CURRENT DIRECTIONS IN PSYCHOLOGICAL S CIENCE Commentary Neuroimaging as a New Tool in the Toolbox of Psychological Science John T. Cacioppo,1 Gary G. Berntson,2 and Howard C. Nusbaum1 1 The University of Chicago and 2Ohio State University ABSTRACT—During the past quarter century, advances in imaging technology have helped transform scientific fields. As important as the data made available by these new technologies have been, equally important have been the guides provided by existing theories and the converging evidence provided by other methodologies. The field of psychological science is no exception. Neuroimaging is an important new tool in the toolbox of psychological science, but it is most productive when its use is guided by psychological theories and complemented by converging methodologies including (but not limited to) lesion, electrophysiological, computational, and behavioral studies. Based on this approach, the articles in this special issue specify neural mechanisms involved in perception, attention, categorization, memory, recognition, attitudes, social cognition, language, motor coordination, emotional regulation, executive function, decision making, and depression. Understanding the contributions of individual and functionally connected brain regions to these processes benefits psychological theory by suggesting functional representations and processes, constraining these processes, producing means of falsifying hypotheses, and generating new hypotheses. From this work, a view is emerging in which psychological processes represent emergent properties of a widely distributed set of component processes. KEYWORDS—functional magnetic resonance imaging; cognitive processes; social processes; clinical processes; developmental processes Address correspondence to John T. Cacioppo, Center for Cognitive and Social Neuroscience, The University of Chicago, 5848 S. University Avenue, Chicago, Illinois 60637; e-mail: cacioppo@uchicago. edu. 62 New imaging technologies are having a demonstrable impact on the landscape of scientific research. The most expensive imaging instrument, and the most vivid example, is the Hubble Space Telescope. The Hubble telescope was deployed in April 1990 and has undergone three major repairs and upgrades since that time. It has also provided data and images at a resolution Galileo Galilei could not have imagined when, early in the 17th century, he discovered the craters on the moon, sunspots, the rings of Saturn, and the moons of Jupiter by gazing through his first crude telescope. The discoveries made possible by the high-resolution data and images from the Hubble nearly four centuries later include massive black holes at the center of galaxies, the existence of precursors to planetary systems like our own, and a greater quantity and distribution of dark matter than expected. As important as were the data provided by the Hubble Telescope, however, these discoveries were dependent on extant theories and methodologies. The discovery of stellar black holes at the centers of galaxies, for instance, was guided by general relativity theory and supported by research using several converging methodologies (e.g., Dolan, 2001). Developments in neuroimaging during the past quarter century have increasingly made it possible to investigate the differential involvement of particular brain regions in normal and disordered thought in humans. Previously, studies of the neurophysiological structures and functions associated with psychological states and processes were limited primarily to animal models, postmortem examinations, electrophysiological measures, and observations of the occasional unfortunate individual who suffered trauma to or disorders of the brain (e.g., Raichle, 2003). The detailed three-dimensional color images provided by neuroimaging, modeling statistical properties of the working brain, have captured the imagination of the public and the scientific community, shaped funding priorities at federal funding agencies and foundations, and produced a dramatic growth in scientific papers and journals in the area (Cacioppo et al., 2007). Copyright r 2008 Association for Psychological Science Volume 17—Number 2 John T. Cacioppo, Gary G. Berntson, and Howard C. Nusbaum This special issue of Current Directions in Psychological Science summarizes recent theoretical advances in various fields of psychological science that are attributable in part to the use of neuromaging technology—most prominently, functional magnetic resonance imaging (fMRI). Although reading these reviews leaves the impression that neuroimaging is an important new tool in the toolbox of psychological science, one cannot help but also be impressed that neuroimaging—like the Hubble Space Telescope—is most productive scientifically when its use is guided by extant theories and complemented by converging methodologies. In the typical neuroimaging study, psychological states and processes are manipulated and activation in different brain regions is measured. The logic of this design is best suited for drawing inferences about the differential involvement of particular brain regions in specific psychological operations. For instance, when Poeppel and Monahan (2008, this issue) ask how speech signals are represented and processed in the brain, they are using neuroimaging along with converging methods from psychological science, and guided by the blueprint of competing theoretical accounts for speech perception, to investigate the differential involvement of particular brain regions in psychological states and processes. It is also possible under certain conditions to draw reasonable inferences about psychological operations based on regions of brain activation (Cacioppo & Berntson, in press; Cacioppo & Tassinary, 1990; Henson, 2006; Poldrack, 2006; Sarter, Berntson, & Cacioppo, 1996). Across the articles in this special issue, evidence from lesion studies, animal studies, neuroimaging, single-cell recording, event-related brain potentials, transcranial magnetic stimulation, computational modeling, and behavior is reviewed to investigate the brain regions involved in perception, attention, categorization, memory, recognition, attitudes, social cognition, language, motor coordination, emotional regulation, executive function, decision making, and depression. Together, the evidence converges on the view that psychological states and processes are mediated by a network of distributed, often recursively connected, interacting brain regions, with the different areas making specific, often taskmodulated contributions (see Poeppel & Monahan, 2008). DISTRIBUTED NETWORKS INVOLVED IN COGNITIVE REPRESENTATIONS AND FUNCTION Humans are visual creatures. The visual properties of scenes drive neurons in the lateral geniculate nucleus of the thalamus, and visual perception has been found to involve a dorsal stream, or ‘‘where pathway,’’ and a ventral stream, or ‘‘what pathway.’’ The dorsal (where) stream includes the areas designated V1, V2, V5/MT, and the inferior parietal lobule and is associated with motion, representation of object locations, and control of the eyes and arms when visual information is used to guide saccades or reaching. The ventral (what) stream includes the areas V1, V2, V4, and the inferior temporal lobe (areas that include the lateral occipital complex and the fusiform gyrus) and is Volume 17—Number 2 associated with form recognition, object representation, face recognition, and long-term memory (Engel, 2008, this issue). Grill-Spector and Sayres (2008, this issue) provide evidence that changes in the size, position, orientation, and other aspects of physical appearance of faces activate the lateral occipital complex; differences in the identity of individuals are related to adaptation responses in the fusiform gyrus; and changes in facial expression and gaze direction involve the superior temporal sulcus. Different theoretical representations and decompositions of speech perception into processing components are described, and the neural outcomes associated with each of these theoretical representations are reviewed, by Poeppel and Monahan (2008). Here, too, the evidence suggests the involvement of a specialized, interconnected set of neural regions that are widely distributed across the temporal, parietal, and frontal lobes. Specifically, the early spectrotemporal analyses involve bilateral auditory cortices and the superior temporal cortex, and phonological analyses involve the middle and posterior portions of the superior temporal sulcus. The processing streams then appear to divide into a ventral stream—which maps auditory and phonological representations onto lexical conceptual representations and involves the middle temporal gyrus and inferior temporal sulcus—and a dorsal stream—which maps auditory and phonological representations onto articulatory and motor representations and involves the Sylvian parietotemporal area, posterior frontal gyrus, premotor cortex, and anterior insula (Poeppel & Monahan, 2008). The attentional modulation of perceptual processes is influenced by motivational states and goals as well as by stimulus properties. Visual attentional control can modulate neural activity in the lateral geniculate nucleus and superior colliculus, as well as in the posterior parietal cortex (specifically, the regions of the superior parietal lobule and the lateral intraparietal area within the intraparietal sulcus) and the frontal eye field and supplementary eye field within the prefrontal cortex (Yantis, 2008, this issue). These perceptual and attentional processes contribute to the acquisition of knowledge about the world that is organized categorically. Barsalou (2008, this issue) notes that the dominant theory in cognitive science posits that this knowledge is represented in an abstract, amodal fashion and constitutes semantic memory. He shows in his review, however, that categorical knowledge includes modal representations using the same neural mechanisms involved in perception, affect, and action. In Barsalou’s view, the representation of a category involves a neural circuit distributed across the relevant modalities, all of which can become activated during conceptual processing. Thus, conceptual processing can be viewed as an embodied rather than purely abstract process. How does categorical knowledge being represented in this distributed, modal fashion square with the neuroscientific evidence for differences in the localization of short- and longterm memory processes? Nee, Berman, Moore, and Jonides 63 Neuroimaging and Psychological Science (2008, this issue) suggest that the evidence supporting the qualitative distinction between short- and long-term storage processes has been misinterpreted, and they suggest that the data instead support a unitary model of memory in which the same regions of the brain that represent perception, action, and affect are involved in both short- and long-term storage processes. That is, short- and long-term memories do not differ in representation but in the activation by attention, which in turn involves a frontal biasing (i.e., maintenance) of representational cortices (e.g., frontal eye fields and intraparietal sulcus for spatial representations; superior temporal sulcus and Sylvian parietotemporal area for phonological and articulatory representations). For instance, damage in the presylvian region produces deficits in short- and long-term memory that depends on phonological material, and the greater prevalence of such material in studies of short- than of long-term storage processes may inadvertently have led to evidence that this region was involved uniquely in short-term memory (Nee et al., 2008). Short- and long-term memory retrieval also activates overlapping regions of the left lateral frontal cortex, whereas the monitoring of retrieved information, whether from short- or long-term memory, is associated with the anterior prefrontal activation (cf. Cabeza, Dulcos, Graham, & Nyberg, 2002). Neuropsychological research dating back several decades suggested structures in the medial temporal lobe (e.g., the hippocampus) were involved in declarative rather than nondeclarative learning. Unlike the distinction between short- and long-term memory, the distinction between declarative and nondeclarative learning is supported by neuroimaging