Bihrle, Amy M., Hiram H. Brownell, John A. Powelson, and Howard Gardner. 1986. Comprehension of humorous and nonhumorous materials by left and right brain-damaged patients. Brain and Cognition 5:399-411.
This study compared the narrative comprehension of people with damage to either the right or left hemispheres (RH vs. LH) of their brains (often from strokes or physical damage, like concussions).
They used a "completion task" where they were shown three panels from a comic strip, and were asked to provide either a humorous ending or a non-humorous ending with a choice of panels. The humorous ending was the original strip's final panel. The non-humorous ending was manipulated to have one of four varying degrees of coherency related with the strip:
1) ordinary endings with non-funny events that still made sense
2) endings associated to the strip but non-sequitur, such as featuring water if the strip had water in it
3) totally non-sequiturs that had no relation to the strip and weren't funny
4) humorous endings but that were non-sequitur given the context of the strip
Overall, they found that people with right hemisphere damage made far more errors than Left hemisphere patients. Left hemisphere damaged patients did far better, and several actually hit the ceiling of good performance.
In errors for the task that asked them to provide humorous endings, Right hemisphere patients generally erred by choosing endings that were surprising but not coherent, while left hemisphere patients chose unsurprising but coherent endings. Right hemisphere damaged patients were most drawn to non-sequitur endings of various types, and not to the ordinary ones. Left hemispheric damaged patients showed the opposite results.
However, when asked to provide a non-humorous ending, both types of patients generally chose endings meaningfully associated to the strip but still non-sequitur.
These results are consistent with the poor understanding of verbal narrative and jokes by right hemisphere damaged patients. Overall, they do seem to be sensitive to the formal properties of jokes (such as that an ending should be surprising), but they seem unable to establish coherency between panels.