The predominant psychophysical methods for investigating multimodal signal detection are reviewed. It is argued that most experimental designs cannot distinguish situations in which the stimulation of one modality affects perception in another from those in which performance is affected as observers divide their attention between modalities. Furthermore, many paradigms are confounded by the effects of temporal and stimulus uncertainty. These problems also may arise in studies of unimodal signal detection. They can be avoided by presenting suprathreshold stimuli to each modality on every trial, and by administering concurrent tasks in conditions involving divided attention.Psychophysical experiments may study unimodal perception (by presenting stimuli to a single sensory modality) or multimodal perception (by presenting stimuli to several modalities simultaneously). Of particular interest in the latter case are crossmodal effects, in which the stimulation of one modality affects perception in another, and attentional effects, in which performance declines when attention is divided among several modalities.The literature on crossmodal and attentional effects is extensive. London's (1954) review of Soviet work alone cites over 500 studies. (For other reviews, see Bernstein, 1970;Gilbert, 1941;Klayman, 1973a;Loveless, Brebner, & Hamilton, 1970;Ryan, 1940;and Taylor, 1974.) However, it is still not clear whether crossmodal or attentional effects exist, because previous research has produced a bewildering array of discrepant findings. This has led recent texts on multimodal perception (e.g., Marks, 1978; Walk & Pick, 1981) to ignore the effects entirely. This is unfortunate, because crossmodal and attentional effects have serious implications, even for studies of unimodal perception. The present paper discusses these implications, reviews various methodological problems, and offers a general methodology that avoids the shortcomings of previous designs. The paper considers only signal detection paradigms, although reaction time paradigms are also useful for studying the effects.
BASIC RESEARCH ISSUES IndependenceAre simultaneous stimuli processed independently of one another? This independence question arises in studies I thankLynn Olzak, James Thomas, Richard Swensson, and Dominic Massaro for providing insightful comments on earlier versions of this manuscript. Requests for reprints should be addressed to Harold Stanislaw,