Needing is related to deprivation of something biologically significant, and wanting is linked to reward prediction and dopamine and usually has more power on behavioral activation than need states alone. The interaction and independence between needing and wanting in terms of brain functioning is not so defined, and it is unclear why sometimes there is association and sometimes dissociation between them. To investigate this question, we used a computational approach. In this approach needing is related to a deviation from preferred states that living things tend to occupy in order to reduce entropy, while wanting is related to precision over policy leading to rewards. We found that: (1) when need states increase, the tendency to occupy preferred states is enhanced independently of wanting (reward prediction), showing a dissociation between needing and wanting. We also found that (2) when need states increase, precision (i.e. saliency) of the reward leading to the preferred state is also increased, suggesting that need can amplify the value of a reward and its wanting. This led to the development of a model showing how needing is dissociated from wanting in that the tendency to be in a preferred state is independent of prediction of reward; and how needing can amplify wanting, by attributing saliency to relevant reward prediction.