Morton Adam (1998) Default assumptions of good behavior
Default assumptions of good behavior Default Assumptions of Good Behavior

Adam Morton
University of Bristol
Department of Philosophy
9 Woodland Rd
Bristol BS8 1TB

Abstract This paper connects Turiel's discovery that small children distinguish between moral and conventional norms with the theory of mind debate and with contemporary work in moral philosophy. My aim is to explain both why pre-schoolers can easily make a moral/conventional distinction, and why at some later age it becomes harder to grasp such a distinction. My answer, in a nutshell, is that there is a simple moral/conventional distinction that is well within the capabilities of very small children, but this distinction is not the right one for adult use, for reasons which are explicit in contemporary work in moral philosophy. So when children begin to grasp these complications, we can expect their earlier simple certainty to vanish. Moreover, the contrast between the kinds of capacities needed to negotiate the earlier moral/conventional distinction and those needed to understand the reasons for its inadequacy are related to those on either side of the famous 'false belief' divide.

Keywords:   moral development, theory of mind

1.  Elliot Turiel has produced a large amount of evidence for a very surprising proposition, that children from an early age can distinguish between simply conventional rules, practices, and conventions, and moral norms. They can distinguish between arbitrary norms and those that are there for a deep reason. One reason this is surprising is that so many adults seem not to understand it. Especially social scientists, who often run together culture-defining moral rules, shifting conventions like tax codes and traffic regulations, and factors which make one society or one individual life better or worse than another.

My aim is to explain both why pre-schoolers can easily make a moral/conventional distinction, and why at some later age it becomes harder to grasp such a distinction. My answer, in a nutshell, is that there is a simple moral/conventional distinction that is well within the capabilities of very small children. But there are reasons why this moral/conventional distinction is not the right one for adult use, reasons which are explicit in contemporary work in moral philosophy. So when children begin to grasp these complications, we can expect their earlier simple certainty to vanish. Moreover, the contrast between the kinds of capacities needed to negotiate the earlier moral/conventional distinction and those needed to understand the reasons for its inadequacy are related to those on either side of the famous 'false belief' divide.

Turiel and his co-workers studied pre-school children's reactions to 'social transgressions', distinguishing between transgressions of moral and conventional norms. Children as young as 2 3/4 years old clearly recognised and remarked on situations where things were going wrong in both ways. Moreover, when children were asked questions designed to elicit a moral/conventional contrast they were often able to comply. For example they were asked if an offending act would have been right if there had not been a school rule forbidding it. The majority of responses suggested that violations of convention, like shouting when one is supposed to be quiet, would not be wrong in the absence of suitable 'rules', but that moral violations, like hitting another child, would be wrong whether or not there happened to be a rule on the topic. Moreover children consistently categorised moral transgressions as serious, as evoking a high degree of disapproval. And older children ranked even minor moral transgressions as more serious than middle-level conventional transgressions. Stealing an eraser was a more serious business, for many of them, than a boy's wearing a dress to school.

The 'moral' transgressions here center on aggression and causing physical pain, lying, and abuses of property rights. These are the topics on which Judy Dunn describes children as understanding, from a very early age, the distress of others. According to Dunn pre- schoolers have no trouble understanding that someone who has been pushed or had a toy grabbed from them will be upset, and are inclined to offer comfort to such victims. And these are also the topics on which Frank Keil reports remarkable conceptual sharpness, so that small children are better able on these topics to separate defining from merely characteristic features than they are when the topic is, for example, kinship relations.

The surprising conclusion that emerges from these studies is that the distinction between what is wrong and what is unconventional is available very early. It is as if the children were little Platonists, construing one another's actions as falling under an intuitively graspable form of The Good (or often more appropriately, The Bad.) But putting it this way suggests an immediate question: what is the conception of right and wrong that pre-school children so readily grasp?

