sign in sign up
<<
Home , Social cognition

OUP UNCORRECTED PROOF – FIRSTPROOFS, Sat Dec 22 2012, NEWGEN

Meltzoff, A.N. (2013). Origins of social : Bidirectional self-other mapping and the “Like-Me” hypothesis. In M. Banaji & S. Gelman (Eds.), Navigating the social world: What infants, children, and other species can teach us (pp. 139-144). New York, NY: Oxford University Press. 3.3 Origins of Social Cognition Bidirectional Self-Other Mapping and the “Like-Me” Hypothesis

ANDREW N. MELTZOFF

uman social cognition begins in the new- self and other suff ers from the same problem, and Hborn period. When the human newborn were precursors to Piaget’s more sophisticated opens its eyes and sees a human act for the fi rst writing on the initial state. time, I have proposed that this engenders a feel- Fourth, although contemporary work on social ing of interpersonal connectedness. To explain primitives has shown that infants have visual pref- the idea, I off ered the “Like-Me” developmental erences for faces (Johnson and Morton), eyespots hypothesis. Th e key suggestion is that perceiving (Baron-Cohen), and self-mobile entities (Leslie), others as like me is a social primitive. Th e new my proposal goes beyond cues that induce infant empirical research shows that the core sense of visual . Infants’ supramodal representa- similarity to others is not the culmination of tion of human action maps external events to the social development, but the precondition for it. self. Crucially, this goes beyond heightened atten- Without this initial felt connection to others, tion, preference, and visual expectations. human social cognition would not take the dis- In this essay I will discuss empirical studies tinctively human form that it does. concerning two social learning mechanisms that In the past, philosophers, social theorists, build on the “like-me” primitive—imitation and and psychologists have considered related ideas, gaze following. Th e studies illuminate three aspects but the foregoing departs from historical discus- of social cognition: (a) origins, (b) mechanisms of sions in four ways. First, the relevant philosophers change, and (c) bidirectional learning between (e.g., Husserl, Merleau-Ponty, Ryle) did not imag- self and other. Th e studies support the view that ine that this process begins at birth. Th ese phi- “like me” is a social primitive that gives rise to losophers postulated that self-other mapping and a life-long ability to connect to other humans, fi rst steps toward cracking the problem of Other which is vital to our survival as a species. Minds were late emerging and mediated by lan- Humans have a long period of infantile imma- guage and/or deliberate introspective reasoning. turity compared to other animals. Th is immatu- Second, many sociologists and social psy- rity has coevolved with powerful social learning chologists working with adults gave priority to mechanisms. Two of these social learning mech- social learning from the outside in: An unformed anisms—imitation and gaze following—are self is given coherence by how others react to it functional in human infancy, rare in the animal (Cooley, Mead)—the “looking-glass self.” Instead, kingdom, and impoverished in . If one as I will show, infants also use self-experience to seeks to understand the birth, growth, and mech- give meaning to the observed of others anisms of change in human social cognition, imi- in ways these theorists missed. tation and gaze following are promising domains Th ird, Piaget’s work with infants led him to to investigate. theorize that they start out as “solipsists” with no Imitation supports rapid learning of behav- links between self and other. Piaget was unable ioral skills, social customs, and causal relations to provide a satisfactory mechanism of change by observation. Rather than relying purely on from initial solipsism to the rich social cognition maturation (which is not responsive to cultural of 5-year-olds, because he underestimated the contingencies), independent invention (which is initial state. Freud’s and Baldwin’s neonatal “adu- slow) or trial and error (which can be dangerous), lism,” merging, or lack of diff erentiation between humans excel in imitative learning from others in

25_Banaji_Ch3.3.indd 139 12/27/2012 9:42:26 PM OUP UNCORRECTED PROOF – FIRSTPROOFS, Sat Dec 22 2012, NEWGEN

