Two experiments compared the effects of four training conditions on proposi
tional reasoning. A syntactic training demonstrated formal derivations, in
an abstract semantic training the standard truth-table definitions of logic
al connectives were explained, and a domain-specific semantic training prov
ided thematic contexts for the premises of the reasoning task. In a control
training, an inductive reasoning task was practised. In line with the acco
unt by mental models, both kinds of semantic training were significantly mo
re effective than the control and the syntactic training, whereas there wer
e no significant differences between the control and the syntactic training
, nor between the two kinds of semantic training. Experiment 2 replicated t
his pattern of effects using a different set of syntactic and domain-specif
ic training conditions.