Assessing language dominance with functional MRI: the role of control tasks and statistical analysis

There is a discrepancy between the brain regions revealed by functional neuroimaging techniques and those brain regions where a loss of function, either by lesion or by electrocortical stimulation, induces language disorders. To differentiate between essential and non-essential language-related processes, we investigated the effects of linguistic control tasks and different analysis methods for functional MRI data. Twelve subjects solved two linguistic generation tasks: (1) a verb generation task and (2) an antonym generation task (each with a linguistic control task on the phonological level) as well as two decision tasks of semantic congruency (each with a cognitive high-level control task). Differential contrasts and conjunction analyses were carried out on the single-subject level and an individual lateralization index (LI) was computed. On the group level we determined the percent signal change in the left inferior frontal gyrus (IFG: BA 44 and BA 45). The conjunction analysis of multiple language tasks led to significantly greater absolute LIs than the LIs based on the single task versus fixation contrasts. A further significant increase of the magnitude of the LIs could be achieved by using the phonological control conditions. Although the decision tasks appear to be more robust to changes in the statistical threshold, the combined generation tasks had an advantage over the decision tasks both for assessing language dominance and locating Broca's area. These results underline the need for conjunction analysis based on several language tasks to suppress highly task-specific processes. They also point to the need for high-level cognitive control tasks to partial out general, language supporting but not language critical processes. Higher absolute LIs, which reflect unambiguously hemispheric language dominance, can be thus obtained.