research. For instance, research by Knowlton and Foerde (2008, this issue) has shown that when performance on a probabilistic classification task is based on declarative memory performance the medial temporal lobe is activated, whereas when performance on the task is based on nondeclarative memory performance the striatum is activated. Knowlton and Foerde (2008) also review evidence showing that nondeclarative skill learning, at least for simple tasks, is associated with repetition suppression—reductions in the regions of neural activation associated with the initial performances of a task (e.g., the premotor region)—a finding that has been interpreted as indicating a greater efficiency of processing in the neural structures involved in novice performance. Priming-related reductions, on the other hand, are found in perceptual and prefrontal regions, with only the latter associated with behavioral facilitation. Knowlton and Foerde (2008) duly note, however, that activation in the perceptual cortices may appear to be less important in the extant literature in part because of the type of priming paradigms that have been used in fMRI research. The complexities of social living, such as recognizing individuals and groups, negotiating nontransitive social hierarchies and shifting alliances, using language to communicate and manipulate, and engaging in social exchanges over extended periods and locales, place special demands on the capacities of the human brain. Mitchell (2008, this issue) reviews evidence 64 that thinking about thinking people (e.g., impression formation, social causality)—in contrast, for instance, to thinking about physical causality—is associated with activation of the medial prefrontal cortex, the right temporo-parietal junction, and the medial parietal region (e.g., the precuneus/posterior cingulate cortex). Mitchell (2008) suggests that the activation of the medial prefrontal cortex appears to be involved whenever people are obliged to consider the psychological characteristics of another person, whereas the temporo-parietal junction appears to be activated when the attentional and perceptual requirements of taking the perspective of another are invoked. The medial parietal region, on the other hand, is activated during the retrieval of episodic memories and self-knowledge, as well as during the viewing of two or more interacting people (e.g., Iacoboni et al., 2004). The brain has evolved to guide behavior in contextually flexible, coordinated, and adaptive ways, and, as with attention, there are top-down as well as bottom-up influences on the orchestration of motor processes. Oliveira and Ivry (2008, this issue) focus on the top-down influences in their discussion of goals as higher-level action representations that connect sensory and motor processes to guide response selection and motor coordination. They review fMRI studies showing that motor planning and externally guided movements are associated with activity in the posterior superior parietal region and, at least for externally guided movements, in premotor regions; internally generated movements are associated with activity in the basal ganglia, the anterior cingulate cortex, and inferior frontal and parietal cortices; and conflicting action goals and effort are associated with activity in medial frontal areas, including the anterior cingulate cortex and presupplementary motor areas. One suggestion that has emerged from this area of research is that goal representation and action planning are not implemented simply as an abstract code but rather involve embodied processes. Not unlike how Barsalou (2008) invokes modal mechanisms, Rizzolati and Arbib (1998) and Skipper, Nusbaum, and Small (2006) review evidence for the role of embodied representations in categorization and language. The fundamental idea that the motor system is important for cognition and perception, through prior experience and mirror neurons, has become an important contribution of neuroscience to bolstering theoretical constructs in the psychology of embodied understanding. However, much of the work on the mirror system in cognition and understanding has been carried out with trained nonhuman primates or with adult humans. For any theory of adult function to be viable, it is critical to understand the development of these mechanisms. Diamond and Amso (2008, this issue) review work on the neural substrates underlying cognitive development, including the mirror-neuron system and neonatal imitative behaviors and maternal touch and gene expression. As the authors note, a major contribution of neuroscience to theories of cognitive development is ‘‘demonstrating the remarkable role of experience in shaping the mind, Volume 17—Number 2 John T. Cacioppo, Gary G. Berntson, and Howard C. Nusbaum brain, and body’’ (p. 136). Such cross-cutting work is necessary to begin to link biological development with learning and experience. Moreover, as cognition can no longer be studied in isolation from the social context of its use, this work suggests the importance of understanding development within its social context of parental interaction. Given the importance of social context, then, it is important to go beyond the treatment of specific processes to understand how such processes depend on the goals they are directed at achieving. To achieve one’s goals, one has to be able to represent the likely rewards (and punishments) associated with different decisions, encode the risk or certitude that the reward will be obtained, update these representations, and act on the basis of these representations. O’Doherty and Bossaerts (2008, this issue) review evidence regarding the brain regions associated with each of these components of decision making. Specifically, they report that the encoding of reward expectation is associated with activation of the orbitofrontal cortex, medial prefrontal cortex, amygdala, and ventral striatum; recognizing greater risk or uncertainty associated with obtaining a reward correlates with increased activity in the anterior insula and lateral orbitofrontal regions; updating of reward expectancies is associated with the ventral striatum and orbitofrontal cortex; and selecting one of several responses to obtain the greatest reward involves the striatum, with the ventral striatum more involved in the prediction of reward across the various options and the dorsal striatum more involved in the selection among the alternatives (O’Doherty & Bossaerts, 2008). The frontal regions have long been thought to be involved in executive functions such as formulating goals and plans; selecting among options to achieve these goals; monitoring the consequences of actions in light of one’s goals; and inhibiting, switching and regulating one’s behaviors accordingly. Aron (2008, this issue) reviews evidence that the initiation of a motor response proceeds from the planning areas of the frontal cortex to the putamen, globus pallidus, thalamus, primary motor cortex, motor nucleus in the spinal cord, and finally to the muscles. Being able to inhibit a motor response once it has been initiated has obvious adaptive value, and Aron (2008) shows that this inhibition involves the right inferior frontal cortex, which projects to the subthalamic nucleus (a region of the basal ganglia that may act on the globus pallidus to block the motor response). Monitoring for response conflicts, in turn, appears to involve the dorsal anterior cingulate and the adjacent presupplementary motor area, which, in turn, is connected to the right inferior frontal cortex and subthalamic nucleus. Switching also involves the presupplementary motor area and the right inferior frontal cortex (Aron, 2008). This work has led to a model in which ‘‘the [presupplementary motor area] may monitor for conflict between an intended response and a countervailing signal . . . Then, when such conflict is detected, the ‘brakes’ could be put on via the connection between the right [inferior frontal cortex] and the [subthalamic nucleus] region’’ (Aron, 2008, p. 127). Volume 17—Number 2 Emotional regulation is another form of executive function in which activity of the amygdala and insula cortex, which are involved in emotional responding, is modulated by activity in the prefrontal cortex (e.g., BA10, ventromedial prefrontal cortex, dorsolateral prefrontal cortex) and anterior cingulate. Ochsner and Gross (2008, this issue) review evidence that different components of reappraisal processing correspond to different areas of prefrontal activation: Selective attention and working memory components are related to dorsal portions of the prefrontal cortex, language or response inhibition are related to ventral portions of the prefrontal cortex, monitoring or control processes are related to the dorsal anterior cingulate cortex, and reflections on one’s emotional state are related to dorsal portions of the medial prefrontal cortex. Although the correspondences proposed by Ochsner and Gross (2008) do not match perfectly those articulated by Aron (2008), the overlapping role for the anterior cingulate is noteworthy in light of the notion that the presupplementary motor area may be especially involved in the monitoring and control of motor conflicts. The complexities of daily living are simplified in part by the formation of preferences and attitudes, which can serve as behavioral guides and simplify decision making. These attitudes can be explicit or implicit. Stanley, Phelps, and Banaji (2008, this issue) review evidence suggesting that the activation of implicit attitudes toward social groups (e.g., minorities) is associated with increased activity in the amygdala, dorsolateral prefrontal cortex, and anterior cingulate cortex. The cumulative evidence to date suggests that the automatic evaluation of a stimulus (e.g., social category) is associated with amygdala activation, the monitoring for response conflicts (e.g., the extent to which the stimulus elicits competing impulses) is associated with anterior cingulate activation, and the regulation of those impulses is associated with dorsolateral prefrontal activation. Failures of effective emotional regulation can become costly in personal, social, and economic terms when these failures become systemic. Depression, for instance, has been estimated to cost more than $43 billion per year in the United States (Greenberg, Stiglin, Finkelstein, & Berndt, 1993). Understanding the variation in biological systems that leads to individual differences in neural mechanisms of emotional regulation is critical to understanding how some systemic failures become chronic and debilitating. Gotlib and Hamilton (2008, this issue) review evidence that depressed individuals show less activity in the dorsolateral prefrontal cortex and greater activation of the amygdala and subgenual anterior cingulate cortex to emotional stimuli than do healthy controls. Parallel findings for basal activity levels in these brain regions are also noted. These findings are consistent with Gotlib and Hamilton’s notion that depression is in large part a disorder of emotion regulation in which the normal inhibitory influence of limbic structures by the anterior cingulate and dorsolateral prefrontal cortex is disrupted, although the subgenual anterior cingulate cortex may play an especially critical role in this dysregulation (Gotlib & Hamilton, 2008). 