Rewards induce learning (positive reinforcement), approach behaviour, economic decisions and positive emotions and mental states (pleasure, desire). We investigate basic neuronal reward signals during learning and decision-making, using behavioural and neurophysiological methods. We use specific behavioural tools to establish formal economic utility functions that constitute mathematical representations of behavioural preferences and predict the animal's choices. We find that the dopamine reward prediction error (RPE) signal codes economic utility, which may explain the maximisation of utility required for evolutionary beneficial behaviour. RPEs have specific valences whereby they act in specific directions; a positive RPE increases, and a negative RPE reduces, the frequency of actions that led to that RPE. Given that electrical and optogenetic activation of dopamine neurones mimicks positive RPE, decision makers would seek situations leading to positive RPEs and avoid negative RPEs, thus increasing the rewards they are getting. Such an ever-increasing reward profile would amount to utility maximisation.