain, most studies have explored children's reasoning about idiosyncratic determinants of behavior (e.g., an idividual's particular beliefs and desires; see Wellman & Gelman, in press; Wellman, 1990, for reviews). Understanding law (rule) following seems closely linked to intention and planning, conceptions which appear difficult for young children (Astington, 1991; Wellman, 1990; see General Discussion section below). Thus children may recognize that voluntary and automatic conformity are different but not understand how those differences arise from the underlying intentional and physical causes involved.
A third possible difference between children and adults is in the phenomena we treat as voluntary and automatic. Assuming children make the distinction they must still learn when conformity is voluntary and when it is automatic. Given the ambiguities of language and experience it would be surprising if they never made mistakes. For example, Carey (1985) raises the possibility that children are misinterpreting biological (pysical) phenomena as social (intentional). Thus the law that "All animals must breathe," may be misconstrued; not seen as an automatic consequence of physical laws, but as a moral or social prescription about proper behavior. In constrast, some have suggested children initially treat social phenomena as biological (see discussion in Hirschfeld, 1995). In particular this claim has been made for children's understanding of gender roles. These roles are linked with biological attributes and children may believe that laws governing gender behavior (gender roles) are biologically based (Smith & Russell, 1984; Ullian, 1976). For example, Taylor (1996) finds that young children think that gender stereotyped behavior may be biologically inherited. The studies reported below assess the possibility that children treat conformity to gender roles as automatic rather than voluntary as a potential instance of a mis-placed distinction.
This paper explores children's beliefs about the processes involved in producing behaviors and outcomes that accord with general laws. Of particular interest is whether children distinguish two different processes; voluntary conformity based on reasons and mental (intentional) causes, and automatic conformity based on physical causes and (natural) necessity. Some laws we decide to follow while others we have no choice about. In addition to asking whether children make this distinction in general it is also important to assess their views about particular cases. Do children agree with our adult intuitions about when conformity is voluntary and when it is automatic?
Participants. Twenty-four children recruited from preschools in a mid-sized midwestern city participated in the study. Twelve children were in a younger group (M = 3:7, range 3:0-4:1), twelve were in an older group (M = 4:9, range 4:7-4:11). Twenty-seven adults recruited from an introductory educational psychology class also participated. Aproximately equal numbers of male and female children participated. Adults were predominantlly (80%) female.
Stimuli & Design. Stimuli for the study were 14 short stories (see Appendix for complete description of items). For children, colored line drawings accompanied the stories. Each story described a young boy who wanted to carry out some action. Participants were asked whether the boy "can" carry out the action. Depending on their first response, participants were either asked how he could or why he could not complete the action. There were three general types of stories. Seven social stories described characters desiring to perform unacceptable actions. Four were actions likely to be proscribed because of negative consequences (e.g,. not dressing warmly enough). Three social stories described boys desiring to violate gender norms (e.g., wearing a dress). Actions in social stories were chosen to be relatively unfamiliar in the hopes that children would not have encountered actual occurrences of these actions (see Pretest). Five physical stories described characters desiring to perform actions which violated physical constraints. Physical actions constituted a mix to parallel the two types of social actions. Some actions were impossible for all people (e.g., a boy who wants to turn into a bird) while others were possible for some people but not the story character (e.g., a young boy who wants to grow a beard). Finally, two stories described suitable actions-- actions which were both possible and permissible (e.g., a boy wanting to open a box after receiving permission).
Pretest. Thirteen children (Mean Age = 4:3) participated in a pretest designed to examine whether children had encountered actions depicted in stories. Participants heard each story (and saw pictures) and were asked whether they had ever seen anyone accomplish the action (violate the law). For example, a child heard the story about the little boy who wants to grow a beard and was asked if s/he had ever seen a little boy grow a beard. After responding, the child was then asked whether there is anybody who could accomplish the action (or who they had seen do it). Children were tested individually and stories presented in random order. Children were unfamiliar with physical and social actions (Mean proportion of "never seen" answers = .89 and .81 respectively, both greater than chance [.5] p<.01, Wilcoxon tests). Both types of actions were equally unfamiliar (did not differ significantly, T(9) = 33, ns). In contrast suitable actions were familiar (Mean "never seen" = .27, below chance, p<.05). Most reports of physical and social violations referred to familiar agents (e.g., parents, girls, silly boys, etc.). Three children cited actions of fantasy figures (e.g., Santa Claus, Peter Pan) for physical violations. However, it is not clear that children view these agents as unreal (Rosengren & Hickling, 1994).
Procedure. Children were tested individually in quite rooms in their daycare centers. The experimenter first read a brief description of the task: "I'm going to show you some pictures of some kids. These kids want to do all sorts of different things. Will you help me figure out which things they can do and which things they can't do?" Following these instructions children heard each story (and saw pictures) one at a time in random order. After each story children were first asked, "Can this boy <accomplish the action>?" Following a yes or no response, children were asked "How?" or "Why not?" (respectively). The procedure for adults was modified to allow data collection using computers. Adults read stories and did not see accompanying pictures. Adults typed their responses to questions.
