Part III · Chapters 9–13

Cognitive Biases

Part III claims that the most damaging analytic errors come from mental shortcuts that work most of the time, fail predictably, and survive being explained to the person using them.

What the word is doing here

The definition is narrow on purpose:

“Cognitive biases are mental errors caused by our simplified information processing strategies.”

That excludes cultural and organizational bias and self-interest (p. 111): a cognitive bias comes from the procedure used to reach any answer, not from wanting a particular one. And these errors are “similar to optical illusions in that the error remains compelling even when one is fully aware of its nature” (p. 112) — which is why the rest of the book is procedure rather than advice.

Biases in evaluating evidence

Chapter 10 adds that people seldom weigh the absence of evidence, that the mind is oversensitive to how consistent evidence is and insufficiently sensitive to how reliable it is, and that impressions persist after the evidence behind them is discredited (p. 115). The middle one repays rereading: a small coherent set of reports feels stronger than a large conflicting one.

Cause, effect, and probability

Chapter 11 argues that causal judgment is inference, not observation: you do not see one billiard ball cause another to move, you conclude it (p. 127). Analysts build a narrative from a single case as historians do, so the criteria are looser than they look (p. 128). Hence its named errors — finding causes for what is random, overestimating how far another government acts on a coherent plan, weighting disposition over situation. Chapter 12 turns to probability, where two shortcuts do most of the damage: anchoring, adjusting too little from your starting figure, and the rule below (p. 147).

Hindsight, and why post-mortems are unfair

Chapter 13 governs how analysis is judged afterwards, which gives it the sharpest institutional consequence. Heuer sets out three forms (p. 161): analysts overestimate the accuracy of their past judgments, consumers underestimate how much a report taught them, and overseers running a post-mortem judge events to have been more foreseeable than they were.

“the receipt of what is accepted as definitive or authoritative information causes an immediate but unconscious restructuring of a person’s mental images”

The chapter then tests its own resistance to correction. One group was briefed on the bias, shown the earlier results and told to compensate. They could not (p. 171). A defect that survives being explained has to be handled in the process, which is why Part IV is a procedure. Heuer’s countermeasure is a question: would the opposite outcome have surprised you?