Subjective value is the single scalar variable that determines the impact of a reward on individual behaviour and welfare. But value cannot be measured; it can only be estimated from observed choice. To estimate value in animals requires that the animal’s choices are meaningful. Here are a few tests for meaningful choice. 

Left: The most simple one is first-order stochastic dominance: the biggest reward has most value and thus dominates the choice. With repeated testing for more accuracy, stochastic processes add to the choice mechanism. To be meaningful, choices should follow the dominating option: in choices between two options, the better option should be chosen in more than half the trials. This requirement holds as long as the value function is positive: more reward is always at least equal to and in at least one instance better than less reward ('more is better'). Monkeys in the laboratory pass the test easily: Top left: the reward in the red option is at least as good as in the blue option and bigger in half the trials; the red option dominates the blue option; top right: now the blue option dominates; bottom: the blue option dominates the red option due to its higher middle reward.