Never Leave Your Room


Scott Alexander

Related to: Priming and Contamination

Psychologists define "priming" as the ability of a stimulus to activate the brain in such a way as to affect responses to later stimuli. If that doesn't sound sufficiently ominous, feel free to re-word it as "any random thing that happens to you can hijack your judgment and personality for the next few minutes."

For example, let's say you walk into a room and notice a briefcase in the corner. Your brain is now the proud owner of the activated concept "briefcase". It is "primed" to think about briefcases, and by extension about offices, business, competition, and ambition. For the next few minutes, you will shift ever so slightly towards perceiving all social interactions as competitive, and towards behaving competitively yourself. These slight shifts will be large enough to be measured by, for example, how much money you offer during the Ultimatum Game. If that sounds too much like some sort of weird New Age sympathetic magic to believe, all I can say is Kay, Wheeler, Bargh, and Ross, 2004.1

We've been discussing the costs and benefits of Santa Claus recently. Well, here's one benefit: show Dutch children an image of St. Nicholas' hat, and they'll be more likely to share candy with others. Why? The researchers hypothesize that the hat activates the concept of St. Nicholas, and St. Nicholas activates an idealized concept of sharing and giving. The child is now primed to view sharing positively. Of course, the same effect can be used for evil. In the same study, kids shown the Toys 'R' Us logo refused to share their precious candy with anyone.

But this effect is limited to a few psych laboratories, right? It hasn't done anything like, you know, determine the outcome of a bunch of major elections?

I am aware of two good studies on the effect of priming in politics. In the first, subjects were subliminally2 primed with either alphanumeric combinations that recalled the 9/11 WTC attacks (ie "911" or "WTC"), or random alphanumeric combinations. Then they were asked to rate the Bush administration's policies. Those who saw the random strings rated Bush at an unenthusiastic 42% (2.1/5). Those who were primed to be thinking about the War on Terror gave him an astounding 75% (3.75/5). This dramatic a change, even though none of them could consciously recall seeing terrorism-related stimuli.

In the second study, scientists analyzed data from the 2000 election in Arizona, and found that polling location had a moderate effect on voting results. That is, people who voted in a school were more likely to support education-friendly policies, people who voted in a church were more likely to support socially conservative policies, et cetera. The effect seems to have shifted results by about three percentage points. Think about all the elections that were won or lost by less than three percent...

Objection: correlation is not causation! Religious people probably live closer to churches, and are more likely to know where their local church is, and so on. So the scientists performed an impressive battery of regression analyses and adjustments on their data. Same response.

Objection: maybe their adjustments weren't good enough! The same scientists then called voters into their laboratory, showed them pictures of buildings, and asked them to cast a mock vote on the education initiatives. Voters who saw pictures of schools were more likely to vote yes on the pro-education initiatives than voters who saw control buildings.

What techniques do these studies suggest for rationalists? I'm tempted to say the optimal technique is to never leave your room, but there are still a few less extreme things you can do. First, avoid exposure to any salient stimuli in the few minutes before making an important decision. Everyone knows about the 9-11 terrorist attacks, but the War on Terror only hijacked the decision-making process when the subjects were exposed to the related stimuli directly before performing the rating task3.

Second, try to make decisions in a neutral environment and then stick to them. The easiest way to avoid having your vote hijacked by the location of your polling place is to decide how to vote while you're at home, and then stick to that decision unless you have some amazing revelation on your way to the voting booth. Instead of never leaving your room, you can make decisions in your room and then carry them out later in the stimulus-laden world.

I can't help but think of the long tradition of master rationalists "blanking their mind" to make an important decision. Jeffreyssai's brain "carefully put in idle" as he descends to a bare white room to stage his crisis of faith. Anasûrimbor Kellhus withdrawing into himself and entering a probability trance before he finds the Shortest Path. Your grandmother telling you to "sleep on it" before you make an important life choice.

Whether or not you try anything as formal as that, waiting a few minutes in a stimulus-free environment before a big decision might be a good idea.



1: I bet that sympathetic magic probably does have strong placebo-type effects for exactly these reasons, though.

2: Priming is one of the phenomena behind all the hype about subliminal advertising and other subliminal effects. The bad news is that it's real: a picture of popcorn flashed subliminally on a movie screen can make you think of popcorn. The good news is that it's not particularly dangerous: your thoughts of popcorn aren't any stronger or any different than they'd be if you just saw a normal picture of popcorn.

3: The obvious objection is that if you're evaluating George Bush, it would be very strange if you didn't think of the 9-11 terror attacks yourself in the course of the evaluation. I haven't seen any research addressing this possibility, but maybe hearing an external reference to it outside the context of your own thought processes is a stronger activation than the one you would get by coming up with the idea yourself.