Rationality Experiments – Herbert Simon The Satisficing vs. Maximizing Studies

For more than half a century, the social sciences have been engaged in a fundamental debate over the computational architecture of human decision-making. At the epicenter of this theoretical and empirical revolution stands Herbert A. Simon, whose foundational concepts of bounded rationality and satisficing dismantled the long-standing hegemony of neoclassical economics. Traditional economic theory predicated its models on Homo economicus—a frictionless, hyper-rational actor endowed with omniscient information-gathering capacities, infinite cognitive bandwidth, and an unyielding commitment to global utility maximization. Simon observed that this idealized agent bore virtually no resemblance to living biological organisms confronting complex, dynamic, and radically uncertain environments. Real human beings do not compute the global optimum across infinite search spaces; instead, they operate within strict neurocognitive and informational constraints, relying on adaptive heuristics to find choices that are simply “good enough.”

The dichotomy between maximizing—searching exhaustively to secure the single best possible outcome—and satisficing—terminating search as soon as an alternative meets or exceeds an internal threshold of acceptability—has evolved from an abstract critique of mathematical economics into one of the most vigorously tested empirical paradigms in cognitive psychology, behavioral economics, organizational behavior, and neurobiology. Over decades of laboratory experiments, verbal protocol analyses, field studies, and psychometric assessments, researchers have mapped the behavioral mechanisms that govern how people navigate choice. The findings are paradoxical: while maximizing strategies are mathematically designed to yield superior objective outcomes, the cognitive, temporal, and emotional costs of pursuing perfection frequently induce decision paralysis, post-decisional regret, counterfactual anguish, and diminished psychological well-being.

This comprehensive treatise examines the theoretical foundations, early experimental breakthroughs, psychometric operationalizations, and contemporary neuroscientific frontiers of the satisficing versus maximizing paradigm. Tracing the lineage of thought from Simon’s early administrative observations at Carnegie Mellon University to modern computerized choice architectures, experimental economics, and artificial intelligence, this paper analyzes the empirical evidence that validates satisficing not as a regrettable cognitive defect, but as an ecologically rational triumph of human problem-solving.

1. Introduction to Herbert Simon and the Foundations of Bounded Rationality

1.1 Historical Context of Neoclassical Economics and the Rational Agent

The neoclassical economic paradigm that dominated the mid-twentieth century was anchored in an axiomatic commitment to the rational actor model. Within this framework, human agents were conceptualized as possessing complete, well-ordered preference structures, capable of performing flawless probabilistic calculations across all conceivable states of the world. This construct of Homo economicus assumed that decision-makers held access to perfect information, or at least an exhaustive probability distribution of possible outcomes, alongside an infinite capacity to process this data without cognitive friction, temporal delay, or computational cost. The overarching mandate of the agent was simple: global utility maximization.

This conceptualization achieved formal mathematical elegance through the expected utility theory articulated by John von Neumann and Oskar Morgenstern in their 1944 work, Theory of Games and Economic Behavior. By establishing the axioms of completeness, transitivity, continuity, and independence, Von Neumann and Morgenstern provided a rigorous mathematical apparatus that allowed economists to model choice under risk as the maximization of expected utility. Neoclassical theorists, notably Milton Friedman in his 1953 essay The Methodology of Positive Economics, argued that the empirical realism of an agent’s internal cognitive process was irrelevant; so long as individuals acted “as if” they were solving complex systems of differential equations to maximize utility, the predictive validity of the equilibrium models remained secure.

Herbert Simon mounted an epistemological challenge against this “as if” methodology. He argued that the neoclassical model was divorced from psychological reality and institutional context. Neoclassical economics assumed a frictionless universe wherein agents possessed instantaneous access to all relevant alternatives, computed their expected payoffs across infinite temporal horizons, and selected the single mathematical optimum without expending cognitive resources. Simon identified this as an impossible computational demand, arguing that true economic science must concern itself with the physical and cognitive limits that shape human judgment. This critique catalyzed the emergence of behavioral economics as an intellectual counter-movement, shifting the analytical focus from normative models of how idealized agents ought to choose to descriptive models of how flesh-and-blood organisms actually make decisions within finite environments.

1.2 Herbert Simon’s Academic Trajectory and Interdisciplinary Approach

Herbert Simon’s scholarly trajectory was uniquely suited to dismantling disciplinary silos. Beginning his intellectual journey in political science and public administration at the University of Chicago under the influence of Harold Lasswell and Charles Merriam, Simon developed an early fascination with how administrative organizations make choices in the absence of complete information. His cross-disciplinary foundation expanded to include mathematical sociology, operations research, computer science, and cognitive psychology, creating an intellectual breadth that resisted orthodox boundaries. This synthesis allowed Simon to view decision-making not merely as a mathematical calculation of equilibrium, but as an information-processing task executed by a biological organism or a computational machine.

The formal crystallization of Simon’s challenge to neoclassical theory appeared in his seminal 1947 monograph, Administrative Behavior. In this work, Simon argued that administrative organizations exist precisely because individual human beings are cognitively incapable of global optimization. Organizations provide the structural scaffolding—division of labor, standard operating procedures, and specialized communication channels—that enables boundedly rational agents to coordinate actions effectively. Simon demonstrated that actual administrative managers do not search for the optimal decision; they seek solutions that allow the organization to function adequately without triggering cognitive or operational collapse.

Throughout the 1950s and 1960s at the Carnegie Institute of Technology (later Carnegie Mellon University), Simon, alongside collaborators such as Allen Newell, pioneered the integration of computational simulation to study human problem-solving architectures. By treating the human brain as a physical symbol system, Simon used computer programs to model the heuristic procedures individuals apply to complex tasks. His work established that human cognition relies on selective search mechanisms rather than brute-force computation. In recognition of his revolutionary contributions, Simon was awarded the Nobel Memorial Prize in Economic Sciences in 1978, honoring his pioneering research into the decision-making processes within economic organizations and validating his lifetime project of replacing ungrounded mathematical abstraction with empirically grounded cognitive reality.

1.3 Conceptualizing Bounded Rationality

Central to Simon’s theoretical architecture is the formal concept of bounded rationality. Simon took pains to clarify that bounded rationality is neither synonymous with irrationality nor indicative of an inherent cognitive pathology. Irrationality implies behavior that systematically violates logical coherence, self-interest, or basic internal consistency. Bounded rationality, by contrast, describes an organism attempting to achieve its goals within the severe constraints imposed by its neurocognitive architecture and the external environment. It is a model of adaptive problem-solving designed to function when classical optimization is mathematically impossible or physically prohibitive.

Simon articulated these constraints as operating across two distinct domains: internal cognitive limits and external environmental structures. The internal limits encompass finite short-term working memory capacity (classically quantified by George Miller as seven plus or minus two chunks of information), limited attentional bandwidth, serial rather than massively parallel conscious processing, and finite computational speed. An individual cannot hold millions of potential permutations in working memory simultaneously. Concurrently, the external environment presents its own structural constraints, characterized by incomplete and noisy information, radical uncertainty, dynamic shifts in environmental states, and strict temporal deadlines that penalize prolonged deliberation.

To illustrate the necessary synthesis of these factors, Simon introduced the celebrated metaphor of the scissors. Rational human behavior, Simon asserted, is shaped by a pair of scissors whose two blades are the cognitive limitations of the actor and the structure of the task environment. To examine only the cognitive capacities of the mind without understanding the environment in which it operates—or conversely, to analyze environmental incentives while assuming an omniscient computational mind—is as futile as attempting to understand how a pair of scissors cuts by examining only one blade. This conceptualization led to Simon’s crucial distinction between substantive rationality and procedural rationality. Substantive rationality, the hallmark of neoclassical economics, evaluates a decision exclusively by its final outcome relative to a theoretical maximum. Procedural rationality, the cornerstone of bounded rationality, evaluates the operational integrity and efficacy of the heuristic search processes, stopping rules, and information-gathering mechanisms used to reach a decision given the agent’s operating constraints.

2. The Theoretical Mechanics: Maximizing Versus Satisficing

2.1 The Maximization Paradigm: Mathematical and Decision Frameworks

The mathematical formulation of global maximization requires identifying the precise element within a choice set that yields the highest possible utility. Formally, given a comprehensive choice set $\mathcal{X}$ and a utility function $U: \mathcal{X} to \mathbb{R}$, a maximizing agent must execute the operation:

$$x^* = arg\max_{x in \mathcal{X}} U(x)$$

In deterministic models with well-behaved, continuous, and convex sets, this requires computing the first-order partial derivatives and solving for zero across all dimensions. In non-deterministic contexts governed by risk or uncertainty, the objective function shifts to the maximization of expected utility, requiring the agent to integrate across all possible states of nature weighted by their subjective or objective probabilities:

$$x^* = arg\max_{x in \mathcal{X}} \mathbb{E}[U(x)] = arg\max_{x in \mathcal{X}} \sum_{s in \mathcal{S}} p(s) U(x, s)$$

This formulation operates under the assumption that the choice set $\mathcal{X}$ is fully known or can be surveyed without cost. However, in any non-trivial real-world scenario, information acquisition is costly. As economists such as George Stigler pointed out, an agent must expend time, energy, and capital to discover each additional element of $\mathcal{X}$. Yet, attempting to optimize the search itself introduces an intractable infinite regress. If an agent must maximize utility by searching for options until the marginal benefit of further search equals its marginal cost, they must first calculate the expected marginal benefit of an unknown piece of information. To optimize the decision of how much to search, one must optimize the allocation of resources to the search-optimization meta-problem, ad infinitum. The meta-decision problem itself becomes computationally non-viable.

Furthermore, human decision spaces are routinely characterized by computational intractability, representing NP-hard combinatorial optimization problems. Consider the classical Traveling Salesperson Problem or the game of chess: while the search spaces are mathematically finite, the number of possible trajectories ($\approx 10^{120}$ in chess) vastly exceeds the number of atoms in the observable universe. Strict maximization under such conditions requires psychological and computational prerequisites that are fundamentally impossible: infinite computational bandwidth, instantaneous parallel calculation, absolute affective detachment, and an unshakeable certainty regarding one’s own future preference landscapes.

