Developmental Social Neuroscience

8/9/2019 Developmental Social Neuroscience

1/6

Introduction to a Special Section of Developmental Social

Cognitive Neuroscience

Rebecca R. Saxe

Massachusetts Institute of Technology

Kevin A. Pelphrey

Yale University

The articles in this special section offer a sample ofthe state of the art in an exploding new field.Noninvasive neuroimaging techniques, includingfunctional magnetic resonance imaging (fMRI),electroencephalograms (EEG), and functional near-infrared spectroscopy (fNIRS), are increasinglybeing used to study development in infants, chil-dren, adolescents, and adults. Applying these tools

to the study ofsocialcognitive development, in par-ticular, is a very new development. Consequently,in this commentary, we wish to step back and con-sider the new intersection of neuroscience andsocial cognitive development. What do these newfields have to offer each other, theoretically andpragmatically? What pitfalls can we expect toencounter along the way?

Even in adults, a neuroscience of social cognitionis a new invention. Traditional neuroscientific tech-niques were applied to the brains of nonhuman ani-mals, to study how brains see, hear, feel, move, and

remember. Uniquely human cognitive capacities,such as language and social cognition, could not bestudied in nonhuman animals. When neuroimagingtechniques were first applied to human adults, thekey test of these technologies was replication, inhumans, of functions known from nonhuman ani-mals. Thus, early neuroimaging focused on thehuman homologues of known regions from otherprimates: early visual cortex, the motion perceptionregion (MT), early sensory, and motor cortices.

Soon, though, human neuroimaging led to fasci-nating novel discoveries. The human brain contains

many cortical regions, previously unknown or littlestudied, which have apparently social functions.The fusiform face area (FFA) is involved in perceiv-ing human faces (Kanwisher, McDermott, & Chun,1997). The extrastriate body area is involved in per-ceiving human bodies (Downing, Jiang, Shuman, &Kanwisher, 2001). The right posterior superior tem-poral sulcus (pSTS) is involved in perceiving and

analyzing human actions (Pelphrey, Viola, &McCarthy, 2004). The right temporoparietal junc-tion (RTPJ) is involved in reasoning about peoplesthoughts (Saxe & Kanwisher, 2003). The medialprecuneus and posterior cingulate and medial pre-frontal cortices are involved in other aspects ofhigher level social cognition (Amodio & Frith,2006). With the exception of the pSTS (Perrett et al.,1985), these regions were unknown from the stud-ies of nonhuman animals.

The discovery of these regions provides an excit-ing challenge for neuroscientists, and a role for

developmental psychology. For neuroscientists, thekey questions are: How are high-level social func-tions implemented in neurons? How do brainregions with these functional roles arise, phyloge-netically and ontogenetically? Ideas, experimentalparadigms, and data from developmental psychol-ogy may help to pose and to address these ques-tions. For example, neuroimaging studies of actionrepresentation in the pSTS have been modeled onstudies of infants action perception (Vander Wyk,Hudac, Carter, Sobel, & Pelphrey, in press); andmost early studies of the RTPJs role in thinkingabout thoughts used versions of the false belief par-adigm originally developed for studying children(e.g., Fletcher et al., 1995; Gallagher et al., 2000). Inthe future, we expect that neuroscientists willcontinue to benefit greatly from the theoreticalconcepts and paradigms generated within develop-mental psychology.

We express our deepest gratitude to the outstanding reviewerswho generously gave of their time in order to help us select thearticles for this special section. Kevin Pelphrey and Rebecca Saxeare each supported by awards from the John Merck ScholarsFund and grants from the Simons Foundation. Additionally, aCareer Development Award from the National Institute of Men-tal Health supports Kevin Pelphrey.

Correspondence concerning this article should be addressed toRebecca Saxe, Department of Brain and Cognitive Science, MIT46-4019, 77 Massachusetts Ave., Cambridge, MA 02139 or toKevin Pelphrey, Child Study Center, Yale University, 230 SouthFrontage Rd., New Haven, CT 06520. Electronic mail may be sentto [email protected] or to [email protected].

