Research Article

Social Evaluation or Simple Association? Simple Associations May Explain Moral Reasoning in Infants

  • Damian Scarf mail,

    Affiliation: Department of Psychology, University of Otago, Dunedin, New Zealand

  • Kana Imuta,
  • Michael Colombo,
  • Harlene Hayne
  • Published: August 08, 2012
  • DOI: 10.1371/journal.pone.0042698

Reader Comments (3)

Post a new comment on this article

The case for social evaluation in infants

Posted by KileyHamlin on 09 Aug 2012 at 16:13 GMT

The case for social evaluation in infants

J. Kiley Hamlin (1), Karen Wynn (2) & Paul Bloom (2)

1. Department of Psychology
University of British Columbia
Vancouver, BC

2. Department of Psychology
Yale University
New Haven, CT

Correspondence concerning this article should be addressed to J. Kiley Hamlin, Department of Psychology, University of British Columbia, Vancouver, BC, Canada. Email:

Do humans possess an innate moral sense? To explore this issue, Hamlin et al. [1] presented evidence that infants judge others based on their social actions—they prefer those who help others and avoid those who hinder others. Scarf et al. [2] raise an alternative interpretation, presenting data that they interpret as suggesting that infants’ preferences in the social scenarios we presented to infants can be explained by an aversion to a ‘colliding event’ and an attraction to a ‘bouncing event’. Scarf et al. suggest, more generally, that babies possess neither a rudimentary moral compass nor any other sophisticated capacity, merely some simple and general associative mechanisms.

One set of concerns has to do with the experiment itself; unfortunately, Scarf et al.’s stimuli fail to match Hamlin et al.’s in several important ways1. First, their Climber looks different from ours: Since even very young infants attend to gaze direction - an important cue to others’ goals and desires [3] - our Climber’s gaze was directed up the hill throughout, indicating the Climber’s intended direction of travel. In contrast, the Scarf et al. Climber had ‘googly’ eyes with unfixed pupils which, due to gravity, usually pointed down the hill and away from the goal2. Second, their Climber acts different than ours: In order to indicate that the climb is a struggle, our Climber decelerates as it attempts to climb, slowly coming to a stop before reversing back down the hill. In contrast, the Scarf et al. Climber moves rapidly uphill, halts abruptly near the top, then moves smartly downwards, all giving an impression of a deliberate change of direction, which is only enhanced by the Climber’s downhill gaze. Third, during Scarf et al.’s Helping events, the Climber resumes its ascent immediately before the Helper contacts it, and continues to ascend between taps from the Helper below, as if able to climb the hill on its own. Finally and most strangely, during Hindering events, the Climber starts to move downwards before the Hinderer makes contact, further clouding its intended goal.

All of these considerations make it plausible, then, that Scarf et al.’s infants responded to perceptual variables because—unlike in our original study—the goal of the Climber was unclear to the infants and therefore the “helping” and “hindering” events did not strike them as helping or hindering.

Independent studies suggest that babies do extend a social interpretation to the Hamlin et al. paradigm. First, in the original, computer-animated version utilized by Kuhlmeier, Wynn and Bloom [4], the Climber did not bounce upon reaching the hilltop (instead, it expanded and contracted upon reaching the hilltop, as well as upon reaching the plateau halfway up - which it did on both helping and hindering trials). When shown these displays, 12-month-old infants expected the Climber to hold different attitudes towards the Helper and the Hinderer [4]. Second, the extent to which individual infants in Kuhlmeier et al [4] held this expectation predicted their theory-of-mind competence at 4 years of age [5] – an indication that these events engaged infants’ reasoning about goals, desires, and social interactions. Third, independent studies using these same computer-animated stimuli (which did not contain any bouncing event), found that infants matched helping events with attractive faces and hindering with unattractive faces, suggesting that they viewed helping as positive and hindering as negative [6-7] Scarf et al’s explanation can account for none of these findings.