To answer this question I suggest we turn to studies of the social skills and propensities of small children. The work of Judy Dunn makes it clear how prevalent tendencies to cooperative play are in very small children, to share food and toys with others and to take account of the feelings of others when cooperation, sharing or kindness is wanting. The importance of these capacities is that they suggest that children begin from infancy to exercize the human aptitude to living in and acting as part of small groups, accomplishing mutually beneficial ends by coordinated behavior. That is at the heart of the human condition, and if a human is going to do what a human's got to do she will have to choose suitable roles with regard to the roles chosen by others, share common resources, and, above all, nurture other people's cooperative activities. So, early in childhood children assume different roles in cooperative play, share food and toys, and are sensitive to disturbance in others. And this generates an ethics, or at any rate an ethos, in a fairly immediate way. Not sharing, not doing what you said you would, not maintaining others in an uncooperative state of mind (by violence or teasing): all of these are violations of an unarticulated imperative 'take a place in common projects'.1

2. There are links here with issues about the development of children's grasp of the conception of mind, in particular with the threshold that is passed between three and four years of age when a child comes to terms with the idea that beliefs can be false. I shall assume that the essential lines of this work are familiar. It leaves us with a deep problem of characterising the knowledge which informs the considerable range of psychological understanding which three year old children do have. Josef Perner, developing the account of Henry Wellman, sees the three year old child using a theory in which the role of 'belief' is played by a simpler concept, which he calls 'prelief'. Prelief does not distinguish between falsity and make believe. Prelief is the state a person is in when they have access to some information construed as reliable, on which to act.
The crucial questions here concern the information in terms of which three-year-old children characterize the content of people's states of mind. What kind of information is it? How do children relate people to this information? How is this information used in planning the child's own action and anticipating the actions of others?

There is a way of thinking of this 'prelief' information that connects these questions with our questions about moral development. Assume, as suggested by Dunn and others, that from a very early age children engage in shared cooperative activity (of which the childhood version is cooperative fantasy and play.) Assume that children from infancy have an interest in and awareness of states of mind which are relevant to such cooperative activity. Then one kind of information that we would expect children to use from a very early age is information needed to coordinate shared activities with others.

More specifically, when a form of social interaction requires that some facts be mutual knowledge then one would expect that children who take part in such interaction have the basis of it. 'Mutual knowledge' is here a technical term drawn from game theory. The mutual knowledge required for an activity is the information which each participant in the activity must not only have but assume that each other person has (and assume that each person can assume that each other person has, and so on.) It is not necessary that each person actually have all this information present in usable form. Instead, each person can know a body of facts from which the required complex information can be derived as required. Or, as David Lewis has pointed out, each person can use a default assumption that in the absence of evidence to the contrary each component of mutual knowledge is present. Thus if person a knows that p she will assume that, for example, person b knows that a knows that b knows that p, unless there is some specific reason to think otherwise.

Mutual knowledge is by its nature shared information: it says 'this is how we are doing the job'. Recent work in game theory shows that without the assumption of mutual knowledge even fairly simple strategic situations become enormously more complex. My suggestion is that young children construe information as mutual knowledge, in two basic respects. They take as a basis for a social activity a body of easily grasped fact, and then assume that each person grasps these facts and that each person may act as if each other person has grasped each of them. The body of fact will very often be the perceptible facts of the environment in which the activity takes place, but it can also be the common tissue of fantasy which makes a game work. In either case the information is manageable as long as the child assumes that all participants share the information and act on the assumption that it is shared, but becomes unmanageable when this assumption cannot be employed. The assumption is not that all beliefs (preliefs) are true, but that they are common. Thus it is that children can engage in sophisticated imaginary play, in which they and others assume roles and construe the play space in a way that transforms it into something quite other, although they have great difficulty with something apparently simple, the possibility that another person may fail to believe something that they know to be true.

