Landing on the Level

Definition

Faced with an ambiguous problem statement, people don't consciously choose an interpretation — they unconsciously "land" on whatever semantic level is most comfortable, usually the one where they already know how to produce an answer. A single word in the statement can shift the level entirely: adding or removing "familiar," or reversing it to "unfamiliar," moves respondents' answers from a confident, narrow guess to hesitation or wild speculation, revealing how much of "solving" a problem is actually just settling on its level before any reasoning starts.

In the Book

Shown a simple circle-like figure with the caption "a very familiar object," nearly everyone answers "circle" instantly. Strip the word "familiar," and the percentage giving that answer drops; strip "very" as well, and it drops further; change "familiar" to "unfamiliar," and confident answers vanish — replaced by "a hole," "a hula hoop," "the cross-section of an oblate spheroid," or no answer at all, because people now feel too much risk in guessing wrong. Adding "think of the most far-out thing it could be" removes that risk and restores fluent answers, because the level has shifted from "give the right answer" to "offer an opinion." The book parallels this with an exam question asking students to "express your views" on why Henry VIII killed his wives: students correctly infer this isn't really asking for their opinion but for what the professor said in lecture — they land on the level that matches what they know how to answer, regardless of the literal wording.

Why It Matters

Two people can read the identical problem statement and silently solve two different problems because each unconsciously selected a different semantic level — the "obvious" reading is often just the reading that happens to fit the reader's existing competence or comfort. Noticing that a problem has multiple possible levels, and deliberately testing which one is intended (by varying the wording and watching how answers shift), catches mismatched expectations before they turn into a solution built for the wrong version of the question.