Goal congruency dominates reward value in accounting for behavioral and neural correlates of value-based decision-making

Abstract

How we engage with a set of options (e.g., items on a menu) is affected both by the rewards they promise and our goal in choosing between them. Typically, our goal is to maximize potential reward and minimize potential punishment. Previous research on value-based decision-making has characterized how people make decisions with this goal in mind, identifying consistent behavioral and neural signatures associated with how rewarding a set of choice options are, overall and relative to one another. However, these studies suffer from a common confound: in each case, more rewarding options were also more congruent with ones goal of choosing the best option. Previous findings can therefore be accounted for by the reward value or the goal congruency of ones options. To compare these accounts directly, we had participants make choices while their goal varied between choosing the best or the worst option, resulting in situations where either high-reward or low-reward options were most goal-congruent. We found that goal congruency uniquely accounts for past observations that decision speed varies with the overall value of ones options. Within brain regions commonly associated with choice value, fMRI activity was associated with both relative and overall goal congruency. These goal congruency correlates were dissociable from separate correlates of the overall reward associated with a choice set (independent of ones goal). Our findings call for a reinterpretation of previous research on value-based choice, and offer an approach moving forward for disentangling the roles of rewards and goals in how we evaluate our options. Significance StatementWhether it is between restaurants or career paths, to make adaptive decisions we must evaluate our options and identify those that are most conducive to our current goal. Dysfunctional decision-making can therefore result from aberrant reward processing (e.g., impulse disorders) or from aberrant goal processing (e.g., OCD, ADHD). By focusing only on how people choose their favorite option in a choice set (when rewards and goals are aligned), past research has been unable to distinguish the degree to which behavior and neural activity are determined by reward versus goal processing. We disentangle these processes and show that behavior and fMRI activity are differentially influenced by the promised rewards versus the degree to which those rewards align with ones current goal.