Substitution
Substitution is what the fast, automatic part of the mind does when a hard question has no ready answer: it answers an easier question instead, and translates the answer back onto the original scale so that it arrives dressed as a reply to the question actually asked. Daniel Kahneman and Amos Tversky built the account. The part that makes it more than a curiosity is that the swap is never noticed. This page is marked mixed because the mechanism is well established and the evidence as the popular source presents it is not checkable.
What it is
Nobody decides to answer a different question. There is no moment of choosing the substitute and no felt sense afterwards that a substitution occurred, which is why the resulting judgement carries the confidence appropriate to the easy question rather than to the hard one. Intensity matching does the translation: a strong feeling about the easy question becomes a high number on the hard one's scale.
This is the general mechanism sitting underneath the named heuristics. Representativeness, availability and the affect heuristic are all substitutions with different easy questions plugged in.
It is worth separating this sense of heuristic from the older one, because the two get conflated constantly. George Pólya's heuristics, from How to Solve It, are deliberate strategies: work backwards, solve a simpler related problem, draw a diagram. They are tools a person chooses to pick up. The heuristics in Kahneman's sense are unchosen. They run before deliberation and produce an answer the person then experiences as their own considered view. Same word, opposite relationship to intention, and anyone writing about heuristics for a general audience should say which they mean.
The worked case in Thinking, Fast and Slow is the mood heuristic for happiness, a survey of German students. Asked first how happy they were and then how many dates they had had recently, the correlation between the two answers was about zero. Reversing the order, so that the dating question came first, made the correlation, in Kahneman's phrase, about as high as correlations between psychological measures can get. The dating question loaded an emotionally charged fact into mind, and the happiness question was then answered by reading off current mood rather than by evaluating a life.
The book names nobody for that study, gives no correlation value and gives no sample size. Three gaps in one paragraph. This page reports the shape of the result and puts no number on it.