Phronesismodule 7 of 9 · free
Module 7

Decision and Bias: Cheap Returns

Try it right now: how many English words start with the letter k, versus how many have k as their third letter? Don't count — just feel which number is bigger. Most people are sure the first-letter words win. They don't; third-letter words are roughly twice as common. You didn't compute anything. You asked a faster question — how easily do examples come to mind? — and answered that one instead.

That swap is the whole story. The mind is rarely at its ground state $S^0_{cog}$ — the configuration of clearest, fully-resolved understanding. Getting there is expensive: it costs attention, working memory, time. So when a hard question displaces you, you usually don't pay full price for the return. You pay for a cheap return instead: a shortcut that lands you near ground for a fraction of the cost.

These shortcuts have a name. Heuristics — fast, frugal rules of thumb that approximate the right answer most of the time. Kahneman's name for the fast, automatic machinery that runs them is System 1: effortless, parallel, always on. A bias is what you see when the cheap return lands in the wrong place systematically — not random error, but a reliable lean in one direction.

Bounded, not broken

Herbert Simon's word for this is bounded rationality. A real mind has finite time, finite memory, finite energy. It cannot afford the optimal answer to every question, so it satisfices — takes the first option that clears the bar. Read against the framework: full optimization means driving displacement all the way to zero, paying the full return cost. A bounded mind refuses that bill. It accepts a small residual displacement in exchange for a much smaller $\Phi_{cog}$. The bias is the residual. It is the price of cheapness, not a malfunction.

The field sets the default

Here is the part that feels like a trick. The cheap return doesn't start from nowhere — it starts from wherever the field happens to put you. Anchoring: show someone a number before asking for an estimate, and their answer drifts toward it. Availability: judge frequency by how readily examples surface (the k trick — and why people overrate plane crashes and underrate the slow killers). Framing: "90% survive" and "10% die" are identical facts that pull different decisions. In each case the wording or the prior number tilts the cognitive landscape, and the cheap return rolls downhill to whatever default the tilt created. You experience the result as your own considered judgment. It was mostly the slope.

What you'll be able to do

The precise version

The rigorous layer. Optional — skip it and the picture above still stands.

Let a decision begin at displacement $\xi_{cog} > 0$: a question whose answer is not yet at ground. The exact return computes the true answer, integrating instantaneous cost $D_{cog}$ along the full path back to $S^0_{cog}$, for accumulated cost $\Phi_{cog}^{\text{exact}}$. A heuristic substitutes a cheaper path with cost $\Phi_{cog}^{\text{heur}} \ll \Phi_{cog}^{\text{exact}}$ but terminates at a nearby state $S'$, not at ground. The leftover gap $\|S' - S^0_{cog}\|$ is the bias.

Crucially the bias is directional. Random error would scatter $S'$ symmetrically around ground and average out. A bias does not: the substituted question has its own ground state, displaced from the true one in a fixed direction, so error accumulates with a sign. That is why biases are predictable, and therefore correctable.

Anchoring, availability, and framing all act on the shape of the field before the return even starts — they move the local minimum the cheap path runs toward. Choice architecture is the engineering inverse: hold the cheap System-1 return fixed and reshape the landscape so its lowest point is the good outcome. Thaler and Sunstein's default is exactly this — the attractor a low-$\Phi_{cog}$ mind falls into when it declines to pay for deliberation. Set the default well and you don't have to make anyone think harder. You move the floor.

Worked example

Effective consent to organ donation differs enormously between near-identical countries — single digits in some, over 90% in others (Johnson & Goldstein, 2003). The gap isn't compassion; it's the form. Opt-in countries make "donor" cost a deliberate act (full return, paid by few). Opt-out countries make "donor" the default and non-donation the act. Same citizens, same System 1, same refusal to pay $\Phi_{cog}$ — the architect simply moved the bottom of the well. The cheap return now lands on donor.

Exercises

  1. Catch one snap judgment today (a price felt "too high," a person felt "trustworthy"). Write the hard question you faced and the easier question System 1 swapped in. Name the anchor or available example that set the default.
  2. Take one decision you face this week and reframe it twice — once as a gain ("keep 80%"), once as a loss ("lose 20%"). Notice which framing makes the choice feel obvious. That pull is the field, not the facts.
  3. (Open-ended.) Pick a default in your own life — a phone home screen, a fridge's front shelf, a recurring calendar slot. Redesign it so the cheap, no-deliberation return lands on the behavior you actually want. Run it for a week and watch whether willpower was ever the real variable.

Sources

The course, $29
The polished textbook PDF and the worked sheets, yours to keep. The lessons stay free to read.
Get the course, $29
← all modulesModule 8: Insight and Problem-Solving →