2.2 The Satisficing Principle: Heuristic Search and Aspiration Levels

Recognizing the mathematical and computational impossibility of unbounded optimization, Simon coined the term “satisficing”—a portmanteau blending the Northumbrian English dialect word satisfice (to satisfy) with suffice. Satisficing shifts the decision architecture from global comparative evaluation to threshold-based local search. Instead of surveying the totality of $\mathcal{X}$ to identify the unique maximum $x^*$, a satisficing agent evaluates alternatives sequentially, terminating search the moment an option is encountered that meets or exceeds an internal, multidimensional vector of aspiration levels, denoted as $boldsymbol{\alpha}$.

Sequential search mechanics dictate that options are processed in the order they are encountered in time and space. The decision-maker does not need to construct a comprehensive ranking of all possibilities; rather, each alternative $x_i$ is evaluated against the aspiration vector:

$$\text{Accept } x_i iff f_k(x_i) ge \alpha_k \quad \forall k in {1, 2, dots, K}$$

where $f_k(x_i)$ represents the evaluated performance of the alternative across attribute $k$, and $\alpha_k$ is the minimum acceptable threshold for that specific attribute. If any attribute fails to meet the requisite threshold, the alternative is rejected, and search proceeds to $x_{i+1}$.

Crucially, Simon’s satisficing framework is not static; it relies on dynamic aspiration levels that continuously adapt to environmental feedback. The aspiration vector $boldsymbol{\alpha}(t)$ fluctuates as a function of the agent’s recent search history and environmental munificence:

$$boldsymbol{\alpha}(t+1) = boldsymbol{\alpha}(t) + \gamma \left( \mathbf{v}(t) – boldsymbol{\alpha}(t) \right)$$

where $\mathbf{v}(t)$ represents the value of the most recently observed outcomes, and $\gamma in (0, 1)$ governs the speed of adaptation. When the environment is rich and acceptable options are discovered with minimal expenditure of search effort, aspirations adjust upward, preventing premature settlement on low-quality outcomes. Conversely, when search encounters persistent failure, cognitive exhaustion, or temporal scarcity, aspirations adjust downward, lowering the threshold to ensure that a choice can be made before resources are depleted. Stopping rules in satisficing algorithms are therefore determined by the intersection of immediate experiential evaluation and dynamic threshold adjustment, guaranteeing decision completion without requiring global search.

2.3 Comparative Formalization: Optimization Versus Aspiration Matching

The structural divergence between optimization and satisficing is clearly observed when comparing their underlying utility functions. Unbounded optimization operates across a continuous, strictly monotonic utility curve wherein every infinitesimal improvement in outcome quality confers a proportional increase in evaluated utility. Aspiration matching, by contrast, functions through a non-linear step-function. An outcome that falls below the aspiration threshold $\alpha$ yields a subjective value of zero (or rejection), whereas any outcome that meets or exceeds $\alpha$ transitions the agent into an undifferentiated state of acceptability:

$$V(x) = \begin{\cases} 1 & \text{if } U(x) ge \alpha \ 0 & \text{if } U(x) < \alpha \end{\cases}$$

This structural difference exposes the trade-offs between decision accuracy and resource expenditure. Maximization allocates infinite or highly disproportionate resources (cognitive load, time, metabolic energy, emotional labor) in pursuit of the marginal delta between a very good outcome and the theoretically perfect outcome. Satisficing deliberately accepts an outcome delta—the distance between the chosen satisficing alternative and the unobserved theoretical maximum—in exchange for an immense reduction in search costs and cognitive computational burden.

In non-ergodic and fundamentally uncertain environments—where future probability distributions are non-calculable, unstable, or shifting—the pursuit of mathematical optimization frequently leads to catastrophic failure. An optimal solution calibrated to past distributions may prove brittle when confronted with structural environmental shocks. Satisficing heuristics, unencumbered by the need to fit noise in complex historical datasets, exhibit robust ecological rationality. Under conditions of environmental opacity, high dimensionality, and computational intractability, satisficing algorithms routinely match or outperform formal optimization models, demonstrating that heuristic simplicity is often an evolutionary safeguard rather than a cognitive vulnerability.

3. Early Experimental Paradigms: Simon’s Problem-Solving and Protocol Analysis

3.1 Verbal Protocol Analysis Methodology

To investigate bounded rationality empirically, Simon rejected both the behaviorist refusal to study internal mental states and the armchair introspectionism of early psychology. Collaborating closely with Allen Newell, Simon developed verbal protocol analysis as a rigorous, empirically grounded methodology for mapping cognitive processes. Protocol analysis required human participants to engage in “concurrent think-aloud” reporting while actively resolving complex, multi-stage logic problems, cryptarithmetic puzzles, or strategic games. Participants were explicitly instructed not to rationalize, justify, or retrospectively explain their reasoning, but simply to verbalize the stream of conscious thoughts moving through their working memory as they engaged with the problem.

Simon and Newell established rigorous epistemological criteria to demonstrate that concurrent verbal reporting does not alter the underlying structure of cognitive processing, unlike retrospective rationalizations, which are notoriously vulnerable to post-hoc fabulation and memory degradation. These audio-recorded verbalizations were transcribed, broken down into discrete informational phrases, and encoded into formal semantic networks known as Problem Behavior Graphs (PBGs). The PBG mapped the step-by-step trajectory of a human subject moving through an abstract “problem space,” delineating the nodes representing knowledge states and the directional vectors representing operators applied to shift from one state to the next.

The empirical data gathered from these protocols revealed that human beings do not execute exhaustive combinatorial searches across problem spaces. Participants did not trace out every possible permutation to find a globally optimal path. Instead, the transcripts illuminated a process of selective, highly constrained search guided by heuristics and threshold rules. Despite early critiques that verbal reporting was inherently subjective or incomplete, Simon and Newell proved that protocol analysis possessed high inter-rater reliability, objective falsifiability, and predictive validity, laying the methodological groundwork for the cognitive revolution and empirical studies of bounded rationality.

3.2 Chess Expertise and Heuristic Search Experiments

To demonstrate satisficing and selective heuristic search in an ecologically complex environment, Simon teamed with William G. Chase in 1973 to conduct landmark experiments analyzing chess expertise. Chess served as the model system for human cognition, much as Drosophila served for modern genetics. Neoclassical maximization applied to chess would require a player to calculate the complete minimax tree across all moves and counter-moves—an operational impossibility given that the game tree complexity contains an estimated $10^{120}$ possible positions. Chase and Simon sought to uncover how grandmasters routinely execute near-optimal moves within seconds without calculating millions of trajectories.

In their classic experimental paradigm, Chase and Simon exposed chess grandmasters, intermediate players, and novices to chessboard positions taken from actual historical master games for brief periods, typically five seconds. The board was then cleared, and participants were instructed to reconstruct the spatial configuration of the pieces from memory. When presented with meaningful, tactically structured board positions from real games, grandmasters demonstrated phenomenal recall accuracy, correctly placing virtually all twenty to twenty-five pieces, whereas novices could only place a handful. However, Chase and Simon introduced an ingenious control condition: they presented the exact same pieces arranged in completely random spatial configurations that violated all tactical rules of chess. Under this randomized condition, the grandmasters’ superior recall vanished completely; their performance degraded to the baseline level of novices.

Chase and Simon demonstrated that the grandmaster’s advantage did not stem from superior raw working memory capacity or an ability to maximize by searching deeper into the decision tree. Rather, the master’s advantage was anchored in cognitive chunking. Grandmasters had spent thousands of hours storing tens of thousands of structural patterns (“chunks”) in long-term memory. When observing a board, their pattern-recognition apparatus immediately grouped discrete pieces into functional configurations. Rather than calculating 20 moves ahead across 30 branches, grandmasters engaged in highly selective search, evaluating only two or three plausible candidate moves that met an immediate aspiration threshold. They satisficed by narrowing the vast search space to a manageable set of high-probability paths, relying on heuristic stopping rules to select a strong move rather than exhaustively computing every failure state.

3.3 The Logic Theorist and General Problem Solver (GPS) Simulations

To prove that satisficing heuristics could solve problems that were computationally intractable via brute-force optimization, Simon, Newell, and Cliff Shaw engineered the Logic Theorist in 1956, widely recognized as the world’s first artificial intelligence program. Running on an early JOHNNIAC computer, the program was designed to prove mathematical theorems from Chapter 2 of Alfred North Whitehead and Bertrand Russell’s Principia Mathematica. Rather than utilizing algorithmic brute-force search—which would exhaustively generate every logically valid proposition until the target theorem was struck—the Logic Theorist implemented heuristic search algorithms that mimicked human bounded rationality.

The core computational engine of this paradigm was refined in their subsequent architecture, the General Problem Solver (GPS). GPS introduced means-ends analysis as a formalized heuristic. In means-ends analysis, the program does not attempt to calculate all paths simultaneously; instead, it evaluates the current state against a target goal state, identifies the most prominent difference between them, searches its heuristic memory for an operator specifically capable of reducing that precise difference, and tests that operator. If a sub-problem or obstacle arises, it sets up an intermediate aspiration threshold—a sub-goal—resolves it satisficingly, and then proceeds. The Logic Theorist famously proved 38 of the first 52 theorems in Principia Mathematica, even finding a shorter, aesthetically superior proof for Theorem 2.85 than Russell and Whitehead had produced by hand.

Crucially, Simon and his colleagues compared the operational traces generated by GPS with the empirical verbal protocol transcripts of human subjects solving the exact same logic problems. The alignment was striking: the computational search paths, sub-goal generation sequences, and heuristic stopping points implemented by the computer program matched the cognitive trajectories of human thinkers. This empirical demonstration validated Simon’s physical symbol system hypothesis and provided undeniable proof that complex, intelligent problem-solving does not require global maximization. Bounded heuristic search, guided by local aspiration thresholds, was shown to be computationally viable, empirically observable, and functionally sufficient for navigating high-dimensional problem domains.