Child Development, July/August 2009, Volume 80, Number 4, Pages 946951

2009, Copyright the Author(s)Journal Compilation 2009, Society forResearchin ChildDevelopment, Inc.

All rights reserved. 0009-3920/2009/8004-0002


2/6

In turn, as neuroscientists begin to link specificbrain regions, and their interactions, to particularsocial cognitive processes, this information mayenhance theory building within developmental psy-chology. In one example, developmental psycholo-gists have debated for decades whether the childsmind contains special-purpose mechanisms forlearning specific contents (domain specificity) orwhether the structure of the mind is the outcome ofdomain-general learning of regularities in experi-ence. The existence of neural regions with highlyspecific response profilesfor example, the FFAresponse to faces and the RTPJ response to thinkingabout thoughtsprovides strong prima facie evi-dence for the corresponding domain-specific cogni-tive mechanisms.

Neuroscientific results can also contribute todevelopmental theory by documenting the continu-

ity of common mechanisms across developmentalstages. For example, does the perception of goal-directed action in 6-month-old infants depend onthe same cognitive mechanism as the perception ofgoal-directed action in adults? Yoon and Johnson(p. 1069) provide evidence that infants perceive thestructure of actions presented as point-light walk-ers, and even follow the gaze of the point-lightperson.Lloyd-Fox and colleagues (p. 986) strengthenthe impact of this evidence for common mecha-nisms across development but provide initial evi-dence that perception of biological motion ininfants depends on the same brain region as inadultsthe pSTS.

A similar approach could make an enormoustheoretical impact on our understanding of theory-of-mind development. The paradigmatic test ofchildrens ability to think about others thoughts isthe false belief task: For example, children areasked to predict where a person will look for anobject that she left in one location but that has sincebeen transferred to a new location. On the tradi-tional version of this task, 3-year-olds predict thatthe character will look for the object in the new(true) location; by contrast, 5-year-old children (like

adults) predict that the character will look for theobject in the old location, where she falsely believes itto be (Sabbagh, Bowman, Evraire, and Ito, p. 1147;Wellman, Cross, & Watson, 2001). The 3-year-oldswho fail the traditional false belief task are notperforming at chance; they make systematicallybelow-chance predictions, with high confidence(Ruffman, Garnham, Import, & Connolly, 2001).

Nevertheless, a rapidly expanding literature hasrecently revealed that much younger childrenasyoung as 15 months oldcan pass the very same

false belief task if their understanding is measuredin terms of looking time (looking longer to un-expected actions) instead of action prediction (e.g.,Scott & Baillargeon, p. 1172). What accounts forthis long developmental lag? One account suggeststhat young infants have a mature understanding offalse beliefs but cannot express this knowledge inthe traditional action prediction format. An alterna-tive view is that infants and adults are using quali-tatively different mechanisms for understandingother minds; for example, the infants may haveonly an implicit understanding of false beliefs,whereas adults use an explicit, and therefore moreflexible and powerful, conception of beliefs. Impor-tantly, these accounts make clearly distinct predic-tions for the corresponding neural signatures. Ifinfants and adults are using the same cognitivemechanism to think about false beliefs, then the

same brain regions should be recruited in the infantand adult versions of the false belief task. Thisprediction has not yet been tested. We hope it willbe very soon.

Neuroscientific evidence does not always havesuch straightforward implications for develop-mental theories. In particular, whereas continuityacross development is fairly easy to interpret, devel-opmental change in neural response patterns isnot. There are a few basic options for patterns ofdevelopmental change in a cognitive neuroscienceexperiment. If the experiment uses a metabolicmeasure of brain function (e.g., fMRI or fNIRS), thenyounger participants could show less activation inthe same regions as older participants, or more acti-vation in the same regions, or less activation in thesame regions and more activation in distinctregions. If the experiment uses an electromagneticmeasure of brain function (EEG or magnetoencepha-logram [MEG]), then younger participants couldshow a slower profile of response to the samestimulus, or a faster response, or could be missing acomponent of the response, or show a differentcomponent not present in the older participants.In all of these cases, these differences could be pres-

ent along with behavioral differences in perfor-mance of the task, or in spite of matched taskperformance across groups.

