Forthcoming in Cognition.

Are children moral objectivists?

Children’s judgments about moral and response-dependent properties

 

Shaun Nichols^a

Trisha Folds-Bennett^b

 

^aDepartment of Philosophy, College of Charleston, Charleston, SC 29424

^bDepartment of Psychology, College of Charleston, Charleston, SC 29424

______________________________________________________________________________

Abstract

Researchers working on children’s moral understanding maintain that the child’s capacity to distinguish morality from convention shows that children regard moral violations as objectively wrong (e.g., Nucci, 2001).  However, one traditional way to cast the issue of objectivism is to focus not on conventionality, but on whether moral properties depend on our responses, as with properties like icky and fun.  This paper argues that the moral/conventional task is inadequate for assessing whether children regard moral properties as response dependent. Unfortunately, children’s understanding of response-dependent properties has been neglected in recent research.  Two experiments are reported showing that children are more likely to treat properties like fun and icky as response dependent than moral properties like good and bad.  Hence, this helps support the claim that children are moral objectivists.

______________________________________________________________________________

Keywords: Moral judgment; response dependent properties; moral objectivism; moral/conventional distinction

 

          Among analytic philosophers, it is widely assumed that people embrace moral objectivism, the view that true moral claims are nonrelativistically true.  Both philosophers who defend moral objectivism (e.g., Smith, 1994; Darwall, 1998) and philosophers who oppose moral objectivism (e.g., Mackie, 1977) maintain that moral objectivism is absolutely central to folk metaethics.  Researchers on moral judgment in children maintain that even young children accept  moral objectivism (e.g., Nucci, 2001).  Here, we will consider this question of childhood objectivism directly. We will argue that the available evidence neglects to explore whether children distinguish moral properties from manifestly non-objective properties that depend on the responses of a population.  Two experiments are presented that show that children do distinguish moral properties from response-dependent properties, thus supporting the claim that children are indeed moral objectivists.

Before we continue, we need a somewhat sharper characterization of commonsense moral objectivism.  J. L. Mackie’s widely influential treatment will work nicely for a start.  “The ordinary user of moral language means to say something about whatever it is that he characterizes morally, for example a possible action, as it is in itself, or would be if it were realized, and not about, or even simply expressive of, his, or anyone else’s relation to it.” (Mackie, 1977, 33). Thus, to claim that an action is objectively immoral is to claim that the action is wrong “as it is in itself” and not in relation to other subjects.  There are various quibbles that might be made over this characterization, but the underlying idea is familiar.  According to the objectivist, if a particular action is morally wrong, then it is wrong simpliciter.  So morally wrong actions are not merely wrong relative to certain populations.  To get some purchase on this, it’s easiest to focus on a particular example.  Let’s say that a teenage boy, Bill, intentionally kicks a small dog.  It cannot turn out, according to the objectivist, that Bill’s kicking the dog was morally wrong for some populations but not for other populations.  If the action is morally wrong, it’s wrong full stop.  Thus, moral objectivism is committed to the view that (i) true moral judgments are nonrelativistically true and (ii) some moral judgments are true.

In developmental psychology, perhaps the most important work on moral judgment over the last two decades has explored the capacity to distinguish moral violations from conventional violations (for discussion see e.g., Blair, 1995, Nichols, 2002, Nucci, 2001, Turiel, 1983).  From a young age, children distinguish moral violations (e.g., pulling someone’s hair) from conventional violations (e.g., talking out of turn) on a number of dimensions.  This work on the moral/conventional distinction has been entered as evidence that children are moral objectivists.  For instance, Larry Nucci writes:  “Preschool-aged children … understand that it is objectively wrong to hurt others” (2001, 86; see also Flanagan, 1991, 348 fn3).  Nucci adduces two key findings as evidence that children are moral objectivists (Nucci, 2001, 86f.): 

i. children regard moral violations as less authority contingent than conventional violations.

ii. children regard moral violations as more generalizably wrong than conventional violations.