4. Behavioral Operationalization: Measuring Maximizing and Satisficing Tendencies

4.1 Schwartz et al. (2002) and the Individual Differences Paradigm

For several decades following Simon’s foundational work, satisficing was treated primarily as a universal descriptive property of human cognition—a shared computational boundary imposed by the physical architecture of the brain. In 2002, an intellectual shift occurred with the publication of a seminal paper by Barry Schwartz, Andrew Ward, John Monterosso, Sonja Lyubomirsky, Katherine White, and Darrin Lehman. Schwartz and his colleagues hypothesized that while bounded rationality limits everyone’s cognitive capacity, individuals differ systematically in their orientation toward choice. They conceptualized satisficing and maximizing not merely as computational protocols, but as stable, measurable individual difference traits along a psychometric continuum.

To capture these behavioral variations, Schwartz et al. developed and psychometrically validated the 13-item Maximization Scale (MS) alongside a concurrent Regret Scale. The Maximization Scale assessed individual tendencies across three operational dimensions:

• Alternative Search: The drive to explore every conceivable option before concluding a choice (e.g., “When I am in the car listening to the radio, I often check other stations to see if something better is playing, even if I am relatively satisfied with what I’m listening to”).
• Decision Difficulty: The subjective experience of cognitive strain and paralysis when confronted with decisions (e.g., “I find that I have a hard time shopping for a gift for a friend”).
• High Standards: The personal insensitivity to “good enough” and the unyielding requirement for perfection (e.g., “No matter what I do, I have the highest standards for myself”).

The empirical findings reported by Schwartz et al. challenged the foundational economic assumption that having more options and pursuing the best inevitably enhances human welfare. Across a series of surveys administered to diverse demographic cohorts, the researchers discovered that maximizing scores were negatively correlated with subjective well-being, happiness, self-esteem, and life satisfaction, while positively correlated with depression, perfectionism, pessimism, and vulnerability to post-decisional regret. Satisficers—individuals whose response patterns favored meeting aspiration thresholds without exhaustive search—consistently reported higher subjective well-being and lower affective distress, establishing maximizing as a significant psychological risk factor.

4.2 Psychometric Refinements and Structural Debates

The introduction of the original Maximization Scale ignited intense psychometric debate within personality psychology and behavioral decision theory. Methodologists rapidly identified structural deficiencies in Schwartz’s 13-item scale, specifically targeting its internal consistency, factor stability, and construct validity. Critics such as Nenkov, Morwitz, Schwartz, and Ward (2008) conducted extensive structural equation modeling and exploratory factor analyses, which revealed that the three sub-dimensions of the MS did not always load coherently onto a single overarching latent construct. In particular, the “High Standards” subscale demonstrated anomalous behavior, frequently correlating positively with positive psychological traits such as need for cognition and conscientiousness, while the “Alternative Search” and “Decision Difficulty” subscales accounted for the bulk of the negative correlations with depression and regret.

To resolve these psychometric anomalies, Nenkov et al. (2008) developed a refined 6-item short-form scale (the MS-S), preserving the tripartite structure while eliminating items that generated statistical noise. Concurrently, other researchers challenged whether Schwartz’s operationalization had conflated the core desire to make an optimal choice with the auxiliary difficulty of decision-making. Diab, Gillespie, and Highhouse (2008) argued that the original scale improperly conflated maximizing with neurotic indecisiveness. They introduced the Maximizing Tendency Scale (MTS), redefining maximizing purely as the pursuit of the best outcome, unburdened by items measuring decision difficulty or consumer search exhaustion:

$$MTS_i = \frac{1}{N} \sum_{j=1}^N S_{ij}$$

where items $S_j$ focused strictly on optimizing standards (e.g., “No matter what it takes, I prefer to find the best possible option”). Under this re-operationalization, Diab et al. found that maximizing was not associated with clinical depression or lower life satisfaction, suggesting that the psychological costs observed by Schwartz were driven by indecisiveness and neurosis rather than high standards per se.

This empirical tension triggered a deeper theoretical debate: does satisficing constitute the polar opposite of maximizing along a single continuous bipolar spectrum, or do satisficing and maximizing represent two orthogonal, independent psychological dimensions? Authors such as Turner, Rim, Betz, and Schmitt (2012) proposed that an individual could possess both high standards (a maximizing trait) and an ability to experience contentment once an internal threshold is crossed (a satisficing trait). This orthogonal conceptualization allowed for the categorizing of individuals into distinct quadrants—such as “maximizing satisficers” versus “maximizing non-satisficers”—resolving decades of conflicting psychometric data and establishing that the psychological penalty lies not in seeking excellence, but in the inability to terminate search once an aspiration threshold has been reached.

4.3 Experimental Validation of Measurement Scales

To confirm that self-reported scores on maximization scales corresponded to observable behavioral differences rather than mere communicative posturing, researchers transitioned from survey metrics to rigorous laboratory experiments. In these empirical settings, participants were placed in controlled information environments where their actual search behavior, cognitive effort, and decision timelines could be precisely quantified. Researchers utilized computerized information boards, mouse-tracking software (such as Mouselab), and high-resolution eye-tracking systems to record how individuals traversed choice architectures.

The behavioral results confirmed the psychometric classifications. When presented with multi-attribute matrices containing products (such as consumer electronics, apartments, or insurance policies), individuals scoring high on maximizing scales spent significantly more time browsing, clicked on a substantially greater percentage of available information cells, and revisited previously inspected options far more frequently than satisficers. Eye-tracking data revealed distinct visual trajectories: satisficers exhibited visual search patterns characterized by quick terminations once an option with acceptable attributes across focal dimensions was identified, whereas maximizers demonstrated chaotic, dense saccadic scanning patterns, repeatedly comparing alternatives to verify that no marginal advantage was overlooked.

Cross-cultural experiments further expanded the validity of these constructs, illustrating how cultural context moderates the psychological outcomes of maximizing. Studies comparing individualistic Western cohorts (e.g., the United States) with collectivist East Asian cohorts (e.g., Japan, China) demonstrated that while maximizing search behavior remains universally taxing on working memory, the depressive sequelae are far more pronounced in individualistic cultures. In societies where individual autonomy and self-realization are paramount, personal responsibility for sub-optimal outcomes is internalized, amplifying regret. Neurocognitive investigations utilizing fMRI and EEG during choice tasks confirmed that high-maximizing traits correlate with sustained activations in the dorsolateral prefrontal cortex (dlPFC) and anterior cingulate cortex (ACC)—neural regions associated with intense cognitive control, conflict monitoring, and the distress of choice ambiguity—confirming the biological reality of maximizing costs.

5. Experimental Studies in Consumer Choice: The Paradox of Choice

5.1 Choice Overload and Assortment Size Experiments

The intersection of Simon’s satisficing principle and contemporary behavioral science culminated in the empirical documentation of “choice overload,” popularly termed the paradox of choice. Standard neoclassical microeconomics assumes that assortment size is monotonically related to consumer welfare; expanding choice sets strictly increases the probability that an optimizing agent can locate an alternative that matches their indifference curves, while leaving the agent free to ignore irrelevant options. In 2000, Sheena Iyengar and Mark Lepper published a series of classic experiments that dismantled this assumption.

In their renowned field experiment conducted in an upscale grocery store, Iyengar and Lepper established an active tasting booth featuring either a limited assortment (6 flavors of Wilkin & Sons exotic jam) or an extensive assortment (24 flavors). While the extensive display attracted a larger crowd of browsers (60% of passersby stopped at the 24-jam display versus 40% at the 6-jam display), the subsequent purchasing behavior revealed a dramatic inversion: 30% of the consumers exposed to the small assortment actually bought a jar of jam, compared to a mere 3% of those exposed to the 24-jam condition. Consumers exposed to choice overload were ten times less likely to make a purchase. Subsequent laboratory experiments utilizing gourmet chocolates replicated this effect, showing that participants navigating extensive assortments found the decision process significantly more frustrating and exhausting, and paradoxically reported lower post-consumption satisfaction with the chocolates they chose.

The differential impact of assortment size on satisficers versus maximizers proved decisive in clarifying the mechanics of this phenomenon. Maximizers, bound by the internal mandate to evaluate all available alternatives to guarantee the selection of the optimal product, experience exponential cognitive strain as the choice set expands. For a satisficer, an increase in assortment size from 6 to 24 simply raises the probability that an acceptable option will be encountered early in the sequential search; once an option exceeds their aspiration threshold $\alpha$, search ceases. For the maximizer, however, an increase from 6 to 24 options increases the number of bilateral comparisons from $\binom{6}{2} = 15$ to $\binom{24}{2} = 276$. The subjective experience for the maximizer is not one of abundance, but of decision exhaustion, cognitive depletion, and creeping anxiety.

5.2 Information Search and Regret in E-Commerce Environments

The explosion of digital commerce and algorithmic retail platforms converted the laboratory findings of choice overload into widespread societal realities. In e-commerce environments where thousands of competing products can be accessed within milliseconds, the operational distinction between satisficing and maximizing search patterns becomes acutely apparent. Using digital clickstream analysis, researchers have systematically mapped how consumers navigate algorithmic filtering, pagination, and dynamic sorting mechanisms (e.g., sorting by price, customer ratings, or relevance).

Maximizers in digital spaces demonstrate expansive search breadths, opening dozens of competing browser tabs, filtering and re-sorting across multiple criteria, and spending hours comparing technical specifications. Yet, this exhaustive digital search triggers intense counterfactual thinking. As an agent evaluates more features, the distinct non-selected positive attributes of unchosen alternatives become highly salient. Maximizers engage in pervasive upward counterfactual thinking (“If only I had bought the other model with the longer battery life…”), which retroactively undermines their subjective evaluation of the selected good. The digital abundance transforms every choice into a minefield of foregone benefits.

