《认知神经科学》课程教学资源（学习资料）书籍《认知神经科学神经学基础》PDF电子版（Neurological Foundations of Cognitive Neuroscience，edited by Mark D'Esposito）

Neglect 13 the anterior cingulate were also active (BA 24).but tied to distinct sensory and motor systems.To not consistently address this question,Wojciulik and Kanwisher Nohre and colleagues (obre Sebestver (1999)used fMRI in three different tasks of visual Gitelman,Mesulam,1997).also using PET. attention.These tasks involved shifting attention found that an exogenous shift in attention wa matching obiects in different locations.and con associated with activation around the intraparieta joining visual features of an object at a specific loca sulcus.Taking advantage of fMRI's bette spatial tion.They found that the intraparietal sulcus wa resolution Corbetta and colleagues (Corb ett activated in all three tasks.While one cannot pr 1998)confirmed activation of the intr the null hypothesis that the intraparietal sul sulcus as well as the postce and pre involved in all attenti suleus with shifts of attention This act this area might mediate a eneral atte found even when explicit motor r selection module.Similarly.Coull and Frith (1998) ired that n be ged without motor in a PET study found that while the on The tial than prep. etal lobule s in the r 么 ietal sulcus nd precentral e whe direc nost striking aspect of neglect synd rather thar just shifting to a of contra solater in which ortex is also acti that spac into a d The stream (Ungerleider 982 Th am proc the spa the loca of objec s of inter n of eye mover ent might be dorsolatera y to ic corte and co-worke 1999)showed tha ormation to be aware of both the "where"and areas on attentional tasks is what”of objects probably not due to these processes However Attention modulates the activity of neural struc the studies did not completely control for tures in the ventral stream dedicated to identify- movements.which could be contributing to these ing objects.Patients with prefrontal damage are activations.Nonetheless. given that dorsolatera impaired in discriminating contralesional visual prefrontal cortex lesions also produce disorders of targets.This impairment is associated with dimin- attention.it is likely that these areas are linked to ished event-related potentials at 125 ms and lasting the posterior parietal regions involved in directing for another 500ms (Barcelo,Suwazono,Knight, spatial attention. 2000).These event-related potentials are linked to extrastriate processing.which is associated with Supramodal,Space-Based,and Object-Based tonic activation as well as the selection of features Attention and the postselection analyses of objects. The earliest point in visual processing at which A long-standing question about the organization attentional modulation can occur is not clear of attention is whether there is a supramodal all- Several studies suggest that the primary visual oaa“ohm cortex might be modulated attention (Brefczynski DeYoe,1999:Gandhi,Heeger

the anterior cingulate were also active (BA 24), but not consistently. Nobre and colleagues (Nobre, Sebestyen, Gitelman, & Mesulam, 1997), also using PET, found that an exogenous shift in attention was associated with activation around the intraparietal sulcus. Taking advantage of fMRI’s better spatial resolution, Corbetta and colleagues (Corbetta, 1998) confirmed activation of the intraparietal sulcus as well as the postcentral and precentral sulcus with shifts of attention. This activation was found even when explicit motor responses were not required, suggesting that these areas can be atten￾tionally engaged without motor preparation. They also found similar blood flow increases in the right intraparietal sulcus and precentral cortex when attention was directed at a peripheral location in a sustained manner, rather than just shifting to a peripheral location. The dorsolateral prefrontal cortex is also acti￾vated in most studies in which visual attention is shifted to different locations. These activations seem to center around the frontal eye fields (BA 6/8) and the adjacent areas. Working memory or in￾hibition of eye movement might be associated with dorsolateral prefrontal cortex activity. Gitelman and co-workers (Gitelman et al., 1999) showed that activation of these areas on attentional tasks is probably not due to these processes. However, the studies did not completely control for eye movements, which could be contributing to these activations. Nonetheless, given that dorsolateral prefrontal cortex lesions also produce disorders of attention, it is likely that these areas are linked to the posterior parietal regions involved in directing spatial attention. Supramodal, Space-Based, and Object-Based Attention A long-standing question about the organization of attention is whether there is a supramodal all￾purpose attention module, or whether attention is better viewed as a collection of different modules tied to distinct sensory and motor systems. To address this question, Wojciulik and Kanwisher (1999) used fMRI in three different tasks of visual attention. These tasks involved shifting attention, matching objects in different locations, and con￾joining visual features of an object at a specific loca￾tion. They found that the intraparietal sulcus was activated in all three tasks. While one cannot prove the null hypothesis that the intraparietal sulcus is involved in all attentional tasks, they suggest that this area might mediate a general attention and selection module. Similarly, Coull and Frith (1998) in a PET study found that while the superior pari￾etal lobule was more responsive to spatial than nonspatial attention, the intraparietal sulcus was responsive to both. The most striking aspect of neglect syndromes is that patients are unaware of contralesional space and of objects that inhabit that space. A central tenet of visual neuroscience is the relative segregation of visual information into a dorsal “where” stream and a ventral “what” stream (Ungerleider & Mishkin, 1982). The dorsal stream processes the spatial locations of objects of interest, whereas the ventral stream processes features necessary to identify the object. Somehow humans integrate these streams of information to be aware of both the “where” and “what” of objects. Attention modulates the activity of neural struc￾tures in the ventral stream dedicated to identify￾ing objects. Patients with prefrontal damage are impaired in discriminating contralesional visual targets. This impairment is associated with dimin￾ished event-related potentials at 125ms and lasting for another 500ms (Barcelo, Suwazono, & Knight, 2000). These event-related potentials are linked to extrastriate processing, which is associated with tonic activation as well as the selection of features and the postselection analyses of objects. The earliest point in visual processing at which attentional modulation can occur is not clear. Several studies suggest that the primary visual cortex might be modulated by attention (Brefczynski & DeYoe, 1999; Gandhi, Heeger, & Neglect 13

Anian chatteriee Boynton, 1999:Somme Dale Seiffert functional anatomical relationshin They found that Tootell.1999).However.Martine and colleague attention directed in space activated the right (Martnine? et al 1999)us data fron ontal (BA 9)and inferior te ral-occip tal related pote tials point out tha whereas attenti directed at tion in the visual tem is evident only after 70-75 istriate Since the initia th e primar the e(). nd ypes of pa t latera a cortex 9).and h beha cereb ella vermis wha Ac in structure Animal Studies of Spatial attention and ng is emanding Representation activity motion area whe moving st muli are irrel evant to the task Animal studies offer insight into mechanisms of (Rees Frith,& Lavie,1997).Baselin spatial attention that a sily by study ade w early visual processing areas can also be mod Lesioaenotobtai mals can be n ulated by attentional sets.Normally,stimuli sup ons ined press the processing of other stimuli located in close proximity.Similarly.subjects instructed to atten my (in the of stroke) athe to color have increased activity in color areas (V4 nal an and when asked to attend to motion have increase activity in motion areas (V5).even when the stimul single the activi neurons, themselves are not colored or moving (Chawla roimaging offers insigh nto the Rees,Friston,1999).Kastner and colleagues neurophysiology at the level of neural networks (Kastner.De Weerd.Desimone.Ungerleider 1998)showed that fMRI activation of areas within Lesion Studies occipitotemporal regions is associated with this normal suppression.This suppression.however diminishes when spatial attention is directed to locations encompassing both stimuli.suggesting an overall enhancement of processing of stimuli in In rodents ol the postenor parietal those areas onta cortex (medial agranular cortex)or the exnerimental paradigm and dorsolateral striatum produce a syndrome simila to neglect (Burcham.Corwin.Stoll.Reep.1997 spatial attention are still being worked out.In this Corwin Reep,1998).These rodents are more early stage of the field's development.some find likely to orient ipsilesionally than contralesionally ings are difficult to reconcile with the rest of the to visual.tactile.or auditory stimuli.This orienta- literature.One might reasonably surmise that the tion bias recovers to a considerable degree over days to weeks.Dopamine antagonists impede spon and the occipital and temporal cortices taneous recovery and dopamine agonists enhance nd obiects Ho recovery (Corwin et al.,1986).probably by influ in a PET study.Fink and colleas (Fink.Dolan encing striatal function(Vargo,Richard-Smith,& Halligan,Marshall,&Frith,1997)did not find this Corwin.1989)

