Cognitive neuroscience: Development and prospects

MICHAEL I POSNER1, and SHOBINI RAO2

1

Department of Psychology, University of Oregon, Eugene, Oregon 97403, USA 2 Neuropsychology, NIMHANS, Hosur Road, Bangalore 560 029, India e-mail: mposner@uoregon.edu

The eld of cognitive neuroscience has enjoyed an explosive growth following the nding that specic brain areas could be activated during processing of visual and auditory words. The subsequent 20 years have provided a variety of techniques that allow the exploration of brain networks related to a large number of cognitive, emotional and social tasks and environments and that have examined the formation and loss of these networks over the human life span. This chapter reviews the various methods developed for noninvasive exploration of human brain function. Substantive contributions are then examined in several areas of particular importance to the Indian research community. These include language, consciousness, brain development and training. We consider the problems remaining to be explored, and possible practical consequences of the research. Cognitive neuroscience is at the intersection of two prior elds: Cognitive psychology and neuroscience. Both these elds have had a relatively short history under these names, although both have roots in ancient philosophy [1]. In this chapter, we will examine the modern history of cognitive neuroscience, discuss new tools for the noninvasive exploration of the human brain and apply them to several active elds of research. We will attempt to examine both the changes, which cognitive neuroscience has produced in our understanding of the mind, and in the understanding of the brain. In each area, we consider the relevance of the topic to the Indian scientic community.

1. Cognitive neuroscience Cognitive psychology was developed in the 1960s. The work of Herbert Simon and Allen Newell, based upon the general problem solver [2] argued that computer programs simulating human performance could serve as a theory of the mind. The computer metaphor left little scope for studies of the brain. At about the same time psychologists, adapting the mathematical theory of communication [3] were able to develop important empirical demonstrations describing human performance in terms of laws governing the rate of information transfer [4,5]. These studies served as the basis for an approach to the mind based on empirical studies which was synthesized by Ulrich Neisser in his 1967 book Cognitive Psychology [6]. Over the years, cognitive psychology branched out to incorporate

aspects of linguistics, computer science and philosophy under the title Cognitive Science [7]. Neuroscience began in the 1950s as the incorporation of many elds that were interested in the basic study of the brain. While cognitive psychology mainly studied human beings, the study of the brain, incorporated work from simpler organisms whose brains were more amenable to anatomical and physiological methods which were by necessity often very invasive. 1.1 Birth of cognitive neuroscience The name cognitive neuroscience was coined by George Miller and Michael Gazzaniga during the early 1980s, when Gazzaniga developed an Institute with that name as part of the grant program of the Sloan Foundation related to cognitive science.

Keywords. Attention; consciousness; cognitive science; language neuroscience; self regulation.

419

420

MICHAEL I POSNER AND SHOBINI RAO

The term neuropsychology, although it had a broader meaning [8] had come to be associated with studies of the brain and performance of people with various forms of brain damage. Although excellent work was done in neuropsychology, particularly in distinguishing the dierent functions of the two cerebral hemispheres, its use in brain injury cases made it less inuential in understanding normal human cognition. It was not until the late 1980s that neuroimaging allowed the normal human brain to be studied while people carried out the kind of tasks which were typical in cognitive psychology. The rst studies involved the language system [9]. These studies used a subtractive method adopted from cognitive psychology to determine activations in sensory, motor, phonological, semantic, and attentional operations. Probably the most important overall result was that even for higher mental processes like word semantics and attention there were suciently common activations to average across people. From the work of Karl Lashely [10] and Gestalt Psychologists [11], many believed that higher mental processes involved the whole brain and did not show any substantial localization. The brain areas involved specically in semantic processing included left ventral frontal lobe, left posterior tempero-parietal cortex, lateral areas of the cerebellum and the dorsal anterior cingulate. Subsequent studies suggested that each area had a dierent function. 1.2 Implication for brain and mind In the twenty years of subsequent imaging studies, some generalizations have emerged about mental processes and about the human brain that supports them. These may be summarized under the headings of localization, interaction, control, and plasticity. Table 1 summarizes many studies involving most of the important areas of cognition, emotion and social interaction. Indeed sometimes the terms aective and social neuroscience are used but most often all elds of research are included in the generalized use of the term cognitive neuroscience. While there is substantial agreement on the brain areas activated in studies of attention, memory and language, there is much less agreement on the exact function of these areas. In addition to separate localization the areas activated have to be coordinated to carry out the task. Although in agreement with patient studies, many tasks involved mainly one cerebral hemisphere (e.g. left for language), most tasks involved activations in both hemispheres and often in subcortical areas such as the basal ganglia and cerebellum. The interaction among these brain areas in terms of networks has been a major concern of recent studies.