Experimental manipulations have highlighted how platform architecture shapes these outcomes:

• Choice Reversibility and Return Policies: Generous 30-day “no questions asked” return policies—intended to alleviate purchasing risk—paradoxically increase post-purchase dissatisfaction among maximizers. Reversibility prevents psychological closure; it leaves the decision state open, compelling maximizers to continue monitoring alternative products and prices long after the transaction has concluded. Satisficers, conversely, view the return policy merely as an insurance baseline and mentally close the transaction once the product clears their aspiration threshold.
• Algorithmic Default Settings: The implementation of default settings, curated selections, and personalized recommendation engines (“Customers who bought this also liked…”) functions as an externalized satisficing heuristic. These systems artificially truncate the search space, reducing cognitive load for satisficers while provoking defensive resistance among extreme maximizers, who suspect that the algorithm is withholding superior alternatives.

5.3 Subjective Well-Being, Envy, and Post-Decision Regret

The psychological friction experienced by maximizers extends far beyond the immediate purchasing environment; it exerts a persistent drag on long-term subjective well-being. At the core of this malaise is the phenomenon of post-decision regret and buyer’s remorse. When an individual strives for the absolute best, any minor imperfection, mechanical flaw, or aesthetic defect in the chosen item represents an absolute failure of optimization. The satisficer, operating under the principle that the chosen option simply needs to be “good enough,” accepts minor flaws as within acceptable operational tolerances. The gap between expectation and reality is structurally narrower for the satisficer, conferring long-term emotional resilience.

This dynamic is amplified by social comparison orientation. Maximizers are chronically dependent on external, relative social benchmarks rather than internal, absolute criteria. Because the objective “best” in human social ecosystems is continuously shifting, maximizers are hyper-sensitized to the acquisitions, lifestyles, and achievements of their peers. Laboratory social-dilemma and reward tasks demonstrate that maximizers experience acute envy and diminished self-esteem when informed that a peer outperformed them, even if their own objective score was exceptionally high. Satisficers, by contrast, evaluate their outcomes against their internal aspiration level; peer performance leaves their subjective satisfaction largely unaffected so long as their personal standards are fulfilled.

Ultimately, maximizing accelerates the hedonic treadmill. When a maximizer expends extraordinary cognitive and emotional effort to secure an optimal outcome, their psychological expectations adapt almost instantaneously to this elevated baseline. The marginal utility derived from the optimal acquisition evaporates through rapid hedonic adaptation, leaving the agent trapped in a continuous cycle of exhaustive search, fleeting satisfaction, and lingering regret. Satisficing operates as an adaptive psychological defense mechanism: by accepting sufficiency and truncating search, the satisficer short-circuits the hedonic treadmill, protects their cognitive reserves, and insulates themselves against post-choice devaluation.

6. Experimental Economics: Laboratory Testing of Satisficing Rules

6.1 Sequential Search and Optimal Stopping Tasks

Experimental economics has subjected Simon’s satisficing framework to quantitative laboratory testing through sequential search and optimal stopping paradigms. The canonical mathematical benchmark for these experiments is the classical Secretary Problem (also known as the marriage or dowry problem). In the standard mathematical formulation, an agent must interview $N$ applicants sequentially for a single position. The total number of applicants is known in advance, but they arrive in random order. After each interview, the agent must immediately hire or reject the candidate; rejections are definitive and irreversible. The agent’s objective is to maximize the probability of selecting the single best applicant from the pool.

The optimal mathematical stopping rule derived from classical probability theory requires the agent to reject the first $r-1$ applicants without hiring, where:

$$r \approx \frac{N}{e} \approx 0.368 \cdot N$$

The agent then selects the first subsequent candidate who is superior to all preceding candidates observed in the initial $36.8%$ baseline-setting phase. This strategy yields an optimal success probability of $1/e \approx 36.8%$.

When experimental economists place human subjects in laboratory implementations of the Secretary Problem, human empirical behavior systematically deviates from this mathematical benchmark:

• Premature Stopping: Human participants stop significantly earlier than the optimal $N/e$ threshold. Rather than sampling through 37% of the pool to establish a baseline, human decision-makers consistently truncate search around the 15% to 25% mark, hiring a candidate who meets a satisfactory threshold rather than holding out for the optimal applicant.
• Costly Search Adaptations: In costly search paradigms—where each additional draw from an unknown distribution incurs a fixed monetary cost—human subjects do not compute the complex dynamic programming equations required by optimal search theory. Instead, they dynamically adjust an internal reservation wage or aspiration threshold $\alpha$. As cumulative search costs mount, participants rapidly depress $\alpha$, terminating the search early to stem further resource losses.
• Heuristic Stopping under Opacity: In environments characterized by structural opacity—where the total pool size $N$ is unknown or the distribution parameters are unstable—human agents abandon optimization altogether, relying exclusively on satisficing stopping heuristics that preserve capital and cognitive energy.

6.2 Payoff Matrices, Risk, and Expected Value Calculations

Laboratory lotteries and risky choice experiments have provided further empirical arenas for contrasting expected utility maximization with threshold-based aspiration models. When presented with complex payoff matrices, human decision-makers routinely violate the core axioms of expected utility theory, demonstrating that choice under risk is guided by reference points and aspiration boundaries rather than continuous expected value computations.

This behavioral reality is formalized in the intersection between bounded rationality and Daniel Kahneman and Amos Tversky’s Prospect Theory. In Prospect Theory, outcomes are evaluated not in terms of absolute terminal wealth states, but as gains and losses relative to an internal reference point. This reference point functions as a psychological satisficing threshold. Laboratory experiments demonstrate that when participants are tasked with choosing among competing gambles, they do not calculate the global expected value:

$$\mathbb{E}[V] = \sum p_i v(x_i)$$

Instead, they focus on categorical thresholds: does the gamble offer a high probability of exceeding an immediate aspiration level (e.g., breaking even or avoiding a painful loss)? The presence of loss aversion—wherein the psychological pain of a loss is approximately twice as intense as the pleasure of an equivalent gain—drives decision-makers to adopt satisficing rules explicitly structured to minimize the probability of falling below the survival threshold.

Furthermore, as experimental economists systematically manipulate task complexity—by increasing the number of probabilistic branches, introducing compound lotteries, or shortening the time allowed per choice—participants exhibit a predictable shift. In simple, low-dimensional choices, agents may approximate expected value calculations; however, as dimensionality and computational friction rise, they abandon optimizing calculations and adopt satisficing rules. They evaluate gambles based on whether they meet a single dominant aspiration criterion (e.g., “guarantees at least $50 with probability$ge 0.8$”), proving that satisficing is the brain’s default operational protocol under complexity.

6.3 Multi-Attribute Choice Experiments

Real-world decisions rarely involve a single unidimensional variable like monetary payoff; they require navigating multi-attribute spaces where options possess conflicting dimensions (e.g., evaluating a vehicle based on price, safety, fuel efficiency, aesthetic design, and cargo volume). Neoclassical maximization mandates the application of a Weighted Additive Model (WADD), wherein the agent identifies the value of every attribute $j$ for every alternative $i$, multiplies it by an importance weight $w_j$, and computes an aggregate score:

$$U_i = \sum_{j=1}^M w_j a_{ij}$$

The agent then selects the alternative that maximizes this comprehensive sum. This process demands immense computational capacity, requiring the trade-off of attributes across diverse scales (e.g., trading off safety ratings against dollars spent).

Experimental studies using computerized process-tracing environments, such as Mouselab, have dismantled the assumption that humans rely on weighted additive optimization. Pioneered by John Payne, James Bettman, and Eric Johnson in their classic work The Adaptive Decision Maker, process-tracing involves concealing attribute values behind closed digital cells on a display screen. To view a value, the participant must move a cursor or mouse over the cell, recording the exact sequence, duration, and order of information acquisition.

These experiments demonstrated that human participants universally transition to non-compensatory, satisficing heuristics as task complexity grows:

• Elimination by Aspects (EBA): Formalized by Amos Tversky, EBA models sequential satisficing across multiple attributes. The decision-maker selects the most critical attribute, sets an aspiration threshold, and immediately eliminates every alternative that fails to meet that threshold. The agent then takes the second most important attribute, applies a new threshold, and continues this process until a single candidate remains. No trade-offs are calculated; options are filtered out based on failure to satisfy thresholds.
• Lexicographic Heuristics: In lexicographic choice, the agent identifies the single most important attribute and selects the alternative that performs best on that dimension alone, completely ignoring all other data. If a tie occurs, the agent evaluates the second attribute satisficingly.
• Non-Compensatory Evaluation: Laboratory eye-tracking and Mouselab data confirm that humans abandon compensatory optimization (where a high score on one attribute compensates for a failure on another) in favor of non-compensatory satisficing rules whenever choices involve more than three or four alternatives, validating Simon’s assertion that real-world multi-attribute choice is procedural, sequential, and threshold-based.

7. Organizational and Managerial Decision-Making Experiments

7.1 Administrative Behavior in Experimental Simulators

Extending Herbert Simon’s initial administrative insights, management researchers have constructed computerized microworld simulations to study executive decision-making within complex organizational systems. Utilizing dynamic software environments such as John Sterman’s “Beer Distribution Game” or simulated corporate crisis platforms, researchers place business executives and management students in charge of virtual enterprises characterized by non-linearities, feedback delays, noisy market signals, and severe time pressure.

Neoclassical models predict that managers will optimize production schedules, capital investments, and inventory orders by calculating the system’s dynamic equilibrium. In practice, experimental simulations consistently reveal widespread managerial failure when optimization is attempted. Managers routinely succumb to “misperceptions of feedback,” over-ordering inventory in response to demand spikes and initiating wild systemic oscillations (the bullwhip effect). Optimization models fail because the complex, multi-loop dynamic feedback systems are non-linear and computationally intractable for human working memory.

Successful management within these experimental simulators is achieved not by managers pursuing global maximization, but by those implementing robust satisficing rules supported by organizational slack. As formalized by Richard Cyert and James March in their foundational 1963 volume, A Behavioral Theory of the Firm, organizational slack—the accumulation of excess resources, buffer inventories, and reserve financial capital—serves as a vital shock absorber. Slack insulates the firm from unexpected environmental shocks, allowing managers to satisfice. In experimental market simulations subjected to sudden volatility, firms governed by maximizing protocols that run on zero-slack efficiency margins frequently experience catastrophic insolvency, whereas firms operating under satisficing heuristics with built-in operational buffers adapt and survive.

