A loyal Marginal Revolution reader asks about prospect theory [an extension of expected utility theory where local utility is based on a perceived expected outcome]. I have never had occasion to apply it. My gut reaction is that it is a tasteful exercise in numerology: people demonstrably do not maximize expected utility, so let's add another unobservable parameter to increase the flexibility of our theory. On the plus side, though, it provides a good
Since I am following Mr. Cowen's agenda, let me also follow his example and slip into dispensing advice. To wit, your utility function is probably flatter than you might imagine, and you are better off acting to maximize expected utility even when it seems counterintuitive. In short, don't spend much effort avoiding small risks.
[Mr. Cowen's comments are here.]