What do these different patterns of developmen-tal change mean? Unfortunately, there is no simpleand universal rule. Instead, there are at least threeoptions in each case. First, it is possible that theobserved change reflects a purely cognitive differ-ence between groups. Younger and older partici-pants brains may be organized and structured inthe same waysthat is, the same neurons perform

Developmental Social Cognitive Neuroscience 947


3/6

the same computations in the same placesbut theyoung children perform the task using differentstrategies, or perceive different value in therewards. In this case, the neural evidence could beused to make a reverse inference about childrensstrategies or evaluations. (In a reverse inference,the engagement of a cognitive process is inferredfrom the activation of a particular brain region;DEsposito, Ballard, Aguirre, & Zarahn, 1998; Pold-rack, 2006.)

Second, it is possible that the observed changereflects a purely neural difference between groups.Younger and older participants may use exactlythe same strategies, but over time the brain mayreorganize, so that, for example, neurons involvedin a common function are closer together. Or theremay even be no changes in the electrophysiologicalchanges in neural activity, but there may be purely

physical changes in the structure, density, or vascu-larization of the brain, causing changes in theobservable neural signature (Colonnese, Phillips,Kaila, Constantine-Paton, & Jasanoff, 2008).

If so, these groups of neurons would becomemore visible as a focal activation in metabolic mea-sures, and a stronger component in electromagneticmeasures, even though the computations involvedremained exactly the same.

Finally, the observed change may reflect achange in how the childs mind and brain are rep-resenting or computing the task; that is, neuroscien-tific tools could provide a window on developingcognitive mechanisms and their neural mecha-nisms. In many cases, we suspect, developmentalsocial cognitive neuroscientists hope that they arestudying this latter kind of developmental change.Differentiating these three causal models of changethus poses a key challenge for many of the articlesin this special section.

Our own article in this issue illustrates thischallenge (Saxe, Whitfield-Gabrieli, Scholz, &Pelphrey, p. 1203). We report an increase in theselectivity of the response of the RTPJ to talk aboutthe contents of other peoples minds with age in

school-age children. However, without additionalinformation, it is impossible for us to specify theprecise mechanism underlying this finding. On theone hand, the children in our study might processthe stories using the same cognitive mechanisms,across ages. From this perspective, the differencesin selectivity would simply reflect maturational dif-ferences driving brain reorganization. Alternatively,the findings might reflect changes in the cognitivestrategies used by children at different ages. TheRTPJ, from this perspective, would remain orga-

nized in the same fashion across the observed agerange, but younger and older children would usedifferent cognitive strategies (e.g., spontaneouslyinventing mental states for the characters). Finally,our result could truly reflect changes across theschool-age years in how both mind and brain areperforming this theory-of-mind task. That is, ourneuroimaging finding could reflect development ofthe cognitive and neural mechanisms for theory ofmind.

A similar challenge faces the three articles in thisissue that investigate changes over adolescence, inthe medial frontal activation during self-reflection(Guyer, McClure-Tone, Shiffrin, Pine, & Nelson,p. 1000; Pfeifer et al., p. 1016; Ray et al., p. 1239).Are adolescents responding to the stimuli differ-ently, cognitively, or emotionally, using matureneural systems? Or are the neural systems for

self-reflection undergoing maturation? Convergingevidence will be necessary to disentangle these twopossibilities.

How might researchers work their way out ofthis kind of dilemma? We suggest that linkagesbetween carefully constructed behavioral tasks(inside and outside of the magnet) and multimodalimaging data (i.e., data concerning functional andanatomical changes) collected in the context oflongitudinal studies of young children will proveessential. The article bySabbagh and colleagues(p.1147) illustrates the power of combining neuroima-ging data and careful behavioral tasks. They foundthat individual differences in resting function in theRTPJ (among other regions) are associated with theperformance on false belief tasks outside the scan-ner. These data provide the best evidence to datethat changes in the structure of the brain are relatedto the development of social cognition.

