System 1 and System 2 Thinking (Heuristics and Biases)

Overview

Daniel Kahneman, winner of the 2002 Nobel Prize in Economics and author of Thinking, Fast and Slow, describes the two families of mental operations:

• System 1 (Type 1): Automatic, effortless, often unconscious, associatively coherent. Handles perception, intuition, and practiced skills (e.g., driving, understanding language).
• System 2 (Type 2): Controlled, effortful, usually conscious, rule-governed, and deliberate. Often thinks it is the hero, but System 1 does most of the work hidden from us.

Key Properties of System 1

• Detects incongruity and abnormality automatically (e.g., an upper-class British voice saying "I have tattoos" triggers a surprise response in 200-400 ms).
• Infers causes and intentions from infancy (infants expect a chasing object to take the most direct route).
• Neglects ambiguity and suppresses doubt, creating a coherent interpretation of reality—even when the evidence is thin.
• "What you see is all there is" (WYSIATI): It jumps to conclusions based only on available information, ignoring missing data, leading to overconfidence.
• Performs natural assessments automatically, such as averaging (e.g., average length of lines) but not summing (e.g., total length). It excels at average-like variables but struggles with sum-like variables (probability, economic value).
• Matches intensities across dimensions (e.g., matching Julie's reading age to a building's height).
• Mental shotgun: It computes more than intended (e.g., automatically generates spelling when asked about rhymes; interprets metaphors even when instructed not to).

Heuristics and Biases

Heuristics are mental shortcuts that often lead to biases:

• Representativeness heuristic: Judging probability by similarity. Example: "Linda" problem—people say she is more likely a feminist bank teller (similar to her description) than a bank teller (less similar), violating probability rules.
• Base rate neglect: Ignoring statistical base rates in favor of case-specific, causal information (e.g., cab problem: witness reliability vs. 85% blue cabs—people rely on witness unless base rate is presented as causal stereotype).
• Conjunction fallacy: Believing a conjunction is more probable than one of its components (e.g., heart attack example: more say over-55 heart attack victims are more numerous than all heart attack victims).

Marvels vs. Flaws

• Marvels: Expertise (e.g., recognizing a spouse's mood from one word) is System 1—it develops through reinforced practice. Most of our daily functioning is flawless.
• Flaws: The same confidence can arise from flawed heuristics, making it hard to distinguish good intuition from bad.

Notable Experiments

• Poster study in a kitchen: Posters of eyes (watching) increased honesty payments compared to flowers—unconscious priming effect.
• "Vomit" and "banana": Seeing these two words triggers automatic associations, disgust, and a causal story (banana causing vomit), preparing you for related ideas.
• Head movement and persuasion: Nodding ("yes" motion) increases belief; shaking ("no") decreases it.
• Hsee's dishes: Sets of dishes are valued higher if seen together (more pieces) but separately, a set with higher average quality is valued more—showing average vs. sum confusion.
• Frequency vs. percentage: Presenting data as "out of 100" reduces conjunction fallacy because it evokes a spatial representation, making the subset relationship clearer.

Conclusion

Kahneman advocates for a cognitive interpretation of these effects (rather than purely evolutionary), emphasizing that System 1's coherence-seeking, associative nature produces both marvels and flaws. Understanding these mechanisms helps us design better decision-making processes.