Heuristics and Biases

Cognitive psychology is the science of representing and processing information in the human mind. Heuristics are informal algorithms utilized in decision-making to reduce the cognitive load associated with complex evaluations. The concept of dual-system thinking frames heuristics as system one, or reflexive thinking. While heuristics speed up cognition, their reflexive nature makes the human mind prone to making errors. These systematic errors or deviations from rational decisions are termed biases.

Four key judgment heuristics primarily drive human decision-making: representativeness, availability, anchoring, and affect.

Representativeness

The representativeness heuristic is deployed when a decision-maker uses similarity as a proxy for probabilistic thinking. An individual checks whether the data at hand is consistent with a pre-existing mental model, ignoring objective statistical probabilities.

A classic demonstration of this is the Linda problem. Subjects read a description of Linda, detailing her involvement in social justice and anti-nuclear demonstrations. When asked to evaluate probabilities, subjects rank the probability of Linda being a "feminist bank teller" higher than simply being a "bank teller". Statistically, the set of feminist bank tellers is a subset of all bank tellers, meaning the probability of the conjunction cannot exceed the probability of the single descriptor. This error is known as the conjunction fallacy, a conflict between heuristic-based thinking and the logic of probability.

Insensitivity to Base Rates

Individuals frequently ignore the base rate (the underlying population probability) if descriptive information is provided.

In one experiment, subjects evaluated the probability of a massive earthquake causing 1000 deaths in India versus a massive earthquake causing 1000 deaths specifically in the state of Uttaranchal. Subjects routinely judged the Uttaranchal scenario as more likely due to richer, representative details, ignoring the statistical fact that Uttaranchal is contained within India.

In a second experiment, subjects read a personality profile of "Dick" indicating he likes mathematical puzzles. Subjects estimated a 90 percent probability that Dick was an engineer, regardless of whether he was drawn from a sample containing 70 engineers or a sample containing only 30 engineers.

Insensitivity to Sample Size

When assessing the reliability of information, individuals fail to appreciate the role of sample size.

Subjects were asked whether a small hospital (15 births per day) or a large hospital (45 births per day) would record more days where over 60 percent of the babies born were boys. Most subjects assumed both hospitals had an equal chance. Statistically, the smaller hospital is much more likely to observe this deviation, as smaller samples have a higher variance and are less likely to perfectly mirror population parameters. Advertising agencies frequently exploit this bias by quoting survey results from very small sample sizes (e.g., 4 out of 5 dentists).

Misconceptions of Chance

Individuals expect that a short sequence of data generated by a random process will look perfectly random. If an unbiased coin yields five consecutive tails, subjects falsely predict the next toss must be a head to "even out" the sequence.

This heuristic heavily impacts real-world fertility decisions. In households exhibiting son preference, parents with two daughters may opt for another child, falsely believing the next birth has a statistically higher chance of being a boy. A direct implication of this bias is the gambler's fallacy, where individuals assume a long losing streak implies they are "due" for a win, ignoring the concept of sampling with replacement.

Regression to the Mean

Extreme events mathematically tend to regress to the mean on subsequent trials, but individuals frequently assign causal interpretations to these random fluctuations.

Flight instructors in the Israeli defense forces noticed that praising pilots for exceptional landings was followed by worse performance, while harshly criticizing poor landings was followed by improvement. The instructors falsely concluded that punishment was an effective training tool, failing to realize that landing a jet is highly variable. Excellent or terrible landings were primarily products of luck, and subsequent attempts simply regressed back to the average performance level.

Confirmation Bias

Individuals systematically seek confirmatory information for their pre-existing beliefs and fail to search for disconfirmatory evidence. The mental cost of processing a contrarian viewpoint is higher than processing a confirmatory viewpoint, especially when individuals are mentally fatigued.

In an experiment on capital punishment, supporters and opponents of the death penalty evaluated identical sets of mixed evidence. Both factions concluded the opposing evidence was methodologically flawed and left the experiment feeling more convinced of their original positions. In a corporate setting, this explains why managers prefer "Yes Men" over "Devil's Advocates," as the latter demand a higher cognitive load to process.

Availability

The availability heuristic involves judging the probability of an event based on the ease with which instances can be brought to mind. If an event is salient, vivid, or recent, it becomes easily retrievable from memory.

Anchoring

Anchoring occurs when individuals make estimates by starting from an initial value (the anchor) and making insufficient adjustments to reach a final answer. This heuristic is heavily utilized when there is a lack of objective information, causing the anchor to provide a natural starting point.

