Transient and sustained BOLD signal time courses affect the detection of emotion-related brain activation in fMRI

Abstract

A tremendous amount of effort has been dedicated to unravel the functional neuroanatomy of the processing and regulation of emotion, resulting in a well-described picture of limbic, para-limbic and prefrontal regions involved. Studies applying functional magnetic resonance imaging (fMRI) often use the block-wise presentation of stimuli with affective content, and conventionally model brain activation as a function of stimulus or task duration. However, there is increasing evidence that regional brain responses may not always translate to task duration and rather show stimulus onset-related transient time courses. We assume that brain regions showing transient responses cannot be detected in block designs using a conventional fMRI analysis approach. At the same time, the probability of detecting these regions with conventional analyses may be increased when shorter stimulus timing or a more intense stimulation during a block is used.In a within-subject fMRI study, we presented aversive pictures to 20 healthy subjects and investigated the effect of experimental design (i.e. event-related and block design) on the detection of brain activation in limbic and para-limbic regions of interest of emotion processing. In addition to conventional modeling of sustained activation during blocks of stimulus presentation, we included a second response function into the general linear model (GLM), suited to detect transient time courses at block onset. In the conventional analysis, several regions like the amygdala, thalamus and periaqueductal gray were activated irrespective of design. However, we found a positive BOLD response in the anterior insula (AI) in event-related but not in block-design analyses. GLM analyses suggest that this difference may result from a transient response pattern which cannot be captured by the conventional fMRI analysis approach. Our results indicate that regions with a transient response profile like the AI can be missed in block designs if analyses do not account for transient responses. This may bias conclusions from empirical reports and meta-analyses towards an underestimation of these regions and their role in emotion and emotion regulation. The cognitive processes underlying differential time courses are discussed.