7.2 Career Choice and Labor Market Realities

The profound real-world consequences of maximizing versus satisficing behavioral strategies were comprehensively demonstrated in an empirical field study conducted by Sheena Iyengar, Rachel Wells, and Barry Schwartz in 2006. The researchers tracked a cohort of 548 graduating college students from eleven elite universities throughout their senior year job search process, measuring their maximization tendencies early in the cycle and documenting their subsequent employment outcomes, emotional trajectories, and professional tenure.

The empirical findings revealed a stark paradox between objective and subjective outcomes:

• The Objective Maximizer Advantage: Students who scored high on the maximization scale secured jobs with starting salaries that were, on average, 20% higher than their satisficing peers ($$44,515$ versus$$37,084$). Maximizers applied to significantly more jobs, attended more interviews, and exhaustively researched employment landscapes to land top-tier, high-paying corporate positions.
• The Subjective Maximizer Penalty: Despite their higher compensation, maximizers were significantly less satisfied with the job offers they accepted, reported higher levels of anxiety, stress, and self-doubt throughout the interview cycle, and exhibited elevated rates of depressive affect. They were haunted by upward counterfactuals, constantly wondering whether an unpursued interview or unaccepted offer at another firm might have provided superior fulfillment.
• The Satisficer Advantage: Satisficing students, who accepted the first job offer that exceeded their internal aspiration threshold regarding mission, compensation, and geographic location, reported significantly higher job satisfaction, greater vocational commitment, and lower turnover intentions within their first year of employment.

This study remains one of the most powerful empirical demonstrations in the literature: maximizing can yield marginally superior objective metrics at the cost of profound subjective dissatisfaction.

7.3 Strategic Management: Satisficing in Corporate Governance

Corporate governance, strategic capital allocation, and executive management provide real-world testing grounds for Simon’s satisficing models. Neoclassical financial theory asserts that corporate boards and chief executive officers execute capital budgeting decisions by computing the Net Present Value (NPV) across all potential capital projects, funding initiatives down to the point where the marginal cost of capital equals the marginal rate of return. However, empirical field studies of corporate finance (such as Graham and Harvey’s extensive corporate surveys) reveal that actual corporate boards consistently rely on heuristic satisficing rules.

The widespread use of simple hurdle rates and payback periods illustrates this dynamic. Rather than calculating complex, multi-scenario discounted cash flow models that require arbitrary assumptions about discount rates thirty years into the future, corporate boards mandate that a project must clear a straightforward hurdle rate (e.g., “any investment must deliver a return $ge 12%$“) or recover its initial capital outlay within a set payback period (e.g., $le 3$ years). Projects that clear these satisficing thresholds are approved; those that do not are rejected. This threshold rule radically reduces agency conflict, simplifies auditing, and prevents computational paralysis within boardrooms.

Nowhere are the pitfalls of maximizing more evident than in corporate Mergers and Acquisitions (M&A). Neoclassical theory conceptualizes M&A as an optimizing mechanism designed to capture operational synergies. In reality, empirical management studies demonstrate that between 70% and 90% of all corporate mergers fail to create shareholder value, frequently destroying it. This failure is driven by CEO hubris and maximizing behavior: executive leadership pursues empire-building acquisitions in an ungrounded effort to optimize market domination, taking on massive leverage and organizational complexity. Firms that satisfice—engaging in targeted, programmatic acquisitions designed to fill specific operational gaps that meet explicit strategic thresholds—consistently outperform empire-building maximizing firms in long-term total shareholder return.

Similarly, James March’s classic exploration of the tension between exploration and exploitation in organizational learning highlights how satisficing governs successful research and development (R&D) investments. Maximizing firms often over-optimize their current core competencies (exploitation), achieving extreme short-term operational efficiency while leaving themselves vulnerable to disruptive technological shifts. Resilient organizations maintain dynamic aspiration levels, continually allocating capital to exploratory, high-risk innovation not because it is mathematically guaranteed to maximize returns, but because it satisfies the vital institutional threshold of long-term adaptive survivability.

8. Neuroscience and Cognitive Psychology of Bounded Decisions

8.1 Neural Mechanisms of Value Representation and Choice Selection

The biological validity of Herbert Simon’s bounded rationality and the maximizing/satisficing divide has received empirical corroboration from functional neuroimaging and neuroeconomics. Using functional Magnetic Resonance Imaging (fMRI), researchers have mapped the distinct neural circuits engaged during value computation, choice conflict, and stopping rule execution. These investigations isolate the neurobiological mechanisms that govern human decision thresholds.

The ventromedial prefrontal cortex (vmPFC) and the orbitofrontal cortex (OFC) serve as the brain’s central valuation engine, encoding the subjective value of options on a common neural currency scale. In satisficing tasks, when an presented option triggers an activation in the vmPFC that surpasses an internally set threshold, this signal feeds forward to the striatum and motor execution areas, prompting rapid choice selection. However, when individuals adopt a maximizing strategy—or when high-maximizing traits are triggered—neuroimaging reveals sustained, elevated metabolic activity within the dorsolateral prefrontal cortex (dlPFC). The dlPFC is responsible for working-memory maintenance, top-down cognitive control, and complex comparative evaluations. Maximizers force their dlPFC to maintain multiple active representations simultaneously, generating measurable mental exhaustion and cognitive load.

Concurrently, the anterior cingulate cortex (ACC) monitors cognitive conflict, choice ambiguity, and anticipated error states. In maximizing paradigms characterized by extensive choice sets, fMRI scans reveal hyperactivity within the dorsal ACC. This activation corresponds directly to the subjective feeling of decision difficulty and anxiety; the brain recognizes the competing, mutually exclusive merits of dozens of alternatives, generating continuous micro-conflict signals. At the neurochemical level, dopaminergic pathways within the mesolimbic and mesocortical systems track prediction errors. For a satisficer, the discovery of an option that crosses the aspiration threshold triggers a clear, phasic burst of dopamine, reinforcing contentment and signaling search completion. For a maximizer, this reward signal is continually suppressed by counterfactual error coding—the persistent awareness that a superior outcome may remain unobserved in the unexamined search space.

8.2 Cognitive Load and Dual-Process Theory

The cognitive mechanics of satisficing are intimately entwined with the constraints of working memory. Experimental psychologists regularly manipulate cognitive load by requiring participants to maintain complex information (e.g., memorizing an 8-digit sequence or monitoring visual stimuli) while performing simultaneous decision tasks. These cognitive-load experiments provide empirical validation for the dual-process framework of cognition, popularized by Daniel Kahneman as System 1 and System 2.

System 1 operates automatically, quickly, and with minimal cognitive effort, utilizing heuristic associations and threshold-based pattern recognition. System 2 is deliberative, analytical, serial, and metabolically expensive, executing the conscious, step-by-step logic required for formal optimization. Laboratory experiments demonstrate that when participants are subjected to high cognitive load or ego-depletion manipulations (prolonged mental exertion that exhausts executive resources), System 2 processing collapses. Under these conditions, individuals are neurologically incapable of sustaining maximizing behavior. They immediately transition to satisficing heuristics, relying on System 1 threshold evaluations to navigate the task environment.

Time pressure experiments yield identical behavioral shifts. When experimental protocols restrict decision time to fractions of a second, any attempt to execute global optimization guarantees cognitive failure. Under severe time constraints, the brain defaults to fast satisficing rules, evaluating options against baseline survival thresholds. Evolutionary psychologists emphasize that our ancestral environment required immediate, fast-and-frugal evaluations when confronting predation, territorial conflict, or resource scarcity. An early hominid who attempted to maximize foraging utility by calculating the global nutritional optimum across the savanna would be predated long before concluding the calculation. Satisficing evolved as a neurobiologically hardwired survival architecture, enabling effective action within finite, high-stakes temporal windows.

8.3 Affective States and Emotional Framing

Far from being an impediment to rational thought, emotion serves as an indispensable biological mechanism for setting satisficing thresholds. Antonio Damasio’s landmark work on the somatic marker hypothesis provides neurobiological proof for this interaction. Damasio studied patients with focal damage to the ventromedial prefrontal cortex—individuals whose logical, abstract reasoning capacities remained intact, but whose ability to experience and integrate emotion was eliminated. When placed in simple decision environments (such as scheduling an appointment or choosing a restaurant), these patients were completely paralyzed. Lacking the visceral, affective signals (“somatic markers”) that tell a healthy brain that an option is simply “good enough” or unacceptable, they engaged in endless, circular comparative calculations, maximizing obsessively without reaching a conclusion.

Damasio demonstrated that affective somatic states function as an innate biological satisficing mechanism. A subtle visceral feeling—a mild aversion or intuitive contentment—rapidly eliminates vast swaths of choices from the search space, setting the aspiration boundary and allowing the deliberative apparatus to focus only on viable alternatives. Emotion does not degrade rational choice; it provides the operational stopping rules that make procedural rationality computationally viable.

Laboratory manipulations of affective states further clarify how emotional framing alters aspiration thresholds:

• Acute Stress and Anxiety: Inducing acute anxiety via physiological stressors (such as the Trier Social Stress Test or cold-pressor tasks) triggers an immediate shrinkage of the search space. Stress hormones, including cortisol and adrenaline, down-regulate prefrontal comparative search and enforce aggressive satisficing, causing individuals to latch onto the first familiar or acceptable safe option.
• Valence and Search Breadth: Positive affect induces cognitive flexibility and promotes broader heuristic exploration, encouraging agents to set ambitious aspiration thresholds while maintaining a relaxed satisficing approach to termination. Negative affect, conversely, fosters analytical vigilance and rumination, driving agents into defensive maximizing patterns where every detail is scrutinized to prevent catastrophic failure.
• The Biology of Decision Fatigue: Prolonged comparative optimization exacts a measurable physiological toll. As subjects engage in continuous maximizing search across extended experimental sessions, their blood glucose levels decrease, prefrontal cortical efficiency declines, and decision fatigue sets in. The brain eventually revolts against maximizing, terminating search arbitrarily or defaulting to destructive impulsivity.