Coding. Children's justifications were coded by two raters: the author and an assistant blind to the hypotheses of the study. There were three categories: cause, reason, and neutral. Cause justifications included references to physical limitations or changes that would be necessary to complete the action (e.g., "He's not tall enough to touch the ceiling."). Explicit statements of possibility/impossibility were also included in this category (e.g., "It is impossible for a boy to grow a beard."). Reason justifications referred either to social injunctions (e.g., "His parents won't let him.") or to statements of adverse consequences (e.g., "He would ruin his shoes."). A variety of response types could not be coded as either causes or reasons. Neutral responses included ambiguous modals ("He has to take off his shoes." "He needs to sleep.") and category-based responses ("Because he's not a girl." "Boys don't fly."). Justifications of positive judgments were also coded into these three categories. For example, justifications for why a boy could have long hair included: "Grow his hair long." (cause), "If his parents say 'ok'." (reason), and "Because he can." (neutral). The two raters agreed on 90% of coding decisions. Disputes were resolvedby a third rater.
Table 1 presents the proportions of judgments that characters could not perform actions. Comparisons against chance responding (.5 probability of "cannot" response) are also indicated. Gender and other social actions are considered separately because of the possibility that these types may be treated differently (see above). Both types of physical actions are considered together (post hoc tests revealed no differences between these sub-types for any age-group).
Consideration of the justification data suggests that children were not conflating physical and social (including gendered) actions. Though they judged that all three types "can't" happen, children gave different kinds of justifications for their responses. Figure 1 presents the mean proportion of cause, reason, and neutral justifications when participants said actions "can't" happen. Justifications for "can" judgments are not included in the figure (and are discussed separately).
When children gave cause and reason justifications they were usually appropriate. Both younger and older children gave more cause than reason justifications for physical actions (T(7) = 26, p<.05, T(11) = 66, p<.005, respectively). Older children gave signficantly more reason justifications for social actions (T(11) = 0, p<.005). While younger children's responses were in the correct direction, the difference did not reach statistical significance (T(7) = 7.5, ns). There were too few non-neutral justifications to make meaningful comparisons for gendered actions.
While the high number of neutral justifications limits the interpretability of children's justifications, a different caveat applies to adult data. Adults frequently judged that characters could perform social and gendered actions. Thus there were relatively few justifications for why characters could not perform these actions. Nonetheless, adults' responses did show the predicted patterns. They gave more cause than reason justifications for physical actions (T(27) = 378, p<.001) and more reason than cause justifications for social and gendered actions (T(15) = 8, p<.005, T(11) = 14, p=.05, respectively).
Children gave few justifications for why actions could happen (only 8 non-neutral justifications), in part because they answered "can't" so often. Adults allowed a significant number actions, and so provided more positive justifications. The bulk of positive justifications were for social or gendered actions (142 vs. 12 for physical). These justification were split between reason and cause (49 and 86, respectively, with 7 neutral). Both types of justifications are equally appropriate for positive judgments. For example, in explaining how a boy could wear a dress a cause justification was, "Find a dress and put it on." a reason justification was, "His parents would probably let him." A frequent type of justification was, physically can but shouldn't (e.g., wear a dress). This type of response probably reflects adults' explicit recognition of the ambiguity of the word "can" in the questions.
The results from Study 1 suggest that children do distinguish between voluntary and involunatry conformity. While children (unlike adults) tended to deny that both social and physical laws could be violated, this may have been because they were relatively insensitive to the conversational norms of avoiding ambiguity. Adults felt the need to mark the difference between social and physical items, children did not. However, children did appear to believe that different processes would lead to conformity to physcial and social laws. When asked to justify their responses children cited the impossibility of physical violations but the impermissibility of social violations. A boy could not pick up a sofa because he was not strong enough; he could not bathe clothed because of the negative consequences if he did. Children's justifications were appropriate given adult assumptions about when conformity is a matter of choice and when it is automatic.
It is important to note that differential familiarity did not seem to account for differences in justifications. A pretest revealed that violations of physical and social laws were equally unfamiliar. Thus children were not simply judging that laws they have seen violated must be based on voluntary conformity.
There are several caveats to consider when intepreting results from Study 1. The first is that many of children's justifications could not be coded as reflecting reason or cause. These responses were rarely wrong or inappropriate, just un-informative. Thus there is some reason to expect that more careful phrasing or follow-up questions would reveal children's competence. A more fundamental problem is the amount of interpretation required on the part of coders to classify children's (and adults' responses). The volitional or choice component of conformity was almost always implied rather than explicitly stated in justifications. For example, a statement that a boy could not take a bath with his clothes on because "his shoes would get wet." was coded as a reason justification. A sensible interpretation is that the boy would (should) not choose the action because of the consequences. However, the statement is equally consistent with the belief that the consequence renders it impossible to perform the action. Conversely, a statement that a boy "isn't tall enough" to touch the ceiling may reflect a social prescription (as in not being tall enough to ride a rollercoaster alone). Coding was relatively conservative in interpreting justifications (hence the high agreement), however, it is not clear that adult intuitions about meanings of statements are appropriate to apply to children's data. We know what an adult means when s/he says "he's not tall enough," but we must have less confidence in interpreting the statement when made by a child. We might expect that children will have learned to give (socially) appropriate justifications. This does not, however, guarantee they understand those justification in the same way adults do.