3. There is a process aspect to this way of characterising children's social competencies. The strategy that young children use to think through social situations is on this account a non-inferential strategy - call it a 'reading off' strategy' - in which the content of a database of common knowledge is taken as authoritative - stating for example the location of things in the environment or the roles assumed in a game - and searched through for answers. Inference from information thus obtained is minimal. Instead of inference the child relies on knowing her way around the database, and on a number of default assumptions, which allow a further item of data to be assumed in the absence of an explicit contradiction of it. The most basic such default assumption is that if an item of information is available to one participant it is available to others.

Older children and adults rely on default reasoning too, of course. But after early childhood inference becomes more and more important. Different items of information, sometimes from completely different domains, are combined in order to derive novel beliefs as needed for the purpose on hand; these beliefs are far too varied to be held in any manageable database. False belief tasks require substantial inference. In the first place the child has to infer from the fact that some evidence is available to another person, and some evidence is not, that that person will form a different belief from the child's. Then the child has to infer from facts about the person's motives what action they will perform. On the other hand in a situation in which a default assumption of mutual knowledge of facts, motives and the nature of the common activity is reliable, to know what the other person will do the child can look up each person's immediate aim, look up the facts about what the standard way of accomplishing that aim is, and read off a prediction. There may be a lot of looking up here, but there is much less inference. In the simplest case the child can operate in a 'simulating' way, putting themselves into the position of the other person but drawing on the same information available to the child for her own actions.

A read-off strategy suffices for many cooperative activities. To that extent, then, cooperative and uncooperative behavior in others can be identified, with simple resources: with extensive fact-gathering and list-searching resources, but little inference. It should be no surprise, given this, that pre-schoolers have the concept of bad behavior, if by that we mean 'behavior that upsets cooperative ventures', such as lying, hitting, and grabbing people's possessions does. Children should be able to conceptualise these in terms of their characteristic property as behavior which will interfere with the functioning of (nearly) all cooperative activities. And this will allow a contrast with particular acts which interfere with the functioning of specific cooperative activities. To put it differently, from a person's behavior in one cooperative activity the child will be able to classify them as being of a character or in a state which will help or hinder other such activities. So the bad/unconventional distinction can be born.3

4. In understanding that some behavior disrupts cooperative activity and that cooperative behavior is generally to the benefit of all, children are naive moral realists. Why are we not all? Because there is more to right and wrong than small scale cooperation and its disruption. This is a large and tangled topic, but notice the following three facts, commonplaces in the work of contemporary moral philosophers.

First for many social ends the is no unique form of cooperation that will produce the required effects. We must drive on the right or on the left but it does not matter which, as long as we all do the same. We need stable institutions of property rights, but there are a great variety of possible formulations of these rights. There is a conventional element even in most intrinsically valuable forms of cooperation.

Second, when there are alternative forms of cooperation some, and more subtly some combinations of them, will produce 'better' results than others.. Some combinations of institutions will promote enterprise and production more than others. Some institutions of family life will produce less heartbreak and fewer mangled victims than others. But it is extremely hard to know which forms of cooperation are the optimal ones.

Third, the 'better', 'optimal' etc. of the preceding two points are not nearly as simple as they seem. People in different social settings can value experiences, social conditions, and whole lives by very different measures. Though most contemporary philosophers think that there are determinate human goods which underlie the variety of things people can want and value, few of them think that they are simple or obvious, or that our current opinions about them have a very good chance of finality.

The effect of these three points is to blur the line between coordinations for mutual benefit and arbitrary convention. Small children are right that the good and the conventional are different; but they are not so simply different as children assume. Moreover, practical versions of these three points can impinge upon practical life in fairly simple ways. The first point surfaces when people decide what way they are going to do a job, or which rules they are going to follow in a game. The second point surfaces when people consider any kind of social change. The third point surfaces when people come to terms with the differences of their tastes.