Boynton, 1999; Sommers, Dale, Seiffert, & Tootell, 1999). However, Martinez and colleagues (Martninez et al., 1999) using data from event￾related potentials point out that attentional modula￾tion in the visual system is evident only after 70–75 ms. Since the initial sensory input to the primary visual cortex occurs at about 50–55ms, they suggest that primary visual activation may be due to feed￾back activity rather than attentional modulation. The behavioral significance of such feedback, if that is what is being observed, remains to be explored. Activity in neural structures downstream in ventral visual processing is clearly modulated by attention. Cognitively demanding tasks can inhibit activity in visual motion areas, even when the moving stimuli are irrelevant to the task at hand (Rees, Frith, & Lavie, 1997). Baseline activity in these early visual processing areas can also be mod￾ulated by attentional sets. Normally, stimuli sup￾press the processing of other stimuli located in close proximity. Similarly, subjects instructed to attend to color have increased activity in color areas (V4) and when asked to attend to motion have increased activity in motion areas (V5), even when the stimuli themselves are not colored or moving (Chawla, Rees, & Friston, 1999). Kastner and colleagues (Kastner, De Weerd, Desimone, & Ungerleider, 1998) showed that fMRI activation of areas within occipitotemporal regions is associated with this normal suppression. This suppression, however, diminishes when spatial attention is directed to locations encompassing both stimuli, suggesting an overall enhancement of processing of stimuli in those areas. The appropriate experimental paradigms and methods of analysis in functional imaging studies of spatial attention are still being worked out. In this early stage of the field’s development, some find￾ings are difficult to reconcile with the rest of the literature. One might reasonably surmise that the parietal cortex mediates attention directed in space and the occipital and temporal cortices mediate attention directed to features and objects. However, in a PET study, Fink and colleagues (Fink, Dolan, Halligan, Marshall, & Frith, 1997) did not find this functional anatomical relationship. They found that attention directed in space activated the right prefrontal (BA 9) and inferior temporal-occipital (BA 20) cortex, whereas attention directed at objects activated the left striate and peristriate cortex (BA 17/18). Both types of attention activated the left and right medial superior parietal cortex (BA 7/19), the left lateral inferior parietal cortex (BA 40/7), the left prefrontal cortex (BA 9), and the cerebellar vermis. Animal Studies of Spatial Attention and Representation Animal studies offer insight into mechanisms of spatial attention that are not obtained easily by study￾ing humans. Lesions in animals can be made with considerable precision, in contrast to lesions in humans, which are often determined by vascular anatomy (in the case of stroke) rather than by cortical or functional anatomy. Neurophysiological studies in animals can address the activity and responsive￾ness of single neurons, in contrast to functional neu￾roimaging in humans, which offers insight into the neurophysiology at the level of neural networks. Lesion Studies Animal lesion studies confirm the idea that dis￾tributed neural networks involving the parietal and frontal cortices mediate spatial attention and aware￾ness. In rodents, lesions of the posterior parietal or frontal cortex (medial agranular cortex) or the dorsolateral striatum produce a syndrome similar to neglect (Burcham, Corwin, Stoll, & Reep, 1997; Corwin & Reep, 1998). These rodents are more likely to orient ipsilesionally than contralesionally to visual, tactile, or auditory stimuli. This orienta￾tion bias recovers to a considerable degree over days to weeks. Dopamine antagonists impede spon￾taneous recovery and dopamine agonists enhance recovery (Corwin et al., 1986), probably by influ￾encing striatal function (Vargo, Richard-Smith, & Corwin, 1989). Anjan Chatterjee 14

Neglect 15 In macague monkeys,lesions to the frontal peri then be associated with greater contralesional arcuate areas and around the inferior parietal lobule neglect since the"what"and"where"of contrale- result in neglect.at least transiently (Duel.1987 sional obiects are no longer conioined.This Milner 1987)These monkeys are more likely to study highlights the difficulties in establishing the orient toward and act on stimuli in ipsilesiona appropriate homology between the monkey and space.Single-cell recordings of neurons arounc parietal cortex.While the intraparietal sulcus and prefrontal cortices neolect in humans is as (reviewed later)suggest that these regions are crit with lesions to the inferior posterior parietal co ical in the maintenance of spatial representations Brodmann's areas 39 and 40.it is not clear which. and preparation for actions directed at ecific loca if any regions.are the appropriate monkey analog tion 、From this one would expe ect that lesion to these areas these would produce neglec in Finally.Gaffan and Hornak (1997)emphasize in monke They find that beha obvio with eglect is esions which see at which human lesi tract les ad