140 Imitation, Modeling, and Learning From and About Others

the environment. A key to human imitation is that congenital aspect of human social cognition, with it is not slavish or mindless duplication. Children much known at the behavioral-psychological level, and adults selectively regulate who, what, and and rapid progress being made illucidating infant when to imitate. Imitative learning is a chief ave- neural mirroring mechanisms (e.g., Marshall & nue for social learning prior to language, but it Meltzoff , 2011; Saby, Marshall, & Meltzoff , 2012). also plays a role when adults are in novel social Infants’ self-experience may play a role in imi- settings. Visiting a foreign country in which social tation. Even at birth infants have had experience greetings and eating implements diff er from ours with self-generated movements, which we call reminds us of the value of imitating experts rather body babbling (Meltzoff & Moore, 1997). Films of than reliance on trial and error or invention. prenatal behavior, for example, reveal that fetuses Following the gaze of other people directs our make repeated lip, tongue, and hand movements in attention to particular regions of space. Our atten- utero. Body babbling provides infants with experi- tion is directed to locations, objects, and events ence in how their own body moves. Body babbling that are particularly rich in information. Th ese is a mechanism for infants’ learning about control- “information hot spots” do not stand out based ling their body, analogous to how vocal babbling on physical cues alone. Th ey are not necessarily provides experience in vocal maneuvers (Kuhl & bigger, brighter, or more visually conspicuous. Meltzoff , 1982). Tongues move diff erently from Th ey gain psychological merely because a hinged joints such as fi ngers and elbows. Based on social other has looked at them. Perceptual align- self-experience with the felt movements of one’s ment with others undergirds common ground for own body, the kinetic signatures of a tongue pro- communication and supports pedagogy and col- trusion/withdrawal (or mouth opening/closing, laborative learning. or fi nger fl exing) could be recognized as cross- modally equivalent to those produced by others. IMITATION AS SOCIAL COGNITION Personal Identity Th eories of developing social cognition must Bodily Imitation be concerned with personal identity. How does In facial imitation, infants are duplicating a gesture an infant distinguish one person from another they see with a gesture of their own that they can- and reidentify a person as the “same one” again not see themselves make. Developmentalists have despite changes in surface appearances? Infants’ known for more than 50 years that 1-year-olds social worlds would be very diff erent from ours if imitate facial gestures, but it was a surprise to dis- an interpersonal relationship were not maintained cover that newborns as young as 42 minutes old when the surface features of the other were altered. imitate such acts (Meltzoff & Moore, 1983). Th is is related to the topic of object individuation Newborn facial imitation provides informa- and numerical identity (Moore & Meltzoff , 2009; tion about the initial state of social cognition. Spelke, Breinlinger, Macomber, & Jacobson, 1992; Newborns can see another person move but have Xu & Carey, 1996), but considers the identity of never seen their own face. Th is apparent gulf people who disappear and reappear and change between self and other is reminiscent of the Other appearance. Minds problem in philosophy (albeit at the level We discovered that infant imitation contrib- of actions and not intentional states). We know utes to infants’ understanding of personal iden- ourselves from the inside and others from the tity. Infants use both a person’s spatial history outside: How do we understand what it is like to (spatiotemporal parameters) and the prior expe- be another person, to feel what the other person rience of interacting with that person (functional feels? parameters) to determine a person’s identity. I have argued that imitation is based on active When infants are ambiguous about whether this is intermodal mapping, positing a primitive and the same person they saw previously, infants show foundational body scheme that allows infants to increased imitation of the person’s past behavior, unify the seen acts of others and their own felt as if verifying “Are you the one who does x?” Body acts in a shared framework. Meltzoff and Moore actions, mannerisms, and distinctive interactive (1997) postulated that human infants accomplish games and routines played by particular people this via a “supramodal” representation of human are akin to nonverbal shared that can acts. Links between and production do be used as identity markers of people. not have to be forged through associative learn- Th e value of using human acts as criteria for ing but are available to the newborn. Imitation is a identity is that they allow infants to test as well

25_Banaji_Ch3.3.indd 140 12/27/2012 9:42:29 PM OUP UNCORRECTED PROOF – FIRSTPROOFS, Sat Dec 22 2012, NEWGEN