So, for instance, children will say that even if the teacher says it’s okay to pull hair, it’s not okay to do that.  By contrast, children are much more likely to allow that it’s okay to talk out of turn if the teacher says it’s okay.  On the other dimension, children are likely to say that pulling hair is not okay in other places, at other times, in other countries, and so forth.  They are more likely to allow that talking out of turn is okay in other places.

This work on the moral/conventional task does indicate that children reject conventionalism, the anti-objectivist view that what counts as morally wrong varies with the prevailing conventions (e.g., Benedict, 1934).  Children apparently regard some moral claims as true, and they do not take this to be merely a matter of prevailing conventions.  However, in philosophical ethics, a more prominent anti-objectivist position maintains that moral properties are not objective because they are “response dependent” (e.g., Hume, 1739/1964, Stevenson, 1944, Gibbard, 1990).  There are different notions of response dependence, but the basic idea is that a property is response dependent just in case that property is constituted by the responses it elicits in a population; so the same object or event might have different response-dependent properties for different populations (see, e.g., Cohen, forthcoming, Johnston, 1989, Smith & Stoljar, 1998, Wedgwood, 1998).  As a result, a commonsense exemplar of a response-dependent property is icky.  The same object might be icky for one population and not icky for another population, and no thing is icky “as it is in itself”.  Rather, whether something is icky depends on the responses of the focal subjects. 

It’s easy to see how this hooks up with the earlier discussion of objectivism.  If morality is objective, then the moral status of an action cannot be relative to a set of subjects the way icky depends on the responses of a set of subjects.  Hume is sometimes interpreted as maintaining that moral judgments are indeed akin to judgments about properties such as “tasty” and “icky”.  Perhaps, then, in making moral judgments, children regard moral properties as response dependent in a similar way.  If so, then children are not moral objectivists after all.[1]  A related concern runs in the other direction.  If children never treat any properties as response dependent, then one might worry that children really don’t grasp the objective/nonobjective distinction.[2]  And in that case, it might be misleading to say that children are objectivists.  Hence, to sustain the view that children are moral objectivists, one would hope to find that children treat moral properties differently from properties that are obviously response dependent.

The moral/conventional task does a poor job of assessing whether children regard moral properties as dependent on our responses; for simple response-dependent properties might themselves be regarded as “non-conventional”, both by children and adults.  Consider first the authority contingency dimension. We can easily devise an authority contingence question for response-dependent properties.  For instance, one might ask the following:

“If the teacher said that liver is yummy, would liver be yummy?”

Adults, and presumably children as well, regard the teacher as in no position to make liver yummy.  If this is right, then merely showing that children regard moral transgressions as bad in an authority independent way does not show that children regard moral properties as response independent.  In the case of the generalizability dimension, the problem is that in the context of response-dependent properties, generalizability questions are notoriously ambiguous.  If asked whether onions are icky in another country, an onion-hater might well assent.  She might interpret the question as asking whether onions in another country would be icky to her.

Hence, results from the moral/conventional task don’t exclude the possibility that children regard moral violations as bad in a response-dependent manner, so we still don’t know whether children are moral objectivists.  We need to see whether children do distinguish systematically between moral properties and response-dependent properties.  It will be of independent interest to see whether children understand that some properties are response dependent, since there are long-toothed philosophical debates over which properties are response dependent. These philosophical debates focus on properties that are far less straightforward than icky.  For instance, much work has been done on whether properties like red are response dependent.  Debates on that issue have a renewed vigor in the literature, and it’s far from settled whether colors are best regarded as response-dependent properties, or as response-independent properties like spectral reflectance distributions.  Since much of the debate centers on how best to characterize the folk notion of color, determining the development of children’s understanding of response-dependence has the potential to illuminate these issues.  At this point, we don’t yet have an answer even to whether children appreciate that properties like yummy are response dependent.  So we start with this simple question.

 

Experiment 1

This experiment investigated whether children distinguish paradigmatically response-dependent properties (yummy, fun) from moral (good) and aesthetic (beautiful) properties.  They were explored along two dimensions:  preference-dependence and generalizability.

