Why a reward you might get beats a reward you will get

In one condition, finishing the task earns two dollars, guaranteed. In another, finishing earns either one dollar or two, decided by a coin the experimenter flips at the end. The expected value of the gamble is lower, yet the people offered the uncertain reward complete the task at a higher rate and report enjoying it more.

What everyone sees

Standard reasoning says a certain reward should dominate an uncertain one of equal or lower expected value — people are supposed to dislike risk, especially for gains. On that account the gamble should depress effort. It raises it. The uncertainty is not being tolerated despite its cost; under the right framing it is contributing value of its own, and the expected payout understates how motivating the offer is.

What is actually happening

Shen, Fishbach and Hsee reported what they named the motivating-uncertainty effect: across several experiments, people exerted more effort for an uncertain reward than for a certain one of equal or greater expected value — but only when they were focused on completing the task rather than deliberating over whether to start. Schultz, Dayan and Montague’s work on dopamine neurons supplies a mechanism at the neural level: reward-predicting signals respond most strongly to uncertainty, peaking when the probability of payoff is around one half, so the anticipation of an unresolved reward is itself experienced as rewarding.

Why it stays hidden

The effect hides because it reverses only inside the doing, not the choosing. Asked in advance which reward they would prefer, people pick the sure thing and mean it, so the standard intuition is confirmed at the moment anyone would test it. The extra pull of uncertainty appears only once the person is already engaged, where it is felt as fun or momentum rather than as a response to risk.