Moreover, we have now replicated our main finding that infants prefer prosocial actors, across several social scenarios that do not involve climbing, colliding, or bouncing. We find that infants prefer a character who returns a ball to someone who dropped it over a character who runs away with the ball and they prefer a character who helps another open a box over a character who slams it shut against the other’s efforts [8]. Critically, these studies included control conditions in which characters performed actions identical in all respects to those in the experimental conditions, this time enacted upon inanimate objects—and here infants do not prefer one character over the other. In several further studies we have found that toddlers (a) direct their own positive behaviors toward helpful individuals [9, see also 10], (b) direct their own negative behaviors toward unhelpful ones [9], and (c) selectively refrain from emulating unhelpful characters [11]. Again, Scarf et al.’s explanation can account for none of these findings, while our proposal that infants engage in social evaluation accounts for all of them.

Finally, there is accumulating evidence that infants are not merely responding to the perceptual features of an action, but rather they — like adults — interpret an action as positive in some contexts, negative in others, and neutral in yet others. One set of studies finds that the same helpful action is viewed positively when directed towards a prosocial individual, but negatively when directed towards an antisocial individual (Hamlin, Wynn, Bloom, and Mahajan [9]3). Another set of studies find that infants prefer an individual who hinders someone whose attitudes (i.e., tastes in food) differ from the infant’s own, but prefer an individual who helps someone whose preferences match the infant’s (Hamlin, Mahajan, Liberman & Wynn [14]). Finally, infants evaluate an individual performing exactly the same action as positive, neutral, or negative, depending on infants’ understanding of the actor’s own mental states and of the mental states of the character affected by the action (Hamlin, Ullman, Tenenbaum, Goodman, & Baker [15]). Perceptual-associative accounts, by their very nature, cannot explain infants’ differing evaluations of perceptually identical actions.

For the reasons above, we find Scarf et al’s account unpromising, but we should note that we agree with the spirit of their enterprise; the vigorous pursuit of alternative explanations is essential to progress in this domain. Indeed, our current understanding that babies possess sophisticated cognitive adaptations for reasoning about (e.g.) number (Wynn, 1992 [16]; vanMarle & Wynn 2009 [17]) and about physical objects (Baillargeon, Spelke & Wasserman, 1985 [18]; Spelke & Hespos, 2001 [19]) would not be possible without the many critics in the 1980s and the 1990s who tried to explain the initial findings in terms of non-specialized mechanisms. The critics were unsuccessful, but their efforts led to increasingly stronger and more careful demonstrations of babies’ extroardinary cognitive competence.

But we do have our concerns with the general style of Scarf et al’s approach. As an example of what worries us, Scarf et al. [2] mention the studies of Hamlin & Wynn [7], which do not involve bouncing or colliding. But instead of proposing an alternative explanation for these findings, they simply assert that these other studies “are also confounded by salient perceptual events that may be driving the infants’ choices.” They do not say what these confounds are, nor why our control conditions are inadequate. Scarf et al.’stheory is so powerful, then, that, so long as there is any perceptible difference to be had, it can account for the findings in terms of that perceptible difference. Indeed, if we were to have obtained the opposite findings, they could “explain” those too. We hope it is clear that this is not a positive feature of their theoretical framework.

On a more theoretical note, we are bemused by Scarf et al’s attempt to recruit Darwin to their cause. Darwin was the last person to view the human mind as equipped with nothing ore than a set of simple and general mechanisms. He believed (correctly in our view) that the forces of natural selection have led to the emergence of specialized cognitive and social capacities. Indeed, he wrote extensively about the powers (and the limits) of our inborn moral sense [e.g., [20]]; and this was a main focus of his sole developmental paper—a diary study of his son, William, reported in the journal Mind in 1877 [21]. If Darwin were around today, he would be gratified to see how many of his ideas are supported by the modern science of infant cognition.


1. Hamlin et al.’s are videos viewable at:

2. We found suggestive evidence in our early piloting that this was the case: with a Climber whose pupils were not fixed but could freely rotate within the googly eye ‘socket’, infants did not prefer Helper over Hinderer, possibly because this made the Climber’s goal unclear, precluding ‘helping’ and ‘hindering’ interpretations of the other characters’ actions.