optic trac les owing way emote ing to be associa his 987 phere acquire glucos e commissure each en nuclei ithout access to ducing 3 did ne sed glucos abolism in the e pulvinar and opu lateral posterior thalamic nucle and in the tion about contral ona space is acquir deeper layers of the superior colliculus.It is inter the nonlesioned hemisphere becau with multiple est ing that recovery in these animals is also as ocular fixations.objects in contra space sociated with recovery of these remote some mes fall on the ipsilesional side of fixatio abnormalities.This idea that distributed abnormali The idea that short-term memories of contralesional ties are needed to produce neglect is reiterated in stimuli influence spatial behavior had not been con- a more recent study by Gaffan and Hornak (1997). sidered previously in animal models. They found in monkeys that transecting white matter tracts underlving the posterior parietal cortex Single-Cell Neurophvsiological Studies was important in producing more persistent neglect. Watson and colleagues (Watson.Valenstein Single-cell neurophysiological studies record the Day.Heilman.1994)reported that damage to activity of neurons in animals.often monkeys that monkeys'superior temporal sulcus produced more are engaged in various perceptual,motor,or cogni- profound neglect than damage to the inferior pari- tive tasks.These studies support the idea that etal lobule.They suggest that the superior temporal neurons in parietal and frontal association cortices sulcus in the monkey may serve as an importan mediate spatial attention and representations.These convergence zone for processing both the dorsal and neurons form a distributed network dedicated to a the ventral visual streams integrating the "where" variety of spatial behaviors,including attention and and"what"of obiects.Damage to this area might intention regarding spatial locations.memory of

In macaque monkeys, lesions to the frontal peri￾arcuate areas and around the inferior parietal lobule result in neglect, at least transiently (Duel, 1987; Milner, 1987). These monkeys are more likely to orient toward and act on stimuli in ipsilesional space. Single-cell recordings of neurons around the intraparietal sulcus and prefrontal cortices (reviewed later) suggest that these regions are crit￾ical in the maintenance of spatial representations and preparation for actions directed at specific loca￾tions. From this, one would expect that lesions in these areas would produce profound neglect in animals. Yet such cortical lesions produce only mild and transient neglect (Milner, 1987). If anything, biased behavior seems more obvious with frontal lesions, which seems at odds with human lesion studies in which posterior lesions are associated more often with neglect. In monkeys, cortical lesions with remote meta￾bolic abnormalities are more likely to be associated with neglect (Duel, 1987). Frontal lesions produc￾ing neglect are associated with decreased glucose utilization in the caudate nucleus and the ventral anterior and dorsomedial thalamic nuclei. Parietal lesions producing neglect are associated with decreased glucose metabolism in the pulvinar and the lateral posterior thalamic nuclei and in the deeper layers of the superior colliculus. It is inter￾esting that recovery in these animals is also as￾sociated with recovery of these remote metabolic abnormalities. This idea that distributed abnormali￾ties are needed to produce neglect is reiterated in a more recent study by Gaffan and Hornak (1997). They found in monkeys that transecting white matter tracts underlying the posterior parietal cortex was important in producing more persistent neglect. Watson and colleagues (Watson, Valenstein, Day, & Heilman, 1994) reported that damage to monkeys’ superior temporal sulcus produced more profound neglect than damage to the inferior pari￾etal lobule. They suggest that the superior temporal sulcus in the monkey may serve as an important convergence zone for processing both the dorsal and the ventral visual streams integrating the “where” and “what” of objects. Damage to this area might then be associated with greater contralesional neglect since the “what” and “where” of contrale￾sional objects are no longer conjoined. This study highlights the difficulties in establishing the appropriate homology between the monkey and the human posterior temporoparietal cortex. While neglect in humans is associated most commonly with lesions to the inferior posterior parietal cortex, Brodmann’s areas 39 and 40, it is not clear which, if any regions, are the appropriate monkey analog to these areas. Finally, Gaffan and Hornak (1997) emphasize the importance of memory in monkeys’ behavioral manifestations of overt neglect. They find that neglect is associated with complete commissuro￾tomy and optic tract lesions, but not with isolated optic tract, parietal, or frontal cortex lesions. They interpret this finding in the following way: Section￾ing the optic tract makes one