Overwhelming evidence from the cognitive sciences shows that, in simple discrimination tasks (determining what is louder, longer, brighter, or even which number is larger) humans make more mistakes and decide more slowly when the stimuli are closer along the relevant scale. We investigate to what extent these effects are relevant for economic decisions. Strikingly, we find that even when there is an objectively correct answer independently of attitudes toward risk, the same effects obtain as expected values become closer. Contrary to pure discrimination tasks, however, differences in payoff-independent numerical magnitudes play a minor role. When correct answers depend on subjective attitudes toward risk, differences in expected values fail to explain error rates. The gradual effects on error rates and response times subsist but are instead explained by cardinal differences in independently-estimated subjective utilities (“strength of preference”). This is in agreement with assumptions typically made (but seldom validated) in random utility models. We conclude that the gradual effects on choice found in cognitive discrimination paradigms are very much present in economic choices, but depend on purely economic variables. An implication is that even if correct economic choices can be seen as ordinal, actual economic choices carry a cardinal component.