Table 1. A list of areas of cognition and emotion for which neuroimaging studies have indicated neural networks involved. One selected study of the each network is referenced [1222]. Function Arithmetic Autobiographical Memory Fear Faces Music Object Perception Reading and Listening Reward Self Reference Spatial Navigation Working Memory Selected reference [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22,23]

Using high density electrical or magnetic recordings from the scalp in coordination with fMRI, it has been possible to work out the time course of these activations and learn something about the direction of information ow. These studies have shown the importance of re-entrant signals so that sensory areas may be activated in the rst 100 millisecs from sensory input, but may be active again by feedback from attentional networks beyond the rst 100 millisecs. These potentially overlapping signals make it dicult to determine the signicance of any activity without a full understanding of the network involved. These new ndings have shown that the longstanding serial model of information ow in the human brain from primary sensory to secondary and tertiary association areas is incorrect [23]. They also make it more important to understand the control signals that might provide priority to particular pathways in any given task. The human is a learning animal. Ancient networks underlying simpler reex activity present at birth can be reworked through experience allowing much more plasticity than has previously been supposed [24]. The activation of many networks are common to all people, suggesting their genetic origins. However, the eciency with which these networks carry out tasks dier among people and provide ample opportunity for brain plasticity to inuence performance. Recent studies have shown how gene by environment interaction shapes these individual dierences.

2. Probing human brain networks In cellular physiology, the idea of a network involves identied neurons that connect to one another by synapses or in some cases through other

COGNITIVE NEUROSCIENCE: DEVELOPMENT AND PROSPECTS

421

Figure 1. Brain networks underlying attention. One fronto-parietal network is involved in orienting to sensory events (circles), while a cingulo-opercular network relates to the resolution of conict among responses (triangles, executive network), a third network is responsible for achieving the alert state involving norepinepherine from the midbrain (squares). [Adapted from 25].

means of communication. Connectionist models, inspired by neural networks, have considered units at particular levels that inuence each other by direct or reciprocal connections. Imaging of human task performance has identied another level of network function, which is clearly related to both the models and the underlying cellular structure by showing that a number of quite separate brain areas must be orchestrated even in a simple task. Each of these areas may be performing a dierent computation, which taken together allow performance of the task. Typically cognitive neuroscience regards the set of activations and their connections as the network that underlies task performance. In this section, we use attentional networks as an illustration of the methods currently available to probe the many networks featured in table 1. Attentional networks are special in that their primary purpose is to inuence the operation of other brain networks. As illustrated in gure 1, three attentional functions for which brain networks have been imaged are: alerting which is involved in acquiring and maintaining the alert state; orienting to sensory stimuli and executive control involved in the resolution of conict between neural systems and regulating thoughts and feelings [9]. Although the sites at which attention inuence can be demonstrated involve any brain area including primary sensory, limbic and motor cortex, as shown in table 1, the sources of these activations are much more limited. Orienting to sensory events has been better studied of these networks both with imaging [26] and cellular [27] methods. The convergence on the

set of brain areas serving as the source of the amplication of sensory signals has been impressive ([28] for a recent review). It is widely agreed that the frontal eye elds work in conjunction with superior and inferior parietal areas as the cortical nodes of the orienting network. In addition, studies have implicated some subcortical areas including the pulvinar of the thalamus and the superior colliculus. Most of the studies of this network have involved visual stimuli, but the sources of the attention inuences in orienting to other modalities are much the same. Of course the site of amplication of the sensory message is quite dierent for each sensory modality. Evidence to date suggests that both maintained alertness during task performance (tonic) and phasic changes induced by a warning signal involving a subcortical structure, the locus coeruleus that is the source of the brains norepinephrine. A great deal of evidence [29] indicates that the tonic state depends upon an intact right cerebral hemisphere. Lesions in this hemisphere can produce profound diculty in responding to unexpected targets. Warning signals, however, may have their inuence more strongly on the left cerebral hemisphere [30]. This distinction may reect a more general division between the hemispheres where rapidly acting events are left lateralized while more slowly changing states involve right hemisphere activity. Tasks that involve conict between stimulus dimensions competing for control of the output often provide activation in the anterior cingulate gyrus and lateral prefrontal areas. It is thought that the conict, induced by a stimulus, is representative of situations where dierent neural networks compete for control of consciousness or output. Because of this we have termed this the executive attention network because it regulates the activity in other brain networks involved in thought and emotion [31,32]. This network shows a strong development in childhood and its maturation is related to what in developmental psychology has been called self-regulation. Individual dierences are invariably found in cognitive tasks involving attention. The Attention Network Test (ANT) was developed to examine individual dierences in the eciency of the brain networks in alerting and orienting the executive attention discussed above [33]. The ANT uses dierences in reaction time (RT) between conditions to measure the eciency of each network. Each trial begins with a cue (or a blank interval, in the no-cue condition) that informs the participant either that a target will be occurring soon, or where it will occur or both. The target always occurs either above or below xation and consists of a central arrow, surrounded by anking arrows that can either point in the same

422

MICHAEL I POSNER AND SHOBINI RAO

Figure 2. The fronto-parietal network (orienting, online control) and the cingulo-opercular network (executive, set control) are active at rest and undergo a long developmental process shown as graphs for children C (age 9) and adults A [adapted from 35]. Abbreviations: aI/fO, anterior insula/frontal operculum; aPFC, anterior prefrontal cortex; dACC/msFC, dorsal anterior cingulate cortex/medial superior frontal cortex; dlPFC, dorsolateral prefrontal cortex; dF, dorsal frontal; IPL, inferior parietal lobule; IPS, intraparietal sulcus.