Fundamentally the distinction between voluntary and automatic conformity turns on the role that mental states and processes play in producing an outcome. Choosing to obey a law involves psychological states of the actor; for example, knowing the law and intending to follow it. Automatic conformity is independent of these mental states. The justifications gathered in Study 1 were informative to the degree they provided information about participants' understanding of the role of actors' mental states. For example, what we want to infer from the justification that "he can't because he'd get his shoes wet." is that the actor's mental state (e.g., desire for dry shoes) is influencing his choice to conform (e.g., conflicts with, and dominates, the expressed desire to bathe clothed). Since there may be legitimate questions about inferences from justifications, it seems appropriate to assess views about the roles of mental states more directly. This strategy is pursued in Studies 2 and 3 presented below.
Participants. Twenty-four children recruited from local daycares participated in the study; twelve were in a younger group (M = 3:6, range 3:1 to 3:9), twelve in an older (M = 4:9, range 4:3 to 5:9). Twenty-six adults recruited from an introductory educational psychology class also participated. Aproximately equal numbers of male and female children participated. Adults were predominantlly (80%) female.
Design & Procedure. Children and adults were asked to predict the outcomes of stories in which an agent intended to violate a particular law. Children were provided with the following introduction to the task: "I'm going to show you some pictures of some naughty kids. These kids never do what they are supposed to. They always try to do the wrong things. Their moms and dads aren't around to tell the kids what to do. I'm going to show you some pictures of these kids trying to do some things and I'll ask you whether they will really be able to do those things or whether they'll only be able to pretend. OK? Will you help me figure out which things they really will do and which things they will only pretend to do?" (instructions for adults were similiar). In each story, participants were reminded of the law (e.g., "Kids can't take baths with their shoes on"). They then heard about a boy who decided he was going to violate the law (e.g., take a bath with his shoes on). After each story participants indicated whether they thought the agent would really do that ("Will he really take a bath with his clothes on?") or whether he could only pretend. The answer of "pretend"allowed participants to predict the actor would do something (rather than simply predicting failure). In addition, it makes more sense to interpret an agent as intending to pretend an impossible action (e.g., to fly) than as actually intending something they know to be impossible. Children were periodically reminded that no adults were around to influence agents. In all other respects, the design and procedure of Study 2 was identical to that of Study 1. In particular, the same physical, social, gendered, and suitable actions were used in stories (see Appendix)..
Figure presents the mean proportions of judgments accepting that story characters would perform the actions they intended (violate laws). Comparisons against chance (.5) responding are also presented in this table. All participants tended to deny that characters actually would violate physical laws. Adults tended to accept violations of all other types of items. Children's reponses for many items did not differ from chance. Nonetheless, some patterns were discernable in children's data.
Participants in all three age-groups accepted suitable actions more often than physical (younger- T(9) = 43; older- T(11) = 66; adult- T(26) = 351, all p<.01). Suitable actions were also more acceptable than social (younger- T(5) = 15, p<.05; older- T(11) = 60; adult- T(15) = 120, both p<.005). However, younger children did not accept very many suitable actions (responses did not differ from chance, see Table 2). These children seemed to show a bias to deny that characters would perform any actions; they did not differentiate between physical, social, and gendered actions (S(2) = 1.12, ns, Friedman test). Older children and adults did. These participants judged that characters would perform social actions more often than physical (older- T(9) = 45, p<.01; adults- T(25) = 325, p<.001). Gendered actions were also more acceptable than physical (older- T(10) = 54, p<.01; adults- T(25) = 325, p<.001). Neither group differentiated between gendered and social actions (older-T(10) = 34; adult-T(17) = 98, both ns, 2-tailed-tests).
Younger children failed to distinguish the effects of agents' intentions in Study 2. These children generally judged that characters would conform to physical and social laws (only pretend to violate) despite their intentions to act otherwise.. Older children, however, did see intentions as more effective at producing violations of social laws than physical laws. This suggests older children do recognize that conformity to some laws is voluntary while conformity to other laws is independent of agents' intentions. This distinction was displayed most clearly by adults in the study. Adults unanimously rejected the possibility that characters' decision would lead to violations of physical laws, but accepted that characters' could violate social laws if they decided to.