9. Gerd Gigerenzer and the Heuristics Program: The Evolution of Satisficing

9.1 Heuristics: Biases or Adaptive Tools?

While Herbert Simon originated the concept of satisficing, its modern empirical and mathematical evolution was catalyzed by Gerd Gigerenzer and the Center for Adaptive Behavior and Cognition at the Max Planck Institute for Human Development. Gigerenzer initiated an intellectual challenge against the dominant “Heuristics and Biases” tradition established by Daniel Kahneman and Amos Tversky. While Kahneman and Tversky’s research program focused extensively on how heuristics lead to systemic cognitive errors and irrational biases when judged against normative logical benchmarks, Gigerenzer argued that this framework fundamentally mischaracterized Simon’s original vision.

Gigerenzer asserted that heuristics are not cognitive defects, computational shortcuts, or second-best compromises. Instead, they constitute the adaptive toolbox—a specialized collection of domain-specific, evolved heuristic mechanisms that biological organisms deploy to solve complex challenges in the physical world. In the real world, the future is fundamentally uncertain (Knightian uncertainty), not merely risky. In risky environments, the complete probability distribution of events is known, and optimization is theoretically possible; in uncertain environments, the distribution is unknown and uncomputable. Under radical uncertainty, formal optimization models break down, suffering from severe overfitting to historical data.

This reality is formalized by the bias-variance dilemma in machine learning and statistical decision theory. Total prediction error is the sum of three components: squared bias, variance, and irreducible noise:

$$\text{Error} = \text{Bias}^2 + \text{Variance} + \sigma^2$$

Complex optimizing models that attempt to account for every available parameter possess low bias on historical training data, but massive variance when exposed to novel, out-of-sample data; they overfit to noise. Simpler satisficing heuristics, which deliberately ignore significant amounts of information, have higher bias but dramatically lower variance. In uncertain, noisy environments, the variance reduction achieved by heuristic models routinely outweighs the increase in bias, allowing simple heuristics to outpredict complex, parameter-heavy optimizing models. Ignorance of irrelevant data is not a cognitive limitation; it is an ecological virtue.

9.2 The Take-the-Best Heuristic and Lexicographic Choice

To demonstrate the mathematical validity of this evolutionary satisficing perspective, Gigerenzer and his colleagues formulated the Take-the-Best (TTB) heuristic. TTB is an explicit, algorithmic realization of bounded satisficing, designed to make paired comparisons between two alternatives based on multiple predictive cues. The heuristic operates through three sequential, non-compensatory rules:

1. Search Rule: Order all available cues by their ecological validity $v_i$, defined as the conditional probability that an alternative has a higher criterion value given that it possesses cue $i$:
   $$v_i = P(\text{Alternative A is better than B} mid \text{Cue } i \text{ discriminates})$$
   Search through the cues sequentially, beginning with the single most valid cue.
2. Stopping Rule: Determine whether the current cue discriminates between the two alternatives (i.e., one alternative possesses the cue and the other does not). If it does, stop search immediately. Do not compute, collect, or evaluate any further cues.
3. Decision Rule: Choose the alternative favored by this single discriminating cue. If the cue does not discriminate, advance to the next most valid cue and repeat the stopping rule. If no cue discriminates, guess randomly ($p = 0.5$).

Gigerenzer, Goldstein, and their research teams conducted extensive computer simulations and empirical tournaments pitting Take-the-Best against computationally intense, optimizing algorithms—including Multiple Linear Regression, Neural Networks, Exemplar Models, and Classification Trees—across diverse, real-world data environments (e.g., predicting city populations, high school dropout rates, homelessness demographics, and corporate financial health).

The results shocked conventional decision theory. Despite using only a fraction of the data and refusing to calculate cue weights or trade-offs, Take-the-Best consistently matched or outperformed multiple regression and complex neural networks in out-of-sample prediction accuracy. This finding was christened the less-is-more effect: conditions exist where spending less time, expending less computational effort, and ignoring vast amounts of information produces systematically more accurate real-world choices than exhaustive maximization.

9.3 Ecological Rationality: Matching Heuristic to Environment

The triumph of Take-the-Best and related heuristics led Gigerenzer to formally define ecological rationality. Ecological rationality does not ask whether a decision process conforms to an abstract system of mathematical logic or internal consistency axioms. Instead, it asks: is the heuristic structurally matched to the statistical properties of the task environment? Rationality is not an internal, context-free absolute; it is the fitness between the mind’s heuristics and the environment’s cues—precisely Simon’s two blades of the scissors.

Mathematical formalizations have identified the precise environmental structures under which satisficing heuristics outperform optimization:

• Cue Redundancy and Correlation: When environmental cues are highly correlated with one another, computing additional cues yields diminishing or negative marginal returns. A single dominant cue captures the underlying variance, rendering exhaustive data collection wasteful.
• Cue Weight Dispersion: When cue validities drop off exponentially (non-compensatory environments, where the most important cue carries more weight than all subsequent cues combined), Take-the-Best is mathematically guaranteed to equal or exceed the accuracy of any linear additive optimization model.
• Sample Size Scarcity: When an agent must make choices based on small sample sizes (high estimation error), optimizing models suffer catastrophic overfitting, whereas simple, robust satisficing heuristics remain stable.

The real-world efficacy of this paradigm was demonstrated in high-stakes field experiments. In emergency medicine, Green and Mehr (1997) implemented a “Fast-and-Frugal Tree”—a satisficing decision hierarchy containing three yes/no threshold questions—to help emergency room physicians decide whether patients suffering acute chest pain should be allocated to a coronary care unit. The simple heuristic tree significantly outperformed complex logistic regression diagnostic systems, correctly categorizing high-risk myocardial infarction patients while drastically reducing false alarms. Similar field experiments in the British judicial system revealed that judges setting bail follow fast-and-frugal satisficing heuristics that utilize only one or two cues, outperforming statistical models that struggle with the chaotic, unquantifiable variables of human behavior.

10. Algorithmic and Computational Perspectives: Satisficing in Artificial Intelligence

10.1 Satisficing Search in Computer Science and Operations Research

While behavioral scientists were validating satisficing in biological organisms, computer scientists were reaching an identical realization: complete optimization across non-trivial state spaces is mathematically impossible. In computational complexity theory, many real-world routing, scheduling, and strategic problems belong to the complexity classes NP-complete or NP-hard. For these problems, no deterministic polynomial-time algorithm exists; finding the guaranteed global optimum requires search times that grow exponentially with the number of variables, rapidly exceeding the computational capacity of any physical machine.

Consequently, artificial intelligence and modern operations research are built on Simonian satisficing principles. In pathfinding algorithms such as the A* Search Algorithm, the choice of heuristic function $h(n)$ dictates the efficiency of state-space exploration. While an admissible heuristic (one that never overestimates the true cost to reach the goal) guarantees finding the optimal path, it can be computationally slow. In time-critical environments (such as robotics, autonomous vehicular navigation, and real-time game engines), computer scientists deploy inadmissible heuristics or weighted A* search. These algorithms intentionally abandon the requirement of optimality in favor of finding an $epsilon$-satisficing path within milliseconds, trading a marginal deviation from the shortest route for a massive leap in computational efficiency.

Similarly, global optimization algorithms designed to navigate chaotic, multi-modal search spaces rely on stochastic satisficing mechanisms. In Simulated Annealing, the algorithm traverses an objective function landscape searching for a global minimum. To avoid becoming trapped in sub-optimal local minima, the algorithm occasionally accepts worse outcomes based on a probabilistic temperature parameter $T$. As the system cools, it adopts a satisficing threshold, settling into a high-quality local optimum that is functionally sufficient for the task. In operations research, supply chain logisticians solving massive Traveling Salesperson Problems rely on heuristic approximation algorithms (such as the Christofides algorithm or genetic algorithms) that guarantee solutions within a narrow, acceptable percentage threshold of the theoretical optimum, prioritizing operational viability over computational perfection.

10.2 Reinforcement Learning and Exploration-Exploitation Dilemmas

In modern reinforcement learning (RL), autonomous agents face the fundamental exploration-exploitation dilemma. When an agent interacts with a dynamic environment to maximize cumulative reward, it must continually decide whether to exploit its currently known highest-yielding action or explore unvisited states to discover potentially superior actions. This dilemma is modeled mathematically through the Multi-Armed Bandit problem.

Classical optimizing solutions, such as the Gittins Index, provide a mathematically optimal strategy for allocating trials under specific discounted conditions. However, computing the Gittins Index is computationally prohibitive for high-dimensional or non-stationary environments. Autonomous systems instead rely on heuristic exploration policies such as $epsilon$-greedy, Upper Confidence Bound (UCB), or Thompson Sampling. These approaches use satisficing reward thresholds rather than computing infinite-horizon expected returns:

$$a_t = \begin{\cases} arg\max_a Q_t(a) & \text{with probability } 1 – \epsilon \ \text{a random action} & \text{with probability } \epsilon \end{\cases}$$

Recent advances in deep reinforcement learning have integrated explicit satisficing criteria into policy optimization. Rather than training agents to blindly maximize an unconstrained scalar reward function—which frequently leads to pathological edge-case exploitation and catastrophic policy collapse—researchers implement satisficing reward thresholds with entropy regularization. The agent is incentivized to discover policies that ensure performance meets or exceeds a safety and operational threshold $\alpha$ across all stochastic rollout simulations. This satisfies the operational demands of industrial robotics and autonomous flight, where a vehicle must guarantee a robust, safe flight profile across varying atmospheric conditions rather than pursuing a dangerous flight path that is optimal only under a single, highly specific set of environmental parameters.

