Modelling the Inverted-U Effect with Act-r

Properties of the Act-r conflict resolution suggest the possibility of modelling a behaviour at different activation levels of the autonomic nervous system. A model of the classical Yerkes-Dodson experiment was built to test the predictions. The resulting model explains such psychological phenomena as the inverted-U curve, relating performance to arousal, within the mathematical representations of equations in Act-r. Dynamical optimisation of the two crucial parameters, namely goal value and gain noise variance, may produce more optimal solution paths. We argue that subjects also perform such optimisation as including it into the models can produce a better match with the data. The dynamics corresponds to activational and motivational changes, related to experience of emotion during problem solving, which leads to interesting speculations about a role emotion in intelligence.