direction (congruent) or in the opposite direction (incongruent). Subtracting RTs for congruent from incongruent target trials provides a measure of conict resolution and assesses the eciency of the executive attention network. Subtracting RTs obtained in the double-cue condition from RT in the no-cue condition gives a measure of alerting due to the presence of a warning signal. Subtracting RTs to targets at the cued location (spatial cue condition) from trials using a central cue gives a measure of orienting, since the spatial cue, but not the central cue, provides valid information on where a target will occur. 2.1 Network connectivity Neural areas found active in studies of functional anatomy must be orchestrated in carrying out any real task. One approach to studying this connectivity uses fMRI to study the correlations between active areas [34]. An important nding was that even at rest, common brain areas appear to be active together (default state). Studies suggest that the connectivity between these areas change over the course

of development. Figure 2 illustrates two sets of connections active at rest: these are a frontoparietal (related to orienting) and a cinguloopercular (related to executive attention) network. Connections change over the life span. Children show many shorter connections and integration of the dorsal anterior cingulate in both networks. Adults show more segregation of the two networks and longer connections. Some of the same brain areas found active during rest change when the person is given a task. For example, while the organization of anatomical areas in alerting and orienting are not fully known, some promising beginnings have taken place. In alerting, the source of the attention appears in the locus coeruleus (lc). Cells in the lc have two modes of processing. One mode is sustained and is perhaps related to the tonic level of alertness over long time intervals. This function is known to involve the right cerebral hemisphere more strongly than the left [36]. Alertness is inuenced by sensory events and by the diurnal rhythm. However, its voluntary maintenance during task performance may be orchestrated from the anterior cingulate [37]. More phasic shifts of alerting can result from

COGNITIVE NEUROSCIENCE: DEVELOPMENT AND PROSPECTS

423

presenting any environmental signal. However, if the signal is likely to warn about an impending target, this shift results in a characteristic suppression of the intrinsic brain rhythms (e.g. alpha) within a few tens of milliseconds and a strong negative wave is (contingent negative variation) recorded from surface electrodes and that moves from a frontal generator towards the sensory areas of the hemisphere opposite the expected target. An analysis of a number of conict tasks shows that the more dorsal part of the anterior cingulate is involved in the regulation of cognitive tasks, while the more ventral part of the cingulate is involved in regulation of emotion [38]. The microstructure of the organization of the cingulate has been studied in detail using correlations between brain areas while the subject rested (functional connectivity). Dynamic casual modeling (DCM) can then be used to help dene the direction of information within the network. [39]. These data are related to an analysis of white matter connections in humans using diusion tensor imaging to trace physical connections and relate them to similar studies in animals [40]. These converging methods are being applied to trace the connectivity of brain networks both at rest and during task performance [33]. 2.2 Mental chronometry Although the fMRI method has become increasingly useful in understanding the sequence of mental operations, their speed may be too fast for analysis by the relatively slow changes involved in imaging based on vascular changes. Another way to examine network activity is to use scalp EEG electrodes to record neural activity synchronized in dierent frequency bands. This method can be used to separate rapid temporal events, for example, it can separate the cue eects from the target eects in the ANT. There is increasing interest in the various frequencies of electrical activity involved in correlations between neural areas. For example, in one study using the ANT [41], a spatial cue indicating the location of the target produced increased high frequency gamma activity (above 30 Hz) about 200 milliseconds after the cue presentation. When the cue brought attention to the target location, gamma activity was found following the cue, but not following the target. When the cue indicated the center location so that a shift of attention was needed following the target, the gamma activity was present following the target. These data suggest that gamma activity is associated with orienting attention. It may occur 200 millisecs after the cue or only after the target depending upon when attention shifts. Taken together, the fMRI, EEG and DTI methods

can provide a more detailed account of the orchestration of neural networks related to attention or to other networks illustrated in table 1. The activation of brain networks does not mean that all parts of the network are needed to carry out the task. In the past, eects of brain lesions have been a primary way to indicate brain areas which when lost will prevent the persons from carrying out certain tasks. For example, damage to areas of the right parietal lobe have led to neglect of the left side of space in multiple sensory systems. It is now possible to use brief magnetic pulses applied to the scalp overlying the brain area of interest (transcortical magnetic stimulation TMS) to disrupt parts of the network at particular times to observe its inuences on task performance [42]. In this way, one can determine what parts of the network are necessary for task performance. Taken together, the methods reviewed above have provided tools for probing human brain networks. In the case of attention, there has been increasing coordination of these studies with animals studies that allow probing of single units and examination of the micro-circuitry and molecular events related to these activations. 3. Language In the 1970s, behavioral studies using habituation to a repeated stimulus provided evidence that from birth, infants are able to discriminate basic phonemic units not only in their own language but also in other languages to which they have never been exposed [43]. Studies using these behavioral methods together with electrical recording from the scalp have probed some of the early development of the phonemic structure underlying language. Recently infants have been exposed to language while resting in fMRI scanners to examine the brain mechanisms activated by language [44]. The infant language system appears to involve the same left hemisphere language structures found in adults [44]. In one study, infants listened to sentences presented aurally in their language. Brain areas in the superior temporal lobe (Wernickes area) and in Brocas area were active. When the same sentence was presented after a delay of 14 seconds, Brocas area activity increased, as though this area was involved in the memory trace. It has long been supposed that the early acquisition of language might involve very dierent mechanisms than are active in adults [45]. Left hemisphere lesions in infancy do not produce a permanent loss of language function as they might do in adults. Nonetheless, the new fMRI data suggests the left hemisphere speech areas are involved in receptive language in infancy.

