PhronesisTry it right now: you have a candle, a box of thumbtacks, and a book of matches. Mount the candle on the wall so it burns without dripping wax on the floor. Sit with it for ten seconds before you read on. Feel the place where your thinking stalls — that wall you keep hitting is the whole topic of this module.
A problem is a gap. You are here — the start state — and the goal is somewhere there, and between them lies everything you don't yet know how to do. Allen Newell and Herbert Simon gave this its lasting shape: a problem is a state space, a branching tree of every configuration you could reach by applying the available moves, and solving it is search — finding a path from start to goal through that tree.
The trouble is that the tree is enormous. Chess has more possible games than there are atoms in the observable universe. You cannot search it all, so you don't. You use heuristics — cheap rules that prune the branches. The most famous is means-ends analysis: look at the difference between where you are and where you want to be, then pick the move that shrinks that difference most. Reduce the gap, step by step, until it closes.
Why you get stuck
Means-ends analysis is greedy. It always walks downhill toward the goal, which means it can march you straight into a dead end — a basin where every nearby move makes things look worse even though the true solution lies just over the ridge. Worse, the way you represent the problem decides which moves you can even see. Karl Duncker called the candle-problem trap functional fixedness: you see the box as a container for tacks, so the move "empty the box and tack it to the wall as a shelf" is invisible. The solution was in your hand the whole time. Your representation hid it.
This is a stuck attractor. You circle the same few states, applying the same operators, returning to the same wall, and each lap feels like effort without progress. The displacement isn't falling. You're spending and not arriving.
What insight actually is
Then — sometimes — it flips. The box stops being a box-for-tacks and becomes a platform. This is insight: not a faster search through the old space, but a re-representation that builds a different space in which the goal is suddenly close. The "aha" is the felt signature of that collapse — a long path replaced, all at once, by a short one. Janet Metcalfe and David Wiebe showed that insight problems have a distinct phenomenology: people can't tell they're about to solve one — their "warmth" ratings of feeling close stay low right up until the answer arrives whole — whereas on ordinary, step-by-step problems warmth climbs gradually as they near the solution.
Incubation is the quiet partner. Step away — sleep, walk, shower — and the wrong representation, no longer rehearsed, loosens its grip. The fixed association decays; a fresh angle becomes reachable. The break isn't idleness. It's letting the stuck attractor cool so the mind can fall into a better one.
What you'll be able to do
- Diagnose a stuck problem as either a bad search heuristic or a bad representation — and treat them differently.
- Deliberately trigger re-representation: change the units, drop an assumption, or restate the goal to make a shorter path visible.
- Use incubation on purpose — schedule the break instead of grinding into a dead end.
The precise version
This is the rigorous layer. Optional — the plain version above carries the whole idea.
Let the goal be the cognitive ground state $S^0_{cog}$: the configuration of clearest understanding, where the gap is closed. A live problem is a displacement $\xi_{cog} > 0$ from that ground. The return path is the trajectory of moves that carries you back to $S^0_{cog}$, and its total price is the accumulated cost $\Phi_{cog} = \int D_{cog}\,dt$ — every unit of working-memory load and attentional effort spent en route. Search is the act of finding a low-$\Phi_{cog}$ return path through the state space.
Functional fixedness is a region where the instantaneous cost $D_{cog}(\xi_{cog})$ stays high and roughly constant: you apply operators, you spend, but $\xi_{cog}$ does not fall. A stuck attractor is a local basin from which every available move is uphill. Greedy means-ends analysis cannot escape it, because escape requires temporarily increasing displacement — paying more to reach a saddle from which a far shorter descent opens.
Insight is a change of coordinates. The state space is not fixed; it is imposed by your representation. Re-representation rebuilds the space so that, in the new coordinates, the goal sits only a cognitive Planck unit or two away — the smallest distinctions the mind can make now span the whole remaining gap. The "aha" is exactly the steep drop: $\xi_{cog}$ falls from large to near-$\theta$ in essentially one step. (Recall $\theta > 0$: even solved, the brain idles; the ground is never free.) Incubation lowers the activation of the dominant-but-wrong representation over time, raising the probability that search lands in a basin with a steeper, cheaper descent to $S^0_{cog}$.
Worked example
Nine dots, three by three; connect all with four straight lines, the pen never lifting. Inside the represented space — "stay within the square the dots imply" — no four-line path reaches the goal, so $\Phi_{cog}$ climbs with every doomed attempt while $\xi_{cog}$ holds flat. The insight is one dropped assumption: the lines may extend past the dots. That single edit enlarges the state space, and within the enlarged space a four-line return path appears that was never there before. Same goal, same operators — a different space, and the displacement collapses in a single move.
Exercises
- Take a problem you're stuck on and write its state space explicitly: start state, goal state, and the legal moves. Often the act of listing moves reveals one you'd been refusing to see.
- Pick one assumption you've been treating as fixed ("the box is for tacks," "the lines stay inside the dots") and deliberately negate it. Search the space the negation opens. Note whether $\xi_{cog}$ drops.
- (Open-ended.) Recall an "aha" moment from your own life. Reconstruct what re-represented — which coordinate changed so a long path became short — and ask whether you could induce that shift on purpose next time.
Sources
- Rincón, D., alice, & clöe (2026). Cognitive Displacement: A Planck Scale for Human Understanding.
- The Displacement Framework.
- Newell, A., & Simon, H. A. (1972). Human Problem Solving. Prentice-Hall — state-space search and means-ends analysis.
- Duncker, K. (1945). On problem-solving. Psychological Monographs, 58(5) — functional fixedness and the candle problem.
- Metcalfe, J., & Wiebe, D. (1987). Intuition in insight and noninsight problem solving. Memory & Cognition, 15(3) — the "warmth" studies.