Figure 4.

(A) During a two-modality continuous performance task, subjects monitored a letter stream for successive verbal targets (A then X) or successive spatial targets (3 then 6 o'clock positions). Three correct targets within a modality were rewarded. Reward expectations lead to a graduated bias toward verbal or spatial cognitive sets, according to the recent history of spatial versus verbal targets. (B) The effects of this “top-down” modulation from cognitive set were studied using dynamic causal modeling of fMRI data. The figure shows modulatory (bilinear) effects representing psycho–physiological interactions in the most likely causal model (selected by Bayesian model comparison). This model included the medial frontal (MF) cortex, the dorsal (PFd) and ventral (PFv) lateral prefrontal cortex, the superior frontal sulcus (SF), the intraparietal cortex (IP), the fusiform gyrus (FG), and the prestriate cortex (PS), with intrinsic connections indicated by the presence of arrows (of any color). Values are time constants (Hz) for the modulatory influences of task bias for which the group posterior mean was positive (solid lines) or negative (dashed lines) for verbal bias (thick green), spatial bias (thick red), or both (thick black). These modulatory effects have strong evidence that they are non-zero, confirmed by post hoc t-tests. The “top-down” modulation of task set resulting from higher reward expectations was associated not only with changing connectivity of the lateral prefrontal cortical regions, but also the feed-forward connections from pre-striate cortex. Moreover, the feed-forward connections were enhanced to parietal cortex with spatial task set bias, and to temporal cortex with verbal task set bias. This illustrates that domain specific “top-down” control is not restricted to changes in feedback connections from higher cortical areas, but is also manifest by changes in feed-forward connectivity. From Rowe et al. (2008a).