 

Method

Participants

Nineteen children, ages 4 through 6 years, participated (M= 64.3 months; range 50-77 months).  All participants were recruited from the N. E. Miles Early Childhood Development Center at the College of Charleston.  Five participants were female; fourteen were male.

 

Materials

Six questions were used in this study. Two questions involved the moral property good.  In one of these, one monkey helps another hurt monkey, and the child is asked whether that is good.  Two questions involved the aesthetic property beautiful.  In one of these, the child was asked whether roses are beautiful.  Two items involved properties commonly regarded as response dependent (yummy, fun).  For each item, children were asked whether a property applied to something, e.g., “Are grapes yummy?”.  Following a “yes” response[3] the child was asked a preference dependence and a generalizability question.

Preference dependence:  You know, I think grapes are yummy too.  Some people don’t like grapes.  They don’t think grapes are yummy.  Would you say that grapes are yummy for some people or that they’re yummy for real?

 

Generalizability:  Now, think about a long time ago, before there were any people.  There were still grapes, just like the grapes now.  Way back then, before there were people, were grapes yummy?

 

Procedure

Children were tested individually in a familiar room in their daycare by two experimenters.  Children were presented with all six items.  Items were counterbalanced for domain, and the questions about generalizability and preference dependence were also counterbalanced.  The last part of the preference dependence question (for some people/for real) was alternated within subject. 

 

Results

Scoring

For the preference dependency question, each ‘for some’ response was given a score of 1, each ‘for real’ response was given a score of 0, and the scores were summed for each domain (moral, aesthetic, response-dependent); so the cumulative score could range from 0 to 2.  A criterion of 2 out of 2 was used to define preference dependence, and a criterion of 0 out of 2 was used to define preference independence.  For the generalizability question, each “yes” answer was given a score of 1, each “no” answer was given a score of 0, and the scores were summed for each domain.  A criterion of 2 out of 2 was used to define generalizability, and a criterion of 0 out of 2 was used to define nongeneralizability.  The frequency of response patterns is shown in Table 1.

 

***Insert Table 1 about here***

 

Preference dependence questions

To compare questions from two different domains, the differences of scores on these two domains were computed for each child, and a sign test was used.  Pre-planned sign tests were conducted comparing response-dependent cases with moral cases and with aesthetic cases.  Comparing moral with response-dependent cases, there were 13 negative differences and 2 positive differences, indicating that participants were more likely to judge the moral properties as preference independent (p<.01, two-tailed).  Comparing aesthetic with response-dependent cases, there were 12 negative differences and 1 positive difference, indicating that participants were more likely to treat the aesthetic properties as preference independent (p<.01, two-tailed).

 

Generalizability questions

For the generalizability questions, pre-planned sign tests were again used to compare response-dependent cases with moral cases and with aesthetic cases.  Comparing moral with response-dependent cases, there was 1 negative difference and 5 positive differences, yielding no significant difference between domains.  For the comparison between aesthetic and response-dependent cases, there were 0 negative differences and 4 positive differences, again yielding no significant difference between domains.  Most children regarded all of the properties (good, yummy, fun, beautiful) as generalizable.[4] 

 

Experiment 2

The previous experiment shows that children do distinguish moral properties like good from response-dependent properties like yummy and fun.  But the properties explored were exclusively positive, and the bulk of the developmental literature has focused on judgments about negative moral properties (e.g., bad).  Hence, a second experiment was conducted to see whether children would distinguish negative moral properties from negative response-dependent properties like icky and boring.  The prediction was that children would distinguish the moral properties from the response-dependent properties.   We were also interested in whether children would distinguish response-dependent properties from conventional and disgusting transgressions judged bad by children.

 

Method

Participants

Thirteen 5-year old children participated (M=67 months; range 64-71 months). All participants were recruited from the O’Quinn School in James Island, South Carolina.  Six participants were female; seven were male.  An additional participant was excluded for failing to answer several questions.

