An fMRI investigation of strategies for word recognition

A common procedure used to study visual word recognition is the lexical decision task (LDT). Behavioral studies have demonstrated that overall performance in this task is modulated by the type of foils presented. There are divergent claims about the impact of different types of foils on overall processing strategies in the LDT: some researchers claim that pseudohomophone foils (e.g., BRANE) encourage participants to selectively de-emphasize phonological processing, while other researchers claim that pseudohomophone foils encourage participants to engage in more extensive processing of all types (orthographic, phonological, and semantic). To evaluate these claims, we conducted a functional magnetic resonance imaging study. Participants (N = 18) completed three lexical decision tasks, each with a different foil type: consonant strings (e.g., BVRNT), pseudowords (e.g., BLINT), and pseudohomophones (e.g., BRANE). We presented homophones (e.g., MAID) and nonhomophones (e.g., MESS) on word trials in order to be able to calculate the homophone effect as a marker of phonological processing for word stimuli in each foil condition. Comparison of behavioral results in the different foil conditions showed that reaction times were longest, error rates were highest, and homophone effects were largest in the LDT with pseudohomophones. Imaging results showed the greatest magnitude of activity in several regions, including the inferior frontal cortex, during the LDT with pseudohomophone foils. A comparison to inferior frontal activity produced during an additional task (rhyme judgment) supported the conclusion that LDT with pseudohomophone foils is a difficult task in which readers engage in an overall response strategy involving extensive processing of phonological information.