424 3.1 Phonemes

MICHAEL I POSNER AND SHOBINI RAO

3.2 Reading Reading is a high level skill and in alphabetic languages such as English, it has properties related to the phonemic structure of language. There have been many studies of adult reading and much more is known than can be reviewed here [see 25,56 for reviews]. The childs ability in phonemic awareness (e.g. rhyming of auditory words), is a good predictor of their being able to learn to read alphabetic languages such as English. Adult studies of reading have revealed a complex neural network involved in the translation of words into meaning. Two important nodes of this network are the visual word form area, of the left fusiform gyrus and an area of the left temporal-parietal junction for translating visual letters into sounds. The visual word form area is involved in integrating or chunking visual letters into units of words [57]. The visual word form has been localized to the fusiform gyrus of the left hemispheres visual system. Although there has been some dispute about the operation it appears to be a part of the visual system that becomes expert in dealing with letters as reading skill develops in later childhood. It is thought that without the functioning of this area, reading cannot become uent. For example, a patient with a lesion that interrupted the ow of information from the right hemisphere to the visual word form area used letter by letter reading when words were presented left of xation (going to the right hemisphere), while they read words normally when the word was projected to the left hemisphere and thus reached the visual word form area [58]. Children from 7 to 18 who were decient in reading skills failed to activate this area, but were able to do so after extensive training [59]. Languages like English that are highly irregular at the visual level are heavily dependent upon brain areas that translate visual words to sound. These areas are at the temporal parietal boundary of the left hemisphere. Children who have diculty in learning to read show little activation in these phonological areas. However, phonics training even after 20 sessions can produce relatively normal activity in these areas and also improve reading by several grade levels. The time course of development of the visual word form area in English is important for the development of uent reading. Phonics training often leaves the child with improved decoding skill, but with a lack of uency. Evidence that the visual word form develops rather late and rst only for words with which the child is familiar [60], suggests the importance of continuous practice in reading to develop uency [59]. More research is needed on the

It has been possible to study changes in phonemic discrimination due to exposure to the native language at least by ten months of age [46,47]. Infants appear to acquire a sharpened representation of the native phonemic distinctions [48]. During the same period they also lose the ability to distinguish representations not made in their own language [49]. At least a part of the loss occurs when the nonnative language requires a distinction that is within a single phonemic category in the native language. An example is the ra-la distinction important in English is lost because it is within a single category in Japanese [50]. It is as though Japanese no longer hear this distinction and even when exposed to English they may not improve in distinguishing ra from la. Training by several methods [50,51] seems to improve this ability even in adults, although it is not known how well this knowledge can be incorporated into normal daily life communication. It might be useful to nd a way that will preserve the distinctions originally made for the nonnative language during infancy. One study showed that 12 sessions of exposure to a mandarin speaker during the rst year was eective [52]. A similar amount of exposure to a computerized version of the speaker was not eective, suggesting the importance of social interaction in this early form of learning. More needs to be learned about how and whether media presentation can be eective in learning. There is also some reason to believe that the process of phonemic discrimination being developed in infancy is important for later ecient use of spoken and written language [53,54]. Electrical recording taken in infancy during the course of phonemic distinctions [53,54] have been useful in predicting later diculties in language and reading. There is a history of using event related potentials to assess infant deafness early in life and being able to do so reliably has been very useful in the development of sign language and other interventions to hasten the infants ability at communication. Perhaps a similar role will prove to be possible for ERPs in the development of methods to insure a successful phonemic structure in the native language. There have been eorts to develop appropriate intervention in later childhood for diculties in the use of language and reading such as the widely used FASTFORWARD programs [http://www.scilearn.com, 55]. Although there are disputes about exactly why and for what populations this method works, it remains important to develop remedies for language diculties based upon research.

