A few months ago, I published an article on the Healthy Poker site (www.healthypoker.com) which I'm re-posting here.  Here is the introduction, and there is a link below.

Omar: “I don’t know about cards, yo, but I reckon these four fives beat a full house.”

Marlo: “That’s my money.”

Omar: “Man, money ain’t got no owners, just spenders.”

Omar, anti-hero from The Wire, holds up Marlo’s Poker game. Marlo protests that the money in the pot is his. Omar disabuses him of that idea.

We think are comfortably ahead on the flop, the turn is a blank and we fire when checked to. Now villain repops for the rest of his stack and most of ours. We only beat a bluff and we require a dispassionate and rational analysis of the probability that villain is bluffing. Yet if we’ve invested 2/3 of our stack, such unbiased rational analysis is extremely difficult. Loss aversion, a very useful attribute on the Savannah five hundred thousand years ago, inevitably biases our decision.

A rational analysis requires us only to consider the potential return on future costs, the amount required to call villain, (and the expected value of calling that amount, weighing his percentage of bluffs against made hands) yet behavioral economists have repeatedly demonstrated a systematic cognitive bias toward weighting historical costs in our decision making.

What we have put in the pot already is (should be) completely irrelevant to how much we should put in now. Having bet pre-flop, flop, turn and river, giving up to a shove should not be influenced by how much we have put in. Yet, psychologists and economist assure us, it does. 

An unhealthy attachment to the money in the pot remains.

If you think you can and do act optimally in such situations, consider the examples below and see whether they might apply more at the Poker table than you would care to imagine.

This sunk cost fallacy (or error) appears everywhere: the value at which we would sell a car or house should be unrelated to how much we paid; our decision to stay until the end or leave a horrible movie should depend only on the relative value of the remaining time and not what we paid for the tickets; Britain and France’s decision to continue Concorde development should have been based exclusively on the economic prospects for the project and not how much had already been invested; in stock trading, future buy-sell decisions should be made independently of whether we own the stock, or have an unrealized loss or gain.

http://www.healthypoker.com/live/2011/09/systematic-cognitive-error-in-poker-part-1-the-sunk-cost-fallacy/