In conditional reasoning, the reasoner must draw a conclusion based on a conditional or “If…, then…” proposition. Previous studies have reported that reversing the premises can effectively promote modus tollens reasoning (a form of conditional reasoning), but subsequent experimental studies have found no such effect. Therefore, to further examine this issue and reveal the cognitive mechanism of conditional reasoning, we asked two groups of healthy volunteers (traditional and inverted premise order groups) to evaluate a set of visually presented conditional tasks (modus ponens/modus tollens) under fMRI. The results indicated that the inverted condition activated more brain regions associated with working memory, including the angular gyrus (BA 39), precuneus (BA 7), inferior parietal lobe, and middle frontal gyrus. The resulting common activation map was used to define the ROIs and perform dynamic causal modeling for the effective connectivity analysis, containing the medial frontal gyrus, hippocampus, cerebellum, and middle occipital gyrus in the right hemisphere and the inferior occipital gyrus in the left hemisphere. The results of intrinsic connections in the optimal model selected by Bayesian model selection showed that the connection strength was stronger in the inverted group rather than in the traditional group, which may indicate that the reversal of the premise order promotes connectivity between brain regions. Despite the lack of a premise order effect, we did discover a neuronal separation between the inverted and traditional conditions, which lends support to the mental model theory to some extent.