Both older and younger children did often deny that characters would actually violate social laws. On the one hand this may reflect some tendency to treat conformity to these laws as automatic. Alternatively, children may be reluctant to judge that a character will actually act naughty. A characteristic of voluntary action is that actors may always "change their minds" and switch their intentions. Perhaps children thought characters were not serious in their intentions, that they had "second thoughts." Reluctance to predict negative or "bad" actions could not account for children's judgments that actors would conform to physical laws. Children regard possible violations of these laws in a positive light (Levy, et al., 1995). However, a method which did not involve naughty behavior might reveal a clearer distinction between voluntary and automatic conformity. The final study in this paper examines predictions of conformity when agents are ignorant rather than disobedient.
Children's intuitions about the dependence of conformity on knowledge have not been addressed previously. However, there is a long tradition of asking when children's judgments of culpability are sensitive to agents' knowledge and intentions (for review, see Darley & Shultz, 1990).
Participants. Twenty-four children recruited from daycare centers in a mid-sized midwestern city participated in the study. Twelve children were in a younger group (Mean Age = 3:5, Range = 2:11-3:10), and 12 in an older group (Mean Age = 5:0, Range = 4:8-5:7). Approximately equal numbers of males and females participated. Twenty-eight adults recruited from introductory educational psychology classes also participated. Adults were predominantely (80%) female.
Design & Method. Stimuli and procedures were nearly identical to those used in Study 2 above. In this case, participants were asked to make predictions about the actions of characters described as not knowing various laws. Children were provided withthe following introduction to the task: "I'm going to tell you some stories about a some kids. These kids don't know some of the rules about what do do. These are children who aren't from around here so they don't know what they can and can't do. Will you help me figure out what these kids are going to do?" (instructions for adults were similar). Participant then heard stories involving the same content as those used in previous studies (see Appendix). In this case, they were first told about a law the character did not know; for example, "Arnie doesn't know that kids can't wear shoes in the bathtub." The rest of the story described the ignorant character as desiring an outcome inconistent with the law (e.g., "Arnie wants to wear shoes in the bathtub"). After being reminded the character was ignorant ("He doesn't know he can't do that."). Participants were asked to predict the character's action ("Do you think Arnie will really wear his shoes in the bathtub?").
Suitable items Study 2 were modified to present false-belief problems (Perner, Leekam, & Wimmer, 1987). For example, one item was as follows: "Nick's toy is in the red box. Nick doesn't know his toy is in the red box. He thinks it is in the green box. He's looking for his toy. Do you think he will open the red box or the green box?" However, because of discrepancies between two different experimenters in the way the false-belief items were presented, the data from these items are unanalyzable. Therefore, these items will not be considered in the results.
Figure 3 presents the mean proportions of judgments that story characters would perform the actions they intended. Comparisons against chance responding are also indicated. Younger children did not differentiate between physical, social, and gendered actions ([chi]2(2) = 3.5, ns, Friedman test). Older children and adults did ([chi]2(2) = 15.2, p<.005 and [chi]2(2) = 41.0, p<.001, respectively). These participants judged that characters would perform social actions more often than physical (older- T(11) = 66.0, p<.01; adults- T(28) = 406.0, p<.001). Gendered actions were also more acceptable than physical (older- T(11) = 65.0, p<.01; adults- T(26) = 350, p<.001). Neither group differentiated between gendered and social actions (older-T(7) = 8.5; adult-T(17) = 84.0, both ns, 2-tailed tests).
Results from Study 3 were generally similar to the results of Study 2. Younger children did not indicate that mental states and intentions would have different effects on conformity to social rules (including gender roles) and physical laws. Predictions of violation did not differ significantly, though the trend was in the correct direction (fewer physical than social violations). In contrast, older preschoolers and adults made clear distinctions between social and physical laws. Ignorance of the law, and the intention to act in a way inconsistent with the law, would lead a character to violate social laws and to fail to conform to gender roles. However, mental states did not effect conformity to physical laws. These laws were expected to hold despite characters' knowledge and intentions. For older children these intuitions were clarified in comparison with Study 2. There were more predictions of violation for social and gender items. The additional information about characters' knowledge states did affect judgments about these items. In contrast information about knowledge did not lead to different responses for physical items across the two studies. However, these children were probably answering at ceiling for both tasks.
Younger children's failure to distinguish between items may be a result of their difficulty understanding the relationship between mental representations and facts in the world. To respond correctly to the questions posed in this study children needed to understand two things: that certain behaviors depend on the knowledge state of the actor, and that the knowledge state of the actor may differ from what the child knows (or believes). One explanation for poor performance is that younger children failed to distinguish conformity based on mental states and intentions from conformity based on physical necessity. An alternative is that children made these distinctions correclty but mis-construed the information about actors' knowledge. Stories were intended to convey that characters would have false beliefs (or at least be ignoranct of truth, e.g., not knowing that toy food may not be placed in a real oven). Younger preschoolers (up until aproximately age four) have difficulty understanding that people may hold beliefs that are contrary to fact (cf. Wellman, 1990). They often assert that beliefs will be veridical in contexts that do not allow formation of true beliefs (e.g., when a person has no perceptual access to a fact). So, younger children may have ignored the information in the stories and judged that characters would be acting from true beliefs; that characters were not realy ignorant of the laws. This would lead to predictions of conformity for social and gender items. Interestingly, younger children were also predicting that characters would fail to conform to physical laws. This may be the opposite error also based on a lack of understanding of false-belief. For example, Flavell (Flavell, Green, & Flavell, 1986) has noted that a failure to distinguish appearance and reality can lead to two errors. One type involves judging that things will seem the way they really are. This is akin to the classic false-belief error. The other error is judging that things really are the way they seem. This would be analogous to the error that if someone believes something, it must be true. Such an error might have led children to judge that if characters didn't know some physical law then the law must not have been true, and thus could not constrain behavior. Difficulties with false-beliefs could explain young children's failure to distinguish voluntary and automatic conformity. Whether they would show the distinction on tasks not involving false-beliefs is a matter for future study.