65 Neuroimaging and Psychological Science Given the importance of the anterior cingulate and dorsolateral prefrontal cortex in motor control, attention, and emotion, the individual variation in function in these areas that can lead to depression may also explain that disorder’s other associated cognitive symptoms. Indeed, understanding the relationship between biological variation in neural mechanisms and psychological processes is important beyond clinical problems. Kosslyn et al. (2002) and Vogel and Awh (2008, this issue) have argued that, to bridge the gap between psychological phenomena and their underlying biological substrata, such variation should be regarded as important data in its own right. Kosslyn et al. (2002) describe how an idiographic approach can be used to address three types of issues: the nature of the mechanisms that give rise to a specific ability, the role of psychological or biological mediators of environmental challenges, and the existence of variables that have nonadditive effects with other variables. Vogel and Awh (2008) extend this argument in their discussion of three additional ways in which an idiographic approach can contribute to psychological theory: validating neurophysiological measures, demonstrating associations among constructs, and demonstrating dissociations among similar constructs. Thus, an idiographic approach, which complements the more typical nomethetic approach, can be applied in any domain to help elucidate psychological theory. Together, the theory and data summarized in this special issue of Current Directions in Psychological Science highlight the notion that encephalization and the remarkable connectivity in the human brain provide the substrate for the integration of inputs from widely distributed neural regions (only some of which are amenable to current brain-imaging technology) whose activation and organization can be contextually determined. The distributed nature of and substantial overlap among the extant networks calls for a revision in our thinking about basic psychological constructs. The early reliance on introspection as a method of identifying elemental psychological processes led to a recognition of the category error—the intuitively appealing but often erroneous notion that the organization of psychological phenomena maps in a one-to-one fashion onto the organization of underlying neural substrates. Perception, memories, emotions, and beliefs were each once thought to be localized in distinct sites in the brain. The contributions to this special issue clearly indicate that psychological and behavioral concepts do not each map onto clear and identifiable ‘‘centers,’’ but rather that each concept is associated with a distributed, interconnected set of neural regions. What appears at one point in time to be a singular theoretical construct (e.g., memory), when examined in conjunction with evidence from the brain (e.g., lesions, neuroimaging), may reveal a more complex and interesting organization at both levels (e.g., declarative vs. procedural memory processes). Conversely, what appeared to be distinct constructs (e.g., short- vs. long-term memory) may need to be reconsidered in light of new neuroscientific evidence. We suspect we are far from seeing the last of such revisions to psychological theories. It is only 66 through these revisions, and corresponding refinements in our understanding and conceptions of the underlying neural functions, that we can reduce the category error and move toward an isomorphism between the psychological and biological domains. Neuroimaging and work in neuroscience more generally are reshaping the constructs that are being used to build psychological theories. Psychological research during the 20th century resulted in many of the basic psychological elements derived from introspection to be recast as the product of multiple, more specific component processes. As illustrated by the articles in this special issue, many of these component processes involve a network of distributed, often recursively connected, interacting brain regions, with the different areas making specific, often task-modulated contributions. Moreover, a single neural region can often be involved in what have been treated as very different psychological processes. One implication is that what have been considered basic psychological or behavioral processes are being conceptualized as manifestations of computations performed by networks of widely distributed sets of neural regions. How might these neural components be combined to produce distinct psychological processes? One metaphor is the Lego set, in which the computations performed in localized neural regions are fixed (like distinct Lego pieces), but different pieces and configurations of these building blocks produce different psychological processes. An alternative metaphor is the periodic table in chemistry, in which different neural component processes may have properties and affinities whose function (computation) depends on the network of areas with which they are combined. There is no evidence at present to favor either perspective, but the important point here is that they suggest very different ways of thinking about neural activity and psychological function. In sum, neuroimaging work is leading to a rethinking of how psychological and neural functions are parcelled. For instance, the close proximity of motor control, emotional appraisal, attention, working memory, and behavioral regulation suggests that these functions may not be as separable as they are currently treated and studied. We may well need a new lexicon of constructs that are neither simply anatomical (e.g., Brodmann area 6 vs. Brodmann area 44) nor psychological (e.g., attention, memory), as we usher in a new era of psychological theory in which what constitutes elemental component processes (functional elements) are tied to specific neural mechanisms (structural elements) and in which the properties of interrelated networks of areas may indeed be more than the sum of the parts. CONCLUSION Critics who say neuroimaging is costly and has contributed little if anything to psychological theory sometime appear to expect the images of the working brain to come with labels regarding their cognitive functions. Although an adequate specification of Volume 17—Number 2 John T. Cacioppo, Gary G. Berntson, and Howard C. Nusbaum neurobiology should contribute to our understanding of cognitive architecture and function, our understanding of the relevant neurobiology is influenced strongly by our extant theoretical models regarding cognitive architecture and function (see Hagoort, 2008, this issue). The contributions to this special issue demonstrate that neuroimaging is an important new tool in the toolbox of psychological science, but one that is most productive scientifically when its use is guided by psychological theories and complemented by converging methodologies. This approach, in which theory and converging methods are used hand in hand to expand our understanding of the neural mechanisms involved in cognition and the contributions of individual and functionally connected brain regions to these processes, promises to advance psychological theory by suggesting functional representations and processes, by imposing significant constraints on these processes, and by producing not only new behavioral hypotheses but also new means of falsifying theoretical hypotheses. Acknowledgments—Preparation of this paper was supported by grants from the National Institute of Mental Health (Grant No. P50 MH72850) and the John Templeton Foundation. REFERENCES Aron, A.R. (2008). Progress in executive-function research: From tasks to functions to regions to networks. Current Directions in Psychological Science, 17, 124–129. Barsalou, L.W. (2008). Cognitive and neural contributions to understanding the conceptual system. Current Directions in Psychological Science, 17, 91–95. Cabeza, R., Dulcos, F., Graham, R., & Nyberg, L. (2002). Similarities and differences in the neural correlates of episodic memory retrieval and working memory. Neuroimage, 16, 317–330. Cacioppo, J.T., Amaral, D.G., Blanchard, J.J., Cameron, J.L., Carter, C.S., Crews, D., et al. (2007). Social neuroscience: Progress and implications for mental health. Perspectives on Psychological Science, 2, 99–123. Cacioppo, J.T., & Berntson, G.G. (in press). Integrative neuroscience for the behavioral sciences: Implications for inductive inference. In G.G. Berntson & J.T. Cacioppo (Eds.), Handbook of neuroscience for the behavioral sciences. New York: Wiley. Cacioppo, J.T., & Tassinary, L.G. (1990). Inferring psychological significance from physiological signals. American Psychologist, 45, 16–28. Diamond, A., & Amso, D. (2008). Contributions of neuroscience to our understanding of cognitive development. Current Directions in Psychological Science, 17, 136–141. Dolan, J.F. (2001). How to find a stellar black hole. Science, 292, 1079–1080. Engel, S.A. (2008). Computational cognitive neuroscience of the visual system. Current Directions in Psychological Science, 17, 68–72. Gotlib, I.H., & Hamilton, J.P. (2008). Neuroimaging and depression: Current status and unresolved issues. Current Directions in Psychological Science, 17, 159–163. Greenberg, P.E., Stiglin, L.E., Finkelstein, S.N., & Berndt, E.R. (1993). The economic burden of depression in 1990. Journal of Clinical Psychiatry, 54, 405–418. Volume 17—Number 2 Grill-Spector, K., & Sayres, R. (2008). Object recognition: Insights from advances in fMRI methods. Current Directions in Psychological Science, 17, 73–79. Hagoort, P. (2008). Should psychology ignore the language of the brain? Current Directions in Psychological Science, 17, 96–101. Henson, R. (2006). Forward inference using functional neuroimaging: Dissociations versus associations. Trends in Cognitive Sciences, 10, 64–69. Iacoboni, M., Lieberman, M.D., Knowlton, B.J., Molnar-Szakacs, I., Mortiz, M., Throop, C.J., & Fiske, A.P. (2004). Watching social interactions produces dorsomedial prefrontal and medial parietal BOLD fMRI signal increases compared to a resting baseline. Neuroimage, 21, 1167–1173. Knowlton, B.J., & Foerde, K. (2008). Neural representations of nondeclarative memories. Current Directions in Psychological Science, 17, 107–111. Kosslyn, S.M., Cacioppo, J.T., Davidson, R.J., Hugdahl, K., Lovallo, W.R., Spiegel, D., & Rose, R. (2002). Bridging psychology and biology: The analysis of individuals in groups. American Psychologist, 57, 341–351. Mitchell, J.P. (2008). Contributions of functional neuroimaging to social cognition. Current Directions in Psychological Science, 17, 142–146. Nee, D.E., Berman, M.G., Moore, K.S., & Jonides, J. (2008). Neuroscientific evidence about the distinction between short- and long-term memory. Current Directions in Psychological Science, 17, 102–106. Ochsner, K.N., & Gross, J.J. (2008). Cognitive emotion regulation: Insights from social cognitive and affective neuroscience. Current Directions in Psychological Science, 17, 153–158. O’Doherty, J.P., & Bossaerts, P. (2008). Towards a mechanistic understanding of human decision making: Contributions of functional neuroimaging. Current Directions in Psychological Science, 17, 119–123. Oliveira, F.T.P., & Ivry, R.B. (2008). The representation of action: Insights from bimanual coordination. Current Directions in Psychological Science, 17, 130–135. Poeppel, D., & Monahan, P.J. (2008). Speech perception: Cognitive foundations and cortical implementation. Current Directions in Psychological Science, 17, 80–85. Poldrack, R.A. (2006). Can cognitive processes be inferred from neuroimaging data? Trends in Cognitive Sciences, 10, 59–63. Raichle, M.E. (2003). Functional brain imaging and human brain function. The Journal of Neuroscience, 23, 3959–3962. Rizzolatti, G., & Arbib, M.A. (1998). Language within our grasp. Trends in Neuroscience, 21, 188–194. Sarter, M., Berntson, G.G., & Cacioppo, J.T. (1996). Brain imaging and cognitive neuroscience: Toward strong inference in attributing function to structure. American Psychologist, 51, 13–21. Skipper, J.I., Nusbaum, H.C., & Small, S.L. (2006). Lending a helping hand to hearing: Another motor theory of speech perception. In M.A. Arbib (Ed.), Action to language via the mirror neuron system (pp. 250–285). New York: Cambridge University Press. Stanley, D., Phelps, E., & Banaji, M. (2008). The neural basis of implicit attitudes. Current Directions in Psychological Science, 17, 164– 170. Vogel, E.K., & Awh, E. (2008). How to exploit diversity for scientific gain: Using individual differences to constrain cognitive theory. Current Directions in Psychological Science, 17, 171–176. Yantis, S. (2008). The neural basis of selective attention: Cortical sources and targets of attentional modulation. Current Directions in Psychological Science, 17, 86–90. 