Bias and Decision Making PSY2002

Biases

departures from normative models of rational decision-making

Norms of rationality

Logical consistency

Probability theory (e.g., laws of conjunction)

Expected Utility Theory (EUT)

Coherence norms

Internal consistency (e.g., Linda problem → conjunction fallacy)

Correspondence norms:

Accuracy relative to reality (e.g., overestimating rare events like plane crashes = availability bias)

Framing bias

Decision changes based on problem wording, despite identical outcomes.

Availability bias:

Judging likelihood by ease of recall, not actual frequency.

Biases highlight

the gap between actual human reasoning and idealised rationality models.

Expected Utility Theory

A normative model for making decisions under risk/uncertainty:

Core idea: Rational decision-makers aim to maximise expected utility, not just monetary value.

EUT formula

EU = p1 x U1 + p2 x U2 +...

Utility vs Value

Utility is subjective satisfaction or personal value.

Diminishing marginal utility: The first £1,000 means more to a poor student than to a millionaire.

EUT Example

Option A: 50% chance of £1,000 → EV = £500

Option B: Sure £499 → Chosen more often, despite lower EV, due to risk aversion and non-linear utility perception.

EUT helps explains

economic and behavioural decisions under uncertainty, but humans often deviate from its predictions.

What are heuristics

mental shortcuts that are often effective in real-world environments

Ecological rationality

Heuristics are not illogical—they're adapted to how environments work.

Success depends on the fit between heuristic and environment.

Recognition Heuristic Example

If one of two cities is recognised, infer it is larger. Works well where recognition correlates with actual size (e.g., Liverpool > Leicester).

Bonded Rationality (SImon, 1957)

We are cognitively limited; heuristics are "fast and frugal" tools that deliver good-enough answers without exhaustive computation.

Effects of heuristics

Reduce cognitive load and often yield correct outcomes - but may lead to biases

Biases do not equal flaws in reasoning

Biases reflect normal reasoning mechanisms

Adaptive Value

Alternative Interpretations

Biases are not necessarily cognitive flaws—they are often a byproduct of effective strategies in natural environments.

Adaptive value of biases

Strategies that lead to errors in lab settings may be evolutionarily adaptive in real-world contexts.

Alternative Interpretations

Confirmation bias in Wason's 2-4-6 task may reflect a positive test strategy, useful in many real-life situations.

Framing effects may reflect contextual sensitivity, not irrationality.

What do Dual process theories propose

Two distinct systems for reasoning

Dual-process models explain how reasoning can be both efficient and error-prone, depending on which "system" dominates.

System 1

Fast, automatic, intuitive, emotionally driven

Often relies on heuristics

System 2

Slow, deliberate, analytical

Engages when decisions are complex or when errors are detected

Kahneman (1934-2024)

Most biases originate in System 1

System 2 can override System 1, but only when it is engaged and has sufficient resources (e.g., time, attention)