Regional brain activity when selecting a response despite interference: An H2 (15) O PET study of the stroop and an...

The Stroop interference test requires a person to respond to specific elements of a stimulus while suppressing a competing response. Previous positron emission tomography (PET) work has shown increased activity in the right anterior cingulate gyrus during the Stroop test. It is unclear, however, whether the anterior cingulate participates more in the attentional rather than the response selection aspects of the task or whether different interference stimuli might activate different brain regions. We sought to determine (1) whether the Stroop interference task causes increased activation in the right anterior cingulate as previously reported, (2) whether this activation varied as a function of response time, (3) what brain regions were functionally linked to the cingulate during performance of the Stroop, and (4) whether a modified Stroop task involving emotionally distracting words would activate the cingulate and other limbic and paralimbic regions. Twenty-one healthy volunteers were scanned with H2 (15) O PET while they performed the Stroop interference test (standard Stroop), a modified Stroop task using distracting words with sad emotional content (sad Stroop), and a control task of naming colors. These were presented in a manner designed to maximize the response selection aspects of the task. Images were stereotactically normalized and analyzed using statistical parametric mapping (SPM). Predictably, subjects were significantly slower during the standard Stroop than the sad Stroop or the control task. The left mideingulate region robustly activated during the standard Stroop compared to the control task. The sad Stroop activated this same region, but to a less significant degree. Correlational regional network analysis revealed an inverse relationship between activation in the left mideingulate and the left insula and temporal lobe. Additionally, activity in different regions of the cingulate gyrus correlated with performance speed during the standard Stroop. These results suggest that the left midcingulate is likely to be part of a neural network activated when one attempts to override a competing verbal response. Finally, the left midcingulate region appears to be functionally coupled to the left insula, temporal, and frontal cortex during cognitive interference tasks involving language. These results underscore the important role of the cingulate gyrus in selecting appropriate and suppressing inappropriate verbal responses. © 1994 Wiley-Liss, Inc.