COGNITIVE NEUROSCIENCE: DEVELOPMENT AND PROSPECTS

425

best methods for developing uency, particularly in non-alphabetic languages. The multiple languages in India makes of particular importance ndings suggesting superior performance of multi-linguals on executive attention scores, such as the Attention Network Test (ANT) [61,62]. The ties to attention have been supported by the use of ANT to assess dierences in executive attention between mono and bilinguals. A study of Spanish English bilinguals found strong advantages over monolingual controls in a wide variety of executive tasks [63]. One study [64] compared Korean and Chinese native speakers who were bilingual in English with French and Spanish speakers bilingual in English and with English monolinguals. Both bilingual groups showed better executive attention than monolinguals and, the Asian group, whose languages diered the most from English, were superior to the Romance bilinguals. This study shows the close ties between language and attention. It also suggests that the need to select among languages when speaking, as occurs in multilinguals may form one important basis for training improved executive attention. 4. Consciousness There is a long tradition of the study of consciousness within Eastern and Western philosophy, however, cognitive neuroscience provides a somewhat new perspective on awareness and will, both of which have been central to the discussions of consciousness. An important distinction in studies of awareness [65] is between general knowledge of our environment (ambient awareness) and detailed focal knowledge of a scene (focal awareness). As lay people we generally believe that we have full conscious awareness of our environment, even when our focal attention is upon our own internal thoughts. Experimental studies [66] show us how much this opinion is in error. In the study of change blindness when cues such as motion, that normally lead to a shift of attention, are suppressed, we have only a small focus for which we have full knowledge and even major semantic changes in the remainder of the environment are not reported. Change blindness is closely related to studies of visual search which have been prominent in the eld of attention and involve an interaction between information in the ventral visual pathway about the object identity and information in the dorsal visual pathway that controls orienting to sensory information (for a review, see [67]). Visual search tasks have been important for examining what constraints attention provides to the nature

of our awareness of a target event. There is clear evidence that attention to a visual event increases the brain activity associated with it. Most evidence arises from studies using event related electrical potentials with visual stimuli and these have clearly shown that early sensory components of the visual evoked potential P1 and N2 (80150 millisec) are enlarged by the presence of attention [28]. As shown in gure 1 focal attention to the target of a visual search appears to involve a brain network that includes the anterior cingulate and lateral prefrontal areas [33]. Humans have a conviction of conscious control that allows us to regulate our thoughts, feelings and behaviors in accord with our goals and people believe that voluntary conscious choice guides at least a part of the action we take. These beliefs have been studied under various names in dierent elds of psychology. In cognition, cognitive control is the usual name for the voluntary exercise of intentions, while in developmental psychology many of the same issues are studied under the name self-regulation [68]. Imaging studies suggest that whenever we bring to mind information, whether extracted from sensory input or from memory, we activate the executive attention network. In some studies a whole set of frontal areas become activated together forming what is called a global workspace [69]. This global workspace becomes active about 300 millisecs after a target event is presented. It provides the neural basis for the set of information on which a person is currently working in the process of problem solving. The distinction between awareness and control (will) is traditional in studies of consciousness. However, one form of awareness, focal awareness, appears to involve the same underlying mechanism as involved in control. In this sense, even though some forms of consciousness (e.g. ambient awareness) may have diverse sources within sensory specic cortex, there is also a degree of unity of the underlying mechanisms involving focal awareness and conscious control. The distinction between focal and ambient factors in consciousness may help to clarify the sense of awareness that can be present even when detailed accounts of the scene are not possible as in change blindness [66]. 5. Genes and experience build networks The common nature of brain networks such as those in table 1 among people argue strongly for the role of genes in their construction. This has led cognitive neuroscience to incorporate data from the growing eld of human genetics. One method for doing this relates individual dierences to dierent forms of genes (genetic alleles) that

426

MICHAEL I POSNER AND SHOBINI RAO

Figure 3. A strategy for relating brain networks to underlying molecular events [adapted from 70]. Bottom of gure are psychological functions, these are related to neural networks as shown in brain images and then to dierences in protein conguration and genetic variation.

may relate to them. Brain activity can serve as an intermediate level for relating genes to behavior as illustrated in gure 3. As one example, the Attention Network Test (ANT) was used to examine individual dierences in the eciency of executive attention. Strong heritability of the executive network supported the search for genes related to individual dierences in network eciently. The association of the executive network with the neuromodulator dopamine is a way of searching for candidate genes that might relate to the eciency of the network [71]. For example, several studies employing conict related tasks, found that alleles of the catechol-o-methyl transferase (COMT) gene were related to the ability to resolve conict. A number of other dopamine genes have also proven related to this form of attention, in addition, research has suggested that genes related to serotonin transmission also inuence executive attention ([72] for a summary). It was also possible to show that some of these genetic dierences inuenced the degree to which the anterior cingulate was activated during task performance in studies using brain imaging. In the future, as suggested by gure 3, it may be possible to relate genes to specic nodes within neural networks, allowing a much more detailed understanding of the origins of brain networks. While genes are important for common neural networks and individual dierences in eciency there is also an important role for specic

experiences. For example, several genes including the DRD4 gene and the COMT gene have both shown to interact with aspects of quality of parenting. This provides evidence that aspects of the culture in which children are raised can inuence the way in which genes shape neural networks inuencing child behavior. If brain networks are inuenced by parenting and other culture inuences, it should be possible to develop specic training methods that can be used to inuence underlying brain networks. For example, one study tested the eect of training during the period of major development of executive attention, which takes place between 4 to 7 years of age [73]. An improvement in conict resolution as measured by the ANT was found in trained children, along with changes in the underlying network and generalization to other aspects of cognition. EEG data showed clear evidence of improvement in network eciency in resolving conict following training. The N2 component of the scalp recorded ERP has been shown to arise in the anterior cingulate and is related to the resolution of conict [74]. The N2 dierences between congruent and incongruent trials of the ANT were found in trained six year olds, that resembled dierences found in adults. No such N2 dierence was found in the untrained controls. These data suggest that training altered the network for the resolution of conict in the direction of being more like what is found in adults. There was also a greater improvement in intelligence in the trained group compared