67 INAUGURAL ARTICLE Functional specificity in the human brain: A window into the functional architecture of the mind Nancy Kanwisher1 McGovern Institute for Brain Research, Massachusetts Institute of Technology, Cambridge, MA 02139 This contribution is part of the special series of Inaugural Articles by members of the National Academy of Sciences elected in 2005. Is the human mind/brain composed of a set of highly specialized components, each carrying out a specific aspect of human cognition, or is it more of a general-purpose device, in which each component participates in a wide variety of cognitive processes? For nearly two centuries, proponents of specialized organs or modules of the mind and brain—from the phrenologists to Broca to Chomsky and Fodor—have jousted with the proponents of distributed cognitive and neural processing—from Flourens to Lashley to McClelland and Rumelhart. I argue here that research using functional MRI is beginning to answer this long-standing question with new clarity and precision by indicating that at least a few specific aspects of cognition are implemented in brain regions that are highly specialized for that process alone. Cortical regions have been identified that are specialized not only for basic sensory and motor processes but also for the high-level perceptual analysis of faces, places, bodies, visually presented words, and even for the very abstract cognitive function of thinking about another person’s thoughts. I further consider the as-yet unanswered questions of how much of the mind and brain are made up of these functionally specialized components and how they arise developmentally. brain imaging | modularity | functional MRI | fusiform face area U nderstanding the nature of the human mind is arguably the greatest intellectual quest of all time. It is also one of the most challenging, requiring the combined insights not only of psychologists, computer scientists, and neuroscientists but of thinkers in nearly every intellectual pursuit, from biology and mathematics to art and anthropology. Here, I discuss one currently fruitful component of this grand enterprise: the effort to infer the architecture of the human mind from the functional organization of the human brain. The idea that the human mind/brain is made up of highly specialized components began with the Viennese physician Franz Joseph Gall (1758–1828). Gall proposed that the brain is the seat of the mind, that the mind is composed of distinct mental faculties, and that each mental faculty resides in a specific brain organ. A heated debate on localization of function in the brain raged over the next century (SI Text), with many of the major figures in the history of neuroscience weighing in (Broca, Brodmann, and Ferrier in favor, and Flourens, Golgi, and Lashley opposed). By the early 20th century, a consensus emerged that at least basic sensory and motor functions reside in specialized brain regions. The debate did not end there, however. Today, a century later, two questions are still fiercely contested. First, how functionally specialized are regions of the brain? The concept of functional specialization is not all or none but a matter of degree; a cortical region might be only slightly more engaged in one mental function than another, or it might be exclusively engaged in a single mental function. Many neuroscientists today challenge the strong (exclusive) version of functional specialization. As one visual neuroscientist put it, “each extrastriate visual area, rather than performing a unique, one-function analysis, is engaged, as are most neurons in the visual system, in many different tasks” (1). www.pnas.org/cgi/doi/10.1073/pnas.1005062107 The second ongoing controversy concerns the question of whether only basic sensory and motor functions are carried out in functionally specialized regions, or whether the same might be true even for higher-level cognitive functions. Although one might think that Broca settled this matter with his demonstration that the left frontal lobe is specialized for aspects of language, the current status of this debate is far from clear. Indeed, a recent authoritative review of the brain-imaging literature on language concludes that “areas of the brain that have been associated with language processing appear to be recruited across other cognitive domains” (2). The case of language is not unique. Indeed, a backlash against strong functional specialization seems to be in vogue. A recent neuroimaging textbook argues that “unlike the phrenologists, who believed that very complex traits were associated with discrete brain regions, modern researchers recognize that . . . a single brain region may participate in more than one function” (3). In this review, I address these ongoing controversies about the degree and nature of functional specialization in the human brain, arguing that recent neuroimaging studies have demonstrated that at least a few brain regions are remarkably specialized for single high-level cognitive functions. To make my case, I first describe three candidates for such functionally specific brain regions identified in my lab. I then consider how much of the brain is made up functionally specialized regions: are they found only for highlevel perceptual functions or also for components of abstract thought? I then ask how these regions arise developmentally; that is, what are the exact roles of genes and experience in the development of these regions? In SI Text, I address a key challenge to the specificity of the fusiform face area (FFA) and parahippocampal place area (PPA), and I consider the computational advantages that may be afforded by specialized regions in the first place. I conclude by speculating that the cognitive functions implemented in specialized brain regions are strong candidates for fundamental components of the human mind. Neuroimaging Evidence for Functional Specialization in the Ventral Visual Pathway Ever since Broca, neurologists and cognitive neuroscientists have investigated cognitive impairments in people with focal brain lesions, providing extensive evidence for localization of at least some functions in the human brain. The study of neurological disorders is one of the few methods that allows powerful inferences about not just the engagement but also the necessity of a given brain region for a specific cognitive function in humans. However, even if a particular functionally specific region exists, a lesion is unlikely to affect all and only that region, so clean functional dissociations in the patient literature are rare. Brain imaging [and functional MRI (fMRI) in particular] thus provides Author contributions: N.K. wrote the paper. The author declares no conflict of interest. 1 E-mail: ngk@mit.edu. This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1005062107/-/DCSupplemental. PNAS | June 22, 2010 | vol. 107 | no. 25 | 11163–11170 NEUROSCIENCE Contributed by Nancy Kanwisher, April 16, 2010 (sent for review February 22, 2010) a powerful complement to lesion studies, allowing neural activity in the normal human brain to be monitored safely and noninvasively at resolutions approaching the millimeter range. The principle underlying fMRI is that blood flow increases locally in active regions of the brain. Although the precise neural events that fMRI reflects are a matter of ongoing research, the general validity of the method as an indicator of neural activity is clear from studies replicating, with fMRI, the properties of visual cortex previously established by the gold-standard method of single-neuron recording in monkeys. Thousands of papers have used fMRI to ask about the relative contributions of different regions in the human brain to a wide variety of cognitive functions. My lab has focused especially on the question of whether any of these brain regions are specifically engaged in a single high-level cognitive function. Supporting the idea that some brain regions are indeed engaged in specific mental functions, we have identified a number of cortical regions (Fig. 1) that respond selectively to single categories of visually presented objects: most notably, the FFA, which responds selectively to faces (4, 5), the PPA, which responds selectively to places (6), and the extrastriate body area (EBA), which responds selectively to bodies and body parts (7). These three brain regions are not the only ones that have been argued to conduct specific perceptual functions (8). Probably the strongest other case is visual area MT/V5, shown much earlier with neurophysiological methods to play a key causal role in the perception of visual motion in monkeys (9–11), and later, identified in humans with brain imaging (12, 13). However, even this classic example of functional specificity does not process visual-motion information exclusively; this area also contains information about stereo depth (14). Another strong case of functional specificity for a simple visual dimension is color (15), for which recent evidence from both fMRI and single-unit recording indicates the existence of multiple millimeter-sized color-selective “globs” in posterior inferotemporal cortex in macaques (16, 17). Other brain regions have been reported to be selectively engaged in processing information about biological motion (18), visually guided reaching (19), and grasping (20). For most cases in the neuroimaging literature, however, the main claim is one of regional specificity (i.e., that the implicated function activates this region more than other brain regions) rather than of functional specificity (i.e., that the implicated region is more engaged for this function than other functions). In contrast, this article focuses primarily on the question of functional specificity, because this is the question that is critical for understanding the architecture of the human mind (Fig. 1). The evidence we and others have collected on the FFA, PPA, and EBA provides unusually strong support for functional specificity of these regions for three reasons. First, each of these regions has been found consistently in dozens of studies across many labs; although their theoretical significance can be debated, their existence cannot. Indeed, these regions are found, in more or less the same place, in virtually every neurologically intact subject; they are part of the basic functional architecture of the human brain. Second, the category selectivity by which each region is defined is not merely statistically significant, but also large in effect size: Each of these regions responds about twice as strongly to stimuli from its preferred category as to any nonpreferred stimuli.* Although effect size is generally ignored in the brain imaging literature, it should not be, as it determines the strength of the inference you can draw: If you know how to double the response of a region, you generally have a better handle on its function than if you merely know how to change its response by a small amount. Third, the fact that these regions can be found easily in any normal subject makes possible a “region of interest” (ROI) research strategy whereby the region is first functionally identified in each subject individually in a short “localizer” scan, and then the response of that region is measured in any number of new conditions that test specific hypotheses about its exact function. It is precisely the fact that the responses of the FFA, PPA, and EBA have been quantified in each of now dozens of different stimulus and task manipulations that enables us to say with confidence that each of these regions is primarily, if not exclusively, engaged in processing its preferred stimulus class (faces, places, and bodies, respectively). Taken together, these three regions constitute some of the strongest evidence that at least some cortical regions are selectively engaged in processing specific classes of stimuli. Next I summarize the evidence for the specificity of each of these regions for a particular class of stimuli. Fig. 1. This schematic diagram indicates the approximate size and location of regions in the human brain that are engaged specifically during perception of faces (blue), places (pink), bodies (green), and visually presented words (orange), as well as a region that is selectively engaged when thinking about another person’s thoughts (yellow). Each of these regions can be found in a short functional scan in essentially all normal subjects. *fMRI response magnitudes are typically measured as percent signal increases compared with a low baseline condition (e.g., fixating on a cross), so a 2-fold response difference might correspond to a 2% signal increase from fixation versus a 1% signal increases from fixation. Crucially, the magnitude of selectivity must be evaluated using data independent of that used to identify the region (21, 22). Selectivity is underestimated when lowresolution methods are used (e.g., when voxels are large or when spatial smoothing or group analyses are used). 