10.3 Artificial General Intelligence (AGI) and Resource-Bounded Agents

The pursuit of Artificial General Intelligence (AGI) has cemented Herbert Simon’s bounded rationality as a foundational physical law of intelligence. Early theoretical models of universal artificial intelligence, such as Marcus Hutter’s AIXI, assumed an agent capable of executing unbounded computation, considering all computable probability distributions across environmental states. While mathematically profound, AIXI is uncomputable in the real physical universe. As computational physicists point out, information processing is bounded by physical law—including Landauer’s Principle, which dictates the minimum thermodynamic energy required to erase a single bit of information, and the finite limits on processing speed imposed by the speed of light and quantum mechanics.

True artificial intelligence operating within the physical universe must be a resource-bounded agent. This reality has given rise to the field of computational metareasoning—the algorithmic process of reasoning about reasoning. Metareasoning agents treat their own internal cognitive cycles as a scarce, costly resource. The agent does not simply attempt to solve an external task optimally; it solves a dual-level problem: allocating computational time between deliberate search and immediate physical execution. Metareasoning algorithms formulate stopping rules that terminate internal deliberation the instant the expected marginal utility of further computation drops below its operational cost, echoing Simon’s satisficing logic.

This paradigm is reflected in the cognitive architectures of modern artificial intelligence. Foundational cognitive systems such as SOAR (developed by John Laird, Allen Newell, and Paul Rosenbloom) and ACT-R (developed by John R. Anderson) structure their problem-solving around production rules, sub-goal generation, and heuristic threshold terminations derived directly from Simon’s early protocol analysis. Similarly, in modern Large Language Models (LLMs), inference techniques such as chain-of-thought prompting, tree-of-thought search, and speculative decoding deploy heuristic pruning strategies to traverse billions of token probabilities, proving that scalable intelligence—biological or silicon—relies fundamentally on satisficing search architectures to function within a complex world.

11. Critiques, Counter-Arguments, and Theoretical Debates

11.1 Can Satisficing Be Reduced to Maximization Subject to Constraints?

The most persistent theoretical critique leveled against Herbert Simon by neoclassical economists—notably champions of the Chicago School such as Gary Becker and George Stigler—is the argument of reductionism. These critics asserted that Simon’s concept of satisficing does not constitute a genuine paradigm shift. Instead, they claimed that satisficing is merely a mathematically trivial variation of classical utility maximization: specifically, maximization subject to information and search costs.

Stigler’s 1961 paper, The Economics of Information, formalized this argument. Stigler posited that an agent continues searching for lower prices or better options until the marginal expected benefit of an additional search unit equals the marginal cost of acquiring that information:

$$\mathbb{E}\left[ \frac{\partial B}{\partial s} \right] = \frac{\partial C}{\partial s}$$

From this neoclassical perspective, when a consumer stops searching and buys a product that is “good enough,” they are not abandoning optimization; they are calculating the optimal stopping point where further search would cost more in time and energy than it would yield in product quality. Consequently, neoclassical theorists argued that satisficing could be subsumed into standard microeconomic models as ordinary constrained optimization.

Herbert Simon explicitly rejected this reductionist critique, demonstrating that it rests on a profound logical flaw: the problem of infinite regress. To calculate the mathematically optimal point to terminate search under costly information, an agent must compute the expected marginal benefit of an unknown piece of information that they have not yet encountered. To calculate the marginal cost and benefit of searching for that information, the agent must expend cognitive resources to optimize the search-optimization calculation. This necessitates an infinite chain of meta-calculations:

$$\text{Meta-Optimization Level 1} to \text{Meta-Optimization Level 2} to dots to \text{Level } \infty$$

Simon demonstrated that attempting to optimize the search process itself demands vastly more computational capacity than the original choice problem. Humans do not optimize their search costs; they use procedural heuristics and aspiration thresholds to cut the Gordian knot of infinite regress. There is a fundamental substantive difference between formal mathematical equivalence modeled post-hoc by an economist and the real-time cognitive process executed by an agent’s brain.

11.2 Methodological Limitations in Satisficing Empirical Research

Despite its profound contributions, the empirical literature surrounding bounded rationality and satisficing faces methodological challenges. A significant operational limitation lies in the difficulty of directly measuring subjective aspiration levels in real-time environments without disrupting the decision process itself. In many laboratory protocols, asking a participant to state their aspiration threshold before each choice primes the subject, artificially altering their natural search behavior and inducing an analytical mindset that would not otherwise occur in a natural setting.

Additionally, experimental laboratory tasks are susceptible to demand characteristics and artificial constraints. In many experimental economics designs, participants navigate synthesized, abstract matrices with clearly demarcated rows and columns—environments that lack the ambient sensory noise, social pressures, emotional stakes, and unquantifiable ambiguities of real-world choice. An individual who satisfices aggressively in their personal or professional life may exhibit maximizing tendencies in a sterile laboratory task simply because the experimental frame makes the total choice set artificially visible and accessible.

Furthermore, researchers face significant hurdles in establishing the objective “universal optimal outcome” against which to empirically verify maximization success. In multi-attribute personal domains (such as career paths, interpersonal partnerships, or creative endeavors), the “optimal” choice is inextricably tied to dynamic, subjective internal preferences that evolve over time. Finally, the heavy reliance on cross-sectional self-report scales (such as the original Schwartz MS or the MTS) in behavioral operationalization studies introduces vulnerabilities to common method variance, retrospective rationalization, and social desirability biases. Longitudinal tracking and real-world behavioral tracing are continually required to validate survey-derived psychometric constructs.

11.3 The Dark Side of Satisficing: When Suboptimal Becomes Maladaptive

While the behavioral literature frequently champions satisficing as a protective buffer against choice paralysis and regret, satisficing is not universally benign. Under specific environmental conditions, threshold-based heuristic stopping can degenerate into severe, maladaptive outcomes. The most prominent hazard is premature stopping in high-stakes, asymmetric-risk domains. In safety-critical fields such as aerospace engineering, nuclear facility operations, and oncology diagnosis, satisficing can have catastrophic consequences:

A radiologist who scans an MRI, detects a single benign anomaly that satisfies their initial diagnostic threshold, and terminates their visual search early may miss an adjacent, lethal malignant tumor. In engineering systems, settling for a component design that is merely “good enough” without exhaustively investigating stress tolerances can cause systemic structural failures, as witnessed in the Challenger and Columbia space shuttle disasters. In domains where the cost of a false negative is existential, rigorous, systematic maximization is functionally and ethically non-negotiable.

Furthermore, satisficing heuristics are deeply vulnerable to status quo bias and systemic aspiration depression:

• Traps of Sufficiency: Individuals frequently satisfice within abusive relationships, toxic corporate work cultures, or unfulfilling careers. Because their immediate conditions hover just above their baseline tolerance threshold, they do not expend the search effort required to locate vastly superior environments, settling for prolonged mediocrity.
• Systemic Aspiration Depression: Sociological and behavioral economic research indicates that individuals raised in environments of systemic socioeconomic disadvantage frequently depress their internal aspiration thresholds ($boldsymbol{\alpha}$) in response to chronic structural deprivation. By lowering educational, professional, and health aspirations to match their limited immediate horizons, agents achieve a state of psychological contentment that leaves them trapped in cycles of generational poverty. When aspiration levels adapt downwards too efficiently, satisficing ceases to be an adaptive cognitive strategy and becomes a psychological anchor that reinforces systemic inequality.

12. Conclusion and the Future of Rationality Research

12.1 Synthesis of Empirical Findings Across Disciplines

Seventy years of theoretical modeling, empirical testing, and cross-disciplinary investigation have fundamentally altered our understanding of human rationality. The cumulative body of evidence confirms the central premise Herbert Simon advanced in the 1940s and 1950s: human beings are boundedly rational agents whose choices are shaped by the interaction between internal neurocognitive constraints and external environmental structures. The neoclassical fiction of Homo economicus, endowed with infinite computational capacity and pursuing global utility maximization across unconstrained search spaces, has been dismantled as a descriptive model of human action.

The behavioral operationalization of this paradigm has generated a clear empirical finding across domains:

• The Maximizer’s Divergence: Maximizers—individuals driven by the mandate to identify and secure the absolute best alternative—frequently achieve marginally superior objective outcomes (such as higher starting salaries or mathematically refined products). However, they pay a substantial psychological penalty. Their exhaustive search patterns, susceptibility to choice overload, relentless upward counterfactual thinking, and chronic social comparison orientation systematically elevate rates of regret, depression, and decision paralysis.
• The Satisficer’s Advantage: Satisficers—individuals who evaluate options sequentially against dynamic aspiration thresholds and terminate search upon encountering sufficiency—enjoy higher subjective well-being, lower post-decisional regret, and greater emotional resilience. Their heuristic stopping rules protect finite cognitive and temporal resources, short-circuiting the hedonic treadmill.

Crucially, Gerd Gigerenzer’s heuristics program and contemporary computational neuroscience have demonstrated that satisficing is not an embarrassing cognitive shortcut or an indicator of biological deficiency. By validating the concept of ecological rationality, this research confirms that simple heuristics, fast-and-frugal trees, and threshold models routinely match or outperform complex optimizing algorithms in uncertain, noisy, and non-ergodic environments. The mind’s heuristic toolbox achieves high predictive accuracy through variance reduction, proving that the intentional disregard of information is often an evolutionary and computational asset. Rationality is not an absolute mathematical ideal; it is the fitness between the blade of the cognitive apparatus and the blade of the task environment.

12.2 Practical Implications for Modern Decision Architecture

The empirical verification of the maximizing versus satisficing dichotomy carries profound, actionable implications for the design of institutional, corporate, and technological systems. In an era marked by the exponential proliferation of digital choices, algorithmic recommendation engines, and pervasive informational noise, deliberate attention to choice architecture has become essential for preserving human welfare and organizational efficacy.