Origins of Social Cognition 141

as to observe. Instead of being limited to gen- being preserved. Neonates can successfully imi- eralized reactions, such as smiling, cooing, and tate without yet knowing what their acts look like greeting of “humans in general,” infants have a in a purely visual sense from the outside. Classical tool for intervening to actively probe whether developmental and psychoanalytic theory (e.g., they are re-encountering the same individual Lacan) implicated physical mirrors in the onto- aft er a perceptual break or change in appearance. genesis of a visual self, but this cannot be the Social beliefs and attitudes are oft en about specifi c only source. Mirrors are not culturally universal people, not categories. For infants who cannot yet or historically ancient. Mutual imitation provides use language to query a person, keeping track of another mechanism. Th rough such social mirror- individuals requires combining observation (spa- ing, infants gain a better sense of what their own tiotemporal criteria) and action experience with felt acts look like. that person (functional criteria) to make judg- studies of mutual imita- ments about a person’s identity. tion in adults show that the right inferior parietal lobe is activated when people experience them- The Imitation Game: Being Imitated selves being imitated (Decety, Chaminade, Grè zes, Begets Social Bonding & Meltzoff , 2002). A speculation is that the right One of the favorite activities of parents and chil- inferior parietal lobule is involved in diff erentiat- dren are reciprocal imitation games, and I argue ing self and other when they are both performing that these games support social bonding and affi li- the same actions (a neural “who done it”). Related ation. Why are these mutual imitation games so social neuroscience work is now emerging in alluring for infants? Temporal contingencies are infants (Saby, Marshall, & Meltzoff , 2012). part of the story, but it is oft en overlooked that people are special to infants because they can sys- GAZE FOLLOWING tematically match the form of their behavior in a For adults, human attention implicates both generative fashion. Th is structural congruence is external objects and the internal states of the psychologically salient to infants (and adults). viewer. When we see someone turn to look at an In one study we gave infants a choice between object, this is interpreted as more than a simple two adults who were sitting side by side, one imi- physical motion in space. It is recognized as an tating what the infant did and the other acting attempt to acquire information from afar despite contingently on the infant’s actions but produc- the gap between the person and thing. Visual ing a mismatching action. Th e infants selectively perception is a kind of psychological contact at looked longer and smiled more at the matching a distance. adult (Meltzoff , 1990). Infants’ attention and posi- When do infants begin to ascribe visual per- tive emotions were directed at the one who acted ception to others’ acts of looking, and how do “like me.” they come to make this attribution? I believe Th ere are two key social-developmental that infants in part develop an understanding sequelae. First, mutual imitation games deepen a of the visual perception of others through their sense of relationship. Mutual imitation indicates self-generated acts of turning in order to see and “communing” or “being with” someone else, even opening/shutting their eyes to cut off and reinstate prior to the time that linguistic exchanges are visual experience. Self-experience changes their possible. Adults develop positive feelings toward interpretation of the visual of others. another person who is refl ecting their behavior back to them, despite not being aware of the cause Self-Experience: Learning From the of these feelings in psychotherapeutic and every- Inside Out day settings (Ogden, 1982; van Baaren, Holland, Th is idea that infants’ own visual experiences Steenaert, & van Knippenberg, 2003). We see the contribute to their understanding of others’ seeds of this in the reciprocal imitation games in vision emerged from a puzzling fi nding. In a preverbal infants. gaze-following study, an adult sat across from the Second, caretakers’ mirroring serves the func- infant and turned to face objects on a random tions as a physical mirror. Infants can use imita- schedule. Twelve-month-olds followed the adult tive interactions to learn what the self looks like. signifi cantly more when her eyes were open rather Th is provides a lever for developmental change, than closed (Brooks & Meltzoff , 2002). Th ese because early facial imitation is mediated by infants seemed to understand that eye closure supramodal equivalence, and infants can accom- occludes vision, but this understanding was con- plish it without modality-specifi c information strained—the same 12-month-olds turned when

25_Banaji_Ch3.3.indd 141 12/27/2012 9:42:29 PM OUP UNCORRECTED PROOF – FIRSTPROOFS, Sat Dec 22 2012, NEWGEN