All three points can be obscured if one's way of life is very regular and traditional. or if ones role in one's way of life is essentially subordinate. Small children do take the lives they find themselves living to be more regular and traditional than they are, and do occupy generally subordinate roles. So they are to some extent protected from seeing the truth beyond their truth. But there is a more basic fact here. If you don't think through social life on the basis of default assumptions you have to do it by inference. And if you try to think through multi-person strategic interactions inferentially you find they are extremely hard. So we should expect that even children whose inferential capacities are quite developed continue to think through social situations with default reasoning. And one effect of this is evident from Turiel's work: children well past the age at which false belief tasks cease to be difficult still hold on to a naive faith in the difference between right and wrong. Of course they are encouraged in this by their elders, and they are not fundamentally mistaken. But there are facts that ought to shake this naive faith, and the remarkable thing is that their effect is not felt until adolescence or even later.4

5. My method has been to make links between the cognitive resources that children bring to moral situations, and the nature of those questions, as described by recent and not so recent philosophy. In this last section I shall draw out the obviously conjectural side of what I have been saying, trying to isolate topics for possible experimental study. It is time to go out on some limbs.

One likely consequence follows from the fact that the default assumptions for cooperative activities implicitly specify desires or aims as well as factual information. (If you are playing tag you have to try to catch or avoid capture, if you are helping load the car you have to aim at getting your share of the stuff carried out from the house.) Thus we should expect that there were correspondingly difficult 'false desire' or 'bad desire' tasks. This expectation might be tested, as follows.

Version A: A child sees two desirable-looking objects of which one looks more desirable - perhaps a chocolate and a peanut butter sandwich. They then taste each of them and find to their surprise that the more desirable looking one is actually bad tasting. The child is then introduced to a character - why not call him 'Maxi'? - who is going to be shown the two objects, without having tasted them, and is asked 'which one will Maxi choose?' or 'which one will Maxi want?'

 Version B: A child observes a situation in which some other person - call her 'Minny' - is in difficulty, and needs help getting out of it. There are two actions the child can perform to help Minny. One seems more promising than the other. The child chooses that one but then observes that it causes even greater problems for Minny. The apparently less promising action does however get Minny out of her fix. Maxi then enters the scene, knowing only Minny's situation and the appearances of the two actions. The child is asked 'what will Maxi do?' or 'what will Maxi want to do?'

Version C: A child observes a situation in which Minny is in difficulty and needs help. The child can choose either to perform a specified action which will help Minny or an action which will give the child pleasure. (Taking a candy from a jar, say.) But not both. The child chooses one. Maxi then enters and the child is asked about Maxi's actions and preferences.

Version D: A child observes a situation in which Minny is in apparently quite minor difficulty and needs help. The child can choose either to perform a specified action which will help Minny or an action which will give the child pleasure, but not both. If the child chooses not to help Minny then Minny's fate is observed, and it turns out that not helping her was serious, as Minny is now in real trouble. Maxi then enters and the child is asked about Maxi's actions and preferences.

Versions A to D differ in the role that false beliefs play in the generation of inappropriate desires or changed preferences, and in the kinds of inappropriate desires involved. It would be as interesting to know differences in the results between the versions as the results of any one of them. I suggest that small children will in A and B often predict Maxi's actions and describe his motives, in accordance with their updated information. They will say that Maxi will want the non-yukky food and choose the actually effective action. I suggest that in C and D small children's predictions will often show a bias towards cooperative interpretations: they will predict that Maxi will choose to help Minny more often than he will in fact choose to help her, and more often then they will choose to help her. And I suggest that in D the greater the difference between Minny's original plight and the plight the child's action puts her in the greater the tendency to predicting that Maxi will choose to help her will be.

Moreover, I suggest that the effect in versions B, C, D where the emphasis is on cooperation, will persist to a later age than in version A, just as the naive moral/conventional distinction persists much later. And I would expect a slight bias of the kind involved in my prediction about D to last well into childhood.

