Hacker News Reader

> the actual experiments failed to build trust in the experimentees

Psychological experiments involve trickery very often, because they need test subjects to be unaware of what experiments is trying to measure. I expect the most of subjects of psychological experiments to know this because they heard of other experiments. So I wouldn't expect that the trust building will change anything.

This particular experiment with a strange proposition to switch cups probably forces subjects to think what is happening, they inevitably form some hypotheses and run with them. I wouldn't believe that the results has something with a regret before I've read the research paper and have found there what researches did to ensure that they measured regret avoidance and not the ability of a subject to simulate their immunity to manipulations (if they thought that experiment tried to measure their resolve by coercing them to switch...).

Normally there is a post-experiment interview where researcher could try to learn the reasoning behind decisions of a test subject, but in this case a direct answer probably will not do. Though if "untrustworthy" researchers tricked test subject so they would get a reward in any case, then subjects would be happy and proud, and probably they would love to explain their "winning logic", even (or maybe especially) if they were mistaken about goals of experiment. It would be like researchers were too smart by half when devising the game and so they lost $5. I personally would feel an urge to brag about my "superior" thought process.

But in any case we need the original paper to learn how it was done and could we believe the interpretation of the experimental data.

> Psychological experiments involve trickery very often, because they need test subjects to be unaware of what experiments is trying to measure. I expect the most of subjects of psychological experiments to know this because they heard of other experiments. So I wouldn't expect that the trust building will change anything.

Long time ago, the joke was that "subjects are 19-23 years old women studying psychology and who happened to walk in the faculty halls that day".

+1 would be interested to see the actual study and can’t find it. There’s apparently a lot of work on the two effects mentioned here, Endowment Effect & Regret Minimisation. But it’s psych research so last time I heard, you could safely ignore about 40% of published papers.

https://link.springer.com/referenceworkentry/10.1007/978-1-4...

https://www.jstor.org/stable/1942711

The actual study is cited in the notes from which the article is excerpted.

https://journals.sagepub.com/doi/10.1177/0146167295217006?ic...

You might be interested in this review of choice-supportive behavior:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5723021/

I recognize a lot of this in “fan” communities. Where people invest a lot of time/money/effort in their hobby, they minimize regret by doing everything to convince themselves they made the right choice. All the classic tech holy wars are susceptible (Mac vs PC, Vim vs Emacs, Xbox vs PlayStation, dozens more…). It’s almost impossible to get a direct measure of regret (e.g. by asking them) on these emotive topics because admitting regret after spending all that time/money/effort is psychologically devastating.

The person I know who exhibits this pattern the most is constantly trying to get the people around him to buy the same things he has bought, because it validates the decision. I consistently regret buying things on his recommendation. I think he tries hardest to recommend the worst ones because they're the biggest sources of regret.

He's otherwise an amazing friend though. The earnestness that makes him a sucker for limited edition scotch comes out in other nicer ways too.

Yeah, IMO the only reason not to switch at $6 is mistrust at the person conducting the experiment: they're tricking me to switch because they know I won.

If I knew it was double blind, I'd switch instantly.

This seems so obvious that it’s baffling to me that the article doesn’t discuss this issue.

Agreed -- if they changed the experiment so it was verifiably trustworthy, I think most people who could do math would switch.

Basically, it would be a repeated game where the subject would be told by the experimenter.

1. I will put a black ball under exactly one of these two cups without you seeing it.

2. You will pick the first cup -- if you keep this cup and it's right, you get $5.

3. You will have an opportunity to change cups always, if you change cups and the changed cup is correct, you get $6.

I cannot see the above experiment failing to generate most people always choosing change if repeated 10 times, with the above structure being common knowledge ahead of time.

That's like saying "choose a cup with your left or right hand -- if you choose correctly with your left hand, you get $5, and if you choose correctly with your right hand, you get $6". Everyone would use their right hands.

> That's like saying "choose a cup with your left or right hand -- if you choose correctly with your left hand, you get $5, and if you choose correctly with your right hand, you get $6". Everyone would use their right hands.

From the article it seems that placing the chosen cup in front of the subject, to make it feel like they "own it", is an important part of the experiment.

> If I didn't know the experimenter was trustworthy (for example a street hustler in NYC), I definitely would not switch in the equal reward scenario and may even require 3x to switch because I would expect that they would only offer if my initial selection was wrong.

If you don't trust the experimenter at all, then you might as well not switch, but I don't think that is a better than switching, as you have no basis for assuming any outcome in any circumstance.

If you are dealing with a street hustler in NYC, the optimal move is to disengage.

WRT the first (equal rewards) scenario: a weak bias towards sticking with an original decision might be favored by evolution, as constantly changing one's mind without new knowledge is not optimal behavior.

Indeed it is weird to theorize why somebody should switch if they trusted the process to be random. There's no reason to. And if I'm not sure about the process, indeed I may want to stick with my choice. I was wondering how a study would control for this. One could make a group where the price was shared with the person doing the shuffling. If people in that group were more inclined to switch, it would indicate that they don't trust the process to be random.

Make it overtly double blind.

If I assume they are trying to trick me, what blinding procedure could be added to convince me they are not?

> If I knew the experimenter was trustworthy, I would not switch after the initial decision because there is no point to and definitely would for the $6 alternative.

One rational reason not to switch for $6 would be the hope/expectation that the offered incentive will increase. In the experiment, the subjects who held on to their first choice the longest received 10x the reward, while the quick switchers got only 1.2x (assuming the alternative cup was winning).

> the subjects who held on to their first choice the longest received 10x the reward

If the experiment was designed right, there wasn't a second offer to the same subject and it was made clear that there won't be one. Instead separate test groups were offered different amounts. Or at the very least the game was repeated.

I think this is 100% correct, and very surprised the author of the article and/or study didn’t figure this out.

Here’s how you could (possibly) prove that.

Instead of the options being you get the money or not, the options become you win you pocket the money, you lose the money goes to a charity of your choice.

I think 100% the person would always choose the outcome with more money, regardless of the level of trust with the person giving the experiment.

That is, you completely remove trust from the equation.

But you also remove risk and regret (to some extent, at least).

Yeah, fair point.

I wonder if you could do it with a coin, and flip the coin after the decision was made? I think that reintroduce risk and regret, but remove trust.

Yes, and it's hard to imagine feeling much regret at failing to win $5. Maybe if the prize or penalty was something more significant, the results would be different?

> Yes, and it's hard to imagine feeling much regret at failing to win $5.

And that how you revealed yourself as a replicant.

That would actually explain quite a bit.

I came here to say exactly this. Nothing in the article says that the operator doesn’t know where the red die is, and the subject cannot know that every subject is always offered the switch.

So they think they’re being tricked into it. Perfectly rational.