Although such a cross-sectional study is anexcellent first step, there is a great need for longitu-dinal studies of social brain development in infants,children, and adolescents. Indeed, it is regrettablethat none of the studies in our special section arelongitudinal in design. For future research, there

are straightforward methodological reasons toadopt longitudinal designs. Functional neuroimag-ing data are inherently noisy (particularly in chil-dren) because individual brains are different fromone another. This is acknowledged, and it is partof the reason why almost all fMRI studies areconducted within subjects. A longitudinal design isthe only way to study developmental processes andhave the power of within-subject statistics, therebymaking it much more likely to visualize actualdevelopmental changes with reasonable sample

948 Saxe and Pelphrey


4/6

sizes. It also makes it more likely to detect relationsbetween different developmental changes, and thusprovides stronger inferential leverage upon whichto build lasting theoretical contributions.

In addition to the challenges for data interpreta-tion mentioned earlier, we think there are theoreti-cal barriers to clear the communication betweenneuroscientists and developmental psychologists(Sur, 2008). For example, developmental psycholo-gists are often interested in investigating whether aconcept (or group of concepts) is innate or learned.By contrast, neuroscientists generally assume thatall neural systems are shaped by both biologicalpreparation and the structure of experience. Thedetails of these interactions are often fascinatingbut do not translate easily into the theoreticallanguage of developmental psychology. Two ofthe articles presented in this special section (Corina

& Singleton, p. 952; Moulson, Westerlund, Fox,Zeanah, & Nelson, p. 1039) tackle the issue of howunusual environments and biological factors inter-act to generate different developmental pathways.We believe that these articles portray clear pathsforward and highlight the value of mechanisticdevelopmental analyses that consider equally bio-logical and environmental influences.

To further illustrate this point, consider thisexample from the field of neuroscience. Ferretbrains contain separate channels by which auditoryinformation from the ears and visual informationfrom the eyes are normally sent to the brain. By avery precise surgery in a developing ferret, thenerves from the eye can be induced to replace thenormal auditory channel, so that primary auditorycortexthe region that normally receives auditorysignalsreceives purely visual informationthroughout development. When they are mature,these ferrets are taught to look for food near a redlight but not a green light. Later if early visual cor-tex is lesioned, in a way that would normally causetotal blindness, these ferrets continue to performnormally on this visual task, relying exclusively ontheir rewired auditory region. Only if the audi-

tory cortex is also lesioned do the ferrets becomefunctionally blind.

Critically, it is possible to then study, in micro-scopic detail, what happened to the auditory cortexthat was rewired during development. In a normalferret or human brain, visual cortex is organized inan intricate grid depending on the location, orienta-tion, contrast, and color of the visual stimulus; audi-tory cortex is laid out in elongated stripes orderedby pitch. So what would auditory cortex look like ifit received only visual inputs? Like auditory cortex

or like visual cortex? The answer is a mixture of thetwo. Like visual cortex, rewired auditory cortex con-tains patches of neurons with preferences for hori-zontal versus vertical lines, and objects at the centerof the visual field are represented at one end of theregion whereas objects at the periphery are at theother end. On the other hand, rewired auditory cor-tex also retains elements of auditory structure, likelong parallel lines of neurons with similar proper-ties, and the grid of visual properties is more looselypacked than in real visual cortex (Von Melchner,Pallas, & Sur, 2000).

The details of this example defy easy character-ization in the traditional theoretical vocabulary ofdevelopmental psychology. The division of sensoryinput into an auditory and visual stream is clearlybiologically driven. However, the organization ofearly auditory cortex is partly experience driven,

partly experience expectant, and partly resistant tochanges in experience.

Similar, and equally fascinating, results are cur-rently emerging from investigations of the neuralbasis of language. In human adults, linguistic func-tions typically depend dominantly on left hemi-sphere temporal and frontal regions. Damage to theleft hemisphere in adulthood often causes deep andpersistent deficits in language use and comprehen-sion, whereas after damage to the same regions inthe right hemisphere, language is often unaffected(see Goulven & Tzourio-Mazoyer, 2003, for areview). The first few studies to use functionalneuroimaging in human infants suggest that thishemispheric asymmetry emerges early in braindevelopment and is present even in very younginfants (e.g., Dehaene-Lambertz, Dehaene, & Hertz-Pannier, 2002; see also Witelson & Pallie, 1973).These data suggest that the left hemisphere is struc-turally better prepared to process linguistic infor-mation. Nevertheless, children who suffer veryearly damage to their left hemisphere appear toacquire language almost normally (Bates, 1999).

