Biases & Fallacies

Reference of cognitive biases and logical fallacies an analyst needs to recognize, in adversary thinking, in source material, and in their own reasoning. Pairs with the structured analytic techniques in Intelligence Disciplines (ACH, Key Assumptions Check, Devil’s Advocacy, Premortem) which exist precisely to counter these failure modes.

Cognitive biases (selected)

Bias	Notes
Confirmation bias	seeking / weighing evidence that supports a held belief over evidence that refutes it. The single most common intelligence failure mode.
Anchoring	over-weighting the first piece of information seen; subsequent updates are too small.
Availability heuristic	judging probability by how easily examples come to mind.
Recency bias	weighting recent events more than older equally-relevant ones.
Hindsight bias	“I knew it all along”; restructuring memory of what was probable post-hoc.
Survivorship bias	only looking at the cases that survived (visible); ignoring those that didn’t.
Selection bias	sampling that excludes some part of the population.
Sunk-cost fallacy	continuing because of past investment, not future expected value.
Optimism bias	systematic over-prediction of favourable outcomes.
Loss aversion	losses weighted ~2x more than equivalent gains.
Endowment effect	valuing things one owns more than identical un-owned things.
Status quo bias	preferring the current state.
Bandwagon effect	adopting beliefs because others have.
In-group favoritism	preferring the in-group; downplaying out-group.
Out-group homogen.	seeing one’s own group as varied + the other group as uniform.
Fundamental attrib.	attributing others’ behavior to character + one’s own to circumstance.
Self-serving bias	attributing success to self + failure to circumstance.
Dunning-Kruger	lowest-quartile competence + highest-quartile self- assessment confidence.
False consensus	overestimating how much others share one’s beliefs.
Curse of knowledge	inability to imagine not knowing what one knows; affects communication.
Belief bias	judging an argument by whether the conclusion matches prior beliefs.
Mirror imaging	assuming the adversary thinks the way we do.
Cognitive easing	accepting the most-fluent explanation.
Pareidolia	seeing meaningful patterns where there are none.
Texas-sharpshooter	fitting the target to the data after the fact.
Clustering illusion	seeing patterns in random clusters.
Gambler’s fallacy	“due for a run” thinking after streaks.
Hot-hand fallacy	expecting streaks to continue.
Base-rate neglect	ignoring the prior probability when given new evidence.
Conjunction fallacy	“specific + plausible” rated more probable than just “plausible”.
Planning fallacy	underestimating duration / cost of a plan.
Hyperbolic discount	preferring smaller-sooner over larger-later more than is rational.
Cognitive dissonanc	e discomfort when actions and beliefs conflict; resolved by changing one.

Heuer’s biases (intel-specific)

Bias	Notes
Vividness criterion	vivid / concrete data weighted over abstract / statistical.
Absence of evidence	treating the absence of evidence as evidence of absence.
Oversensitivity to consistency	small recent incidents shifting confidence too much.
Persistence of impressions	first impressions surviving disconfirming evidence.
Coping with complexity	simplifying complex situations to known patterns.
Mirror imaging	adversary intent = our intent (above).
Defensive avoidance	avoiding information that contradicts prior judgments.
Wishful thinking	weighting outcomes by preference, not evidence.

Logical fallacies

Fallacy	Notes
Ad hominem	attack on the person, not the argument.
Straw man	refuting a weakened restatement of the argument.
Tu quoque	“you do it too”; doesn’t address the original.
Appeal to authority	“X says so” without independent merit; valid when X has relevant expertise.
Appeal to nature	“natural = good”.
Appeal to tradition	“we’ve always done it this way”.
Appeal to novelty	“new = better”.
Appeal to popularit	y “most people believe it”.
Appeal to fear	argument by threat / consequence.
Appeal to emotion	instead of evidence.
Genetic fallacy	dismissing an argument because of its origin.
Slippery slope	A leads inevitably to Z without justifying the steps.
False dilemma	presenting two options when more exist.
False equivalence	treating two unequal things as equivalent.
Hasty generalisatio	n generalising from a small sample.
Post hoc / cum hoc	“after this therefore because of this”.
Affirming consequen	t if A → B and B, conclude A. Invalid.
Denying antecedent	if A → B and not A, conclude not B. Invalid.
Begging the questio	n premise assumes the conclusion.
Circular reasoning	A because B; B because A.
No true Scotsman	redefining terms to exclude counterexamples.
Composition	what’s true of the parts is true of the whole.
Division	the inverse.
Equivocation	using a word with multiple meanings as if one.
Loaded question	“have you stopped X yet?”
Whataboutism	deflection to a different complaint.
Red herring	introducing irrelevant material.
Bulverism	“you only believe X because of who you are”.
Special pleading	making an exception only for one’s own case.
Moving goalposts	changing the criterion when met.
Fallacy of sunk cos	t (above; both bias + fallacy).
Galileo gambit	“they laughed at Galileo and he was right” doesn’t make you Galileo.
Texas sharpshooter	(above).
Argument from incredulity	“I can’t think of an alternative explanation”.
Argument from ignorance	“I can’t think of an objection”.

Mitigations

• Pre-mortem, “assume the assessment is wrong; reason backward from why”.

• Devil’s Advocacy / Red Team, formally argue the alternative position.

• ACH (Heuer), list hypotheses, evidence; refute, not confirm.

• Key Assumptions Check, surface every assumption; test each.

• Indicator-based forecasting, “if X, hypothesis Y is more likely”; falsifiable triggers.

• Diverse review, multiple analysts; not just for capacity but for cognitive variance.

• Calibrated language, ICD 203 style: “almost certainly”, “likely”, “probably”, “even chance”, “unlikely”, “almost no chance”; avoid weasel words.

• Bayesian discipline, start from base rates; update proportional to evidence’s likelihood ratio.

• Source-quality awareness, single-source / chain-of-reports amplification is a known failure pattern.

References

• Richards J. Heuer, Psychology of Intelligence Analysis (CIA, 1999), the standard text. Free PDF.

• Daniel Kahneman, Thinking, Fast and Slow (2011).

• Phil Tetlock, Superforecasting (2015).

• Carl Sagan, The Demon-Haunted World (1995), “baloney detection kit”.

• Nassim Taleb, The Black Swan / Fooled by Randomness.

• Wikipedia: List of cognitive biases

• Wikipedia: List of fallacies

• Cognitive Bias Codex (Manoogian / Benson)

• Intelligence Disciplines, the SAT framework that counters these.

• Profiling