Logical Fallacies & Cognitive Biases – Quick Review

Aim: hold beliefs only when supported by good reasons (increase likelihood of truth).
Arguments justify beliefs:
- Deductive: if valid + true premises → sound → conclusion must be true.
- Inductive: if strong + true premises → cogent → conclusion likely true.
First check argument form before believing the claim.

Affirming the Consequent: $If\ A\rightarrow C;\ C\Rightarrow A$ (invalid).
Denying the Antecedent: $If\ A\rightarrow C;\ \lnot A\Rightarrow \lnot C$ (invalid).
Appeal to Irrelevant Authority: claiming truth because a (non-expert/unrelated) figure supports it.
Genetic Fallacy: judging a claim by its origin, not its content.
Ad Hominem: attacking the person instead of the argument.
Appeal to Ignorance: “no proof against X, therefore X” (or vice-versa).
False Dichotomy: presenting only two options when more exist.
Post Hoc / Not a Cause for a Cause: mistaking correlation or sequence for causation.
Hasty Generalisation: broad claim from an unrepresentative sample.
Slippery Slope: asserting that one step inevitably triggers extreme consequences.
Straw Man: misrepresenting an opponent’s view to refute it easily.

System 1: fast, automatic, effortless, heuristic-driven; good for routine tasks, prone to bias.
System 2: slow, effortful, analytical; overrides errors but is lazy & resource-heavy.
Many reasoning errors occur when System 2 fails to check System 1.

Availability: judging likelihood by how easily examples come to mind (e.g., dramatic events).
Framing Effect: risk-seeking under loss frames, risk-averse under gain frames; same facts, different wording.
Halo Effect: a single positive trait colours unrelated judgments about a person.
Anchoring: first number/idea encountered sets a reference point for later estimates.
Small-Sample Bias: extreme outcomes & spurious patterns common in small datasets; ignore base rates.
WYSIATI (What You See Is All There Is): mind builds stories from available info, neglecting missing data & reliability.