We report an experimental study on the effects of diagrams on deductiv
e reasoning with double disjunctions, for example: Raphael is in Tacom
a or Julia is in Atlanta, or both. Julia is in Atlanta or Paul is in P
hiladelphia, or both. What follows? We confirmed that subjects find it
difficult to deduce a valid conclusion, such as Julia is in Atlanta,
or both Raphael is in Tacoma and Paul is in Philadelphia. In a prelimi
nary study, the format of the premises was either verbal or diagrammat
ic, and the diagrams used icons to distinguish between inclusive and e
xclusive disjunctions. The diagrams had no effect on performance. In t
he main experiment, the diagrams made the alternative possibilities mo
re explicit. The subjects responded faster (about 35 s) and drew many
more valid conclusions (nearly 30%) from the diagrams than from the ve
rbal premises. These results corroborate the theory of mental models a
nd have implications for the role of diagrams in reasoning.