Children's intuitions about differences in conformity to laws was assessed by asking them to reason about the roles of mental and physical causes. In Study 1, participants explained why violations of laws could not (or could) occur. At all ages, people sometimes gave reason explanations (citing social norms or expected outcomes) and sometimes gave cause explanations (citing physical conditions). Moreover, children and adults tended to agree on the type of explanation appropriate for various laws; reason explanations were given for obedience to social laws (wearing a coat in the cold) and cause explanations were given for obediance to physical laws (a child not growing a beard). However, many of children's explanations could not be coded as either reasons or causes. This was especially true for explanations of conformity to gender roles. A diferent set of concerns involved the validity of the coding scheme. If children's and adults' intuitions and undertandings really are different then it is inappropriate to have adults code children's responses. There is no gaurantee that responses that strike an adult as referring to a reason are necessarily meant that way by the child making the response. While the results of Study 1 are suggestive, it was important to follow-up on these results with methods probing more specific judgments.
Studies 2 and 3 asked participants to reason about the role of mental states (intention and knowledge) in producing conformity. Voluntary conformity is mediated by mental events; instances of this type of conformity are behaviors. Automatic conformity is mediated by physical events and ocurrs independent of actors' mental states. Study 2 asked whether or not intentions to violate a law would be effective. Could a character who decided to break a law (to act naughty) actually carry through on his intention? Children tended to deny that characters would violate laws. They may have been reluctant to predict that someone will act in a naughty way. While there was some general bias to predict conformity, older children did predict violation more often for some types of laws than others. Younger children showed a similar, but non-significant, trend. In particular, children's judgments accorded with adults'; intentions were more effective in cases of social laws and gender roles than in cases of physical laws. These findings were replicated and clarified in Study 3. In this study characters were described as ignorant of the laws, rather than as cognizant but disobedient. Younger children still failed to see different consequences for violations; their predictions of violation did not differ from chance. This pattern may have been due to young children's notorious difficulty reasoning about false beliefs (Perner, Leekham, & Wimmer, 1987; see Wellman, 1990 for review). Older preschoolers, however, made some clear discrimiations. They generally predicted violation of social and gender laws, but conformity to physical laws. These children seemed to share adults' understanding that conformity may be produced by intentional causes in some cases and independent of mental state in others.
The older preschoolers included in the studies above made at least one important distinction between sorts of laws. The claim that children conflate different types of laws is usually interpreted and investigated with regard to the source or authority of laws (e.g., Gabbenesch, 1993; Turiel, 1983). However, this claim is itself often conflated with a different argument; that children do not ascribe different types of force or necessity to laws (different sorts of conformity, e.g., Taylor, 1996). These different ideas about laws are well illustrated in the case of norms for gender specific behavior. Ideas about these norms (gender laws) were specifically investigated in the above studies. While it is one thing to ask whether people believe that gender norms are universally applicable and not (legitimately) alterable, it is quite another thing to ask whether they believe conformity to these norms is automatic or is under voluntary control. Studies asking whether gender norms are seen as akin to moral laws address the first question (Carter & Patterson, Levy, et al., 1995). Studies asking whether gender norms are seen as akin to biological laws address the second question (Taylor, 1996).
In some ways it is unsurprising that children understood non-intentional mechanisms of conformity. Several researchers have argued that young children are not animists, they recognize impersonal, mechanistic causality (Bullock, Gelman & Baillargeon, 1982; White, 1995). Similarly, studies have demonstrated that young children give non-psychological accounts for biological phenomena (e.g., morphology of animals; Gelmen & Kremer, 1991; Inagaki & Hatano, 1991; Springer & Keil, 1991). For example, Kalish (in press) found that preschool-aged children judged physical contact but not mental state to be a causal factor in the production of illness (see also Johnson & Wellman, 1982 for similar findings). While there seems to be good evidence that children recognize non-psychological (physical) causation there is less evidence they appreciate such relationships embody necessity. White (1995) argues that as young as age two, children believe in physical necessity; that a given set of physical conditions must (automatically) lead to a one outcome rather than another. However, these modal judgments have not been directly assessed. The studies reported above begin to address this point. In Study 1 children indicated that physical laws couldn't be violated and provided justifications referring to physical attributes. This suggests the modal force of children's denial was not based in social or psychological considerations but rather involved some other sort of impossibility. That physical facts were cited in justifications also suggests children were not thinking about logical impossibility. An important direction for future work on children's understanding of physical-mechanical causality is to assess whether children share adult intuitions about the necessity of causal relationships.
