A few snippets from this interesting study:
1. The brain processes concrete information (like "A man goes through a house") very differently from abstract information like metaphors ("A man goes through a sorrow").
2. Comprehending the concrete description activates visual and visual association areas ("I see"), whereas comprehending the abstract or metaphorical sentence, activates previous experience (temporal lobes) and more classic language pathways (BTW, we desaturated the color of the "concrete" figure to make it easier to compare).
The "I see" pictures are much more like the view from children's patterns of activation (see second link below) than adult's patterns on verbal fluency tasks. When a child doesn't understand, it probably means we should backtrack to see if a change in our words (or physical example) can really help them "see" what we mean.
3. Another interesting result from this study was the finding that abstract language was much better understood by reading than by listening. We've seen this. It's probably a lot more common than people realize (i.e. the I'd-rather-read-a-book person).
Concrete vs. Abstract Language and fMRI
Eide Neurolearning Blog: Children Think in Pictures, Teachers Think in Words