When subjects were asked to quickly estimate the product of an ascending sequence (1 to 8), the average guess was 512. Conversely, subjects given a descending sequence (8 to 1) started with a higher anchor and produced an average guess of 2250. Similar experimental results are observed when subjects guess the invention year of the telephone or the height of Mount Everest after being exposed to randomized numerical anchors.

Anchoring severely impacts consumer and negotiation behavior. If an initial job salary offer is systematically lower (historically observed in offers made to women), this serves as a lower anchor that permanently depresses the trajectory of future wage negotiations.

Affect

Affect is defined as a rapid, emotional (good or bad) response to a stimulus that occurs before higher-level reasoning can initiate.

The tobacco industry historically exploited the affect heuristic by using the "Marlboro Man" to associate smoking with a positive, masculine lifestyle, thereby nudging consumers to unconsciously lower their perceived risk of the product. Conversely, modern public policy utilizes the affect heuristic in reverse by mandating gory, graphic health warnings on cigarette packaging to trigger an immediate negative emotional response.

Overconfidence

Overconfidence is considered the most robust finding in the psychology of judgment, defined as a cognitive bias where a person's judgment is reliably better in their own mind than objective accuracy dictates.

To mitigate over precision, researchers suggest asking individuals to explicitly articulate the percentage chance of alternative outcomes or failure. Alternatively, forcing individuals to generate multiple distinct estimates (the "crowd within oneself" technique) leads to a more accurate averaged final judgment.

Exponential Growth Bias

Exponential Growth Bias (EGB) is the pervasive cognitive tendency to linearize exponential functions when evaluating them intuitively, leading to severe underestimations of future values. This applies to population growth, traffic escalation, and compounding interest on debt.

In predicting COVID-19 trajectories, subjects suffering from EGB consistently underestimated future case numbers. High EGB negatively correlates with compliance to health norms, as individuals falsely perceive future risk to be low.

Behaviorally informed nudges can reduce EGB. Breaking prediction tasks into smaller, iterative steps reduces the bias, while providing direct numerical feedback, graphical feedback, or statistical forecasts can eliminate EGB entirely.

Hindsight Bias and Curse of Knowledge

Hindsight Bias

Hindsight bias is the tendency to perceive past events as having been highly predictable before they took place. Subjects falsely believe they "knew it all along".

In an experiment analyzing historical military conflicts, subjects were provided with a passage and randomized historical outcomes (e.g., British win, Gurkha win, stalemate). Subjects across all treatments falsely reported they would have independently predicted their randomly assigned outcome. Hindsight bias limits an individual's ability to learn from the past and makes it impossible to objectively evaluate the pre-hoc decisions of others (e.g., juries evaluating medical malpractice).

Curse of Knowledge

The curse of knowledge is the psychological inability to ignore or "un-know" possessed information when attempting to evaluate the perspective of someone lacking that knowledge.

Subjects who were secretly informed that a diner hated a restaurant interpreted a highly positive written review from that diner as "sarcastic". Subjects who were informed the diner enjoyed the meal interpreted the identical review as "sincere". This inability to project from a state of ignorance causes severe communication breakdowns in organizational and design settings.

Prospect Theory

Prospect theory serves as the behavioral alternative to the classical expected utility framework. While neoclassical economics assumes satisfaction depends on absolute levels of wealth, prospect theory relies on reference-dependent preferences.

Principle 1: Changes vs. Levels

Fundamental human senses of pleasure and pain react to changes rather than absolute levels. These changes are calculated based on a reference point. The reference point can be established by expected outcomes, status quo, fair outcomes, or legal entitlements.

Principle 2: Loss Aversion

Losses matter more than gains. The psychological pain inflicted by a loss is quantifiably larger than the pleasure generated by a gain of the identical magnitude. The loss aversion coefficient (the slope of the value function in the loss domain relative to the gain domain) is generally estimated to be around 2 to 2.5.

This dynamic directly creates the endowment effect. In university experiments, sellers of an object operate in the loss domain (giving the object up) while buyers operate in the gain domain (acquiring the object). Because the value function is steeper in the loss domain, the seller's willingness to accept is systematically higher than the buyer's willingness to pay.

Principle 3: Risk Preferences

Humans are risk-averse in the gain domain (concave value function) and risk-seeking in the loss domain (convex value function).

This preference reversal violates classical expected utility, proving that the curvature of the utility function depends strictly on the baseline reference point (0 lives saved vs. 600 lives saved).