COGNITIVE NEUROSCIENCE: DEVELOPMENT AND PROSPECTS

427

to the control children. This nding suggested that training eects had generalized to a measure of cognitive processing that is far removed from the training exercises. Given the wide range of individual dierences in the eciency of attention, it is expected that attention training could be especially benecial for those children with poorer initial eciency. These could be children with pathologies that involve attentional networks, children with genetic backgrounds associated with poorer attentional performance, or children raised in dierent degrees of deprivation. 6. Brain plasticity Training studies discussed in the last section show evidence that performance of networks can be altered by experience. This idea has been important in rehabilitation in cases of brain injury and psychiatric disorders. It has also led to suggestions for improved education for normal people based on ideas coming from cognitive neuroscience. Brain networks are frequently damaged by neurological or psychiatric disorders and head trauma. Neuro-imaging has made it easier to determine the extent and location of brain damage, so a major area of application of cognitive studies is to the treatment of such disorders. The National Institute of Mental Health and Neuroscience (NIMHANS) in Bangalore has been a pioneer in these eorts. The NIMHANS neuropsychology battery consists of 21 neuropsychological tests. This battery was administered to 540 normal males and females, aged between 1665 years, including illiterates and literates whose educational levels diered widely [75]. There have been a series of studies using this battery before and after training. Improved basic cognitive functions were found using both hospital based computerized and home based paper and pencil training [76,77]. A number of randomized studies have also shown improvements in attention and memory in normal children and adults [7881]. All these methods involve practice in some cognitive skills including repetitive trials on tasks similar to what might be tested in schools or cognitive laboratories. All involve a training group showing signicantly more improvement than control groups. The control groups often involve the same number of laboratory sessions with training or experiences not thought to involve the elements of attention working memory used in the experimental group. These methods dier considerably from mindfulness training, exposure to nature settings or integrative body-mind training (IBMT), all of which also aect the state of attention. Recently, both IBMT (using body-mind optimization) [82], and

nature exposure [83] (using attention restoration theory) have conducted similar randomized design using similar before and after assays as those used in attention training. These attention state methods appear to achieve similar success to attention training in the attention network test and also help control of stress, improve self-regulation, etc. These two training streams represent the dierent traditions/cultures and methodologies in the East (IBMT, mindfulness) and West (practice and brain plasticity). These techniques provide support for the idea that training can provide general changes in brain networks that can lead to widespread improvement in cognitive processes. Since it is likely that the various methods activate dierent brain networks, imaging may be used to combine methods to produce improved eectiveness. 7. Future developments Cognitive neuroscience has developed a number of methods than in concert can link important aspects of human behavior to underlying neural networks and to the cellular and molecular levels that underlie them. It is likely that new methods of imaging will eventually provide more details about the functioning of the human brain when active and in the resting state. One of the major accomplishments of cognitive neuroscience is to attract the attention of physical and mathematical scientists who are capable of contributing to imaging apparatus and new algorithms. Imaging and cognitive theory may also contribute to new forms of compensation for brain injuries and pathologies. Already deep brain stimulation, guided by imaging theories have been used in treatments of depression [84] and in an eort to improve the integrative behavior for patients in vegetative neurological states [85]. We can also expect additional training and state change methods, guided by imaging of what specic brain network(s) are inuenced by the method, designed to assist people with brain injuries or other forms of pathology. Interfaces between humans, computers and prosthetic devices are playing an important role in extending the sensory and motor capacities of patients of all kinds [86]. A better understanding of the neuronal networks related to human capacities could lead us toward improvements in these methods and the development of methods to extend the cognitive range of normal people in directions of human improvement. This chapter is a brief overview of the achievements and potential of cognitive neuroscience during the twenty years of its existence as a scientic discipline. In addition to the technical

