How does an excess in a neurochemical lead someone to being paranoid about the intentions of their neighbour? And why does blocking a dopamine receptor improve this symptom? In this article we present a heuristic framework which attempts to link the biology, phenomenology and pharmacology of psychosis. Focussing on dopamine's role in reward prediction and motivational salience we propose that psychosis arises from an aberrant assignment of novelty and salience to objects and associations. Antipsychotics block dopamine receptors and decrease dopamine transmission, which leads to the attenuation of aberrant novelty and salience. This 'salience' framework accounts for existing data and questions several current assumptions about the speed of onset phenomenological effects of antipsychotics and their behavioral effects in animal models. We review new data to show that in contrast to the prevailing idea of a "delayed onset" of antipsychotic action, the improvement is evident in the first few days. Antipsychotics do not eradicate symptoms, but create a state of "detachment" from them. And the actions of antipsychotics in the conditioned avoidance response model, one of the best established animal models for identifying antipsychotic action, are consistent with the idea that they dampen aberrant as well as normal motivational salience. The article discusses the caveats, limitations as well as the clinical implications of the salience framework.