Dopamine neurons encode trial-by-trial subjective reward value in an auction-like task; Data and code

Personal choices fluctuate from moment to moment and thus reflect instantaneous subjective reward value. Owing to the constaints of choice-based tasks, there is no direct evidence documenting the neural basis of moment-to-moment changes in subjective value. The well-established Becker-DeGroot-Marschak (BDM) auction-like mechanism, allows participants to place bids that reveal their true subjective reward value on a trial-by-trial basis; untruthful bidding results in suboptimal reward (‘incentive compatibility’). In our experiment, male rhesus monkeys trained for several years to accurately place BDM bids for juice rewards without specific external constraints. Their bids for physically identical rewards varied from trial to trial with higher bids for larger reward amounts. In these highly experienced animals, responses of midbrain dopamine neurons reflected the trial-by-trial variations of bids despite constant, explicitly predicted reward amounts. Conversely, dopamine responses for similar bids were statistcally identical even for different physical reward amounts. Support Vector Regression accurately predicted animals' bids using as few as twenty dopamine neurons. Thus, phasic dopamine reward signals reflect instantaneous subjective reward value.