142 Imitation, Modeling, and Learning From and About Others

the adult wore a blindfold, which also occludes one control group saw the same adult wearing the vision. Why? same blindfold and making fumbling reaches as if Infants control their own vision by clos- she could not see. Infants in another control saw ing their eyes. We hypothesized that extensive the adult wearing the black cloth as a headband self-experience turning off /on visual access and successfully grabbing the objects. through eye closing/opening provides a basis for Infants who saw the blindfolded adult make understanding similar behaviors in others. Th is visually guided reaches were signifi cantly more predicts that if infants are given experience that likely to follow the adult’s line of regard than the blindfolds block their view they should make new control infants. Th is suggests that infants can draw attributions to others, and they did. Meltzoff and inferences about whether the person can or can- Brooks (2008) gave 12-month-olds experience not see based on observing a structured pattern of with blindfolds occluding their vision. Aft er this goal-directed behavior. Th e infants inferred sight experience, infants treated adults wearing blind- because the blindfolded adult could systematically folds diff erently. Infants who received fi rst-person reach out and pick up distal objects. Th us, infants experience with the blindfold treated the adult followed her gaze to the external object. Th is is as though her view was blocked, suggesting that signifi cant, because it contradicts the infants’ own their fi rst-person experiences changed their attri- self-experience with opaque cloths. (For related butions to others. Self-experience provided a work, see Williamson & Meltzoff , 2011.) “like-me” framework for interpreting the other’s experience. “SOCIAL ROBOTS” We also discovered that self-experience can Infants do not “gaze follow” in response to every teach infants novel information that violates movement in the visual fi eld. If the wind blows the notions of everyday visual opacity. It is well the door opens, they are unlikely to follow where established that by 18 months infants know that the doorknob is pointing. We investigated who opaque barriers such as walls, barriers, and blind- or what infants would gaze follow. We showed folds occlude the adult’s line of regard. We pro- 18-month-old infants a robot and experimentally vided 18-month-olds with biologically deviant varied how it interacted with others. self-experience. We designed a trick blindfold that Th e eff ective manipulation was witnessing looked opaque from the outside but was made of the robot having a social interaction: If the robot special material that could be seen through when engaged in imitative exchanges with a person, the held close to the eyes. Infants given an intense infants seemed to attribute sentience to the robot bout of this self-experience generalized from self and to follow its gaze. When a robot simply moved to others and treated the blindfolded adult as if its head (complete with eyespots) to the side, she too could see through it: Th ey followed her infants did not gaze follow. When the robot exhib- gaze to distal objects (Meltzoff & Brooks, 2008). ited identical motions as the adult, but not as part As infants gain fi rst-person experience, they fl ex- of a reciprocal social interaction, gaze following did ibly transform their understanding of others. Self- not occur. (Meltzoff , Brooks, Shon, & Rao, 2010). experience can induce a change in understanding It is not solely the visual features or move- the other. ments of an entity, but the interactions it engages in, that are cues of social cognition. Reciprocal Self-Experience Is Not the Royal Road imitation is interpreted as a marker of psychologi- to Social Cognition cal agency. An entity that systematically imitates Is self-experience the sole path for coming to is one that perceives. Infants draw social-cogni- understand the inner workings of other people? tive inferences from watching structured patterns My research and theory, and those of others, of behavior—in this case social interaction—and strongly suggests not. Infants also come to under- do not require self-experience with the robot (in stand others as bearers of psychological states by closely related studies such as Johnson, Slaughter, observing structured patterns of their behavior. & Carey, 1998, there was contingent interaction To test this, we modifi ed the blindfold para- between the infant and the nonhuman entity). digm used with 18-month-olds. Infants were not given self-experience. Th ey watched a blindfolded CONCLUSIONS adult reach out and grab objects on the table in Th e primitive on which social cognition rests is front of them. We thought that this “visually the perceived “like me” equivalence between self guided behavior” would be suffi cient to support and other. Th is provides infants, even newborns, the inference that the adult is seeing. Infants in with a feeling of kinship with fellow humans, and

25_Banaji_Ch3.3.indd 142 12/27/2012 9:42:29 PM OUP UNCORRECTED PROOF – FIRSTPROOFS, Sat Dec 22 2012, NEWGEN

