Connectionist investigations of individual differences in Stroop performance

A simple connectionist model for the Stroop task was implemented to investigate a defective inhibition hypothesis first postulated by Fox in 1994 for individuals of high anxiety. Simulations with this model indicated that Stroop interference was positively, instead of negatively, related to inhibition. An alternative excitation hypothesis assuming that high anxiety corresponds to stronger verbal connections to the distracting material could also be investigated in the model. Stronger verbal connections indeed corresponded to larger interference. It is argued that this type of simulation could be useful, not because it proves in any way the supposed mechanisms, but because it clarifies the relation between theory and empirical data and so stimulates further specification of theoretical assumptions and invites empirical research into these assumptions.