These insights translate directly into practical systemic interventions:

• Public Policy and Choice Architecture: As demonstrated by behavioral economists Richard Thaler and Cass Sunstein, expanding public option menus—such as presenting citizens with hundreds of competing, complex retirement savings plans or healthcare policies—routinely induces cognitive overload, driving individuals into total inaction. Effective public policy requires choice architects to curate menus, implement robust, beneficial default options, and design transparent, threshold-based paths that enable citizens to satisfice safely without navigating complex computational spaces.
• Personal Well-Being and Digital Hygiene: At the individual level, cultivating deliberate satisficing habits serves as an essential psychological defense against modern burnout. By setting explicit, unyielding stopping rules for consumer purchases, information consumption, and career transitions, individuals can actively insulate their working memory from choice paralysis. Deliberately refusing to optimize every minor micro-decision preserves metabolic energy and emotional reserves for the high-stakes decisions that truly warrant deep, deliberate cognitive focus.
• Organizational and Institutional Design: Corporate enterprises must institutionalize decision frameworks that prevent maximizing paralysis and hubris. By embedding structural buffers (organizational slack), relying on explicit capital allocation hurdle rates, establishing fast-and-frugal operational protocols, and incentivizing procedural integrity over post-hoc outcome bias, organizations build systemic resilience. Firms designed around satisficing principles adapt dynamically to environmental volatility, whereas over-optimized, frictionless enterprises collapse under unforeseen structural shocks.

12.3 Emerging Frontiers in Rationality Experiments

The empirical investigation of bounded rationality and satisficing is advancing into new methodological and technological frontiers, fueled by breakthroughs in artificial intelligence, digital telemetry, and mobile neuroimaging. Researchers are no longer confined to static survey instruments or artificial laboratory environments; they can now track and analyze decision processes in real time across dynamic, high-dimensional spaces.

Among the most promising frontiers are:

• Human-AI Hybrid Decision Systems: As autonomous generative agents and LLM-powered analytical tools integrate into medical, legal, and financial workflows, the procedural division of labor between human and machine is shifting. Emerging experimental paradigms explore architectures where the AI conducts broad, high-speed computational search (maximizing across millions of unstructured data points), while the human expert applies contextual, value-driven aspiration thresholds (satisficing) to select and authorize final actions. Mapping the friction points and cognitive handoffs in these human-AI systems is critical to preventing algorithmic complacency and catastrophic system failures.
• In-Vivo Neurotechnological Tracking: Advances in portable, high-density functional near-infrared spectroscopy (fNIRS), mobile EEG headsets, and real-time pupillometry allow cognitive neuroscientists to monitor the biological markers of cognitive load, choice conflict, and stopping-rule execution in naturalistic settings. Researchers can now observe the precise millisecond a consumer’s or executive’s brain shifts from exploratory evaluation to threshold-induced choice selection, deepening our understanding of how stress, fatigue, and environmental architecture alter neural thresholds.
• Big Data Field Experiments in Digital Ecosystems: The analysis of multi-million-node digital clickstreams across e-commerce platforms, streaming algorithms, and algorithmic labor markets (such as rideshare and freelance platforms) provides unprecedented empirical arenas for testing sequential search models at population scale. These massive datasets allow researchers to measure the real-time adaptation of human aspiration levels in response to platform design interventions, pricing shifts, and dynamic choice assortments.

Seven decades after Herbert Simon challenged the neoclassical establishment, his work continues to illuminate the study of mind, behavior, and machine intelligence. Simon taught us that human rationality is neither omniscient nor broken; it is a finely tuned, adaptive instrument shaped for survival in a complex, uncertain world. In a twenty-first-century landscape characterized by infinite informational options and finite human attention, the satisficing principle remains an enduring guide: true rationality lies not in calculating the unattainable optimum, but in possessing the wisdom to recognize when an outcome is truly good enough.

References

• Anderson, J. R. (1993). Rules of the mind. Lawrence Erlbaum Associates. https://doi.org/10.4324/9780203772713
• Chase, W. G., & Simon, H. A. (1973). Perception in chess. Cognitive Psychology, 4(1), 55–81. https://doi.org/10.1016/0010-0285(73)90004-2
• Cyert, R. M., & March, J. G. (1963). A behavioral theory of the firm. Prentice-Hall. https://www.wiley.com/en-us/A+Behavioral+Theory+of+the+Firm%2C+2nd+Edition-p-9780631180708
• Damasio, A. R. (1994). Descartes’ error: Emotion, reason, and the human brain. G.P. Putnam’s Sons. https://www.penguinrandomhouse.com/books/33139/descartes-error-by-antonio-damasio/
• Diab, D. L., Gillespie, M. A., & Highhouse, S. (2008). Are maximizers really unhappy? The measurement of maximizing tendency. Judgment and Decision Making, 3(5), 364–370. https://doi.org/10.1017/S1930297500000843
• Friedman, M. (1953). The methodology of positive economics. In Essays in Positive Economics (pp. 3–43). University of Chicago Press. https://press.uchicago.edu/ucp/books/book/chicago/E/bo24590458.html
• Gigerenzer, G., & Goldstein, D. G. (1996). Reasoning the fast and frugal way: Models of bounded rationality. Psychological Review, 103(4), 650–669. https://doi.org/10.1037/0033-295X.103.4.650
• Gigerenzer, G., Todd, P. M., & the ABC Research Group. (1999). Simple heuristics that make us smart. Oxford University Press. https://doi.org/10.1093/oso/9780195121568.001.0001
• Green, L., & Mehr, D. R. (1997). What alters physicians’ decisions to admit to the coronary care unit? The Journal of Family Practice, 45(3), 219–226. https://pubmed.ncbi.nlm.nih.gov/9300001/
• Hart, P. E., Nilsson, N. J., & Raphael, B. (1968). A formal basis for the heuristic determination of minimum cost paths. IEEE Transactions on Systems Science and Cybernetics, 4(2), 100–107. https://doi.org/10.1109/TSSC.1968.300136
• Hutter, M. (2005). Universal artificial intelligence: Sequential decisions based on algorithmic probability. Springer. https://doi.org/10.1007/b13823
• Iyengar, S. S., & Lepper, M. R. (2000). When choice is demotivating: Can one desire too much of a good thing? Journal of Personality and Social Psychology, 79(6), 995–1006. https://doi.org/10.1037/0022-3514.79.6.995
• Iyengar, S. S., Wells, R. E., & Schwartz, B. (2006). Doing better but feeling worse: Looking for the ‘best’ job undermines satisfaction. Psychological Science, 17(2), 143–150. https://doi.org/10.1111/j.1467-9280.2006.01677.x
• Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux. https://us.macmillan.com/books/9780374533557/thinkingfastandslow
• Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263–291. https://doi.org/10.2307/1914185
• Laird, J. E. (2012). The Soar cognitive architecture. MIT Press. https://doi.org/10.7551/mitpress/9780262122962.001.0001
• March, J. G. (1991). Exploration and exploitation in organizational learning. Organization Science, 2(1), 71–87. https://doi.org/10.1287/orsc.2.1.71
• Nenkov, G. Y., Morwitz, V. G., Schwartz, B., & Ward, A. (2008). A short form of the Maximization Scale: Factor structure, reliability and validity studies. Judgment and Decision Making, 3(5), 371–388. https://doi.org/10.1017/S1930297500000855
• Newell, A., & Simon, H. A. (1956). The logic theory machine: A complex information processing system. IRE Transactions on Information Theory, 2(3), 61–79. https://doi.org/10.1109/TIT.1956.1056797
• Newell, A., & Simon, H. A. (1972). Human problem solving. Prentice-Hall. https://psycnet.apa.org/record/1973-09419-000
• Payne, J. W., Bettman, J. R., & Johnson, E. J. (1993). The adaptive decision maker. Cambridge University Press. https://doi.org/10.1017/CBO9781139173933
• Schwartz, B. (2004). The paradox of choice: Why more is less. Ecco/HarperCollins. https://www.harpercollins.com/products/the-paradox-of-choice-barry-schwartz
• Schwartz, B., Ward, A., Monterosso, J., Lyubomirsky, S., White, K., & Lehman, D. R. (2002). Maximizing versus satisficing: Happiness is a matter of choice. Journal of Personality and Social Psychology, 83(5), 1178–1197. https://doi.org/10.1037/0022-3514.83.5.1178
• Simon, H. A. (1947). Administrative behavior: A study of decision-making processes in administrative organization. Macmillan. https://www.simonandschuster.com/books/Administrative-Behavior-4th-Edition/Herbert-A-Simon/9780684835822
• Simon, H. A. (1955). A behavioral model of rational choice. The Quarterly Journal of Economics, 69(1), 99–118. https://doi.org/10.2307/1884852
• Simon, H. A. (1956). Rational choice and the structure of the environment. Psychological Review, 63(2), 129–138. https://doi.org/10.1037/h0042769
• Simon, H. A. (1976). From substantive to procedural rationality. In S. J. Latsis (Ed.), Method and Appraisal in Economics (pp. 129–148). Cambridge University Press. https://doi.org/10.1017/CBO9780511760082.008
• Simon, H. A. (1978). Rational decision-making in business organizations. Nobel Memorial Lecture. Nobel Foundation. https://www.nobelprize.org/prizes/economic-sciences/1978/simon/lecture/
• Sterman, J. D. (1989). Modeling managerial behavior: Misperceptions of feedback in a dynamic decision making experiment. Management Science, 35(3), 321–339. https://doi.org/10.1287/mnsc.35.3.321
• Stigler, G. J. (1961). The economics of information. Journal of Political Economy, 69(3), 213–225. https://doi.org/10.1086/258464
• Thaler, R. H., & Sunstein, C. R. (2008). Nudge: Improving decisions about health, wealth, and happiness. Yale University Press. https://yalebooks.yale.edu/book/9780300122237/nudge/
• Turner, B. M., Rim, H. B., Betz, N. E., & Schmitt, M. J. (2012). The maximization inventory. Judgment and Decision Making, 7(1), 48–60. https://doi.org/10.1017/S193029750000185X
• Tversky, A. (1972). Elimination by aspects: A theory of choice. Psychological Review, 79(4), 281–299. https://doi.org/10.1037/h0032955
• Von Neumann, J., & Morgenstern, O. (1944). Theory of games and economic behavior. Princeton University Press. https://press.princeton.edu/books/paperback/9780691130613/theory-of-games-and-economic-behavior