hemisphere blind to visual information. This hemisphere acquires visual information from the other hemisphere through interhemispheric commissures. If each hemisphere maintains a representation of contralateral space, then a monkey without access to information about contralesional space will act as if this space did not exist. With an isolated optic tract lesion, informa￾tion about contralesional space is acquired through the nonlesioned hemisphere because with multiple ocular fixations, objects in contralesional space sometimes fall on the ipsilesional side of fixation. The idea that short-term memories of contralesional stimuli influence spatial behavior had not been con￾sidered previously in animal models. Single-Cell Neurophysiological Studies Single-cell neurophysiological studies record the activity of neurons in animals, often monkeys that are engaged in various perceptual, motor, or cogni￾tive tasks. These studies support the idea that neurons in parietal and frontal association cortices mediate spatial attention and representations. These neurons form a distributed network dedicated to a variety of spatial behaviors, including attention and intention regarding spatial locations, memory of Neglect 15

Anian chatteriee 16 Reference Frames The integration of different sensory modalities in Parietal Neurons ctedifereatn点Psabyiaoled al con in co In the 1970s,Mountcastle and co-workers found (Brotchie.Anderson.Snyder.Goodman,1995 neurons in the parietal cortex of monkeys that were responsive when the animals attended to From studies of people with neglect,we know that viewer-centered reference frames can be anchored lights in their peripheral vision despite gazing toward a central location (Mountcastle.Lynch to retinotopic,head-,or body-centered coordinates Georgonolous Sakata Acuna 1975)They From animal studies it appears unlikely that a found that neurons in the posterior parietal cortex different pool of neurons code retinal and head- responded to a variety of spatial behaviors,includ- centered coordinates.Andersen and colleagues suggest that head-centered coordinates are derived ing fixation.smooth pursuit.saccades.and reachins (Mountcastle,1976).Neurons in different region from the interaction of retinal and eye position signals.The amplitude of a neuron's response to (ventral.medial.lateral)of the stimulation of a retinal location is modulated by 5.7a.and 7b.seem to be critical to the mediation eye position.Within area 7a,neurons compute the location of a stimulus in head-centered coordinates of spatial attention These ne fo linked to different sensory and motor ystem from these interactions (Andersen,Essick,Siegel Fo 1985).Anderson et al.suggest that other areas ietal (LIP) onsive to tactile stimuli or the dir including the lateral intraparietal sulcus,area V3 aspects of moving visual tral in the pulvinar,nucleus and parts of the premotor and (VIP) neurons (Duhar prefrontal cortex may code different kinds of spatial 1998 reference frames in a similar fashion (Andersen. and 1995a)Pouget and Seinowski (1997)use hasis Man functions to offer a slightly different computational of and 1991).VIP neurons aligned In addition to reference frames divided along viewer-cent ered co are especi tioned as concentrie shells around the body an from distant ex space 1998.正 ons in kevs paretal are and proprioceptive ated by the link n nput from neck muscle (Andersen,1995b:Snyder. (Snyder.Batist Grieve,Brotchie,&Ande rsen.1998) .R Generally,neurons within the posterior parietal adopts th cortex link specific sensations to different moto space.He claims that the systems.although there is disagreement on whethe pe neurons within the LIP sulcus are purely attentional m or whether these neurons are necessarily linked R 1993:R to eye movements (Andersen,Bracewell,Barash Gnadt,Fogassi,1990:Colby Goldberg.1999). Neuron withi ain the mon are tuned acti ving diffe systems,such as the mouth,eyes,or hands

spatial locations, and facilitation of perception of objects in different locations. Parietal Neurons In the 1970s, Mountcastle and co-workers found neurons in the parietal cortex of monkeys that were responsive when the animals attended to lights in their peripheral vision despite gazing toward a central location (Mountcastle, Lynch, Georgopolous, Sakata, & Acuna, 1975). They found that neurons in the posterior parietal cortex responded to a variety of spatial behaviors, includ￾ing fixation, smooth pursuit, saccades, and reaching (Mountcastle, 1976). Neurons in different regions (ventral, medial, lateral) of the posterior intra￾parietal sulcus and nearby regions, such as areas 5, 7a, and 7b, seem to be critical to the mediation of spatial attention. These neurons form a mosaic linked to different sensory and motor systems. For example, lateral intraparietal (LIP) neurons are less responsive to tactile stimuli or the directional aspects of moving visual stimuli than ventral in￾traparietal (VIP) neurons (Duhamel, Colby, & Goldberg, 1998). Many posterior