The relevant comparison should be closer to "unchanged" rather than "chance."
No, what kind of restaurant I want to go to changes based on my daily mood.
Yeah, it's interesting that they used restaurants here, out of all possible things to have opinions about. It's definitely plausible that peoples' preferences are about this variable about restaurants.
In the first session, participants saw two triples of restaurants, one that was used to devise a test pair for the second session and one that was used as a control pair. For simplicity, we label the three alternatives in each triple A, B, and C. For the first triple, participants chose between A and B. Then the loser of this pair—say, B—was compared with C. After choosing between B and C, participants performed a matching task, changing C (creating Restaurant C') so that it equaled B in overall attractiveness. Because B and C' were equally attractive and A was preferred to B, this process yielded a pair for the second session, A and C' , in which one alternative (C') was known to be inferior to the other (A).
So the second comparison wasn't between the original two choices. Instead, subjects were asked to modify a third choice to be "equal" to the one they rejected, and it was this third choice that got reworded and compared with their original preference. Restaurant preferences aren't all that solid to begin with, so I wouldn't read too much into this sleight-of-hand.
"Choosing an Inferior Alternative" (ungated version) is a 2006 paper1 by J. Edward Russo, Kurt A. Carlson and Margaret G. Meloy. It is presented as an example of cognitive biases leading to poor decision-making. In this case, participants were asked to rate options based on several characteristics, then the presentation of those options' characteristics was rearranged so that the inferior alternative's good points came first. With the new presentation, a majority of the same participants selected the alternative they had previously rated as inferior. You can read full details in the links above, and feel free to discuss methodological problems.2
This is an interesting result, and perhaps not too surprising here at Less Wrong. People have cognitive biases and can become committed to poor decisions, and you can manipulate those biases and decisions if you know what you are looking for.
What I want to focus on is the first sentence under "Results": "Just as they should have, a minority of participants (.41) selected the inferior [alternative] in the control choice, in which neither [alternative] was intentionally targeted." Either the methodological problems make it moot, or that is a fairly scary sentence.
Why is that scary? 41% of the participants picked what they identified as the inferior alternative in the previous round. The authors note it is below chance and move on. That is (1) not much below chance and (2) 41% of people whose preferences flipped when the presentation was rephrased. This is the control group, so their presentations were not manipulated to make the inferior alternative look good. Preferences change over time but having 41% change in two weeks in the control group seems like a big deal. The relevant comparison should be closer to "unchanged" rather than "chance." The treatment group saw 62% of participants change, so the treatment had about half as big an additional effect as the control.
The effect of rephrasing the options was almost twice as large as the additional effect of manipulating the order of that rephrasing. I am saying that a second time slightly differently because apparently that causes a huge effect in how we perceive things.
"Rationalists should win" has a problem if what we want to win can change that easily or quickly. Intentional manipulation is one thing, but many people switched choices without even being manipulated.
Part of me wants to dismiss this. Methodological problems! Unintentional re-ordering in the control group! Participants were not taking it seriously! The alternatives were restaurants, and I go to different restaurants all the time, not always the "superior alternative," so of course that changes on any given day! Hmm, that last one is pretty compelling, and it would leave the authors' original conclusion intact because the experimental group was significantly different than the control group, although it weakens the coherence of the idea of "an inferior alternative" for restaurants.
But part of me also thinks this is a fairer way to worry about things than our common, univariable considerations. It is rarely the case that Omega offers us a choice with a single, dollar-denominated outcome. We do that on purpose to focus discussion on the important point. Most of our real world choices are a complex mix of factors that are not directly comparable, such as between job options when one pays more, another has less commute, one of them offers more time off, and then their cultures differ in assorted ways large and small that might sound better or worse depending on how you feel that day.
So, on decisions more important than where to eat lunch, you should pause and weigh the alternatives, including rearranging the alternatives ("the enemy gate is down") and deciding which factors are most important to you and considering those first, rather than letting the salespeople for the alternatives present their strong points first. You should try to identify which of your preferences are more variable and put less weight on ones that might change even without manipulation. You should try to focus on your decision-making process in a way that counters biases rather than reinforcing them.
I remain somewhat troubled by the sheer size of that 41% change in a control group. But hey, this is a "Discussion" post, so please toss out your thoughts on our poorly mannered human preferences. Bonus points for working through how this affects our Coherent Extrapolated Volition.
1: It was published in Psychological Science, and I am aware of some issues with their quality control. I don't know if they apply here, but it seemed worth mentioning.
2: Please, go for it. I would except that it would distract from my point here.