Accumulation of virtual tokens towards a jackpot reward enhances performance and value encoding in dorsal anterior cingulate cortex

Normatively, our decisions ought to be made relative to our total wealth, but in practice, we make our decisions relative to variable, decision-time-specific set points. This predilection introduces a major behavioral bias that is known as reference-point dependence in Prospect Theory, and it has close links to mental accounting. Here we examined neural activity in the dorsal anterior cingulate cortex (dACC) of macaques performing a token-based risky choice task, in which the acquisition of six tokens (accumulated over several trials) resulted in a jackpot reward. We found that subjects make faster and more accurate choices, and that they are less prone to risk-taking as offer contingencies are easier and the jackpot reward becomes more likely to be achieved. By comparing alternative models that accounted for progressive token accumulation, we found that subjective evaluations are best explained by a reference-dependent value 'RDV' model where offer values are considered as potential gains or losses with respect to a token-dependent reference. The reference-dependent model allows to implement a dynamical comparison of the two offered values to each other and to the number of missing tokens to reach the six-tokens threshold as jackpot approaches. In dACC, we find that gains in subjective values entail higher fractions of encoding cells than losses, and that the encoding tuning of expected utility variables is best aligned with upcoming choices in gains than in losses. These results suggest a neural basis of reference dependence biases in shaping decision-making behavior and highlight the critical role of value representations in dACC in driving evaluations.