Should You Trust Your First Instinct?

Grant's Thinking book

Source: J. Krueger

But now I might be mistaken
Hmm, hmm, hmm - Top, ZZ

There is a folk psychological belief that says it is best to go with one’s first instinct when making a decision or considering a course of action. Yet, we know that this is not always true, and so there is a lingering sense of needing to think more and more carefully. The task of science is to sort out when first instincts are best and when they are not. Once we are informed about the proper conditions, we can make better decisions and be happier.

Complex stories about the conditions of judgment are difficult to tell and it is hard to find a receptive audience. Our first instinct is to wish for first instincts to be valid. Because if it is so, we gain much while paying little, and this satisfies our desirability bias (i.e., wishful thinking). With Think Again, Adam Grant (2021) has written an excellent and wide-ranging trade book on the matter, which I have reviewed in full elsewhere (Krueger, in press). Here, I want to add a note on the first-instinct fallacy in its narrow sense.

What Is the First-Instinct Fallacy?

The term first-instinct fallacy was introduced by Justin Kruger and colleagues (2005) in an article showing that students who revised their test answers obtained better scores than students who did not. The first-instinct fallacy is not just that the first answers were worse than the second answers, but that the students believed the opposite.

Asymmetrical counterfactual thinking explains the phenomenon. If you revise your first-instinct response and find out that the first response was correct, you experience regret and you can anticipate this regret. If you stick with your first response and an alternative turns out to be correct, there is also foreseeable regret, but it is weaker and more likely to fade with time.

The reason is that the first kind of regret bemoans the committed action of having switched from a correct to an incorrect response, whereas the second kind of regret bemoans the inaction of not having switched from an incorrect to a correct response. Action looms larger than inaction. It is more painful to tell oneself "If only I had not changed my mind" than it is to say "If only I had changed my mind." The former carries a stronger element of guilt.

Why were the second responses better in the Kruger study? The student participants were free to decide whether they wanted to change their answers. If a test is not too difficult, more than 50 percent of responses are correct, including first ones. Most tests are designed to allow this. If all answers were changed, accuracy would fall below 50 percent. Hence, the selectivity of when and where to make a change is crucial.

In discussing this research, Grant notes that the decision to change likely depends on the confidence gained from rethinking. That is, students change their answers only after convincing themselves that the change brings an improvement, and this is more likely true than false. In short, thinking again and changing one’s response is beneficial if this additional thinking increases confidence and if confidence is correlated with accuracy.

When Is Guessing Again Helpful?

This reconstruction of the research suggests a nuance to the first-instinct fallacy. It is not the case that people think their first instincts are always best—as already noted above. Were it so, they would never take up the invitation to think again.

Let’s rephrase the first-instinct fallacy by looking at an estimation task as opposed to a binary (or multiple) choice task. Suppose you are asked to estimate Martin Luther’s birth year. You might reasonably narrow the window to the second half of the 15th century, but after that, you’re guessing. If your guessing is random, your second guess will be no better than your first, on average. If there are remnants of valid knowledge, you can improve; if not, your second guess might steer you away from the truth simply because you wish to generate a different number.

With Leonard Chen, I looked into this question, and we found that first guesses were better than second guesses (Krueger & Chen, 2014). No first-instinct fallacy there. But the second guesses were not useless either. When first and second guesses are averaged, the result is often more accurate than either guess alone. This process of resampling one’s own mind is known as dialectical bootstrapping, and it reveals the wisdom of the crowd within one’s own mind.

Trust Essential Reads

What the Flint Water Crisis Revealed About Trust

Algorithm Aversion: Who Trusts Algorithms to Make Decisions?

Supposing now our first instinct says that Luther was born in 1475, a number we obtain by splitting the range from 1450 to 1500 (Grüning & Krueger, 2021). We only later learn the actual year of his birth was 1483. Before we do, though, we generate a second estimate. If the second estimate is lower than 1475, it is worse than the first and there is no wisdom of the crowd. If the second estimate is higher than the first—e.g., the year 1500 looms as a salient anchor—it is also better as long as it does not err more on the other side of truth, that is, as long as it does not exceed 1501. Yet, the second estimate may be as high as 1507 before the average of the two estimates is farther away from the true value than is the first estimate [1]. In short, the second estimate is not necessarily better than the first when we are forced to produce one, but it is likely to improve accuracy when averaged with the first estimate.

Following Grant’s advice to not forget humor, I close with a nod to Seinfeld’s "The Opposite" episode. Doing the opposite of what your first instinct tells you, as a matter of principle, makes for great comedy.

My name is George. I'm unemployed and I live with my parents.

—George Costanza making contact with the other gender while ignoring his instinct not to.

More of the show's dialogue can be read here.

[1] The second estimate, Y, can be as large as 4T - 3X, where T is the true value and X is the first estimate.

For more on first instincts and the fidelity of you inner voice, see Krueger 2009.