11164 | www.pnas.org/cgi/doi/10.1073/pnas.1005062107 FFA. The FFA is the region found in the midfusiform gyrus (on the bottom surface of the cerebral cortex just above the cerebellum) that responds significantly more strongly when subjects view faces than when they view objects (4, 5, 23). This region responds similarly to a wide variety of different kinds of face images (24), including photos of familiar and unfamiliar faces, schematic faces, cartoon faces, and cat faces as well as faces presented in different sizes, locations, and viewpoints (25, 26). Crucially, when relatively high-resolution imaging methods are used (including individual– subject analyses without spatial smoothing), no nonface object has been reported to produce more than one-half the response found for faces in this region. Further, the evidence (27, 28) allows us to reject alternative hypotheses proposed earlier that the FFA is not specifically responsive to faces but rather is more generally engaged in fine-grained discrimination of exemplars of any category or of any category for which the subject has gained substantial expertise. Importantly, the magnitude of the FFA response is correlated trial by trial with success both in detection of the presence of faces and in identification of individual faces (29, 30). Thus, as discussed further in SI Text, the FFA seems to play a central role in the perception of faces but to play little if any role in the perception of nonface objects. This hypothesis is consistent with evidence that (i) face-selective responses have been observed in approximately this location in subdural electrode recordings from the brains of subjects undergoing presurgical mapping for epilepsy treatment (31–33) and (ii) lesions in approximately this location can produce selective deficits in face perception (34). Answering the question of what exactly the FFA does with faces has been more difficult. Current evidence indicates, however, that it is sensitive to multiple aspects of face stimuli including face parts Kanwisher PPA. The PPA is defined functionally as the region adjacent to the collateral sulcus in parahippocampal cortex that responds significantly more strongly to images of scenes than objects (6). The PPA responds to a wide variety of scenes, including indoor and outdoor scenes, familiar and unfamiliar scenes, and even abstract scenes made of Legos (38, 39). The PPA is primarily responsive to the spatial layout of one’s surroundings: its response is not reduced when all of the objects are removed from an indoor scene, leaving just the floor and walls (6). This response profile is tantalizingly reminiscent of the geometric module (40, 41), inferred from behavioral data in which rats and human infants (and adults whose language system is tied up by a concurrent verbal task) rely exclusively on the layout of space, not on objects or landmarks, to reorient themselves in an environment after they are disoriented. Evidence that the PPA is not only activated when information about spatial layout is processed, but that it is further necessary for this function, comes from patients with damage in or near the PPA, who have difficulty encoding information about spatial layout and more generally, in knowing where they are (42, 43). The precise role of the PPA in place perception and navigation is a topic of ongoing investigation (38, 39). EBA. The EBA is a region on the lateral surface of the brain adjacent to (and sometimes partly overlapping with) visual motion area MT, which responds significantly more strongly to images of bodies and body parts than to images of objects or faces. This region responds equally to visually very different images of bodies and body parts, from a photograph of a hand to a photograph of a body (human or animal) to a schematic stick figure of a person. Evidence that this region is not only activated during but is also necessary for the perception of bodies comes from studies in which disruption of the EBA by a brain lesion (44) or transcranial magnetic stimulation (TMS) (45, 46) impairs the perception of body form but not the perception of faces or object shape (45). Further, current evidence indicates that the EBA is more involved in perceiving other people’s bodies than one’s own (47, 48) and that it is more engaged in the perception of the form/identity of bodies than in the actions they are carrying out (44, 49–51). Ovals, Gradients, or Archipelagoes? For simplicity, I have discussed functionally specific regions in the cortex as if they are discrete entities with sharp, well-delineated edges, like the kidney, liver, and heart. Indeed, some functional divisions in the cortex are remarkably sharp, such as the border between retinotopic visual areas V1 and V2. However, there is no reason to assume all functional distinctions in the brain have perfectly sharp edges. Similarly, there should be no requirement that these regions must be simple convex shapes. Irregular-shaped regions with long tenKanwisher Generality: How Much of the Brain Is Composed of Functionally Specific Regions? The evidence for functional specificity within several brain regions (FFA, PPA, EBA) invites a return to the broader questions raised by Gall, Fourens, and Broca: how much of the brain is composed of PNAS | June 22, 2010 | vol. 107 | no. 25 | 11165 INAUGURAL ARTICLE drils or even multiple nonadjacent but nearby (and presumably connected) subregions might be expected. If it becomes clear at higher resolutions that the FFA is in fact a set of distinct noncontiguous regions (a “fusiform face archipelago”?), that will strain the organ analogy but still leave viable a meaningful sense in which these noncontiguous patches constitute a functionally distinct system, much as Maui and Lanai share deep geological, biological, and cultural similarities in virtue of being part of the Hawaiian islands, despite the channel of water between them. However, the more a region turns out to be extensively interdigitated with other functionally distinct entities and the more its borders resemble an arbitrary cutoff point on a gradual functional change across the cortex (52), the less this case will follow the classic idea of a functionally distinct brain region. Most questions about biological systems are matters of degree, and so too is the question of functional specialization in the cortex. Currently available evidence suggests an impressive degree of compartmentalization in at least a few cortical regions (53). Further experiments using new tasks and higher resolution will provide more precise quantitative tests of the anatomical distinctness of these regions. In sum, evidence is now strong that each of at least three cortical regions in humans are selectively (perhaps even exclusively) engaged in specific cognitive functions: the FFA in representing the appearance of faces, the PPA in representing the appearance of places, and the EBA in representing the appearance of bodies. (See SI Text for my reply to an important challenge to the functional specificity of these regions.) Although I have emphasized the role of each of these regions in visual perception, their response is not determined solely by the stimulus that the subject is viewing. The activity of these regions can be strongly modulated by visual attention (54), and they can even be activated when no stimulus is present at all. Simply imagining a face (with eyes closed) selectively activates the FFA and imagining a place activates the PPA (55). Of course, no complex cognitive process is accomplished in a single brain area, and arguments for the specificity of these regions by no means imply that other brain regions play no role. Earlier cortical regions such as primary visual cortex are obviously crucial in the perception of faces, places, and bodies, and higher areas (e.g., in parietal and frontal regions) are also probably necessary for information in the FFA, PPA, and EBA to be used by other cognitive systems and to reach awareness (56–58). Further, none of these regions is the only one with its defining selectivity. For faces, selective responses are found not only in the FFA but also in a nearby but more posterior occipital face area, as well as other regions in the superior temporal sulcus (34, 59), and anterior temporal pole (60). For bodies, selective responses are found not only in the EBA but also in the fusiform body area (FBA). For scenes, selective responses are found not only in the PPA but also in retrosplenial cortex (RSC) and the transverse occipital sulcus (TOS). These other selective regions have not been studied in the same detail as the FFA, PPA, and EBA, so their functions are less clear. Still, the existence of multiple selective regions for each of these three stimulus classes raises the exciting possibility that we may ultimately understand how the percept of a face, for example, emerges from the joint activity of a number of functionally distinct regions, each conducting a different aspect of the analysis of the face stimulus. In the subsequent sections of this article, I discuss four major questions raised by the work on the FFA, EBA, and PPA concerning their specificity, generality, origins, and computational significance. NEUROSCIENCE (eyes, noses, and mouths), the T-shaped configuration of those features, and external features of faces like hair (35) and that representations extracted in the FFA show some invariance across changes in stimulus position and less invariance across changes in viewpoint (25), mirroring comparable behavioral results. The FFA further exhibits neural correlates of long-known behavioral signatures of perception (28), including disproportionate inversion effects (36) and sensitivity to holistic information in upright but not inverted faces (37). Despite these initial insights, important open questions about the FFA remain to be addressed, including a more precise characterization of the representations that it extracts and the computations that it performs, whether it plays some (albeit lesser) role in the perception of any nonface objects, whether it is cytoarchitectonically distinct from its neighbors, what other regions it is connected to, whether and how interactions with other regions modulate or participate in the computations conducted in the FFA and whether it constitutes a single contiguous region on the cortical surface. regions that are selectively engaged in specific cognitive functions? We consider this question by asking whether other specialized brain regions exist for (i) other object categories in the ventral visual pathway and (ii) components of high-level thought. Other Category-Selective Regions? Do we have cortical regions selectively involved in the perception of snakes? Weapons? Vegetables? As Pinker asks in The Language Instinct, does the brain have a produce section (61)? What about categories of objects that may not have been crucial to the survival of our ancestors but that play central roles in modern daily lives, like cars and cell phones? There hardly seems room in the brain for all of these categories, or even all of the important ones, and it is not clear what would be accomplished computationally by such extreme compartmentalization anyway. Happily, we are not restricted to mere speculation; we can simply test empirically for other specialized brain regions. Downing and I did just that (62), screening broadly for 20 different categories of objects selected for their (arguable) evolutionary importance (spiders and snakes, predators, prey, tools, food), their experiential frequency in modern life (cars, chairs), or their implication from prior studies of patients with focal brain damage (fruits and vegetables, musical instruments). Despite replicating the existence of cortical regions selective for faces, places, and bodies in each subject, we found no evidence of cortical specialization for any of the other object categories tested. The previously reported selectivity for tools (63) was not evident in our data, and any partial dissociations between responses to living and nonliving things (or animate versus inanimate objects) were restricted to the already