For a complete treatment of the Candle Problem and several other cases of balderdash masquerading as science, go here.

In summary:

The Candle Problem is an interesting experiment, and probably valid with respect to the performance of specific tasks against tight deadlines. I think the results apply whether the stakes are money or any other kind of prize. The experiment illustrates the “choke” factor, and nothing more profound than that.

I question whether the experiment applies to the usual kind of incentive (e.g., a commissions or bonus), where the “incentee” has ample time (months, years) for reflection and research that will enable him to improve his performance and attain a bigger commission or bonus (which usually isn’t an all-or-nothing arrangement).

There’s also the dissimilarity of the Candle Problem — which involves more-or-less randomly chosen subjects, working against an artificial deadline — and actual creative thinking — usually involving persons who are experts (even if the expertise is as mundane as ditch-digging), working against looser deadlines or none at all.