Putting the brain back together: Mechanisms of interhemispheric integration in face perception

The fact that the two hemispheres use different strategies to process information is well established. It is still unclear, however, how they integrate their representations. The current study used a face matching task to test two hypotheses: First, whether the benefit of interhemispheric integration is greater when the two hemispheres process visual information differently (upright faces) than similarly (inverted faces). Second, whether interhemispheric integration is interrupted when the visual input is bilaterally inconsistent. Four types of facial stimuli were presented centrally in an upright or inverted orientation: Bisymmetric faces, which project consistent facial information to the two hemispheres; chimeric faces, which project inconsistent facial information to the two hemispheres and right and left hemifaces, which, respectively, project relevant facial information to the left and right hemisphere. The union of the accuracy levels of left and right hemifaces was used to estimate the performance for bisymmetric and chimeric faces of two independent hemispheres. Performance for bisymmetric faces surpassed the predicted performance of independent hemispheres and to a greater extent for uptight than inverted faces. This finding confirms the hypothesis that interhemispheric integration is beneficial to a greater extent when the two hemispheres process information differently than similarly. Performance for chimeric faces revealed interference relative to the predicted performance of two independent hemispheres. The magnitude of interference for chimeric faces was smaller than the magnitude of facilitation for bisymmetric faces, which suggests that regulatory mechanisms interrupt integration when the two hemispheres are presented with inconsistent information. Overall, our findings suggest that interhemispheric integration is adaptively regulated and is advantageous, in particular when the two hemispheres make different contributions to information processing.