Original title: The three biggest obstacles to solving motivated behavior, and how to overcome them
Amitai Shenhav · September 2026
Why do animals and humans choose one action over another? Historically, scientists used three ideas to explain motivation: goals, value, and affectThe feeling component of emotion, like pleasure or displeasure.. Affect means how an action feels. But each idea has a major flaw. Goals direct actions, but we do not know what directs goals. Value labels actions as good or bad, but how those values get assigned across rational and emotional systems remains a mystery. Affect suggests which actions feel best, but it is typically not specific or reliable enough to pick an action. To solve these problems, the author introduces the Affective Gradient Hypothesis. This framework explains how motivation works through two processes: prediction and optimization. The brain constantly predicts future states and their emotional feelings. It then adjusts actions and mental control to reach outcomes expected to be the most positive or least negative. This perspective updates past definitions to fully explain motivated behavior.
You are choosing a movie to watch with friends. You try to pick based on what each person likes, but their preferences shift depending on which movie you suggest—someone who said comedy suddenly wants action when you pick a comedy. In this paper, the traditional ideas about why we act—like having a goal or a value—each break down because they assume stable guides that don't exist. The new hypothesis instead focuses on predicting how each possible action will feel and adjusting choices moment by moment.
For anyone studying why humans or animals choose one action over another—psychologists, neuroscientists, economists—the usual tools (goals, value, affect) each break down in key ways. This paper identifies those breakdowns and offers a new hypothesis that could replace them with a single, coherent account based on predicting future feelings and adjusting actions accordingly. If correct, it would change how we think about decision-making, addiction, and even everyday choices.
The abstract presents a theoretical proposal without empirical data or experimental tests. It does not specify how to measure the affective gradient or how to distinguish AGH from existing theories in practice. The hypothesis is offered as a framework, not a set of testable predictions, so its empirical grounding remains to be demonstrated.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
The three biggest obstacles to solving motivated behavior, and how to overcome them — Amitai Shenhav — PsyArXiv — doi:10.31234/osf.io/6xwbd_v1
Abstract licence: CC BY
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.