Functional MR Imaging during Auditory Word Perception: A Single-Trial Presentation Paradigm

Functional magnetic resonance imaging (fMRI), while still in relatively early stages of its development, has become a widespread tool for addressing issues in the neurobiology of language and other cognitive domains (e.g., Belliveau et al., 1991; Binder et al., 1994; Cohen et al., 1994; Hickok et al., 1995; Shaywitz et al., 1995; for a review of the technique see Bandettini et al., 1995). In a typical fMRI experiment, stimuli are presented in blocks of say 20–40 sec that alternate through several cycles with an equally long baseline/control period. So for example, a subject may see individual words presented one-at-a-time for 30 sec, then non-words for 30 sec, and so on for five cycles. This presentation scheme provides a serious constraint on experimental design in fMRI studies and consequently limits the kinds of questions one can address with this methodology. One cannot, to cite just one example, use lexical decision as a behavioral measure in the magnet because the item type (word/non-word) is fully predictable from the experimental design. Relatedly, under these experimental conditions it is unclear to what extent fMRI activations are due to expectation-driven (controlled) processes versus lower-level perceptual (automatic) processes. Here we present data from an alternative fMRI presentation scheme used with auditory speech perception. The paradigm is modeled on standard methodologies in event related potential (ERP) studies where test items are ran-