Introduction

To adaptively control its behavior, an organism requires information about the location of external objects and events. This information is conveyed simultaneously and independently by different sensory systems, which initially represent object properties within modality-specific coordinate systems (retinotopic in vision, somatotopic in touch, tonotopic in audition). In order to localize and identify relevant stimuli or events, spatial information delivered by different input systems needs to be integrated. Although the neural mechanisms underlying intersensory integration have been extensively studied (see Stein & Meredith, 1993), the question of how attention can be directed to objects and events when these are initially represented in different sensory modalities has only recently begun to be addressed (see Driver & Spence, 1998, for an overview). Directing attention to specific locations enhances the processing of sensory information originating from attended locations relative to other locations. To select visual, auditory, and tactile information originating from the same object, spatial attention must be coordinated across different sensory modalities. This cross-sensory coordination may have important implications for mechanisms of selective attention, which could involve spatial synergies (cross-modal links) in the attentional processing of information across sensory modalities.

Until recently, experimental investigations of spatial attention were confined to the study of spatially selective processing within single sensory modalities (see Parasuraman, 1998, for a recent review). Numerous studies have shown that shifts of attention can be triggered both involuntarily (stimulus-driven or exogenous attention) and voluntarily (intentional or endogenous attention). In the former case, attention is attracted to the location of salient but spatially nonpredictive peripheral events. In the latter case, attention is directed to the expected location of relevant stimuli, as when this location is indicated by a spatially predictive symbolic cue. Both exogenous and endogenous attention result in faster and more accurate responses when visual, auditory, or tactile stimuli are presented at attended locations.

Because research on attentional selectivity has traditionally focused almost entirely on unimodal attention, issues related to multisensory spatial attention have not yet been investigated systematically. Very few experiments have studied whether there are any cross-modal links in spatial attention between and among vision, audition, and touch, and which mechanisms are responsible for such links. Does directing attention to the expected location of a visual event affect the processing of auditory or tactile stimuli at visually attended versus unattended locations? Does the automatic capture of attention by an irrelevant peripheral sound or touch have any spatially selective impact on the processing of subsequently presented visual stimuli?

Results from several recent behavioral studies (e.g., Butter, Buchtel, & Santucci, 1989; Spence & Driver, 1996; Spence, Pavani, & Driver, 2000) have suggested that there are cross-modal links in endogenous spatial attention between and among vision, audition, and touch. In these experiments, participants directed attention to the expected location of target stimuli within one primary modality. On a minority of trials, stimuli of a different (secondary) modality were presented, which were equally likely (or even more likely) to appear on the side opposite to the expected location in the primary modality. Performance benefits for stimuli at the location attended in the primary modality were observed not only for that primary modality, but also for secondary modality stimuli, thus demonstrating that the locus of attention within one modality (that is, a spatial expectancy specific to a particular modality) affects the processing of information in other modalities.

Along similar lines, other recent studies have demonstrated that involuntary shifts of spatial attention triggered by stimuli in one modality can affect performance in response to subsequently presented stimuli in a different modality, thereby reflecting cross-modal links in exogenous spatial attention. Responses to visual stimuli are faster or more accurate (or both) when these stimuli are presented on the same side as a previous uninformative auditory event (Spence & Driver, 1997; McDonald & Ward, 2000), and responses to auditory stimuli can be facilitated by previous visual events at the same location (Ward, 1994; Ward, McDonald, & Lin, 2000). Irrelevant auditory events influence not only the speed of responses to subsequent visual targets but also signal detection performance, as masked visual stimuli are detected more accurately when preceded by an auditory event at the same location (McDonald, Teder-Sälejärvi, & Hillyard, 2000).

Although such observations from behavioral studies demonstrate cross-modal links in endogenous and exogenous spatial attention, they do not provide any direct insight into the neural processes underlying such links. For example, performance benefits for secondary modality stimuli at attended locations could result from the effects of cross-modal endogenous attention on perceptual processes or from attentional modulations of postperceptual stages (see Spence & McDonald, Chap. 1, this volume, for further discussion). Do cross-modal links in spatial attention affect modality-specific sensory-perceptual processes, or are the effects of such links restricted to later, postperceptual processing stages? In addition, the existence of cross-modal links in endogenous attention may also have implications for our understanding of covert processes involved in the control of attentional shifts. Such attentional control processes are activated in anticipation of and in preparation for stimuli expected at a specific location. The presence of cross-modal links in spatial attention could indicate that endogenous attentional orienting processes are controlled by a single supramodal system that directs spatial attention to the location of relevant external stimuli, regardless of their modality (Farah, Wong, Monheit, & Morrow, 1989). Alternatively, shifts of spatial attention may primarily be controlled by modality-specific mechanisms, although the effects of such attentional shifts will spread to other sensory modalities (Spence & Driver, 1996).

To obtain further insights into the neural basis of cross-modal attention, measures of behavioral performance need to be combined with different brain imaging methods. This chapter reviews recent electrophysiological studies investigating multisensory spatial attention with event-related brain potential (ERP) measures. ERPs reflect phasic modulations of brain activity that are time-locked to the onset of external or internal events. They can be obtained by averaging the EEG activity measured in response to such events over a number of recording epochs. ERP waveforms consist of positive- and negative-going deflections that are assumed to be generated at least partially by synchronous postsynaptic activities at the apical dendrites of pyramidal cells in cortical layer IV. ERP components are labeled with respect to their polarity (positive or negative) and latency, and are assumed to reflect different stages in the processing of external or internal events. In contrast to other measures of brain activity, such as single-cell recordings, ERPs provide a noninvasive method of monitoring brain events that take place during cognitive processing. ERPs can thus be obtained in normal voluntary subjects under standard experimental conditions. Compared with functional brain imaging measures such as positron emission tomography (PET) and functional magnetic resonance imaging (fMRI), the spatial resolution of ERP recordings is relatively poor, but their temporal resolution is excellent. ERPs provide a continuous on-line measure of cerebral activities, and so are especially useful as markers of the time course of information processing.

This chapter discusses ERP studies investigating a number of questions and issues related to cross-modal links in endogenous and exogenous spatial attention. The next section considers how cross-modal links in endogenous attention affect the processing of visual, auditory, and tactile information. Do such links modulate modality-specific sensory processing stages, or are cross-modal effects primarily located at later postperceptual stages? The third section discusses which spatial coordinates frames are involved in cross-modal links in endogenous spatial attention. In the fourth section, we consider whether cross-modal links in endogenous spatial attention are strategy dependent or reflect genuine constraints on the processing of sensory information within and across modalities. The fifth section reviews ERP studies investigating effects of cross-modal links in exogenous (involuntary) attention. The final section discusses whether covert attentional control processes active during shifts of endogenous spatial attention are supramodal or modality-specific.