747Recent work on the coding of spatial information in the brain hassignificantly advanced our knowledge of sensory to motortransformations on several fronts. The encoding of informationreferenced to the retina (eye-centered) but modulated by eyeposition, called a gain field representation, has proved to be verycommon throughout parietal and occipital cortex. The use of aneye-centered representation as a working memory of spatiallocation is problematic if the eyes move during the memoryperiod. Details regarding the manner in which the brain solvesthis problem are beginning to emerge. Finally, the discovery ofeye-centered representations of ongoing or intended armmovements has changed the way we think about the order ofoperations in the sensory to motor coordinate transformation.AddressesMcDonnell Center for Higher Brain Function, Department of Anatomy& Neurobiology, Box 8108, Washington University School of Medicine,660 South Euclid Avenue, St Louis, MO 63110, USA; e-mail: [email protected] Opinion in Neurobiology 2000, 10:747–7540959-4388/00/$ — see front matter© 2000 Elsevier Science Ltd. All rights reserved.AbbreviationsFEF frontal eye fieldsLIP lateral intraparietal areaPM premotor cortexPMd dorsal premotor cortexPMv ventral premotor cortexPPC posterior parietal cortexPRR parietal reach regionSC superior colliculusSEF supplementary eye fieldsV1 visual area 1IntroductionA major brain function is the manipulation of spatial infor-mation. Much of what we do involves extracting spatialinformation from sensory input and then using that spatialinformation to direct a motor response. A simple exampleis reaching for something we have seen. This act requiresthat spatial information be transformed from a retinal coor-dinate frame to the coordinate frame of the muscles orjoints involved in moving the arm. A major endeavor ofsystems neuroscience has been to discover the algorithmsand the sites at which such transformations are per-formed [1–4,5•,6•].This review will focus on several recent issues related to thisendeavor. It will consider only transformations of visual infor-mation used to drive eye and arm movements in primates,and will focus on the involvement of the parietal cortex. It haslong been known that the lateral intraparietal area (LIP) inparietal cortex is involved in coding targets for saccadic eyemovements. Recently it has become clear that cortical regionsmedial to LIP are involved in coding arm movements. Thedetails of how spatial information is represented and manipu-lated in these regions is of critical importance to ourunderstanding of sensory to motor transformations.Eye-centered coding and gain fieldsMany occipital and parietal areas involved in early process-ing of visual information do so in an eye-centered frame ofreference [7]. In many areas, eye-centered spatial signalsare modulated by postural information. Cells in these areashave receptive fields whose locations are fixed withrespect to the retina. For example, the receptive field of aparticular cell may always lie 10° above the fovea, regard-less of the positions of the eyes, head and body. At thesame time, however, the magnitude of the activity evokedby any given stimulus depends on eye, head, arm or bodyposition. The original description of this modulation,called a ‘gain field’, was a monotonic influence of eye posi-tion on visually evoked and saccade-related activity inareas 7a and LIP [8]. Since then, gain fields have beendescribed in numerous other brain areas as well.Effects of eye position have been described in visualarea 1 (V1), the cortical area closest to the retinal inputand therefore the most likely to be purely eye-centeredor retinotopic [9,10,11•,12]. However, these observationsare controversial [13•,14]. Elsewhere, the data for eyeposition gain fields are well established. Spatial respons-es in areas V3a, MT (middle temporal area), MST(medial superior temporal area), VIP (ventral intraparietalarea), V6, V6a, 7m, PMv (ventral premotor cortex [PM]),PMd (dorsal PM) and SEF (supplementary eye fields)have all been shown to be modulated by the position ofthe eyes in the head [15–22,23•,24•]. Head position gainfields have been reported in both areas LIP and 7a [25].In LIP, the gain fields are referenced to the body, where-as in 7a they are referenced to the inertial frame, that is,to the fixed world around you. Recently, eye positioneffects have been identified in V4 [12,26••], demonstrat-ing that eye position gain fields occur even in the ventral(‘what’) visual pathway.Several roles for gain fields in coordinate frame transfor-mations have been proposed. Zipser and Andersen [27]first showed that retinotopic signals modulated by eyeposition could be used as an intermediate stage in com-puting head-centered location of visual targets. As very fewneurons in parietal cortex code in explicit head-centeredcoordinates [28,29], an alternative proposal is that gainfield representations themselves constitute a distributedrepresentation of head-centered spatial information.Consistent with this idea, Bremmer et al. [30] recentlyshowed that a network using actual recorded responsescould be used to generate a reliable and precise estimate ofthe head-centered location of a visual stimulus.Coordinate transformations for eye and arm movements in the brainLawrence H SnyderRemarkably, despite 15 years of experimental and theoret-ical work on gain fields, only circumstantial evidence existsthat gain field information is in fact used in a spatial com-putation: gain fields are ubiquitous, and the distributedrepresentation contains spatial information that is rarely, ornever, explicitly represented elsewhere. At one time, anargument could be made that gain field information is notused by the brain. This argument was based on the ideathat LIP, one of the first areas in which gain fields werewell characterized, serves to identify the location of visualtargets for saccadic eye movements and to transfer thatspatial information to the superior colliculus (SC). As such,LIP would appear to have no use for the head-centeredposition information contained in the eye-position gainfield. However, recent studies using animals
View Full Document
Unlocking...