parietal and frontal neurons are responsive to combinations of visual and tactile stimuli (Colby & Duhamel, 1991). VIP neurons are responsive to aligned visual and tactile receptive fields when they move in specific directions. Medial intraparietal (MIP) neurons are especially respon￾sive to joint rotations and movements of limbs. Other neurons in area 7a integrate visual and vestibular input, and neurons in the lateral intra￾parietal area integrate visual and proprioceptive input from neck muscles (Andersen, 1995b; Snyder, Grieve, Brotchie, & Andersen, 1998). Generally, neurons within the posterior parietal cortex link specific sensations to different motor systems, although there is disagreement on whether neurons within the LIP sulcus are purely attentional or whether these neurons are necessarily linked to eye movements (Andersen, Bracewell, Barash, Gnadt, & Fogassi, 1990; Colby & Goldberg, 1999). Reference Frames The integration of different sensory modalities in the posterior parietal cortex is presumably involved in constructing different kinds of reference frames (Brotchie, Anderson, Snyder, & Goodman, 1995). From studies of people with neglect, we know that viewer-centered reference frames can be anchored to retinotopic, head-, or body-centered coordinates. From animal studies it appears unlikely that a different pool of neurons code retinal and head￾centered coordinates. Andersen and colleagues suggest that head-centered coordinates are derived from the interaction of retinal and eye position signals. The amplitude of a neuron’s response to stimulation of a retinal location is modulated by eye position. Within area 7a, neurons compute the location of a stimulus in head-centered coordinates from these interactions (Andersen, Essick, & Siegel, 1985). Anderson et al. suggest that other areas, including the lateral intraparietal sulcus, area V3, the pulvinar, nucleus and parts of the premotor and prefrontal cortex may code different kinds of spatial reference frames in a similar fashion (Andersen, 1995a). Pouget and Sejnowski (1997) use basis functions to offer a slightly different computational solution to the mediation of different reference frames encoded within the same array of neurons. In addition to reference frames divided along viewer-centered coordinates, space can be parti￾tioned as concentric shells around the body, with close peripersonal space being coded distinctly from distant extrapersonal space (Previc, 1998). In monkeys, this segregation of space may be medi￾ated by the link between attentional neurons and multiple motor systems (Snyder, Batista, & Andersen, 1997). Rizzolatti adopts the strong posi￾tion that all attentional circuits organize movements to specific sectors of space. He claims that the facil￾itation of perception by attention is a consequence of circuits activated in preparation for moving (Rizzolatti & Berti, 1993; Rizzolatti et al., 1988). Neurons within the monkey intraparietal sulcus are tuned to actions involving different motor systems, such as the mouth, eyes, or hands. In Anjan Chatterjee 16

Neglect 17 combination with their connections to frontal and"where"of things.Single-cell monkey physio- regions,these neurons integrate the visual fields logical studies also support such modulation. with the tactile fields of specific body parts and with Neurons in area V4 are sensitive to specific stimuli the actions of these body parts(Gross Graziano. located within their receptive fields (Moran 1995).The parietal and frontal interconnections are Desimone.1985).Their firing increases when the anatomically segregated along a ventral-to-dorsal animal attends to that location.This stronge axis(Petrides Pandya.1984).Neurons within the response to the stimulus for which the neuron VIP sulcus are respe sive to visual stimuli within already tuned.when the animal attends to it.sug 5cm of the monkey's face (Colby.Duhamel.& gests a physiological correlate of the enhanced pe Goldberg.1993).These neurons project to area F4 eption of obiects when attention is directed to the of area 6 in the pre otor that co ation of those objects. tributes to head and mouth m nts (Fogass et al..1996:Grazi 1994)and mediate the co of very y close Conelusions and Future Directions e Na. s in the MIP sul reaching distance Grazi Convergence There is a remarkable convergence of some ideas across different disciplines with highly varied tradi- Gentilucci et al1988) and ar tions and methods.Four related ideas about spatial 994 attention and representation recur and are summa- Thi rized here. ections to as Distributed Networks arm-centere s(G et a Thes ved in the decisio Neural networks involving different and noncon- processes by whic h different kinds of movements tiguous parts of the brain mediate spatial attention are selected(Merchant,Zainos,Hernandez,Salinas. Rather than being localized to a single brain loca &Romo,1997) tion.spatial attention is mediated by the parietal and Neuron within the monkey LIP sulcu frontal