 

Materials

Eight items were used in this study.  The four key items involved paradigmatically response-dependent properties (icky, boring) and standard moral transgressions (hitting, pulling hair) judged to be bad.  The study also included two items involving conventional transgressions (standing during story-time; drinking soup out of a bowl) and two items involved disgusting transgressions (spitting in water before drinking it; wiping nose on hand and sleeve).  As in the previous study, for each item, children were asked whether a property applied to something.  For example, in one of the response-dependent items, the child was asked, “Is it boring to clean house?”.  Following a “yes” response[5] the child was asked a preference dependence question and a generalizability question. 

Preference dependence:  You know, I think it’s boring to clean house  too.  Some people like to clean house.  They think it’s not boring.  Would you say that cleaning house is boring for some people or that it’s boring for real?

 

Generalizability:  In another country or in some place far away from here, is it boring to clean house?

 

Procedure

Children were tested individually in a room in their daycare by two experimenters.  Children were presented with all eight items.  To guard against response-set, after the first four items were completed, one experimenter played a memory game with the child before continuing with the remaining items. Items were counterbalanced as in Experiment 1.

 

Results

Scoring

Same as in Experiment 1.  The frequency of response patterns is shown in Table 2.

 

***Insert Table 2 about here***

 

Preference dependence questions

As in experiment 1, to compare questions from two different domains, the differences of scores on these two domains were computed for each child, and a sign test was used.  Pre-planned sign tests were conducted comparing response-dependent cases with moral cases, disgusting cases, and conventional cases.  Comparing moral with response-dependent cases, there were 8 negative differences and 1 positive difference, yielding a significant difference between domains (p<.05, two-tailed).  Comparing disgusting with response-dependent cases, there were 8 negative differences and 1 positive difference, yielding a significant difference between domains (p<.05, two-tailed).  For the comparison between conventional and response-dependent cases, there were 8 negative differences and 2 positive differences, yielding no significant difference between domains. 

 

Generalizability questions

Again on the generalizability questions, pre-planned sign tests were conducted comparing response dependent cases with moral cases, disgusting cases, and conventional cases.  There were 0 negative and 6 positive differences between the moral cases and the response dependent cases, yielding a significant difference between domains (p<.05, two-tailed).  Similarly, there were 0 negative and 6 positive differences between the disgusting cases and the response dependent cases, showing a significant difference between disgusting cases and response-dependent cases (p<.05, two-tailed).  Comparing the conventional and response-dependent cases, there were 2 negative differences and 6 positive differences, yielding no significant difference between these domains.

 

Discussion

            The findings of both experiments support the claim that children do not regard moral properties as response dependent.   In both experiments, children show an appreciation that properties like yummy, icky, fun, and boring are response dependent.  Children were significantly more likely to treat such properties as preference dependent than they were to treat moral properties or aesthetic properties as preference dependent.  Like moral violations, disgusting violations were treated as bad in a preference independent way.  Furthermore, children were less likely to generalize the response-dependent properties (icky, boring) than the badness of disgusting and moral violations.  Together with previous findings that moral violations are not merely conventionally bad, these results suggest that children are indeed moral objectivists. 

Of course, the findings in this paper do not complete the case for childhood objectivism.  For one thing, the experiments probe only a simple kind of response dependence. There are more sophisticated kinds of response dependence (see, e.g., Gibbard 1990), and it’s possible that children regard moral properties as response dependent in some more sophisticated way.  Thus, there are many further empirical questions about whether children regard moral properties as entirely independent of responses.  In addition, it will be important to explore both the early development of the child’s appreciation that some properties are response dependent as well as the extent to which children will treat other important properties (e.g., color, temperature) as response dependent. 

