Decision‐Making Biases & Remedies Study Notes

"How Did I Fall For That?" – General Foundations

• Human cognition is as vulnerable to mental illusions as human vision is to optical illusions.
• No person – regardless of intelligence, education, or experience – is naturally immune.
• Cluster of remedies:
  • AWARENESS → know the biases exist.
  • TRAINING → practice identifying them in real and simulated decisions.
  • MINDFULNESS → pause, reflect, apply critical thought.
• Biases rarely work in isolation; e.g., overconfidence often pairs with hindsight bias (“I knew it all along”).
• Meta-rule: ASSUME NOTHING…QUESTION EVERYTHING!

Overconfidence

• Definition: Systematic tendency to overestimate the accuracy of one’s knowledge, judgments, and forecasts.
• Heightened when we leave our comfort zone or domain of expertise.
• Danger sign: phrases like “I’m 99\% sure.”
• WHY it happens
  • Illusion of superiority (better-than-average effect).
  • Illusion of control over random events.
  • Ignorance of the full set of possible outcomes.
  • Selective memory: successes easily recalled, failures fade.
  • Confirmation-seeking search (feeds confirmation bias).
• Practical consequence: overly narrow confidence intervals, underestimated risk buffers, and surprise when outcomes deviate.

Inertia Bias (Procrastination)

• Disposition to postpone or avoid decisions/action.
• Mechanisms
  • Conflict-avoidance → “analysis paralysis” (continuously gathering data because choice feels painful).
  • Unpleasantness avoidance → delay when short-term costs loom larger than long-term gains.
• Antidote: make the solution clear-cut and reduce psychological cost of initiating it.

Immediate Gratification (Present Bias)

• Mirror image of inertia but rooted in the same self-control weakness.
• Short-term rewards overweighted; future outcomes discounted to near zero.
• Manifestations: impulsive purchases, under-saving, unhealthy consumption.
• Economic parallel: hyperbolic discounting V=\frac{A}{1+k\,t} where a high k signals steep present bias.

Anchoring Bias

• People fixate on an initial value (anchor) and insufficiently adjust for subsequent data.
• Strongest when no objective benchmarks exist (e.g., pricing exotic derivatives like credit-default swaps pre-2008).
• Trivial or irrelevant anchors still pull estimates (classic "last 2 digits of SSN" experiment).
• Guardrail: generate anchors from data, not from arbitrary numbers, and force multiple adjustment rounds.

Selective Perception

• Under ambiguity, interpretation reflects attitudes, interests, experience, and cultural background more than the stimulus itself.
• Consequence: two observers can see the same event and walk away with incompatible narratives.
• Skill: intentionally adopt another perspective (“see through their eyes”).

Confirmation Bias

• Data collection is subjective—we seek, notice, and recall evidence that supports existing beliefs.
• Psychological roots
  • Cognitive consistency is pleasurable.
  • Conflicting information creates dissonance (mental discomfort).
  • Reduces complexity: easier to work with confirming evidence than conflicting data.
• Creates echo chambers and amplifies polarisation.

Framing Bias

• “Frames” are mental structures that shape meaning; wording or context steers decisions.
• Losses vs. gains
  • People are risk-averse in the domain of gains, risk-seeking in the domain of losses (Prospect Theory).
• Three classic types
  1. Risky-choice framing (Asian disease problem)
  • Program A: “Save 200 people.”
  • Program B: \tfrac{1}{3} chance save 600 / \tfrac{2}{3} chance save 0.
  • Equivalent negative frame: C “400 will die” vs. D \tfrac{1}{3} chance none die / \tfrac{2}{3} all 600 die.
  • Majority choose A over B (gain frame) but choose D over C (loss frame)—pure wording shift.
  1. Attribute framing: “75\% lean” vs. “25\% fat.” Same product, different perception.
  2. Goal framing: Breast-exam messages
  • Positive: “Women who DO exams have an increased chance of detecting tumors early.”
  • Negative: “Women who do NOT do exams have a decreased chance of detecting tumors early.”
• Managerial implication: craft frames ethically; test multiple frames before deciding.

Availability Bias

• Decisions lean on information that is vivid, recent, or emotionally charged rather than statistically sound.
• Example: Fear of flying spikes after a plane crash, despite P(\text{car death}) > P(\text{plane death}).
• Experiences (news, anecdotes) dominate abstract statistics.
• Leads to distorted risk assessment and resource misallocation (e.g., over-investing in low-frequency hazards).

Representation Bias (Representativeness Heuristic)

• Probability judged by similarity to a prototype rather than by base rates or laws of randomness.
• Patterns inferred where none exist: “I’m on a lucky streak.”
• Regression to the mean
  • Extreme outcomes tend to move toward average next period.
  • Formal: E[X{t+1}\,|\,Xt]=\mu+\rho\,(X_t-\mu),\;|\rho|<1.
• Sample-size neglect: treating “3/5 successes” as equally credible as “1200/2000 successes.”

Coping with Randomness

• Humans over-interpret chance events; we crave causal stories.
• Practical guidelines
  • Accept that coincidences happen without deeper meaning.
  • Resist attributing random events to “fate.”
  • Avoid forming superstitions from illusory patterns.

Sunk Cost Fallacy

• Treating past, irrevocable costs as reasons to continue an endeavour.
• Rational rule: only future costs and benefits matter—today’s decisions can’t rewrite the past.
• Psychological drivers
  • Desire to appear consistent (“saving face”).
  • Aversion to waste.
• Tip: reframe decision as “fresh project starting today.”

Limited Search & Bounded Rationality

• Cognitive limits → we simplify by searching only a slice of the alternative space.
• May satisfice: accept first option that meets a minimally acceptable threshold.
• Simon’s Bounded Rationality: optimisation is impossible; aim for “good enough.”
• Check: Does the option align with values, goals, and plans?
• Einstein reminder: “Everything should be made as simple as possible, but not simpler.”

Emotional Involvement

• High arousal (stress or excitement) narrows attention and speeds up choices, often impulsively.
• Negative emotions
  • Constrict cognition, encourage rapid, oversimplified decisions.
• Positive over-excitement
  • Can lead to overcommitment, ignoring downside risk.
• Strategy: acknowledge emotion, delay key decisions until emotional intensity subsides or use pre-commitment devices.

Self-Serving Bias

• Personal outcomes
  • Success → internal attribution (“I’m skilled”).
  • Failure → external attribution (“Bad luck”).
• Social reversal
  • When judging others’ failures, we downplay situational factors and blame internal flaws.
• Organizational impact: fosters blame culture and hampers honest post-mortems.

Hindsight Bias

• After the fact, outcomes appear obvious (“I knew it all along”), masking past uncertainty.
• Mechanisms
  • Faulty memory reconstruction: we misremember prior probability estimates as more extreme.
• Consequence: reduces learning—if outcome seemed inevitable, we never probe why it actually happened.

Integrative Checklist – Avoiding Decision Traps

• Deliberately search for disconfirming evidence.
• Perform a “bias audit” on yourself: impatience? overconfidence? risk appetite?
• Prioritise long-term over myopic goals.
• Experiment with alternative frames—rewrite problem statements.
• Conduct an empathy exercise: “Walk in someone else’s shoes.”
• Diversify experiences to expand mental models.
• Treat extreme performances as temporary; remember regression to the mean.
• Accept randomness: not every pattern is causal.
• Push for “outside-the-box” options; widen the choice set beyond the obvious.
• Manage emotion: build pauses, ask neutral parties, or use decision pre-mortems.