The Paradigm Shift: From Homo Economicus to Bounded Rationality

The emergence of behavioral economics represents one of the most significant paradigm shifts in the social sciences of the twentieth century, fundamentally challenging the neoclassical assumption of *Homo economicus*—the perfectly rational, self-interested, utility-maximizing agent that has dominated economic thought for over a century. This chapter traces the epistemological and historical foundations of that shift, moving from the abstract axioms of classical rationality to the empirically grounded, psychologically informed model of bounded rationality. The transition was not a sudden revolution but a cumulative process, drawing from early critiques within economics itself, the mid-century cognitive revolution in psychology, and a sustained program of experimental research that systematically documented the "anomalies" in human judgment and choice. We will critically examine the rational actor model, introduce Herbert Simon’s concept of bounded rationality as the cornerstone alternative, define the key terminology that structures the field (bias, heuristic, anomaly, descriptive versus normative), and establish the evidence-based, interdisciplinary methodology that defines the modern study of irrationality. This foundation is essential: it reframes irrationality not as random error or moral failing, but as the predictable output of a cognitive machinery operating under severe constraints of time, information, and computational capacity.

The Neoclassical Rational Actor Model: Axioms and Limitations

The neoclassical framework, crystallized in the works of Marshall, Walras, and later formalized by Samuelson and others, rests on a compact set of axioms about the decision-maker. This model, often termed the *Homo economicus* or *rational actor* model, is first and foremost a normative theory: it prescribes how decisions *should* be made to be coherent and utility-maximizing. Its descriptive power, however, has long been assumed by many economists, who treated its prescriptions as an accurate, if idealized, description of actual human behavior. The core axioms can be distilled as follows:

1. Complete and Transitive Preferences: An agent can rank any two sets of goods or lotteries (completeness) and if they prefer A over B and B over C, they must prefer A over C (transitivity). This imposes a consistent, well-ordered preference structure. 2. ​Utility Maximization: Given a set of options and their associated outcomes, the agent will choose the option that yields the highest expected utility. Utility is a cardinal or ordinal representation of preference. 3. ​Perfect Information and Costless Processing: The agent knows all relevant alternatives, their probabilities, and their outcomes. Acquiring and processing this information incurs no cognitive or temporal cost. 4. Unbounded Computational Capacity: The agent can perform any logical or mathematical operation required to compute the optimal choice, regardless of complexity. 5. Stable and Exogenous Preferences: Preferences are consistent over time and are not influenced by the context or manner in which options are presented (i.e., they are independent of framing).

These axioms yield elegant, mathematically tractable models, most notably Expected Utility Theory (EUT). For a simple lottery $L$ with outcomes $x_i$ occurring with probabilities $p_i$, the expected utility is: $$EU(L) = \sum_{i} p_i u(x_i)$$ where $u(\cdot)$ is the von Neumann-Morgenstern utility function. The rational agent chooses the lottery with the highest $EU$.

The elegance of this model is undeniable. It provides a clear benchmark for rational choice and has generated profound insights in market theory, welfare economics, and finance (e.g., the Efficient Market Hypothesis). However, its limitations as a *descriptive* theory—a theory of how people *actually* decide—became increasingly apparent from multiple fronts.

Logical and Conceptual Critiques: Even within economics, the model’s internal consistency has been questioned. The assumption of transitive preferences, for instance, struggles to explain common phenomena like preference reversals (e.g., preferring A to B in one choice set but B to A in another, even when the options are identical). The assumption of stable, context-independent preferences is directly challenged by the very phenomena this book will explore—anchoring, loss aversion, and framing effects—where the *description* of an option alters choice.

Empirical Anomalies: Long before the formal heuristics-and-biases program, field observations and experimental economics began accumulating "anomalies"—systematic deviations from EUT predictions. The Allais Paradox (1953) demonstrated that people’s choices violate the independence axiom of EUT. Faced with a choice between a certain $1 million (Option 1) and a 10% chance of $5 million, 89% chance of nothing (Option 2), most prefer the certainty. However, when the same certainty is embedded in a larger lottery (e.g., choosing between an 11% chance of $1 million and a 10% chance of $5 million, both with 89% chance of nothing), the majority preference flips to the higher-risk option, violating the axiom that the common outcome (the 89% chance of nothing) should be algebraically cancelled out. This demonstrated that certainty effects and probabilistic reasoning errors were not noise but systematic.

The Problem of Complexity: The most fundamental limitation is computational. Real-world decisions—buying a house, choosing a career, investing in a startup—involve astronomical numbers of possible states of the world, complex probabilistic dependencies, and outcomes that unfold over decades. The notion that a human mind could, in real time, compute the optimal solution according to EUT is not just implausible; it is computationally intractable. As computer scientists would later formalize, many decision problems are NP-hard, meaning their solution time grows exponentially with problem size. The human brain, with its finite neurons and metabolic energy, is a bounded computational device operating in a world of infinite complexity. The *Homo economicus* model, in its strict form, asks the impossible.

These cracks in the edifice did not immediately topple the paradigm. The dominant response within mainstream economics was often to treat observed deviations as "noise" around a rational core, to be explained away by transaction costs, asymmetric information, or the peculiarities of laboratory experiments with students. A deeper, more fundamental critique required a new foundation—a theory of how minds *actually* work under scarcity. That foundation was provided by Herbert Simon.

Herbert Simon and the Architecture of Bounded Rationality

Herbert Simon, a polymath who won the Nobel Prize in Economics (1978) and the Turing Award in Computer Science, provided the conceptual scaffolding for the entire field of behavioral economics with his concept of bounded rationality. In his seminal 1955 paper, "A Behavioral Model of Rational Choice," and later in works like *Administrative Behavior* (1947), Simon argued that rationality is always *bounded* by three constraints:

1. The Information Available: The decision-maker does not know all possible alternatives or all consequences. Information is costly to acquire and is often ambiguous or conflicting. 2. The Cognitive Limitations of the Mind: The human brain has limited working memory, attention, and computational power. It cannot perform complex integrations or exhaustive searches. 3. The Time Available: Decisions are often required under severe time pressure, precluding prolonged deliberation.

Given these bounds, Simon proposed that agents do not maximize but satisfice. They search through the set of available alternatives only until they find one that meets an acceptable threshold of adequacy—a "satisficing" solution. The aspiration level itself adapts based on