The ultimatum game

One of the things it has taken economists a surprisingly long time to realize is that people don't always act rationally. And yet there's a simple little game psychologists have used for some time that demonstrates beautifully why we can't be relied on to do the sensible thing. It's a game in which people are offered money with no strings attached, and turn it down.

Here's how it works. The person running the game offers a sum of money, say £10, to two people. The first person's role is to decide how the money is split between the two of them. They might go for a fair 50:50 or could decide to keep 99% for themselves - the choice is theirs. But here's the thing. The second person can then say 'Yes' or 'No'. If player 2 he says 'Yes', the money is awarded according to the split the first person decided on. If player 2 says 'No', neither of the players gets any money. This is a one-off game - it's not repeated.

Now, if player 2 was making a rational decision, they would say 'Yes' whatever the split. Because, however small the amount, they are getting money for nothing. In practice, though, unless player 1 offers the second person around 30% of the cash, the tendency is to say 'No'. They'd rather lose the money than be screwed.

That's as far as the game is usually reported. I must admit, I'd like to see more - specifically a range of games with different stakes. While I would certainly say 'No' if I was offered 1p out of a pound, I would definitely say 'Yes' to £1 million out of £100 million. It would be interesting to see where different people's breakpoint between 'Yes' and 'No' was - and to try to identify why that breakpoint is at that level.

It's probably quite a sophisticated decision. For instance, I would say 'Yes' to £1 if the other player was getting £9, but I would say 'No' if they were getting £999,999. What would you do? (And does anyone know of research going into this more detailed analysis of behaviour?)