The good understanding of social rules children demonstrated is also consistent with existing research. Several researchers have argued that young children are adept at reasoning about rules or laws involving permission and obligation (deontic modality; e.g., Harris & Nunez, 1996). Children (and adults) are better at reasoning about these social rules than about other types involving logical or causal relationships; for example, on Wason's (1968) selection task (see Cummins, 1996, for review). One explanation for this superior performance is that the deontic nature of the rule hightlights the possibility that people might disobey; because conformity is optional violation is a salient possibility (Cosmides, 1989). This understanding triggers a disconfirmation strategy of hypothesis testing which leads to success on the Wason task. However, children's interpretation of modality has not been directly assessed. For example, children might be asked to reason about a mother's rule stating that if a girl plays outside she must put a coat on (Harris & Nunez, 1996). While adults see this rule as involving permission and obligation, it is not clear that children share this interpretation. Put another way, children may not understand the obligation involved in the rule in the same way adults do; as involving voluntary rather than automatic conformity.
The studies described above do shed some light on children's understanding of deontic modality. Voluntary conformity involves two attributes; mental representations (what someone thinks) and intentional choice. A prototypical example is choosing to obey what is believed to be a law. Voluntary conformity seems central to deontic modality and to notions of permission and obligation . For example, having permission to do something means being allowed to choose that action. One can only act out of obligation if one is aware of the obligation and decides to obey. The studies described above suggest that preschool-aged children do understand the role that mental states play in producing conformity to some laws. Thus at least one component of deontic modality-that permission and obligation involve thoughts- is in place. Study 2 above suggests that young children also recongize the chosen aspect of volunatry conformity. A character could decide to act naughty and violate some laws. However, whether children interpreted those decisions in the same way as adults would, as abritrary and freely chosen, has not been demonstrated. In the discussion of conceptions of automatic conformity above, the question was raised whether children actually view physical relationships as necessary. In the context of children's understanding of voluntary conformity and deontic modality a similar question arises: Do young children view intentional decisions as freely-chosen?
Several studies have demonstrated that young children distinguish between intentional and non-intentional actions (Shultz & Wells, 1985; Wellman & Schult, in press) but it is nonetheless unclear what children mean when they identify some action as intentional (e.g., "on purpose"). Intentions may be things that happen to one, similar to desires. For example, we think a desire for an apple just occurs, but acting on that desire, forming an intention to reach for an apple, is voluntary. Astington (1991, 1990) has argued that preschool-aged children do not see intentions as akin to plans, as choices to engage in one behavior rather than another. This kind of active consideration of behavior, of intention separate from belief or desire, might be absent from the young child's theory of mind (Wellman, 1990; Astington, 1991). Intention is also central to our adult understanding of voluntary conformity to a rule or obligation. A rule can be thought of as a plan for action. In voluntary conformity there is a sense of explicitly considering some course of action and deciding whether to accept or reject the rule or obligation. Children may understand that rules need to be mentally represented in order to be followed without further believing there must be some deliberate, voluntary decision to obey the rule. Actions may follow from beliefs (including beliefs about rules) and desires without need for intention, perhaps automatically. Or, to the extent intentions (e.g., decisions) are admitted into the causal chain they may be viewed as mental products like any other, wiithout the unique role as carriers of free-will ascribed to them in adult common sense.
In characterizing voluntary conformity, intention or choice was said to be central. Exisiting work on young children's understanding of the mind casts doubts on whether they share adults' conceptions of intentional decision; doubts not erased by the studies reported above. However, it is important to point out that an account of voluntary action (rule-following) based on deliberate choice is only one construal of commonsense beliefs. We can see someone as voluntarily following a rule without believing they actually chose or decided to follow the rule (cf. Ryle, 1954). For example, action may be habitual; not consciously chosen but nonetheless voluntary and dependent on mental representations. A baseball player running to first-base (rather than third) after hitting the ball is voluntarily following a rule but probably not making any choices (e.g., not considering the rules of baseball and deciding to follow them). Prior intention may not be a necessary component of voluntary action (Taylor, 1966; White, 1995). Thus it remains unclear whether, despite their difficulties with the notion of intention, young children view some actions as voluntary and undetermined.
