Amitai Shenhav

Humans and other animals have a wide repertoire of actions they can engage in at any given moment, ranging from moving their limbs to generating utterances to controlling the flow of information from perception to action. What motivates them to engage some of these actions and not others at any given moment? To answer this question, motivation scientists have historically leveraged three kinds of explanatory variables: goals (what directs action), value (what determines a good or bad action), and affect (what it feels like if an action is or isn’t taken). I will argue that each of these variables faces significant challenges in explaining motivated behavior: Goals can explain how actions are directed, but not what directs goals; value can be assigned to an action, but it is a mystery how these values are assigned and how they are shaped by both “cold” and “hot” systems; and affect can explain what actions will make you feel best, but under most definitions affect is neither specific nor reliable enough to guide action selection. My Affective Gradient Hypothesis (AGH) proposes a way out of this entanglement, by re-centering motivation on two sets of processes: prediction and optimization. AGH suggests that motivated behavior arises from the continual representation of potential future states and their affective value, and the dynamic adjustment of available actions and control states to achieve states that are anticipated to be most positive and/or least negative. This account revises traditional definitions of goals, value, and affect to overcome existing obstacles and to provide a complete, coherent, empirically grounded, and generalizable account of what drives motivated behavior.

The fact that people often avoid making decisions is well known, and past research has helped to identify some of the conditions and reasons for doing so. For instance, people may forgo choices between bad options because they prefer not to end up with one of those options. It is much less clear when and why people avoid choosing in cases where they eventually will have to make a given decision. To study such instances of choice deferral, we presented participants with a series of choices, and for each choice they were allowed to either choose immediately or defer the decision until later in the experiment. Across six experiments and three choice domains (choices among consumer goods, artwork, and political candidates), we find that the strongest predictor of choice deferral is the overall value of a given set of options, with relative value (i.e., how hard it is to identify the best option) counterintuitively playing a smaller role. We show that the influence of overall value on choice deferral can be accounted for by a dynamic decision model according to which participants appraise the option set relative to a criterion before deciding whether to choose or defer, comparing this to a previous model whereby participants make such a decision based on a predetermined decision time limit. We further reveal that the influence of overall value on choice deferral is determined by how congruent options are with a given choice goal (choose-best or choose-worst) rather than simply how bad those options are. Collectively, our findings shed new light on how people decide to put off the inevitable.

To behave adaptively, people need to integrate information about probabilistic outcomes and balance drives to approach positive outcomes and avoid negative

Representing the will of voters is challenging when many opt out of indicating their preferences. Such opt-out behavior has been explained by voters lacking a preference and/or disliking their options. We provide evidence for a third account: people opt out of choosing between undesirable candidates because bad options are incongruent with their typical goal of selecting the best one. Using a voting task, we show across two lab-based studies that the tendency to opt out of choices between bad candidates is eliminated when participants are asked to reject the worst candidate. Leveraging our experimental findings, we simulate elections and show that rejection-based voting can produce election outcomes that are more representative of the preferences of the electorate. To validate this prediction, we conduct two Prolific surveys of self-reported US Independents before the 2024 US presidential election, and show that people are less likely to respond “undecided” when asked who they will vote against rather than who they will vote for. Our findings help understand when and how people vote, and how to better reveal the preferences of voters who know which candidates they like least but are unwilling to endorse the one they like most.

Decision scientists have grown increasingly interested in how people adaptively control their decision making, exploring how metacognitive factors influence how people accumulate evidence and commit to a choice. A recent study proposed a novel form of such adaptive control, whereby the values of one's options contribute to both the formation of a decision and the effortful invigoration of a response. In this framework, the control process was operationalized in a drift diffusion model as the lowering of the decision threshold on difficult trials. Reanalyzing the data from this experiment, we establish alternative explanations for these findings. We show that the reported evidence for controlled threshold adjustments can be explained away by task confounds, time-dependent collapses in decision thresholds, and stimulus-driven dynamics in an alternative form of evidence accumulation. Our findings challenge the specific evidence for this new theory of motivated control while at the same time revealing paths and pitfalls in computational approaches to a more general understanding when and how control guides decision-making.

Declines in cognitive flexibility are a hallmark of cognitive aging, but their causes remain elusive. Here, we examine a previously untested source of aging-related cognitive inflexibility, building on a dynamical systems framework for flexible cognition. We propose that cognitive inflexibility can arise in part from slower transitions through the space of distinct configurations of cognitive control needed to pursue different goals. To test this model, we had participants across the lifespan (aged 19-88) perform a cognitive task under different performance goals, each of which induced a different configuration of cognitive control. Using computational modeling, we measured dynamic changes in two control signals (attentional focus and response caution) as participants pursued distinct goals. This allowed us to test three potential sources of age-related decreases in cognitive flexibility: 1) diminished control capacity in environments that require more goal switching; 2) diminished range of control adjustments; and 3) slower transitions between control configurations. Of these, we found that age was only associated with transition speed. When given sufficient time to maintain a given goal, older adults were able to adjust control to a similar extent as younger adults. However, older adults were more likely to undershoot their target control configuration when frequently switching between goals, consistent with longer transition times between configurations. Our findings demonstrate the critical role that cognitive dynamics play in explaining the mechanisms through which cognitive inflexibility arises in older adulthood. Significance: Meeting the shifting demands of daily life requires a person to constantly adjust the way in which they allocate their cognitive resources. As people get older, it becomes increasingly difficult to achieve the same levels of cognitive flexibility. We show that these deficits can be explained by age-related changes in the dynamics governing adjustments between cognitive states. Older adults take longer to adapt their cognitive state to a new goal, despite demonstrating preserved ability to reach desired states when given sufficient time. These findings validate a new model of cognitive flexibility and its variability across individuals, including forms of inflexibility seen in healthy aging and across a range of psychiatric and neurological disorders.

To behave adaptively, people need to integrate information about probabilistic outcomes and balance drives to approach positive outcomes and avoid negative outcomes. However, questions remain about how uncertainty in positive and negative outcomes influence approach-avoid decision-making dynamics. To fill this gap, we developed a novel Probabilistic Approach Avoidance Task (PAAT) and characterized behavior in this task using sequential sampling models In this task, participants (Study 1: blinded mixed clinical sample N=34; Study 2: online nonpsychiatric sample N = 58) made a series of choices between pairs of options, each consisting of variable probabilities of reaching a positive outcome (monetary reward) and of reaching a negative outcome (aversive image). Participants tended to choose options that maximized the likelihood of reward and minimized the likelihood of aversive outcomes. Moreover, the weights they placed on each of these differed for choices where these likelihoods were in opposition (i.e., the riskier option was also more rewarding; incongruent trials) relative to when these were aligned (congruent trials). Computational modeling revealed that the relative influence of rewarding and aversive outcomes on choice was captured by differences in the rate of decision-relevant information accumulation. These modeling results were validated with a series of model comparisons and posterior predictive checks, demonstrating that our sequential sampling models reliably captured our behavioral data. Together, these findings improve our understanding of the influence of motivational conflict, outcome type, and levels of uncertainty on approach-avoid decision-making.