Orienting Attention to Locations in Perceptual Versus Mental Representations A. C. Nobre1, J. T. Coull2,*, P. Maquet2, C. D. Frith2, R. Vandenberghe3,y, and M. M. Mesulam3 Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/16/3/363/1758100/089892904322926700.pdf by guest on 18 May 2021 Abstract & Extensive clinical and imaging research has characterized the system for directing spatial attention to locations in the the neural networks mediating the adaptive distribution of external world. We found that these two crucial aspects of spatial attention. In everyday behavior, the distribution of spatial cognition are subserved by extensively overlapping attention is guided not only by extrapersonal targets but also by networks. However, we also found that a region of right parietal mental representations of their spatial layout. We used event- cortex selectively participated in orienting attention to the related functional magnetic resonance imaging to identify the extrapersonal space, whereas several frontal lobe regions neural system involved in directing attention to locations in selectively participated in orienting attention within on-line arrays held as mental representations, and to compare it with mental representations. & INTRODUCTION working memory, the function that mediates the on-line Attention biases the selection of targets according to maintenance and manipulation of mental representa- changing motivation and volition to guide perception tions for guiding behavior. and action. To date, the extensive research on attention Attentional orienting and working memory are sub- has focused on the selection of items located in the served by partially overlapping distributed neural sys- extrapersonal world. However, as humans, much of our tems. Spatial orienting to external events is coordinated world is internal. We constantly build and manipulate by a distributed network including areas in the posterior mental representations based upon experiences and parietal, lateral premotor and dorsal prefrontal, and expectations. It may be equally important, therefore, to cingulate cortices (Nobre, 2001; Mesulam, 1981, 1999). direct attention within these internalized representa- Overlap with the neural system underlying working tions. Orienting to mental representations of extraperso- memory occurs in the intraparietal sulcus and frontal nal space is a common aspect of our daily activities. Some eye fields (Awh et al., 1999; LaBar, Gitelman, Parrish, & examples are looking back at a relevant target in the Mesulam, 1999; McCarthy, 1995). Prefrontal regions, room after having been distracted or reaching toward anterior to these areas of overlap, are additionally en- your coffee cup on the desk while not breaking gaze from gaged during tasks requiring working memory. the computer monitor. The known anatomical overlap between neural sys- Here we used event-related functional magnetic reso- tems involved in orienting attention and in working nance imaging (ef MRI) to investigate the neural system memory supports the functional similarities between involved in orienting spatial attention to previous stim- the two cognitive domains (Awh & Jonides, 2001). Down- ulus events held on-line within internalized memory ing has shown that the contents of working memory can representations, and to compare it with the neural guide the orienting of attention to optimize performance system involved in orienting spatial attention to upcom- in behavioral tasks (Downing, 2000). Similar patterns of ing extrapersonal events. This investigation was stimu- activation and modulation have been reported in poste- lated by several lines of research that have emphasized rior extrastriate visual areas by attention and working the close interplay between attentional orienting and memory (Awh, Anllo Vento, & Hillyard, 2000; Awh et al., 1999; Chelazzi, Miller, Duncan, & Desimone, 1993). Based on these types of observations, influential theoretical 1University of Oxford, 2Institute of Neurology, 3Northwestern models posit a close link between attention and working University Medical School *Current address: Laboratoire de Neurobiologie de la Cogni- memory (Desimone & Duncan, 1995; Baddeley, 1993). tion, Centre National de la Recherche Scientifique, Marseille, The overlap between spatial attention and working France memory also supports the intuitive notion of the ability yCurrent address: Neurology Department, University Hospital to scan a working memory scene in a spatially ad- Gasthuisberg, Louvain, Belgium dressed fashion in a similar way as a perceptual scene D 2004 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 16:3, pp. 363–373 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892904322926700 by guest on 24 September 2021 in the extrapersonal world. The neural mechanisms for relative to noninformative neutral cues. Spatially mislead- orienting attention to locations in mental representa- ing invalid cues led to behavioral costs. Attentional tions has not yet been investigated systematically, but the benefits and costs were equivalent whether cues ap- influence of spatial attention upon the access to and peared before or after the array, showing that retrocues manipulation of mental representations is illustrated by were just as potent as precues in optimizing behavioral the deficits of patients with left hemispatial neglect, a performance. neurological syndrome characterized by a severe deficit In our present ef MRI experiment, the spatial precue of spatial attention. Some patients with this syndrome specified a focused spatial orientation toward one quad- are unable to direct attention to the left side of internal rant of an upcoming rectangular array. The retrocue Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/16/3/363/1758100/089892904322926700.pdf by guest on 18 May 2021 representations while others cannot introduce the rele- instructed an equally focused spatial orientation toward vant detail into the left side of newly compiled mental one quadrant of an internal representation of the array representations (Bisiach & Luzzatti, 1978; Bisiach, Luz- held in working memory. Event-related analysis of f MRI zatti, & Perani, 1979). Neglect of internal representations data revealed brain areas activated by spatially informa- can sometimes occur in the absence of neglect for the tive precues and retrocues within the trials, controlling extrapersonal space (Ortigue et al., 2001; Beschin, Basso, for aspects of cue processing that were unrelated to & Della Sala, 2000; Coslett, 1997; Guariglia, Padovani, their spatial orienting nature (see Methods). We com- Pantano, & Pizzamiglio, 1993), suggesting that the neural pared activations associated with the precue and retro- systems for orienting attention to external versus internal cue to identify areas of common and differential representations may differ, at least in part. engagement by the two types of attentional orienting. We developed an experimental task to reveal brain Areas of common activation helped to determine the areas involved in orienting attention toward locations in neural substrates shared by the two types of attentional mental representations (Figure 1). The task required orienting whereas areas of differential activation helped retrieving the color of an item from a briefly presented to identify regions potentially involved in selective func- array, where each quadrant contained one stimulus. In tions supporting the orienting of attention to extra- some conditions, spatially informative cues occurred personal locations versus their internal representations. seconds after the array had appeared and disappeared, pointing to the relevant quadrant location of the inter- RESULTS nalized representation from which the color should be retrieved. These retrocues oriented attention toward a Behavioral Performance specific location in the array that existed only as a Participants performed the behavioral task adequately mental representation in working memory. In other (see Figure 1). The average level of accuracy to decide conditions, spatially informative precues preceded ar- whether the probe stimulus was the same color as the rays and directed attention to one quadrant of the stimulus at the cued location was 80%. Accuracy during extrapersonal array, as in traditional investigations of ‘‘yes’’ and ‘‘no’’ trials was statistically equivalent. The attentional orienting (Posner, 1980). Orienting attention percentage of correctly identifying matching stimuli (hits to locations in mental representations of previously in yes trials) was 79% and the percentage of correctly presented stimulus arrays was compared to orienting dismissing nonmatching probes was 81% (19% false attention to locations in upcoming perceptual arrays. alarms). Accuracy of task performance did not differ The experimental paradigm built upon the work of between precue (81%) and retrocue (79%) trials, in Sperling (1960), who used spatial cues to prompt either yes or no trials. The location of the probe stimulus retrieval of items from immediately preceding arrays in the array during matching trials did not have a main held in very short-term iconic memory. effect on accuracy. The fact that spatial retrocues do orient attention to Reaction times were on average 754 msec. Participants locations within on-line representations of arrays held in were faster to identify matching probes in yes trials working memory was confirmed by an independent set (712 msec) than to dismiss