The studies reported in this paper addressed young children's understanding of different ways events come to conform to general laws. Older and younger preschoolers gave different types of explanations for conformity. They sometimes cited reasons and sometimes causes. In general children and adults gave the same types of explanations for the same types of events. At least the older children in these studies also seemed to share with adults the intuition that confomity may be dependent on knowledge and mental states. Social rules and norms for gender appropriate behavior were seen in these terms. Other instances of confomrity were seen to be mediated by physical conditions. Physical laws were judged to involve this kind of conformity. While older preschooler's conceptions seem adult-like in these regards it is less clear whether they share the modal intuitions that also underlie a distinction between voluntary and automatic conformity. That children view some outcomes as naturally necessary and others as undertermined and freely chosen has not been clearly demonstrated. While these modal intuitions are often quite subtle, and difficult even for adults to express clearly, they seem to be at the heart many of the distinctions drawn between mental and physical phenomena (between the mind and the body). Thus it will be important for future studies to further explore children's understandings of these different modalities of causal relationships.
Bullock, M., Gelman, R., & Baillargeon, R. (1982). The development of causal reasoning. in W. J. Friedman (Ed.), The developmental psychology of time (pp. 209-254). New York: Academic Press.
Carey, S. (1985). Conceptual change in childhood. Cambridge, MA: MIT Press.
Cosmides, L. (1989). The logic of social exchange: Has natural selection shaped how humans reason? Cognition, 31, 187-276.
Cummins, D. D., (1996). Evidence for the innateness of deontic reasoning. Mind and Language, 11, 160-190.
Darley, J. M., & Shultz, T. R. (1990). Moral rules: Their content and acquisition. Annual Review of Psychology, 41, 525-556.
Davidson, D. (1963). Actions, reasons, and causes. Journal of Philosophy, 60, 685-700.
Davidson, P., Turiel, E., & Black, A. (1983). The effect of stumulus familiarity on the use of criteria and justifications in children's social reasoning. British Journal of Developmental Psychology, 1, 49-65.
Flavell, J. H., Green, F. L., & Flavell, E. R. (1986). Development of knowledge about the appearance-reality distinction. Society for Research in Child Development Monographs, 51, No. 1, Series No. 212.
Gelman, S. A., & Kremer, K. E. (1991). Understanding natural cause: Children's explanations of how objects and their properties originate. Child Development, 62, 396-414.
Gopnik, A. (1993). How we know our minds: The illusion of first-person knowledge of intentionality. Behavioral and Brain Sciences, 16, 1-14.
Harris, P. L., & Nunez, M. (1996). Understanding of permission rules by preschool children. Child Development, 67, 1572-1591.
Hirschfeld, L. (1995) Anthropology, psychology, and the meaning of social causality. In D. Sperber, D. Premack & A. J. Premack (Eds.), Causal cognition: A multidiscipinary approach. (pp. 313-344). Oxford: Clarendon Press.
Inagaki, K., & Hatano, G. (1993). Young children's understanding of the mind-body distinction. Child Development, 64, 1534-1549.
Johnson, C. N. & Wellman, H. M. (1982). Children's developing conceptions of the mind and brain. Child Development, 53, 222-234.
Kalish, C. W. (in press). Children's understanding of mental and bodily reactions to contamination: What you don't know can hurt you but cannot sadden you. Developmental Psychology
Komatsu, L. K., & Galotti, K. M. (1986). Children's reasoning about social, physical and logical regularities: A look at two worlds. Child Development, 57, 413-420.
Kuhn, D., Garcia-Mila, M., Zohar, A., & Andersen, C. (1995). Strategies of knowledge acquisition. Society for Research in Child Development Monographs, 60, Serial No. 245.
Levy, G. D., Taylor, M. G., & Gelman, S. A. (1995). Traditional and evaluative aspects of flexibility in gender roles, social conventions, moral rules, and physical laws. Child Development, 66, 515-531.
Lockhart, K. L., Abrahams, B., & Osherson, D. N. (1977). Children's understanding of uniformity in the environment. Child Development., 48, 1521-1531.
Nicholls, J. G., & Thorkildsen, T. A. (1988). Children's distinctions amon matters of intellectual convention, logic, fact, and personal preference. Child Development, 59, 939-949.
Perner, J., Leekam, S. R., & Wimmer, H. (1987). Three-year-olds' difficulty with false belief. British Journal of Developmental Psychology, 5, 125-137.
Piaget, J. (1965). The moral judgement of the child. New York: Free press (originally published 1932).
Rosengren, K., S., & Hickling, A., K.(1994). Seeing is believing: children's explanations of commonplace, magical, and extraordinary transformations. Child Development 65, 1605-26
Searle, J. R. (1969). Speech acts : an essay in the philosophy of language, London, Cambridge University Press
Shatz, M. & Wilcox, S. A. (1991). Constraints on the acquisition of English modals. in S. A. Gelman & J. P. Byrnes (Ed.s), Perspectives on language and thought: Interrelations in development (pp. 319-353). New York: Cambridge University Press.
Shultz, T. R., & Wells, D. (1985). Judging the intentionality of action-outcomes. Developmental Psychology, 21, 83-89.
Smith, J., & Russel, G. (1984). Why do males and females differ? Children's beliefs about sex differences. Sex Roles, 11, 1111-1120.
