Pii: S0191-8869(99)00169-5

We evaluated the validity of Gray's and Fowles' three-arousal model in two studies of criminal o€enders using a continuous motor task involving rewards and punishments. Consistent with predictions for the behavioral approach/activation system (BAS), o€enders displayed signi®cant ( p < 0.00001) increases in response time and heart rate (HR) from a no-incentive practice phase to a reward-only (experiment 1) and active-avoidance (experiment 2) phase. Trait impulsivity was correlated with the response time index of BAS activation in experiment 1 but not experiment 2. Consistent with predictions for the behavioral inhibition system (BIS), o€enders showed a signi®cant increase in number of skin conductance responses (SCRs) ( p < 0.05) from a reward-only to a mixed-incentive phase in experiment 1 and a signi®cant increase in SC amplitude to punishment cues in both experiments. Consistent with predictions for the dynamics of the model, participants showed signi®cant slowing of response time ( p <0.0001) from reward-only (experiment 1) or active avoidance (experiment 2) to mixed-incentive phases despite showing an initial tendency toward response facilitation to the onset of the punishment cue signifying the beginning of the mixed-incentive phases. Participants also showed signi®cant ( p < 0.002) decreases in HR between these phases in both studies, but this e€ect was only evident on trial 1. The BIS-in ̄uenced response time and HR indices were signi®cantly ( p < 0.05) correlated with anxiety in experiment 1, but unexpectedly, anxiety was not correlated with SC indices of the BIS in either study. Although much of the data support the validity of the Gray/Fowles model, particular ®ndings suggest that further re®nement of this theory is indicated. 7 2000 Elsevier Science Ltd. All rights reserved.