Origins of Social Cognition 143

it supports bidirectional learning from and about I thank P. Kuhl, R. Williamson, R. Brooks, D. Liu, people. and P. Marshall for helpful comments on an ear- Going from the direction of inside out, infants’ lier draft , and K. Moore and A. Gopnik for useful own self-experiences change their understanding discussions. of others. I have discussed instances of this projec- tion from self to others. Infants who experience REFERENCES that a blindfold occludes their own vision make Brooks, R., & Meltzoff , A. N. (2002). eTh importance new attributions to others who wear it. Another of eyes: How infants interpret adult looking behav- example comes from infants’ heightened atten- ior . Developmental , 38 , 958–966 . tion and positive emotion at being imitated. Cvencek, D., Greenwald, A. G., & Meltzoff , A. Infants prefer those who act “like me” as manifest N. (2011). Measuring implicit attitudes of through prosocial acts such as smiling at the imi- 4-year-olds: Th e Preschool Implicit Association tating adult. Later in development, children and Test. Journal of Experimental Child Psychology , adults exhibit positive attitudes toward those who 109 , 187–200 . share features of the self that go beyond action. Decety, J., Chaminade, T., Grè zes, J., & Meltzoff , “Like-me” preferences (initially based on shared A. N. (2002). A PET exploration of the neural actions) are building blocks for the in-group mechanisms involved in reciprocal imitation. positivity that later develops based on gender and NeuroImage , 15 , 265–272 . other characteristics (e.g., Cvencek, Greenwald, & Dunham, Y., Baron, A. S., & Carey, S. (2011). Meltzoff , 2011; Dunham, Baron, & Carey, 2011). Consequences of ‘minimal ’ group affi liations in Adults prefer and feel for those who are children . Child Development , 82 , 793–811 . “like me” along abstract dimensions such as reli- Johnson, S. C., Slaughter, V., & Carey, S. (1998). gion, nationality, and social class, but a sense of Whose gaze will infants follow? Th e elicitation of “us” versus “them” has its fi rst instantiation in the gaze-following in 12-month-olds . Developmental felt kinship for those who act “like me.” Science , 1 , 233–238 . Infants also acquire new information about Kuhl, P. K., & Meltzoff , A. N. (1982). Th e bimodal themselves and the world simply by observing perception of speech in infancy . Science , 218 , others—demonstrating outside-in learning. I 1138–1141 . discussed that observing a structured pattern of Marshall, P. J., & Meltzoff , A. N. (2011). Neural mir- behavior supports infants’ inferences about the roring systems: Exploring the EEG mu rhythm psychology of others. At a more basic level, studies in human infancy . Developmental Cognitive of infant imitation show that they do not require Neuroscience , 1 , 110–123 . Meltzoff , A. N. (1990). Foundations for developing a extensive prior experience with an action to imi- of self: Th e role of imitation in relating tate it. Infants readily imitate novel acts (Meltzoff , self to other and the value of social mirroring, Kuhl, Movellan, & Sejnowski, 2009). Th ey also social modeling, and selfpractice in infancy. In D. learn novel causal relations based on observing Cicchetti & M. Beeghly (Eds.), Th e self in transi- the outcomes of others’ goal-directed interven- tion: Infancy to childhood (pp. 139–164 ). Chicago, tions (Meltzoff , Waismeyer, & Gopnik, 2012). The IL : University of Chicago Press . human capacity for the imitation of novel acts and Meltzoff , A. N., & Brooks, R. (2008). Self-experience observational causal learning are noteworthy— as a mechanism for learning about others: A they free us from learning about people and things training study in social cognition . Developmental only via self-experience. Psychology , 44 , 1257–1265 . Developing social cognition, even in infancy, Meltzoff , A. N., Brooks, R., Shon, A. P, & Rao, R. P. N. requires a bidirectional mapping between self (2010). “Social” robots are psychological agents for and other, and children with autism have spe- infants: A test of gaze following . Neural Networks , cifi c defi cits in this bidirectional system. An ini- 23 , 966–972 . tial “like-me” equivalence between self and other Meltzoff , A. N., Kuhl, P. K., Movellan, J., & Sejnowski, allows social cognition to get off the ground. Later T. J. (2009). Foundations for a new science of bidirectional learning—from self to other and learning . Science , 325 , 284–288 . other to self—allows it to soar. Meltzoff , A. N., & Moore, M. K. (1983). Newborn infants imitate adult facial gestures. Child ACKNOWLEDGMENTS Development , 54 , 702–709 . Supported by the National Science Foundation Meltzoff , A. N., & Moore, M. K. (1997). Explaining (SMA-0835854), the National Institutes of Health facial imitation: A theoretical model. Early (HD-22514), and the ONR (N00014-0910097). Development and Parenting , 6 , 179–192 .

25_Banaji_Ch3.3.indd 143 12/27/2012 9:42:29 PM OUP UNCORRECTED PROOF – FIRSTPROOFS, Sat Dec 22 2012, NEWGEN

144 Imitation, Modeling, and Learning From and About Others

Meltzoff , A. N., Waismeyer, A., & Gopnik, A. (2012). Spelke, E. S., Breinlinger, K., Macomber, J., & Jacobson, Learning about causes from people: Observational K. (1992). Origins of knowledge. Psychological causal learning in 24-month-old infants. Review , 99 , 605–632 . , 48, 1215–1228. van Baaren, R. B., Holland, R. W., Steenaert, B., & Moore, M. K., & Meltzoff , A. N. (2009). Numerical van Knippenberg, A. (2003). Mimicry for money: identity and the development of object perma- Behavioral consequences of imitation. Journal of nence. In S. Johnson (Ed.), Neoconstructivism: Th e Experimental , 39 , 393–398 . new science of cognitive development (pp. 61–83 ). Williamson, R. A., & Meltzoff , A. N. (2011). Own and New York : Oxford University Press . others’ prior experiences infl uence children’s imi- Ogden, T. H. (1982). Projective identifi cation and psy- tation of causal acts. Cognitive Development , 26 , chotherapeutic technique. New York: Aronson . 260–268 . Saby, J. N., Marshall, P. J., & Meltzoff , A. N. (2012). Xu, F., & Carey, S. (1996). Infants’ metaphysics: Th e Neural correlates of being imitated: An EEG study in case of numerical identity . , preverbal infants. Social Neuroscience, 7 , 650–661. 30 , 111–153 .

25_Banaji_Ch3.3.indd 144 12/27/2012 9:42:29 PM