documented properties of the face, place, and body areas. Although null results can always be trumped by later discoveries made with higher spatial resolution or greater statistical power, the resolution and power that was sufficient for robust replication of the FFA, PPA, and EBA did not turn up any new categoryspecific regions. A central conceptual puzzle arises, however, in the search for brain regions selective for new object categories: how do we decide which categories to test? If we proceed by testing only the categories that seem plausible to us, then we risk getting trapped within the confines of our own theoretical preconceptions. This concern is underscored by the fact that the brain specializations already described for faces, places, and bodies are reminiscent of two of the mental faculties proposed by Gall: the sense of people, and the sense of place. Given that Gall arrived at these categories without real evidence, the fact that we have arrived at the same categories is worrisome. Are we, like the phrenologists, allowing our cultural biases to determine what we find in the brain? Are specializations we discover in the brain a kind of hightech projective test? With rigorous experimental methods, we can reduce the chance that the outcomes of our experiments are determined by our cultural/theoretical predispositions. However, how can we ever prevent our conceptual baggage from biasing the space of hypotheses that we consider? My colleagues and I are developing methods to circumvent these biases by searching for structure in the functional responses of the ventral visual cortex in a hypothesis-neutral fashion (64–66). This method searches large datasets composed of the response of each voxel to a large number of stimuli and discovers dominant response profiles in that dataset. Importantly, the method knows nothing about the location of each voxel, so it makes no assumption that functionally related voxels are adjacent. Even more importantly, the method does not look only for selectivity for single-object categories but instead, for any profile of response across the stimuli that best characterizes a large number of voxels (e.g., a high response to all categories except one or a high response to one-half of the categories and a low response to the other one-half, etc.). 11166 | www.pnas.org/cgi/doi/10.1073/pnas.1005062107 For our first test of this method, we scanned subjects while they viewed eight different categories of stimuli. Remarkably, the method spontaneously identified face-, place-, and body-selective response profiles among the top five most robust profiles (Fig. S1 and SI Text). Even more impressively, when we split the data in half to produce 16 different conditions (two per category), without telling the algorithm which pairs of conditions belonged to the same category, the algorithm discovered response profiles characterized by high responses to both face conditions compared with everything else, although these conditions were not labeled as the same category. We found the same for scenes and bodies. These results suggest that face, place, and body selectivity are not simply our own cultural projections onto the brain but are actually inherent in the brain’s response to visual stimuli. Also, they suggest that we do not have similar specificity in the brain for lots of other categories; face, place, and body selectivity are probably special cases. We are now conducting a stronger test of this hypothesis by generating a larger set of stimuli more representative of human visual experience and asking whether face, place, and body selectivity still emerge from the data, even when no stimulus categories are presumed in advance and even when we do not start by constructing a stimulus set that contains a sizable proportion of faces, places, and bodies. It will be most exciting if this new test not only (re)discovers face, place, and body selectivity but also discovers new, previously unknown, response profiles. Selective Cortical Regions for Aspects of Thought? Perhaps it is not surprising that discrete cortical regions can be found that are selectively engaged in processing specific aspects of high-level vision. After all, we are highly visual animals who allocate onethird of our cortex to various aspects of vision, and some division of computational labor within this broad expanse of cortex would seem to make sense. But what about the rest of cognition? Do we have specialized brain machinery for specific components of thought? Indeed, we do. Several years ago, Rebecca Saxe made the astonishing discovery of a region at the junction of the temporal and parietal lobes of the right hemisphere that is selectively engaged when one thinks about what another person is thinking (67, 68). Using the ROI method, Saxe and colleagues (67, 68) have identified this region (known as the rTPJ) in hundreds of subjects and measured its response to a wide array of tasks. These data show that the rTPJ responds strongly when people read scenarios that describe what a person knows or thinks but not when people read scenarios describing physical, as opposed to mental, representations (e.g., in maps or photographs) or vivid descriptions of a person’s physical appearance that do not refer to the contents of the person’s mind. This region is so selective that it does not even respond when people think about another person’s bodily sensations (e.g., thirst, hunger, pleasure), which are mental states but which do not have propositional content like thoughts and beliefs. Most impressively, this region is more strongly activated when people make decisions about what another person knows than when they make the identical response to the identical stimuli but do not construe the task as pertaining to another person’s thoughts (69). The rTPJ is the most functionally selective highlevel cortical region yet described in humans. The discovery of the rTPJ, and the characterization of its functional specificity, serves as an existence proof that functionally specific cortical regions are not restricted to primary sensory and motor areas, or high-level perceptual regions, but can be found for at least one very abstract and high-level aspect of human cognition. This finding invites the question of whether other aspects of high-level cognition may also be computed in specialized cortical regions. Perhaps the most obvious case here is the one proposed by Gall and Broca: language. Surprisingly, despite two centuries of investigation, no consensus has emerged on the Kanwisher Kanwisher Origins: How Do Functionally Specific Regions Arise Developmentally? Although it is obvious that genes and experience both play crucial roles in the development of all brain structures, it is less clear which of the precise details of the circuitry of each brain region are specified in the genome and which are derived from experience. At first glance, the existence of brain regions selective for faces, places, and bodies would seem to fit nicely with the view held by many of the most prominent advocates of modularity of mind and brain—from Gall to Chomsky, Fodor, and Pinker—that organs of mind and brain are innate (i.e., the products of natural selection). Indeed, it seems plausible that the rapid and accurate recognition of faces, places, and bodies had such survival value to our ancestors that detailed instructions for wiring up the specific neural circuitry of the FFA, PPA, and EBA may have become specified in the genome. However, alternative accounts are also plausible. Quite apart from the experience of our ancestors, each of us modern-day humans probably looks at (and attends to) faces, places, and bodies more frequently than almost any other stimulus class. Given that cortical organization can be affected by experience, the existence of regions specialized for processing these visual categories could result from the extensive experience each of us has with these categories during our lifetime, without any specific genetic predilection for these categories per se. Recent evidence, discussed next, suggests that the cortical machinery of face perception may be primarily genetically specified, whereas the selectivity of another nearby cortical region may be primarily determined by the individual’s experience. Specific Role of Genes in Face Perception. Until very recently, we had almost no relevant data on the degree to which the existence, location, and fine-grained circuit details of the FFA were genetically specified versus derived from experience, leaving the topic wide open for passion and polemic. In just the last few years, however, several new lines of evidence point to a specific role of genes in determining the neural machinery of face perception. First, a congenital disorder in face perception, developmental prosopagnosia, has been shown to run in families (76, 77). Second, face-perception ability is heritable (i.e., more strongly correlated for identical than fraternal twins), and this effect is independent of the heritability of domain-general abilities like IQ or global attention (78, 79). Third, the spatial distribution of fMRI responses across the ventral visual pathway to faces is more similar between monozygotic than dizygotic twins; the same is true for scenes but not for chairs or words (80). Although all three findings implicate genes in face-specific processing, they do not tell us which genes are involved or by what causal pathway they affect face perception. Perhaps these genes simply increase social interest and hence, experience with face perception, enhancing ability through training. Or perhaps they directly specify the detailed wiring of the neural circuits for face perception. Evidence that genes may be largely responsible for wiring up much of the face system, with little or no role of experience with faces, comes from recent reports that impressive face discrimination abilities are present in human newborns (81) and even in baby monkeys reared for up to 2 years without ever seeing faces (82). These findings support the hypothesis that the specific instructions for PNAS | June 22, 2010 | vol. 107 | no. 25 | 11167 INAUGURAL ARTICLE all of these questions in nonhuman primates. Therefore, the discovery of functionally specific brain regions that are present in both humans and macaques, such as face- and body-selective regions, opens up fantastic opportunities to address the biological mechanisms of cognition in a way that is nearly impossible in humans. The discoveries (72) of face- and body-selective regions in macaque cortex and the investigation of these regions using the powerful tools of systems neuroscience (73–75) provide a stunning illustration of the insights that can be gleaned from work in primates on the neural machinery of high-level vision. NEUROSCIENCE question of whether any brain regions are specialized for language (or components thereof). The problem arises in part from a conflict between the findings from studies of patients with focal brain lesions, which suggest considerable functional specificity of some cortical regions for some aspects of language, versus the findings from the large neuroimaging literature on language, which suggest considerable overlap between linguistic and nonlinguistic processing. Evelina Fedorenko and I have argued that one possible explanation of the conflict between these two types of studies is that the methods that have been used in virtually all prior neuroimaging studies of language (group analyses) are not well-suited for detecting functional specificity. Group analyses underestimate functional specificity, because different individuals’ brains are anatomically quite different from each other, so alignment across brains is necessarily imperfect. As a result, functionally different regions will sometimes be aligned to the same location in the group space (70, 71). Fedorenko and I are now revisiting the question of functional specificity of the language system using the same individual–subject ROI method that has enabled us to discover the functional specificity of the other regions described above. Note that the failure to discover functionally specific brain regions for a given cognitive process can also be informative. Suppose, for example, that we discover that no brain region is selectively engaged in any aspect of language processing but