428

MICHAEL I POSNER AND SHOBINI RAO

ndings discussed in this review, imaging studies are having a wide inuence in the general culture by giving people who read or watch television, a picture of brain activity during many human tasks and situations. As a result, there is likely to be a large continuing interest in the application of cognitive neuroscience perspective to many societal issues. References
[1] Finger S 1994 Origins of neuroscience: A history of explorations into brain function (Oxford: Oxford University Press). [2] Newell A, Shaw J C and Simon H A 1958 Psychol. Rev. 65 151. [3] Shannon C E and Weaver W 1949 The mathematical theory of communication (Urbana: University of Illinois Press). [4] Hick W E 1948 Q. J. Exp. Psychol. 4 11. [5] Hyman R 1953 J. Exp. Psychol. 45 188. [6] Neisser U 1967 Cognitive psychology (Englewood Clis: Prentice Hall). [7] Posner M I (ed.) 1989 Foundations of cognitive science (Cambridge: Bradford Books/MIT Press). [8] Hebb D O 1949 Organization of behavior (New York: John Wiley & Sons). [9] Posner M I and Raichle M E 1996 Images of mind (New York: Scientic American Library). [10] Lashley K S 1929 Brain mechanisms of intelligence (Chicago: University of Chicago Press). [11] Kohler W 1947 Gestalt psychology (New York: Liveright). [12] Dehaene S 1997 The number sense (Oxford: Oxford University Press). [13] Fink G R, Markowitsch H J, Reinkemeier H, Bruckbauer T, Kessler J and Heiss W D 1996 J. Neurosci. 16 4275. [14] Ochsner K N, Ludlow D H, Knierim K, Hanelin J, Ramachandran T, Glover G C and Mackey S C 2006 Pain 129 69. [15] Haxby J V 2004 In: Functional neuroimaging of visual cognition: Attention and performance XX (eds) Kanwisher N and Duncan J (Oxford: Oxford University Press) 83. [16] Levitin D J 2006 This is your brain on music (London, U.K.: Dutton [Penguin Group]). [17] Grill-Spector K 2004 In: Functional neuroimaging of visual cognition: Attention and performance XX (eds) Kanwisher N and Duncan J (Oxford: Oxford University Press) 169. [18] Knutson B, Fong G W, Bennett S M, Adams C M and Homme D 2003 Neuroimage 18 263. [19] Johnson S C, Schmitz T W, Kawahara-Baccus T N, Rowley H A, Alexander A L, Lee J H and Davidson R J 2005 J. Cognitive Neurosci. 17 1897. [20] Shelton A L and Gabrieli J D E 2002 J. Neurosci. 22 2711. [21] Smith E E, Jonides J, Marshuetz G and Koeppe R A 1998 Proc. Natl. Acad. Sci. USA 95 876. [22] Ungerleider L G, Courtney S M and Haxby H V 1998 Proc. Natl. Acad. Sci. USA 95 883. [23] Luria A R 1973 The working brain (New York: Basic Book).

[24] Dehaene S and Cohen L 2007 Neuron 56 384. [25] Posner M I and Rothbart M K 2007 Educating the human brain (Washington: American Psychological Association Books). [26] Corbetta M and Shulman G L 2002 Nat. Rev. Neurosci. 3 201. [27] Busse L, Katzner S and Treue S 2008 Proc. Natl. Acad. Sci. USA 105 16380. [28] Hillyard S A, DiRusso F and Martinez A 2003 In: Functional neuroimaging of visual cognition (eds) Kanwisher N and Duncan J (Oxford: Oxford University Press) 381. [29] Fan J, McCandliss B D, Fossella J, Flombaum J I and Posner M I 1990 Ann. Rev. Neurosci. 13 25. [30] Fan J, McCandliss B D, Fossella J, Flombaum J I and Posner M I 2005 Neuroimage 26 471. [31] Crottaz-Herbtte S and Mennon V 2006 J. Cognitive Neurosci. 18 766. [32] Etkin A, Egner T, Peraza D M, Kandel E R and Hirsch J 2006 Neuron 51 871. [33] Fan J, McCandliss B D, Sommer T, Raz M and Posner M I 2002 J. Cognitive Neurosci. 3 340. [34] Dosenbach N U F, Fair D A, Cohen A L, Schlaggar B L and Petersen S E 2008 Trends Cogn. Sci. 123 99. [35] Fair D A, Dosenbach N U F, Church J A, Cohen A L, Brahmbhatt S, Miezin F M, Barch D M, Raichle M E, Petersen S E and Schlaggar B L 2007 Proc. Natl. Acad. Sci. USA 104 13507. [36] Coull J T, Frith C D, Buchel C and Nobre A C 2000 Neuropsychologia 38 808. [37] Mottaghy F M, Willmes K, Horwitz B, Muller H W, Krause B J and Sturm W 2006 Neuroimage 29 225. [38] Bush G, Luu P and Posner M I 2000 Trends Cogn. Sci. 4 215. [39] Friston K J, Harrison L and Penny W 2003 Neuroimage 19 1273. [40] Posner M I, Sheese B, Odludas Y and Tang Y 2006 Neural Networks 19 1422. [41] Fan J, Byrne J, Worden M S, Guise K G, McCandliss B D, Fossella J and Posner M I 2007 J. Neurosci. 27 6197. [42] Schenkluhn B, Ru C C, Heinen K and Chalmers C D 2008 J. Neurosci. 28 11106. [43] Eimas P D, Siqueland E R, Jusczyk P and Vigorito J 1971 Science 171 303. [44] Dehaene-Lambertz G, Hertz-Pannier L, Dubois J, Meriaux S, Roche A, Siman M and Dehaene S 2006 Proc. Natl. Acad. Sci. USA 103 14240. [45] Vicari S, Albertoni A, Chilosi A M, Cipriani P, Cioni G and Bates E 2000 Cortex 36 31. [46] Kuhl P K 1994 Curr. Opin. Neurobiol. 4 812. [47] Saran J R 2002 J. Mem. Lang. 47 172. [48] Kuhl P K, Stevens E, Hayashi A, Deguchi T, Kiritani S and Iverson P 2006 Developmental Sci. 9 F13. [49] Werker J F, Gilbert J H V, Humphrey K and Tees R C 1981 Child Dev. 52 349. [50] McClelland J L, Fiez J A and McCandliss B D 2002 Physiol. and Behav. 77 657. [51] Iverson P, Hazan V and Bannister K 2005 J. Acoust. Soc. Am. 188 3267. [52] Kuhl P K, Tsao F M and Liu H M 2003 Proc. Natl. Acad. Sci. USA 100 9096. [53] Guttorm T K, Leppanen P H T, Poikkeus A M, Eklund K M, Lyytinen P and Lyytinen H 2005 Cortex 41 291. [54] Molfese D L 2000 Brain Lang. 72 238.