Bibliography: section by section

1. Recent work on moral development
Dunn, Judy 1988 The beginnings of social understanding Blackwell

Keil, Frank 1989 Concepts, Kinds, and Cognitive Development MIT Press

Turiel, Elliot The development of social knowledge Cambridge University Press 1983. chs 3, 4, 7

2. Theory of mind .
Davies, Martin and Tony Stone eds. 1995a Folk psychology: the theory of mind debate Blackwell. 1995b. Mental Simulation: evaluations and applications Blackwell 1995.

Freeman, Norman & Hazel Lacohee 1996 Making explicit three-year-olds' implicit competence with their own false beliefs. Cognition, to appear.

Goldman, Alvin 1992: 'In defence of the simulation theory' Mind and language 7, pp. 104-119.

Gopnik, Alison 1993 'How we know our minds: the illusion of first person knowledge of intentionality' Behavioral and Brain Sciences 16, 1-14.

Gordon, Robert 1995a 'The simulation theory: objectives and misconceptions' in Davies and Stone 1995a. 1995b Simulation without introspection or inference from me to you. In Davies and Stone 1995b.

Heal, Jane 1995 How to think about thinking. In Davies and Stone 1995b.

Perner, Josef and Deborah Howes 1995 He thinks he knows. in Davies and Stone 1995a.

Wellman, Henry 1990 The Child's Theory of Mind MIT Press.

3. Default assumptions and mutual knowledge
Binmore, Ken and Shin, Hyon Song 1992 Algorithmic knowledge and game theory. In Biccieri and Dalla Chiara, (eds.), Knowledge, belief, and strategic interaction, Cambridge University Press.

Bratman, Michael 1992 Shared cooperative activity. Philosophical Review 102, pp. 327-342.
Heal, Jane 1978 Common Knowledge. Phil Quarterly 28, 116--131

Hargreaves Heap, S., Hollis, M., Lyons, B., Sugden, R. and Weale, A.1992: The theory of choice. Blackwell.

Lewis, David 1969 Convention. Harvard University Press

Morton, Adam 1996a 'Folk Psychology is not a predictive device' Mind
Rabinowicz, Wlodzimierz 1992: 'Tortuous labyrinth: noncooperative normal-form games between hyperrational players' in Christina Biccieri and Maria Luisa Dalla Chiara, (eds.), Knowledge, belief, and strategic interaction, Cambridge University Press.

Skyrms, Brian 1990: 'Equilibrium and the dynamics of rational deliberation' in Biccieri and Dalla Chiara, (eds.), Knowledge, belief, and strategic interaction, Cambridge University Press.

4. Cooperation and convention in moral philosophy
Cohen, G.A. 1993 'Equality of what? On Welfare, Goods, and Capabilities' in Sen and Nussbaum 1993.

Danielson, Norman 1976 'Wide reflective equilibrium and theory acceptance in ethics' Journal of Philosophy 76, 256-82.

Danielson. Peter 1991: 'Closing the compliance dilemma' in Vallentyne, P. (ed.) Contractarianism and rational choice. Cambridge U. P.

Elster, Jon & John E Roemer 1991 Interpersonal Comparisons of Well-being Cambridge University Press
Gauthier, David 1986: Morals by agreement Oxford University Press

Hammond, Peter 'Interpersonal comparisons of utility: why and how they are and should be made' in Elster &Roemer
MacIntyre, Alasdair 1984: After Virtue second edition, Duckworth

Morton, Adam 1996b' The disunity of the moral' to appear in The problematic reality of values' Jan Bransen and Marc Slors, eds. van Gorkum.

Nussbaum, Martha C and Amartya Sen 1993 The Quality of Life Oxford University Press
Railton, Peter 1986: 'Moral Realism' Philosophical Review 95, pp.163-207.

Rawls, John 1971 A theory of justice. Harvard University Press.

Scanlon, Thomas 1991 The moral basis of interpersonal comparisons. In Elster & Roemer 1991

Sen, Amartya 1993 Capability and Well-Being. In Sen and Nussbaum.
Smith, Holly 1991 Deriving morality from rationality. In Vallentyne, P (ed.) Contractarianism and rational choice Cambridge U. P.