rather that all regions that support language processing also contribute substantially to nonlinguistic functions. Such a discovery would offer powerful clues into what language is all about. Specifically, we would want to know: what are those nonlinguistic functions that overlap with (say) syntactic processing? What would it tell us about syntax, if it shares neural machinery with (say) music perception, social cognition, or arithmetic? Such possibilities illustrate the exciting prospect of discovering components of mind and brain defined not by the content of the information they operate on, but rather by the computational structure of the problems they solve. Indeed, evidence of domains of cognition that are not computed in cortical tissue selective for that function would offer clues about the broader questions of which mental functions get their own private patch of real estate in the brain, which do not, why some do and others do not, and what the computational advantages might be of functional specialization in the first place (discussed further in SI Text). In some sense, the discovery and characterization of components of the mind and brain that are uniquely human are the most exciting. The fact that our minds and brains have a special circuit just for figuring out what another person is thinking tells us something deep about what it means to be a human being. If we are lucky enough to discover brain machinery specialized for other uniquely human cognitive abilities, such as syntax or a component thereof, it will provide a similarly thrilling insight into human nature. Further, such discoveries might enable us to trace the evolutionary origins of the function in question. For example, if we discover cytoarchitectonic or gene-expression markers for the brain region for understanding other minds, we could then look for the homologous region in primates and investigate its function. Discovering functionally specific components of mind and brain that are not uniquely human, but that are shared with other animals, offers different scientific opportunities. Most current methods available with humans do not enable us to determine precisely the time course of engagement, the causal role, or the connectivity of a given cortical area. (Important exceptions are studies using TMS in normal subjects and electrodes implanted for surgical purposes in humans.) We cannot study in humans the development of a given region under controlled rearing conditions, and we have no good tools for studying the actual neural circuits that implement the cognitive ability in question. However, methods exist to answer constructing the critical circuits for face perception are in the genome. Note that despite this recent evidence that the face system can develop with little or no experience with faces (81, 82), it is nonetheless clear that experience with faces does affect the faceperception system. First, in the other race effect, psychophysical studies have demonstrated what most people know from daily life: we are better able to distinguish individuals from a more familiar than less familiar race (aka “they all look alike”). Second, in perceptual narrowing, face-discrimination abilities that are initially effective on face stimuli of all races or primate species become restricted within a few months of life to only the race/species that the subject has experienced (82–84). This tuning is entirely consistent with the view that the basic face-perception system can arise with virtually no face experience, even if it is subsequently fine tuned by experience, a phenomenon paralleled in language development (85, 86). What do developmental studies in humans tell us about the origins of the face system? A long-standing view has held that face perception develops very slowly in humans, not reaching adult levels until adolescence or later (87, 88). Consistent with this view, several imaging papers (89, 90) have argued that the FFA increases in size through and even beyond adolescence. Some have suggested that this slow development implies that experience plays a critical role in constructing the face-perception system (89, 90). This conclusion does not follow, however, because some developmental changes that occur long after birth are primarily genetically, not experientially, determined (as in the case of puberty). Further, more recent behavioral results show that every aspect of face-specific perceptual processing tested so far (inversion effects, measures of holistic processing, etc.) is present at the earliest ages ever tested; several signatures of face processing are present within the first 3 days of life (91). Ongoing studies in our lab and others are finding adult-sized FFAs in the majority of children scanned at age 5 and 6 years. Thus, despite the widespread claims to the contrary, current developmental data do not argue for slow development of face-specific perceptual mechanisms. In sum, although the precise roles of genes and experience in the construction of category-selective regions of cortex are not yet clear, several studies suggest that the face system may be largely innate: experience with faces may not be necessary for the initial development of the face-perception system, although experience apparently fine tunes it. Still, if new evidence strengthens this view, it would not necessarily imply that all functionally specific regions of cortex are constructed in the same way. Indeed, the functional selectivity of at least one region of the brain, the visual word form area, is derived from the individual’s experience, not their genes, as discussed next. At Least One Functionally Specific Cortical Region Derives Its Specificity from Experience. Visual word recognition provides a powerful test case of the origins of cortical selectivity. Everyone in our study population has extensive experience looking at visually presented words, so if experience is ever sufficient to specify the selectivity of a cortical region for a particular class of stimuli, we would expect to find one for visual words. However, crucially, human beings have only been reading for a few thousand years, which is not thought to be long enough for the evolution of a complex structure. Thus, if a brain region is found that responds selectively to visually presented words, that would suggest that cortical selectivity can be specified by experience (92). What does the evidence show? A number of studies going back almost two decades have argued for the existence of a visual word form area. However, many of these studies contrasted the cortical response to visually presented words with the response to very simple baseline tasks (93, 94), leaving unanswered the question of whether the region is 11168 | www.pnas.org/cgi/doi/10.1073/pnas.1005062107 specific to visual word recognition or whether it plays a more general role in the recognition of any complex visual stimuli. We searched for several years for a brain region that responded more strongly to visually presented words than to line drawings of familiar objects. Although we failed initially to find such a region in many studies, when technical advances enabled us to scan at higher resolution, we then found it in the majority of subjects (95). This region is tiny, about one-tenth the volume of the FFA, which explains why we did not see it with standard imaging resolutions (Fig. S2 and SI Text). To further test the selectivity of this region, we used the same localize-and-test procedure that was effective in characterizing the FFA, PPA, and EBA. In independent tests of the response of the region, we replicated the fact that it responded severalfold higher to words than to line drawings (Fig. S2A). Further, we showed that the response was low, in this region, to stimuli that shared many of the visual properties of words: strings of digits and letters in an orthography unfamiliar to the subject (Hebrew). The response to consonant strings was the same as that to words, which suggests that meaning and orthographic regularity are not required to activate this region. In contrast, when we scanned subjects who read both English and Hebrew, we found a high response to words written in both languages (and orthographies) in this region (Fig. S2B). Thus, the response of this region is determined by the individual’s experience. An even stronger demonstration of the experience dependence of this region comes from a before-and-after study of Chinese illiterates, who showed a character-selective response in this region after being trained for several months to read but not before (96). Many important questions about this cortical region remain to be answered, such as whether it can develop in an alternate location if damage to this region occurs in childhood (97) or adulthood (98, 99) and whether it reflects a discrete, functionally homogeneous module or a gradient of selectivity (52). Whatever the answers to these questions, the current evidence indicates that the particular selectivity of this region depends on the specific experience of the individual and not the experience of his or her ancestors. In sum, recent studies are beginning to shed light on the roles of genes and experience in the origins of cortical regions selectively engaged in specific cognitive functions. Multiple lines of evidence indicate a specific role for genes in wiring up the face system, yet at least one other region derives its selectivity from experience. Much remains to be understood about how exactly genes and experience shape neural circuits. Conclusions What a great privilege it is to have access to technology that Gall and Broca never dreamed of, technology that enables us to discover fundamental components of the human brain. Already, the evidence is strong for cortical regions that are selectively engaged in the perception of faces, places, bodies, and words and another region for thinking about what other people are thinking. Possible cortical specializations for other domains, including aspects of number (100), music (101), and language (70), are under active investigation. The possibility is within reach of obtaining a cognitively precise parts list for the human brain. The most exciting aspect of this enterprise is not where each component is found in the brain but which functions get their own brain region and ultimately, why some do and others apparently do not. But even a complete parts list, exciting as it would be, is only a first step. A wide landscape of exciting new questions has opened up. What are the exact neural circuits that enable each region to conduct its signature function? Why do these regions arise so systematically where they do in the brain, and are there ever circumstances in which a region arises in a different locus or moves over after damage to its original locus? Is there some hardware constraint (cytoarchitecture, connectivity, proximity to other areas, etc.) that Kanwisher 1. Schiller PH (1996) On the specificity of neurons and visual areas. Behav Brain Res 76:21–35. 2. Blumstein S (2009) The Cognitive Neurosciences, ed Gazzanica MS (MIT Press, Cambridge, MA). 3. Huettel SA, Song AW, McCarthy G (2004) Functional Magnetic Resonance Imaging (Sinauer Associates, Sunderland, MA). 4. Kanwisher NG, McDermott J, Chun MM (1997) The fusiform face area: A module in human extrastriate cortex specialized for face perception. J Neurosci 17:4302–4311. 5. McCarthy G, Puce A, Gore JC, Allison T (1997) Face-specific processing in the human fusiform gyrus. J Cogn Neurosci 9:605–610. 6. Epstein R, Kanwisher N (1998) A cortical representation of the local visual environment. Nature 392:598–601. 7. Downing PE, Jiang Y, Shuman M, Kanwisher N (2001) A cortical area selective for visual processing of the human body. Science 293:2470–2473. 8. Zeki SM (1978) Functional specialisation in the visual cortex of the rhesus monkey. Nature 274:423–428. 9. Zeki SM (1974) Functional organization of a visual area in the posterior bank of the superior temporal sulcus of the rhesus monkey. J Physiol 236:549–573. 10. Albright TD (1984) Direction and orientation selectivity of neurons in visual area MT of the macaque. J Neurophysiol 52:1106–1130. 11. Newsome WT, Wurtz RH, Dürsteler MR, Mikami A (1985) Deficits in visual motion processing following ibotenic acid lesions of the middle temporal visual area of the macaque monkey. J Neurosci 5:825–840. 