St Petersburg Paradox and Risk Premium

The St. Petersburg Paradox features a coin-flip gamble where the payout doubles with every consecutive head. Mathematically, the expected value of this gamble is infinite. However, individuals are unwilling to pay large sums to play. The paradox is resolved by realizing humans maximize expected utility rather than expected value.

Due to a concave utility function, the utility of an expected value is greater than the expected utility of a gamble. Consequently, risk-averse individuals prefer a certain, sure income over a risky lottery with the exact same expected value. The risk premium is the maximum financial amount an individual is willing to sacrifice to avoid taking on risk.

Probability Weighting Function

The expectation principle fails to perfectly describe human evaluations of risky probabilities. Objective probabilities are processed via subjective decision weights.

The resulting probability weighting function forms an S-shaped curve relative to a linear expectation line.

• Possibility Effect: Individuals heavily overweight highly improbable outcomes near 0%. This qualitative leap from impossibility to possibility explains behaviors like buying lottery tickets, entering the acting industry, or low-level drug dealing.
• Certainty Effect: Individuals underweight highly probable events relative to absolute certainty (100%).

Mental Accounting

Mental accounting is the subjective process by which individuals organize, evaluate, and track their financial activities in separate, distinct categories. This violates the classical economic principle of fungibility, which dictates that money should be perfectly interchangeable regardless of its origin or intended use.

Consumers frequently maintain separate informal limits for mental accounts such as entertainment, food, and clothing. For example, unexpectedly spending money on an expensive dinner causes individuals to severely under-consume in their food and entertainment accounts for the rest of the week, but has zero impact on their clothing account.

Acquisition Utility vs. Transaction Utility

Humans calculate purchases using two distinct forms of utility.

Consumers are willing to pay vastly different prices for the exact same beverage on a beach depending on whether the runner purchases it from a luxury resort or a rundown grocery store. The acquisition utility is identical, but the transaction utility shifts because the consumer sets a higher reference price for the luxury resort. Marketers engineer transaction utility by artificially inflating Maximum Retail Prices (MRPs) to establish high reference points before applying discounts.

Bundling Losses and Gains

Based on the mathematics of the value function, mental accounting provides rules for how individuals should evaluate multiple outcomes to maximize satisfaction.

Managers utilize these principles when designing bonus and penalty structures to manipulate employee motivation.

Sunk Costs

The sunk cost fallacy occurs when an individual exhibits reluctance to abandon an asset or endeavor simply because unrecoverable investments (time, money, effort) have already been made.

While rational entities ignore sunk costs, humans do not, because failing to utilize the investment forces the closure of a mental account in a severe negative deficit. Consumers will endure physical pain to wear uncomfortable shoes longer if the financial purchase price was high.

Consumers also experience payment depreciation, where the psychological memory of the sunk cost fades over time. Gym attendance predictably spikes immediately after membership fees are paid, and steadily declines as the memory of the payment depreciates over the subsequent months.

---

Ultra-Quick Revision (Exam Essentials)

Key Concepts & Distinctions

• Dual-System Thinking: Heuristics operate in System 1 (reflexive, fast, low cognitive load) rather than System 2 (analytical, slow).
• Expected Value vs. Expected Utility: Humans do not maximize absolute financial expected value (as proven by the St. Petersburg Paradox); they maximize expected utility, factoring in risk aversion.
• Classical Economics (Econs) vs. Prospect Theory (Humans): Classical economics relies on absolute levels of wealth and assumes fungibility. Prospect theory proves humans rely on reference points, track changes rather than levels, and utilize non-fungible mental accounts.
• Acquisition vs. Transaction Utility: Acquisition focuses on the product's actual consumption value. Transaction utility focuses on the psychological pleasure of a "good deal" against a reference price.
• Bundling Rules: To maximize subjective utility, segregate gains (to maximize concavity) and integrate losses (to minimize convexity and loss aversion penalties).

Must-Know Terms

• Conjunction Fallacy: Falsely assuming the combination of two specific events is more probable than a single, general event.
• Base Rate Neglect: Ignoring underlying statistical population data in favor of descriptive, representative data.
• Gambler's Fallacy: Believing that independent random events will self-correct in the short term.
• Possibility Effect: Disproportionately overweighting highly unlikely probabilities (near 0%).
• Loss Aversion: The psychological principle that losses inflict significantly more pain than equivalent gains provide pleasure.
• Endowment Effect: Setting a higher valuation on an item simply because one currently possesses it (seller's WTA > buyer's WTP).
• Curse of Knowledge: The inability to ignore possessed information when evaluating the perspective of an uninformed party.
• Fungibility: The standard economic principle that all money is identical and interchangeable, which is frequently violated by human mental accounting.