3. Scarf et al. [12] have elsewhere proposed an associative account of these findings. We believe their suggestion reflects misunderstandings of our methodology, our conclusions, and their own associative proposal (see [13]).

Hamlin et al. [1] was supported by NIMH #081877 and NSF #BCS-0921515 awards to KW and PB.

1) Hamlin, J.K., Wynn, K., & Bloom, P. (2007). Social evaluation by preverbal infants. Nature 450: 557-560.
2) Scarf, D., Imuta, K., Columbo, M., & Hayne, H. (in press). Social evaluation or simple association? Simple associations may explain moral reasoning in infants. PLoS One.
3) Hood, B.M., Willen, J.D., & Driver, J. (1998). Adult’s eyes trigger shifts of visual attention in human infants. Psychol Sci 9: 53-56.
4) Kuhlmeier, V.A., Wynn, K., & Bloom, P. (2003). Attribution of dispositional states by 12-month-olds. Psychol Sci 14: 402-408.
5) Yamaguchi, M., Kuhlmeier, V.A., Wynn, K., & vanMarle, K. (2010). Continuity in social cognition from infancy to childhood. Dev Sci 12: 746-752.
6) Taylor-Partridge, T., Griffin, A.M., Rosen, L., Langlois, J.H., & Principe, C. (2006). Precursors of attractiveness stereotypes: Infants link facial attractiveness with valence (Conference of the Am. Psych. Assoc., Washington, D.C.).
7) Langlois, J. H., Griffin, A. M, Rubenstein, A. J., Rosen, L., Taylor-Partridge, T., Ramsey, J., Hoss, R., & Principe, C. Infants link valence with facial attractiveness: Rudiments of a stereotype? (in revision).
8) Hamlin, J.K. & Wynn, K. (2011). Young infants prefer prosocial to antisocial others. Cog Dev 26: 30-39.
9) Hamlin, J.K., Wynn, K., Bloom, P., & Mahajan, N. (2011). How infants and toddlers react to antisocial others. Proc Nat Acad Sci USA 108: 19931-19936.
10) Dunfield, K.A., & Kuhlmeier, V.A. (2010). Intention-mediated selective helping in human infants. Psychol Sci 21: 523-527.
11) Hamlin JK, & Wynn, K. (2012). Who knows what’s good to eat? Infants fail to match the food preferences of antisocial others. Cog Dev. DOI: 10.1016/j.cogdev.2012.05.005.
12) Scarf D, Imuta K, Colombo M, Hayne H (2012). Golden rule or valence matching? Methodological problems in Hamlin et al. Proc Nat Acad Sci USA 109: E1426.
13) Hamlin JK, Wynn K, Bloom P (2012). Reply to Scarf et al.: Nuanced social evaluation: Association doesn’t compute. Proc Nat Acad Sci USA 109: E1427.
14) Hamlin, J.K., Mahajan, N., Liberman, Z., & Wynn, K. (in press). Not like me = bad: Infants prefer those who harm dissimilar others. Psychol Sci.
15) Hamlin, J.K., Ullman, T., Tenenbaum, J., Goodman, N., & Baker, C. (in press). The mentalistic basis of core social cognition: experiments in preverbal infants and a computational model. Dev Sci.
16) Wynn, K. (1992). Addition and subtraction by human infants. Nature, 358, 749-750.
17) vanMarle, K., & Wynn, K. (2009). Infants’ auditory enumeration: Evidence for analog magnitudes in the small number range. Cognition, 111, 302-316.
18) Baillargeon, Spelke & Wasserman (1985). Object permanence in 5-month-old infants. Cognition, 20, 191-208.
19) Spelke, E. S., & Hespos, S. J. (2001). Continuity, competence, and the object concept. In E. Dupoux (Ed.), Language, brain, and cognitive development: Essays in honor of Jacques Mehler (pp. 325-340). Cambridge, MA: Bradford/MIT Press.
20) Darwin, C. (1871). The descent of man. London: J. Murray.
21) Darwin, C. (1877). A biographical sketch of an infant. Mind, 2, 285-294.

Competing interests declared: We are the authors of the original article.