If you reran the candle experiment but gave the participants the problem statement 24 hours in advance, I would wager that the incentivized group would be more motivated to think through the problem the night before.
Motivation is a vector, not a scalar. The problem isn't that the people who are incentivized by reward aren't motivated, it is that the motivation they are given makes them behave a different way. In the case of largely creative work, the motivation they are given modifies their behavior in entirely suboptimal ways.
It's not just that you motivate people, it is what you motivated them toward.