Affective priming studies have demonstrated that most stimuli are unintentionally, and in that sense automatically, evaluated. One functional explanation for this automatic evaluation is that it exists to predispose appropriate behaviours, allowing them to be executed rapidly and efficiently. Using a lever based experiment Chen and Bargh (1999) have shown that pleasant evaluations predispose approach movements and unpleasant evaluations predispose avoid movements. A key aim of this thesis was to develop an approach/avoid paradigm suitable for patient testing. Whilst the evaluation stage has been extensively studied in patients, the behavioural stage has received less attention. Understanding the approach/avoid mechanism and the brain structures involved is important in predicting and treating the behavioural problems occurring after brain injury. Thus, I created an approach/avoid task using a touch-screen, and lateralized the stimuli to make it suitable for patient testing. This task produced large reliable congruency effects; following explicit evaluation healthy participants were unintentionally faster to approach pleasant and avoid unpleasant stimuli than vice versa. Despite several attempts, no congruency effects were seen following automatic evaluation. Using healthy participants I determined the time-course of the approach/avoid predisposition following explicit evaluation. Further, by reversing the meaning of the response effects in the task I reversed the congruency effect, and showed that the predisposed behaviours following explicit evaluation are goal based and semantically determined, rather than being specific inflexible movements. The performance of a confabulatory right frontal lobectomy patient on the task suggested that confabulation might be caused by an imbalance in the approach/avoid system, hypothesized to be lateralized across the frontal lobes. I also showed that unilateral left-sided amygdala damage slows the evaluation of contra-lesional unpleasant stimuli but does not impair the subsequent approach/avoid behaviour. Thus, this thesis consolidates two previously unconnected strands of psychological research; approach/avoid research and neuropsychological patient based work.