Taken together, these examples suggest thatneuroscientific data may not fit neatly into the pre-

existing theoretical categories of developmentalpsychology. Instead, we think both fields willchange and adapt in response to their marriage,and new theoretical questions will arise that are ofinterest to both groups of scientists. We are veryexcited to see where these questions will lead, inour specific field of developmental social cognitiveneuroscience, and more generally.

Of course, neuroscientific measures have prag-matic as well as theoretical benefits. Direct accessto brain development could be used to provide



5/6

critical diagnoses andor earlier predictors oflong-term outcomes, including for neurodevelop-mental disorders such as autism and schizo-phrenia. The inclusion of three articles examiningvarious aspects of autism in this special section(Akechi et al., p. 1134; Van Hecke et al., p. 1118;White, Hill, Happe, & Frith, p. 1097) is a reflec-tion of the intense interest in this neurodevelop-mental disorder within the field of developmentalsocial neuroscience. Properly acquired neuroimag-ing data are very reliable, inherently quantitative,and offer unparallel density and measurementprecision. Neural measures are presumably moreproximal to the genetic and neural causes ofautism, and therefore could provide earlier andmore sensitive markers of disease onset. Also, thedensity of neural data may allow for betterdifferentiation of subgroups within heterogeneous

populations that are lumped together by standardbehavioral diagnostic criteria.

For example, one could adopt the fNIRSapproach taken by Lloyd-Fox and her colleagues(p. 986) to study the response to biological versusnonbiological motion in very young children withand without autism, or even infants at increasedrisk for developing autism. A neurobiological mar-ker for individual differences in social cognitiveabilities would be important for improving earlyidentification. It could also offer advantages interms of guiding interventions. Targeted treatmentscould be designed to target the earliest markers,which might then guarantee the most effectivecourse of intervention possible. Neuroimaging datacould be used both to identify treatment targetsand to evaluate treatment outcomes. In particular,neuroimaging data could help to distinguishdifferent mechanisms of treatment effectiveness,including whether posttreatment improvementsin social cognition correspond to rehabilitatingnormal neural mechanisms for social cognition,or to the construction of novel compensatorymechanisms.

We are also optimistic about a new trend in

developmental social cognitive neuroscience: com-bining neural data with genetic markers (Hariri &Weinberger, 2003). These studies would provide forthe empirical realization, in humans, of a transac-tional approach to the study of mechanisms fordevelopmental changes in social cognition (Gott-lieb, 1997; Sameroff & Chandler, 1975). Such a per-spective emphasizes the necessity to characterizethe development of social cognition as an emergentproperty reflecting transactions occurring acrosslevels of an active, developing organism in its envi-

ronment. The goal of theory building is then tospecify mechanisms through the identification ofcritical transactions between two, three, or morelevels of analysis (e.g., geneM brain M behaviortransactions over a specific developmental period).We embrace this as a way to generate exciting anddeeply influential theoretical advances in our field.

In sum, we stand at the beginning of a goldenera for work at the interface of developmentalpsychology and social cognitive neuroscience. Thisexistence of elegant behavioral techniques forstudying change at the cognitive level of analysisand the explosion of neuroimaging techniques forstudying brain development in even the very youn-gest children have provided the opportunity forstudies of social cognitive development, and partic-ularly individual differences in developmentalpathways, which span multiple levels of analysis

from molecules to mind. We hope you enjoy thearticles in this special section as much as we haveenjoyed working with the individual authors,the Editor-in-Chief of Child Development, Dr. JeffreyJ. Lockman, and the Managing Editor, Ms. DetraDavis, to make them available to you.

References

Amodio, D. M., & Frith, C. D. (2006). Meeting of minds:The medial frontal cortex and social cognition. NatureReviews Neuroscience,7, 268277.

Bates, E. (1999). Language and the infant brain. Journal ofCommunication Disorders,32, 4.