COGNITIVE NEUROSCIENCE: DEVELOPMENT AND PROSPECTS

429

[55] Temple E, Deutsch G K, Poldrack R A, Miller S L, Tallal P, Merzenich M M and Gabrielli J D E 2003 Proc. Natl. Acad. Sci. USA 100 2860. [56] Schlaggar B L and McCandliss B D 2007 Ann. Rev. Neurosci. 30 475. [57] McCandliss B D, Cohen L and Dehaene S 2003 Trends Cogn. Sci. 7 293. [58] Cohen L, Henry C, Dehaene S, Martinaud O, Lehericy S, Lemer C and Ferrieux S 2004 Neuropsychologia 42 1768. [59] Shaywitz B A, Skudlarski P, Holahan J M, Marchione R N, Constable R T, Fullbright R K, Zelterman D, Lacadie B S and Shaywitz S E 2007 Ann. of Neurology 61 363. [60] Posner M I and McCandliss B D 1999 In: Converging methods for understanding reading and dyslexia (eds) Klein R and McMullen P (Cambridge: MIT Press) 305. [61] Bialystok E 2006 In: Encyclopedia on early childhood development (eds) Tremblay R E, Barr R G and Peters R De V (Montreal: Centre of Excellence for Early Childhood Development) 1. [62] Bialystok E and Martin M M 2004 Psychol. Sci. 7 325. [63] Carlson S M and Meltzo A N 2008 Developmental Sci. 11 282. [64] Yang S and Lust B Boston Conference on Lang. Dev. 31 (in press). [65] Iwasaki S 1993 Cognition 49 211. [66] Rensink R A, ORegan J K and Clark J J 1997 Psych. Science 8 368. [67] Driver J, Eimer M and Macaluso E 2004 In: Attention and performance XX: Functional brain imaging of visual cognition (eds) Kanwisher N and Duncan J (Oxford: Oxford University Press) 267. [68] Posner M I, Rothbart M K, Sheese B E and Tang Y 2007 J. of Cog., Aective and Soc. Neurosci. 7 391. [69] Dehaene S 2004 In: Attention and performance XX: Functional brain imaging of visual cognition (eds)

[70] [71]

[72] [73]

[74] [75] [76] [77]

[78] [79] [80] [81]

[82]

[83] [84] [85] [86]

Kanwisher N and Duncan J (Oxford: Oxford University Press) 205. Green A E, Munafo M R, DeYoung C G, Fossella J A, Fan J and Gray J R 2008 Nature Neurosci. Rev. 9 710. Fossella J, Sommer T, Fan J, Wu Y, Swanson J M, Pfa D W and Posner M I 2002 BMC Neurosci. 3 14. Posner M I, Rothbart M K and Sheese B E 2007 Developmental Sci. 10 24. Rueda M R, Fan J, Halparin J, Gruber D, Lercari L P, McCandliss B D and Posner M I 2004 Neuropsychologia 42 1029. van Veen V and Carter C S 2002 J. Cognitive Neurosci. 14 593. Rao S L, Gangadhar B N and Hegde A S 1985 NIMHANS J. 3 151. Rao S L 1998 Q. J. Surgical Sci. 34 47. Rao S L Cognitive neuroscience and neuropsychology. Status report on neurosciences in India, Indian Academy of Sciences. Diamond A, Barnett S, Thomas J and Munro S 2007 Science 318 1387. Green C S and Bavelier D 2003 Nature 423 434. Jaeggi S M, Buschkuehl M, Jonides J and Perrig W J 2008 Proc. Natl. Acad. Sci. USA 105 6829. Rueda M R, Rothbart M K, McCandliss B D, Saccamanno L and Posner M I 2005 Proc. Natl. Acad. Sci. USA 102 14931. Tang Y Y, Ma Y, Wang J, Fan Y, Feng S, Lu Q, Yu Q, Sui D, Rothbart M K, Fan M and Posner M I 2007 Proc. Natl. Acad. Sci. USA 104 17152. Berman M, Jonides J and Kaplan S 2008 Psychol. Sci. 19 1207. Krishnan V and Nestler E J 2008 Nature 455 894. Giacino J T, Hirsch J, Schi N and Laureys S 2008 Arch. Phys. Med. Rehab. 87 S67. deCharms R C 2008 Nat. Neurosci. Rev. 9 720.