12. Zeki S, et al. (1991) A direct demonstration of functional specialization in human visual cortex. J Neurosci 11:641–649. 13. Tootell RB, et al. (1995) Functional analysis of human MT and related visual cortical areas using magnetic resonance imaging. J Neurosci 15:3215–3230. 14. DeAngelis GC, Cumming BG, Newsome WT (1998) Cortical area MT and the perception of stereoscopic depth. Nature 394:677–680. 15. Hadjikhani N, Liu AK, Dale AM, Cavanagh P, Tootell RBH (1998) Retinotopy and color sensitivity in human visual cortical area V8. Nat Neurosci 1:235–241. 16. Conway BR, Moeller S, Tsao DY (2007) Specialized color modules in macaque extrastriate cortex. Neuron 56:560–573. 17. Conway BR, Tsao DY (2009) Color-tuned neurons are spatially clustered according to color preference within alert macaque posterior inferior temporal cortex. Proc Natl Acad Sci USA 106:18034–18039. 18. Grossman E, et al. (2000) Brain areas involved in perception of biological motion. J Cogn Neurosci 12:711–720. 19. Connolly JD, Andersen RA, Goodale MA (2003) FMRI evidence for a ‘parietal reach region’ in the human brain. Exp Brain Res 153:140–145. 20. Culham JC, et al. (2003) Visually guided grasping produces fMRI activation in dorsal but not ventral stream brain areas. Exp Brain Res 153:180–189. 21. Vul E, Kanwisher N (2010) Begging the question: The non-independence error in fMRI data analysis. Foundations and Philosophy for Neuroimaging, eds Hanson S, Bunzl M (MIT Press, Cambridge, MA). 22. Kriegeskorte N, Simmons WK, Bellgowan PS, Baker CI (2009) Circular analysis in systems neuroscience: The dangers of double dipping. Nat Neurosci 12:535–540. 23. Puce A, Allison T, Asgari M, Gore JC, McCarthy G (1996) Differential sensitivity of human visual cortex to faces, letterstrings, and textures: A functional magnetic resonance imaging study. J Neurosci 16:5205–5215. 24. Kanwisher N, Yovel G (2006) The fusiform face area: A cortical region specialized for the perception of faces. Philos Trans R Soc Lond B Biol Sci 361:2109–2128. 25. Grill-Spector K, et al. (1999) Differential processing of objects under various viewing conditions in the human lateral occipital complex. Neuron 24:187–203. 26. Schwarzlose RF, Swisher JD, Dang S, Kanwisher N (2008) The distribution of category and location information across object-selective regions in human visual cortex. Proc Natl Acad Sci USA 105:4447–4452. 27. Yovel G, Kanwisher N (2004) Face perception: Domain specific, not process specific. Neuron 44:889–898. 28. Mckone EM, Robbins RR (2010) Are faces special? The Handbook of Face Perception, eds Calder AC, et al. (Oxford University Press, Oxford). 29. Grill-Spector K, Malach R (2004) The human visual cortex. Annu Rev Neurosci 27:649–677. 30. Rotshtein P, Henson RN, Treves A, Driver J, Dolan RJ (2005) Morphing Marilyn into Maggie dissociates physical and identity face representations in the brain. Nat Neurosci 8:107–113. 31. McCarthy G, Puce A, Belger A, Allison T (1999) Electrophysiological studies of human face perception. II: Response properties of face-specific potentials generated in occipitotemporal cortex. Cereb Cortex 9:431–444. 32. Allison T, Puce A, Spencer DD, McCarthy G (1999) Electrophysiological studies of human face perception. I: Potentials generated in occipitotemporal cortex by face and nonface stimuli. Cereb Cortex 9:415–430. 33. Puce A, Allison T, McCarthy G (1999) Electrophysiological studies of human face perception. III: Effects of top-down processing on face-specific potentials. Cereb Cortex 9:445–458. 34. Kanwisher AC, Barton JB (2010) The functional architecture of the face system: Integrating evidence from fMRI and patient studies. The Handbook of Face Perception, eds Calder AC, et al. (Oxford University Press, Oxford). 35. Liu SRA, Chiarello C, Quan N (1999) Hemispheric sensitivity to grammatical cues: Evidence for bilateral processing of number agreement in noun phrases. Brain Lang 70:483–503. 36. Yovel G, Kanwisher N (2005) The neural basis of the behavioral face-inversion effect. Curr Biol 15:2256–2262. 37. Schiltz C, Rossion B (2006) Faces are represented holistically in the human occipitotemporal cortex. Neuroimage 32:1385–1394. 38. Epstein R (2005) The cortical basis of visual scene processing. Vis Cogn 12:954–978. 39. Epstein RA (2008) Parahippocampal and retrosplenial contributions to human spatial navigation. Trends Cogn Sci 12:388–396. 40. Cheng K, Gallistel CR (1984) Testing the geometric power of an animal’s spatial representation. Animal Cognition: Proceedings of the Harry Frank Guggenheim Conference, eds Roitblat HL, Bever TG, Terrace HS (Erlbaum, Hillsdale, NJ), pp 409–423. 41. Hermer L, Spelke E (1996) Modularity and development: The case of spatial reorientation. Cognition 61:195–232. 42. Epstein R, De Yoe E, Press D, Kanwisher N (2001) Neuropsychological evidence for a topographical learning mechanism in parahippocampal cortex. Cogn Neuropsychol 18:481–508. 43. Habib M, Sirigu A (1987) Pure topographical disorientation: A definition and anatomical basis. Cortex 23:73–85. 44. Moro V, et al. (2008) The neural basis of body form and body action agnosia. Neuron 60:235–246. 45. Pitcher D, Charles L, Devlin JT, Walsh V, Duchaine B (2009) Triple dissociation of faces, bodies, and objects in extrastriate cortex. Curr Biol 19:319–324. 46. Urgesi C, Berlucchi G, Aglioti SM (2004) Magnetic stimulation of extrastriate body area impairs visual processing of nonfacial body parts. Curr Biol 14:2130–2134. 47. Chan AW, Peelen MV, Downing PE (2004) The effect of viewpoint on body representation in the extrastriate body area. Neuroreport 15:2407–2410. 48. Saxe R, Jamal N, Powell L (2006) My body or yours? The effect of visual perspective on cortical body representations. Cereb Cortex 16:178–182. 49. Urgesi C, Calvo-Merino B, Haggard P, Aglioti SM (2007) Transcranial magnetic stimulation reveals two cortical pathways for visual body processing. J Neurosci 27:8023–8030. 50. Saxe R, Xiao DK, Kovacs G, Perrett DI, Kanwisher N (2004) A region of right posterior superior temporal sulcus responds to observed intentional actions. Neuropsychologia 42:1435–1446. 51. Candidi M, Urgesi C, Ionta S, Aglioti SM (2008) Virtual lesion of ventral premotor cortex impairs visual perception of biomechanically possible but not impossible actions. Soc Neurosci 3:388–400. Kanwisher PNAS | June 22, 2010 | vol. 107 | no. 25 | 11169 INAUGURAL ARTICLE ACKNOWLEDGMENTS. Many people provided useful comments on this manuscript, especially Bevil Conway, Sue Corkin, Ev Fedorenko, Charles Jennings, Eric Kandel, Hans Op de Beeck, John Rubin, Liz Spelke, and Bobbie Spellman. The writing of this paper was supported by National Institutes of Health Grant EY13455 (to N.K.) and a grant from the Ellison Medical Foundation. NEUROSCIENCE tracted in the place area, as suggested by extensive research on the perception of faces and spatial layouts? Is it involved only in the representation of the physical characteristics of a face, or does it contain information about the sex, age, race, mood, or identity of the person? Methods such as fMRI adaptation and fMRI pattern analysis have started to answer these questions, although each method has limitations and progress to date has been modest. Satisfyingly precise characterizations of the mental functions implemented in each region will require extensive further work using not only fMRI and other brain-based methods but also increased efforts to relate these findings to behavioral and computational work on the representations and algorithms entailed in different aspects of cognition. forces these regions to arise where they do? How do these regions work with each other—and with more general-purpose brain regions (102)—to support complex real-world cognition? How did these regions evolve, and what functions did they conduct in our primate ancestors? Can each region be recruited to perform new tasks? For example, can the neural machinery of social cognition be used to think about the mood of a financial market or to understand why a computer program fails to understand what we want it to do, and can the PPA be used to understand maps, architectural diagrams, or graphs depicting 3D landscapes of data? But what psychologists like me most want to do is discover fundamental components not just of the brain but also of the mind. For the discoveries of functionally specific brain regions to be useful in this enterprise, we need much richer understandings of the role of each of these regions in cognition. We need not just loose descriptions of the function of a region (e.g., face perception) but precise characterization of the computations and representations conducted in each region. Does the face area extract qualitatively different kinds of representations from those ex- 52. Vinckier F, et al. (2007) Hierarchical coding of letter strings in the ventral stream: Dissecting the inner organization of the visual word-form system. Neuron 55:143–156. 53. Spiridon M, Fischl B, Kanwisher N (2006) Location and spatial profile of categoryspecific regions in human extrastriate cortex. Hum Brain Mapp 27:77–89. 54. Wojciulik E, Kanwisher N, Driver J (1998) Covert visual attention modulates facespecific activity in the human fusiform gyrus: fMRI study. J Neurophysiol 79:1574–1578. 55. O’Craven KM, Kanwisher N (2000) Mental imagery of faces and places activates corresponding stiimulus-specific brain regions. J Cogn Neurosci 12:1013–1023. 56. Kanwisher N (2001) Neural events and perceptual awareness. Cognition 79:89–113. 57. Baars BJ, Newman J (1994) A neurobiological interpretation of global workspace theory. The New Science of Human Experience: Cognitive Neurobiology and the Quest for Consciousness, eds Revonsuo A, Kamppinen M (Oxford University Press, New York), pp 211–226. 58. Dehaene S, Kerszberg M, Changeux JP (1998) A neuronal model of a global workspace in effortful cognitive tasks. Proc Natl Acad Sci USA 95:14529–14534. 59. Puce A, Allison T, Bentin S, Gore JC, McCarthy G (1998) Temporal cortex activation in humans viewing eye and mouth movements. J Neurosci 18:2188–2199. 60. Rajimehr R, Young JC, Tootell RB (2009) An anterior temporal face patch in human cortex, predicted by macaque maps. Proc Natl Acad Sci USA 106:1995–2000. 61. Pinker S (2000) The Language Instinct: How the Mind Creates Language (Harper Collins, New York). 62. Downing PE, Chan AW, Peelen MV, Dodds CM, Kanwisher N (2006) Domain specificity in visual cortex. Cereb Cortex 16:1453–1461. 63. Chao LL, Martin A, Haxby JV (1999) Are face-responsive regions selective only for faces? Neuroreport 10:2945–2950. 64. Lashkari D, Vul E, Kanwisher N, Golland P (2008) Discovering Structure in the Space of Activation Profiles in fMRI, eds Metaxas D, et al. (Springer Verlag, Berlin), pp 1016–1024. 65. Lashkari D, Vul E, Kanwisher N, Golland P (2010) Discovering structure in the space of fMRI selectivity profiles. Neuroimage 50:1085–1098. 66. Kriegeskorte N, et al. (2008) Matching categorical object representations in inferior temporal cortex of man and monkey. Neuron 60:1126–1141. 67. Saxe R, Kanwisher N (2003) People thinking about thinking people. The role of the temporo-parietal junction in “theory of mind.” Neuroimage 19:1835–1842. 68. Saxe R, Powell LJ (2006) It’s the thought that counts: Specific brain regions for one component of theory of mind. Psychol Sci 17:692–699. 69. Saxe R, Schulz LE, Jiang YV (2006) Reading minds versus following rules: Dissociating theory of mind and executive control in the brain. Soc Neurosci 1:284–298. 70. Fedorenko E, Kanwisher K (2009) Neuroimaging of language: Why hasn’t a clearer picture emerged? Lang Linguist Compass 3:839–865. 71. Fedorenko E, Hsieh P-J, Castañón AN, Whitfield-Gabrieli S, Kanwisher N (2010) A new method for fMRI investigations of language: Defining ROIs functionally in individual subjects. J Neurophysiol, 10.1152/jn.00032.2010. 72. Tsao DY, Freiwald WA, Knutsen TA, Mandeville JB, Tootell RB (2003) Faces and objects in macaque cerebral cortex. Nat Neurosci 6:989–995. 73. Tsao DY, Freiwald WA, Tootell RB, Livingstone MS (2006) A cortical region consisting entirely of face-selective cells. Science 311:670–674. 74. Moeller S, Freiwald WA, Tsao DY (2008) Patches with links: A unified system for processing faces in the macaque temporal...
Purchase answer to see full attachment