# Reality Conditions

## Wednesday, July 19, 2006

### A Psychology Experiment

I took part today in a psychology experiment at university, and thought it would be interesting to describe it here.

The session in which I participated had about 12-15 participants. (There were several other sessions at other times as well). Each of us had to answer 6 question sheets, independently of all the others and without knowing anything about their answers (with a couple of exceptions I'll get to). Each question sheet would decide a payoff (between £0 and £20 in the lowest and greatest possibilities) and at the end of the session (which lasted for 1 hour) we rolled a dice and recieved the payoff according to that sheet, plus £10 just for participation. The question sheets were in many cases versions of classic experiments I knew about, like the Prisoner's Dilemma and the Ultimatum Game.

1) The first sheet obviously had the purpose of measuring risk-aversion. There were ten cases, and for each case two options to choose between. Option A always involved a payoff of either £4 or £5; option B always involved a payoff of either £9.60 or £0.25 (if I remember the amounts correctly). The probabilities varied between the ten cases: Case 1 was a decision between £4 certain against £9.60 with probability 0.1 and £0.25 with probability 0.9; Case 2 was a decision between £4 with probability 0.9 and £5 with probability 0.1 against £9,60 with probability 0.2 and £0.25 with probability 0.8; and so on until case 10 was an almost certain £5 against a certain £9.60. Obviously in the first cases one must choose A, and in the last ones B; there is an point in the middle where the mathematical expectated gain of B becomes larger, but risk-adverse persons will wait more than that before switching, for fear of the possibility of £0.25. I was tempted to take the mathematically "correct" choice just for the sake of it, but finally my risk-aversion won over me and I marked option A for all cases which had probability less than 0.7 for the £9.60. (If the dice at the end rolled 1, the payoff would be chosen by two more randomizations -one to choose which of the ten cases applied and one to choose the payoff according to the case's probabilities).

2) This was an interesting asymmetrical variation on the Prisoner's Dilemma. Half of the participants were assigned to be Senders, the other half Recievers. Each Sender had £4 and had to decide how much of it to send to a randomly paired Reciever. The Reciever would recieve that amount multiplied by 3, and could choose to send back any amount to the Sender. So for example if I choose to send £3, my partner (which, like I, remains anonymous) recieves £9 and if he chooses to send back £4, we will get each a payoff of £5 (I get the £4 plus the £1 I kept at the beginning). I was a Sender, and the choice was to "play sure" and not send anything (I get £4 certain) or to be nice and trustful sending money that will be multiplied, hoping that my partner will be grateful and reward me sending back more than the mere £4. I choose to be extremely "nice" and send the whole £4, risking to lose it all (nothing forces my partner to send back anything; we will never even know who each other was) but trusting that happy at recieving £12 the other person would send back, if not £6 for a fair share, at least £5 to make it worth for me. We Senders also had to fill in a form which amount we expected to recieve back; I wrote £5. Unfortunately my dice didn't fall on 2, so I will never know what the other person did.

3) This one was a simple Ultimatum Game. Senders had £10, and could offer any fraction of the amount to Recievers. Recievers had to accept, in which case the amount was shared as the Sender had proposed, or decline, in which case nobody got anything. Obviously a "rational" reciever would accept any amount, which is always better than nothing; but in practice if offered a low share people will refuse it out of spite to make the greedy Sender lose his part as well. I was a Sender again, and offered £4, which was the lowest amount I was sure anybody would accept.

4) This was a Future Discount experiment. We had again 10 cases, and for each case we had to choose option A or B. Option A was always recieving £10 within 2 weeks, and option B was recieving within 8 weeks an amount that progressed from £10.25 in Case 1 to... I can't remember how much, but probably as much as £20, for Case 2. I didn't care too much for the amounts and just circled option B in all cases! I'm not in a pressing need for money, and I guess the subjective probability I assign to suddenly needing ten extra pounds between the 2 weeks time and the 8 weeks time is significantly lower than 2.5%.

5) An Ultimatum Dilemma but without any option for the reciever. The sender just decides at will how to split £10, and the reciever has to accept it. Previously the reciever writes in his sheet how much he expects to recieve. (I was a reciever here, and I expected to recieve a paltry £1. Fortunately I didn't roll a 5 so this low confidence in the unadultered altruism of my fellows was not tested).

6) A Prisoner's Dilemma. Each of us was playing with a randomly slected (and anonymous) partner, and had to choose whether to "contribute" £0, £1, £2, £3 or £4 to a deal. The payoffs were escalated in the expected way, with the lower contributions always dominating the higher ones. For example, payoff if both contributed £0 was £4 for each, if one contributed £0 and the other one £4 it was £6 for the "defecter" and £3 for the "cooperator" and if both contributed £4 it was £5.30 for each. My decision, surely influenciated by my many readings of accounts that put the "Cooperation or Defection" dilemma at the very centre of Ethics, was to cooperate with £4 to keep my self-image high. I rolled a 6 at the end of the session, found that my partner had done almost as well as I did contributing £3, and I won £4.80.

So that was it. As to the purpose of the experiment (and its global results) I will only learn them when a paper is written (we were promised to be informed when this happens). As I thought that each of this particular experiments had been done thousands of times already, my idea is that the study aims at finding correlations -e.g between being risk averse and being a Defector. Any other ideas...? Any comments or violent disagreements with my decisions...?

• In an experiment as you describe, I would be unable to resist making choices to minimize my monetary return, both in terms of expected payoff, and sending all money to the reciever regardless of payoff. The money really doesn't matter - I'd be far more curious about the integrity of the researchers: would they discard all of my data as an outlier not fitting any of their hypotheses or would the final paper reflect the peculiar behavior?

By  Anonymous, at 9:10 PM, July 26, 2006

• Good post ! thanks for sharing

• excellent text I really liked the way you write the information, to have easier access and friendly, it helps me a lot for my college work