Colonnese, M. T., Phillips, M. A., Kaila, K., Constantine-Paton, M., & Jasanoff, A. (2008). Development of hemo-dynamic responses and functional connectivity in ratsomatosensory cortex.Nature Neuroscience,11, 7279.

Dehaene-Lambertz, G., Dehaene, S., & Hertz-Pannier, L.(2002). Functional neuroimaging of speech perceptionin infants.Science,298, 20132015.

DEsposito, M., Ballard, D., Aguirre, G. K., & Zarahn, E.(1998). Human prefrontal cortex is not specific forworking memory: A functional MRI study. NeuroImage,8, 274282.

Downing, P., Jiang, Y., Shuman, M., & Kanwisher, N.(2001). A cortical area selective for visual processing ofthe human body. Science,293, 24702473.

Fletcher, P. C., Happe, F., Frith, U., Baker, S. C., Dolan, R.J., Frackowiak, R. S., et al. (1995). Other minds in thebrain: A functional imaging study of theory of mindin story comprehension. Cognition,57, 109128.

Gallagher, H. L., Happe, F., Brunswick, N., Fletcher, P.C., Frith, U., & Frith, C. D. (2000). Reading the mind incartoons and stories: An fMRI study of theory ofmind in verbal and nonverbal tasks. Neuropsychologia,38, 1121.

950 Saxe and Pelphrey


6/6

Gottlieb, G. (1997). Synthesizing nature-nurture: Prenatalroots of instinctive behavior. Mahwah, NJ: Erlbaum.

Goulven, J., & Tzourio-Mazoyer, N. (2003). Review:Hemispheric specialization for language. Brain ResearchReviews,44, 112.

Hariri, A. R., & Weinberger, D. R. (2003). Imaging ge-

nomics.British Medical Bulletin,65, 259270.Kanwisher, N., McDermott, J., & Chun, M. (1997). Thefusiform face area: A module in human extrastriate cor-tex specialized for the perception of faces. Journal ofNeuroscience,17, 43024311.

Pelphrey, K. A., Viola, R. J., & McCarthy, G. (2004). Whenstrangers pass: Processing of mutual and averted gazein the superior temporal sulcus. Psychological Science,15, 598603.

Perrett, D. I., Smith, P. A. J., Potter, D. D., Mistlin, A. J.,Head, A. S., Milner, A. D., et al. (1985). Visual cells inthe temporal cortex sensitive to face view and gazedirection. Proceedings of the Royal Society of London: B.Biological Sciences,223, 293317.

Poldrack, R. A. (2006). Can cognitive processes beinferred from neuroimaging data? Trends in CognitiveSciences,10, 5963.

Ruffman, T., Garnham, W., Import, A., & Connolly, D.(2001). Does eye gaze indicate knowledge of false

belief: Charting transitions in knowledge? Journal ofExperimental Child Psychology,80, 201224.

Sameroff, A. J., & Chandler, M. J. (1975). Reproductiverisk and the continuum of caretaker casualty. In F. D.Horowitz (Ed.), Review of child development research(Vol. 4, pp. 187245). Chicago: University of Chicago

Press.Saxe, R., & Kanwisher, N. (2003). People thinkingabout

thinking people. The role of the temporo-parietal junc-tion in theory of mind.NeuroImage,19, 18351842.

Sur, M. (2008). The emerging nature of nurture. Science,322, 1636.

Vander Wyk, B. C., Hudac, C. M., Carter, E. J., Sobel, D. M.,& Pelphrey, K. A. (in press). Action understandingin the superior temporal sulcus region. PsychologicalScience.

Von Melchner, L., Pallas, S. L., & Sur, M. (2000). Visualbehaviour mediated by retinal projections directed tothe auditory pathway.Nature,404, 871876.

Wellman, H. M., Cross, D., & Watson, J. (2001). Meta-

analysis of theory-of-mind development: The truthabout false belief.Child Development,72, 702707.

Witelson, S. F., & Pallie, W. (1973). Left hemisphere spe-cialization for language in the newborn: Neuroanatomi-cal evidence of asymmetry. Brain,96, 641.


Developmental Social Neuroscience

Documents

Transcript of Developmental Social Neuroscience