and probably cingulate cortices,as well as (Duhamel,Colby,&Goldberg.1992)may be con- the basal ganglia,thalamus,and superior colliculus. nected to saccadic mechanisms of the frontal eye fields and the superior colliculus.Neurons in the Multiple Representations of Space superior colliculus are responsive to behaviorally The brain constructs multiple representations of relevant stimuli when linked to saccadic eve move space.de ments (Wurtz Goldberg.1972:Wurtz Munoz. intuitions of space as a homoge- um tha 1995).These networks probably link sensations to round tions eye movements and construct distant extrapersonal involve the body and different kinds of extrapersonal space space can be space. viewed as concentric shells around the body.close Space-Based and Object-Based Attention to the trunk,within reach of our limbs,or further away in more distant space.Extrapersonal space car Neuroimaging studies in humans have shown tha also be partitioned into retinotopic,head-centered ntion can influence and trunk-centered coordinates that all have the viewer as the primary referent.Viewer-independent ence is presumably involved in binding the "what reference frames are anchored to the spatial axes

combination with their connections to frontal regions, these neurons integrate the visual fields with the tactile fields of specific body parts and with the actions of these body parts (Gross & Graziano, 1995). The parietal and frontal interconnections are anatomically segregated along a ventral-to-dorsal axis (Petrides & Pandya, 1984). Neurons within the VIP sulcus are responsive to visual stimuli within 5cm of the monkey’s face (Colby, Duhamel, & Goldberg, 1993). These neurons project to area F4 of area 6 in the premotor cortex, an area that con￾tributes to head and mouth movements (Fogassi et al., 1996; Graziano, Yap, & Gross, 1994) and may mediate the construction of very close peripersonal space. Neurons in the MIP sulcus are responsive to visual stimuli within reaching distance (Graziano & Gross, 1995). These neurons project to ventral premotor cortices that mediate visually guided arm movements (Caminiti, Ferraina, & Johnson, 1996; Gentilucci et al., 1988) and are sensitive to stimuli in arm-centered rather than retinotopic coordinates (Graziano et al., 1994). This area has direct con￾nections to the putamen, which also has such arm-centered neurons (Graziano et al., 1994). These putamenal neurons may be involved in the decision processes by which different kinds of movements are selected (Merchant, Zainos, Hernandez, Salinas, & Romo, 1997). Neurons within the monkey LIP sulcus (Duhamel, Colby, & Goldberg, 1992) may be con￾nected to saccadic mechanisms of the frontal eye fields and the superior colliculus. Neurons in the superior colliculus are responsive to behaviorally relevant stimuli when linked to saccadic eye move￾ments (Wurtz & Goldberg, 1972; Wurtz & Munoz, 1995). These networks probably link sensations to eye movements and construct distant extrapersonal space. Space-Based and Object-Based Attention Neuroimaging studies in humans have shown that visual or spatial attention can influence the pro￾cessing of objects in the ventral stream. This influ￾ence is presumably involved in binding the “what” and “where” of things. Single-cell monkey physio￾logical studies also support such modulation. Neurons in area V4 are sensitive to specific stimuli located within their receptive fields (Moran & Desimone, 1985). Their firing increases when the animal attends to that location. This stronger response to the stimulus for which the neuron is already tuned, when the animal attends to it, sug￾gests a physiological correlate of the enhanced per￾ception of objects when attention is directed to the location of those objects. Conclusions and Future Directions Convergence There is a remarkable convergence of some ideas across different disciplines with highly varied tradi￾tions and methods. Four related ideas about spatial attention and representation recur and are summa￾rized here. Distributed Networks Neural networks involving different and noncon￾tiguous parts of the brain mediate spatial attention. Rather than being localized to a single brain loca￾tion, spatial attention is mediated by the parietal and frontal and probably cingulate cortices, as well as the basal ganglia, thalamus, and superior colliculus. Multiple Representations of Space The brain constructs multiple representations of space, despite our intuitions of space as a homoge￾neous medium that surrounds us. These representa￾tions involve the body and different kinds of extrapersonal space. Extrapersonal space can be viewed as concentric shells around the body, closer to the trunk, within reach of our limbs, or further away in more distant space. Extrapersonal space can also be partitioned into retinotopic, head-centered, and trunk-centered coordinates that all have the viewer as the primary referent. Viewer-independent reference frames are anchored to the spatial axes Neglect 17