            The results on childhood objectivism also raise broader theoretical questions.  In particular, we might consider more systematically how children acquire an objectivist notion of morality. One possibility is that children have an innate concept of moral violation, which carries with it a commitment to moral objectivism.   At the other end of the spectrum, one might maintain that children learn, through instruction or discovery, moral objectivism.  In light of recent work on the emotional underpinnings of moral judgment (e.g., Blair 1995, Nichols 2002), a rather different alternative is that emotional response plays an important role in leading children to treat moral violations as objectively wrong.  Renewed attention to the ontogeny of moral objectivism will, we hope, help illuminate these issues.

 

 

Acknowledgements

We would like to thank the teachers, parents, and children of the N.E. Miles Early Childhood Development Center at the College of Charleston and of the O’Quinn School on James Island.  Thanks also to Abigail Nuse and Emily Askey for assistance in collecting data.  Finally, we are grateful to Jonathan Cohen, Tim German, Michael Gill, Ned Hettinger, Jim Hittner, Kim May, and to several anonymous referees for comments on a previous draft of this paper. 

 

 

References:

 

Benedict, R. (1934). Patterns of culture.  New York: Penguin Books.

 

Blair, R. (1995).  A cognitive developmental approach to morality: investigating the psychopath,  Cognition, 57, 1-29.

 

Cohen, J. forthcoming.  Color properties and color ascriptions: a relationalist manifesto.

 

Darwall, S. (1998). Philosophical ethics.  Boulder, CO:  Westview.

 

Flanagan, O. (1991).  Varieties of moral personality:  ethics and psychological realism.  Cambridge, MA:  Harvard University Press.

 

Flavell, J., Flavell, E., Green, F., and Moses, L. (1990).  Young children’s understanding of fact beliefs versus value beliefs.  Child Development, 61, 915-928.

 

Gibbard, A. (1990).  Wise choices, apt feelings.  Cambridge, MA:  Harvard University Press.

 

Gopnik, A. and A. Meltzoff (1997).  Words, thoughts and theories.  Cambridge, MA:  MIT Press.

 

Hume, D. (1739/1964).  A treatise of human nature.  Oxford:  Clarendon Press.

 

Johnston, M. (1989).  Dispositional theories of value, Proceedings of the Aristotelian Society Supplementary Volume, 63, 139-174.

 

Mackie, J. (1977).  Ethics: inventing right and wrong.  London: Penguin.

 

Nichols, S. (2002). Norms with feeling:  towards a psychological account of moral judgment,  Cognition, 84, 221-236.

 

Nichols, S. (forthcoming).  Sentimental rules: on the natural foundations of moral judgment.  New York: Oxford University Press.

 

Nichols, S. and S. Stich (2003).  Mindreading.  Oxford:  Oxford University Press.

 

Nucci, L. (2001). Education in the moral domain.  Cambridge:  Cambridge University Press.

 

Smith, M. (1994).  The moral problem,  Oxford:  Blackwell.

 

Smith, M. & D. Stoljar (1998).  Global response dependence and noumenal realism, The Monist, 81, 85-111.

 

Stevenson, C. (1944).  Ethics and language.  New Haven:  Yale University Press.

 

Turiel, E. (1983).  The development of social knowledge: morality and convention,  Cambridge: Cambridge University Press.

 

Wedgwood, R. (1998).  The essence of response-dependence, European Review of Philosophy, 3, 331-54.

Table 1. Frequency of Response Patterns for Preference Dependence and Generalizability Questions across Domain

 

	Preference Dependence Questions			Generalizability Questions
Domain	Preference dependent	Ambiguous	Preference independent	Generalizable	Ambiguous	Nongeneralizable
Response   dependent	9	7	3	14	3	2
Moral	3	4	12	17	2	0
Aesthetic	2	8	9	17	1	1

 

Table 2. Frequency of Response Patterns for Preference Dependence and Generalizability Questions across Domain

 

	Preference Dependence Questions			Generalizability Questions
Domain	Preference dependent	Ambiguous	Preference independent	Generalizable	Ambiguous	Nongeneralizable
Response   dependent	7	4	2	7	4	2
Moral	2	2	9	13	0	0
Disgusting	2	4	7	13	0	0
Conventional	3	2	8	10	2	1