Spelke, E. S., Phillips, A. & Woodward, A. L. (1995). Infants knowledge of object motion and human action. In D. Sperber, D. Premack & A. J. Premack (Eds.), Causal cognition: A multidiscipinary approach. (pp. 44-78). Oxford: Clarendon Press.
Springer, K., & Keil, F. C. (1991). Early differentiation of causal beliefs appropriate to biological and non-biological kinds. Child Development, 62, 767-781.
Taylor, M. G. (1996). The development of children's beliefs about social and biological aspects of gender differences. Child Development, 67, 1555-1571.
Taylor, R. (1966). Action and purpose, Englewood Cliffs, NJ: Prentice-Hall.
Turiel, E. (1983). The development of social knowledge: Morality and convention. Cambridge: Cambridge University Press.
Turiel, E. (1989) Domain-specific social judgments and domain ambiguities, Merrill-Palmer Quarterly, 35, 89-114.
Ullian, D. (1976). The development of conceptions of masculinity and femininity. In B. Lloyd and J. Archer (Eds.), Exploring sex differences. (pp. 25-47). New York: Academic Press.
Wason, P. C. (1968). Reasoning about a rule. Quarterly Journal of Experimental Psychology, 20, 273-281.
Wellman, H. M. (1990). The child's theory of mind. Cambridge, MA: MIT Press.
Wellman, H. M., & Gelman, S. A. (in press). Knowledge acquisition in foundational domains. In D. Kuhn and R. Siegler (Eds.), Cognition, perception and language. V. 2 of the Handbook of Child Psychology (5th ed.). Editor-in-cheif: William Damon. New York: Wiley.
Wellman, H. M., & Schult, C. A. (in press). Explaining human movements and actions: Children's understanding of the limits of psychological explanation. Cognition.
White, P. A. (1995). The understanding of causation and the production of action: from infancy to adulthood. Hove (UK) ; Hillsdale.
I do mean to exclude logical laws (e.g., -[A & -A]) because these laws are not causal; they constrain descriptions not events.
In fact, the distinction between laws and events is often unclear in this work. For example, Nicholls & Thorkildsen (1988) asked children, "Could dogs ever be called something different, like 'rabs'?" It is unclear whether this question refers to a particular event (someone saying 'rab' to mean dog) or to a general law ('rab' comes to mean dog).
Whether the lights go on when I raise the switch in my office does not depend on anyone's decision to conform; though clearly the person wiring the circuit origninally had to intentionally conform to the building code. While we would not normally consider, "The lights go on when you push the switch up." to be a law (certainly not a law of nature) it does meet the very general criteria outlined above. The statement has modal force (the lights have to, must, should come on) and we can think of a particular event (a given switching) as conforming or not. The purpose of this paper is not to argue for a particular definition of "law", but to explore a set of intuitions that underly our judgments about conformity to causal regularities.
A post hoc test revealed that younger children judged that more gender than impermissible actions could happen. However, this test was only significant at p<.05 as a 1-tailed test (and not corrected for family-wise error). There was no apriori reason to guess this direction of difference rather than the other. All tests reported are one-tailed and corrected for familywise error using Holm's test unless otherwise noted.
 A related body of work has examined children's understanding and use of excuses (e.g, Weiner & Handel, 1985). Children seem to distinguish controllable and uncontrollable excuses (and expect less blame for latter). However, "controllable" as used in this work includes both automatic conformity (e.g, missing a date because a car breaks down) and voluntary choices which seem acceptable (e.g., missing a date because something more important comes up).
One experimenter presented the items as indicated (to approximately 1/2 the children). Another experimenter worded false-belief items as: "Nick doesn't know where his toy is. He wants to look in the green box. Will he really look in the green box?" Responses to this item are not informative about understandings of false-beliefs.
A biological law view of gender roles would probably imply that they are universal and unalterable. It should be noted that the studies above tested only one construction of a natural (biological) view of gender roles. A more familiar, and plausible, claim that gender roles are biologically based would assert that the preferences for certain behaviors are biologically determined. Probably no-one would claim that it is biologically determined that girls play with dolls. However, some might maintain that it is biologically determined that girls like to play with dolls. Further investigation of children's beliefs about gender roles should focus on their beliefs about the origins of preferences. Whether or not children believe that preferences (or any other mental state) may be voluntarily chosen has not been explored (though see Kalish, in press for children's beliefs about the controllability of emotions).
There is no agreed upon strict definition of "deontic modality." In this paper (as in the literature on reasoning about rules, see Cummins, 1996 and in linguistics, Lyons, xx), I use the term to refer to notions of permission and obligation in the context of social-intentional interactions. In a broader sense, deontic may be taken to mean "involving goods" (i.e., as opposed to truths). In this sense permission and obligation need not be social or intentional. For example, Cosmides and Tooby write: "the cognitive programs of an organism that confers benefits on kin cannot violate the [Cost to self < (Benefit to kin members) X (coefficient of relatedness to kin member)] constraint of Hamilton's kin selection theory." (1994, p. 97 emphasis added). This kind of obligation need not involve intention on the part of the organisms involved.