Infographic for "Lessons from Herbert A. Simon".

Core Concepts: Bounded Rationality, Satisficing, and Decision-Making

Herbert A. Simon challenged models that describe economic actors as perfectly informed maximizers. His work asked how people and organizations make decisions when search, attention and computational capacity are limited. — Simon — Rationality as Process and Product.

  • Bounded Rationality: A decision maker cannot survey every alternative or compute every consequence. Simon studied rational choice within limits of information, search and mental computation. — Simon — Rationality as Process and Product.
  • Satisficing: Rather than search indefinitely for an optimum, a decision maker can stop when an alternative exceeds an aspiration level. Simon used this as a model of choice under limited search. — Simon — Rationality as Process and Product.
  • Use Heuristics to Search Selectively: Simon treated heuristics and approximation procedures as ways to search huge problem spaces selectively when exhaustive calculation is infeasible; their value depends on how well they help solve the task. — Simon — Rationality as Process and Product.

Key Learnings and Quotes

On Bounded Rationality and Decision-Making

  1. Look Beyond the Thinker to the Environment: Simon argued that much apparent complexity in human behavior may come from the complexity of the environment and the search for workable designs, rather than from an equally complicated inner mechanism. He framed this as a qualified explanatory proposal, not a proved universal law. — Simon — The Science of Design.
  2. "What information consumes is rather obvious: it consumes the attention of its recipients. Hence a wealth of information creates a poverty of attention." — Simon — Designing Organizations for an Information-Rich World.
  3. Understand Skilled Intuition as Recognition: Simon explained expert intuition as rapid recognition: cues in a situation retrieve relevant information from a well-trained memory, even when the expert cannot describe every intermediate step. — Simon — What Is an Explanation of Behavior?.
  4. Decision-Making Runs Through Administration: Simon argued that administration is not merely carrying out a fixed policy. Decisions about what to do and how to act continue throughout an organization, including at its operative level. — Administrative Behavior — Chapter 1.
  5. Question the Maximizer Assumption: Simon contrasted the economic model of an actor who always seeks the best option with a satisficing model in which search can stop once an alternative clears an aspiration threshold. — Simon — Rationality as Process and Product.
  6. Judge Both Outcomes and Procedures: Simon distinguished substantive rationality—choosing an action appropriate to a goal—from procedural rationality—the effectiveness of the process used to choose under human cognitive limits. — Simon — Rationality as Process and Product.
  7. Make Assumptions About Human Nature Explicit: Simon argued that a research program changes substantially depending on whether it models people as nearly omniscient optimizers or as boundedly rational decision makers. The choice shapes both what researchers ask and how they investigate it. — Simon — Human Nature in Politics.
  8. Ground Models in Actual Goals and Beliefs: Simon warned that rationality principles have little power to predict political behavior without empirical knowledge of actors' goals, beliefs and values. A useful model cannot simply assume those contents are already known. — Simon — Human Nature in Politics.
  9. Distinguish Intuition from Stress Responses: Simon separated skilled intuition, built through learning and experience, from stress-driven emotional responses that can lead managers to postpone decisions or choose short-term comfort over effective action. — Simon — Making Management Decisions.
  10. Treat Problem Solving as Guided Search: Simon and his coauthors described human problem solving as searching for possibilities, generating partial solutions, and evaluating cues that guide what to try next. This is a process model, not the unauthenticated maze quotation in the old profile. — Newell, Shaw and Simon — Elements of a Theory of Human Problem Solving.
  11. Look for Sparse Structure in Complex Systems: Simon argued that many complex systems are easier to understand than a count of all possible interactions suggests: most potential connections are absent or negligible, and the remaining structure can often be analyzed hierarchically. — Simon — How Complex Are Complex Systems?.

On Organizations and Administration

  1. Treat Decisions as Interdependent: In organizations, one person's output becomes another's input. Simon described decision-making as an organized network of relations supported by both formal and informal communication. — Administrative Behavior — Chapter 1.
  2. Combine Specialized Knowledge: Simon argued that decisions can draw on facts and goals supplied by people with different expertise; organizing determines where those contributions come from and how they are assembled. — Administrative Behavior — Chapter 1.
  3. Define Authority Through Accepted Decisions: Following a definition he said was substantially equivalent to Chester Barnard's, Simon described authority as a subordinate allowing a superior's decision to guide action without independently reassessing its merits. — Administrative Behavior — Chapter 1.
  4. Respect the Limits of Authority: A superior's decision must reach the people expected to act on it, and authority has limits: Simon described a subordinate's zone of acceptance beyond which disobedience follows. — Administrative Behavior — Chapter 1.
  5. Design the Premises of Decisions: For Simon, organizing meant finding where relevant knowledge resides and assigning responsibility for the goals and constraints that shape decisions. Facts and values become premises from which choices are assembled. — Administrative Behavior — Chapter 1.
  6. Use Loyalty Without Losing Perspective: Simon argued that organizational identification helps people evaluate choices against a manageable set of goals, but identification with a department can also distort decisions when its interests conflict with the wider organization. — Administrative Behavior — Chapter 1.
  7. Coordinate Specialized Work: Dividing work allows specialized skills to develop, but the resulting decisions must still fit together. Simon saw vertical specialization as one way to coordinate people carrying out horizontally specialized tasks. — Administrative Behavior — Chapter 1.
  8. Counter Departmental Budget Bias: Simon warned that department heads may be too identified with their own units to weigh competing budget needs fairly; a central budget agency can help counter that narrow bias. — Administrative Behavior — Chapter 1.
  9. Use Hierarchy as One Coordination Tool: Simon treated formal hierarchy and authority as ways to coordinate decisions, but not the only ones: advice, information, training and organizational loyalties also influence action. — Administrative Behavior — Chapter 1.
  10. Set the Agenda to Focus Attention: Simon argued that a manager can shape an organization's problem-solving habits by repeatedly directing attention to defining problems, finding solutions and learning from mistakes. — Simon — Making Management Decisions.

On Artificial Intelligence and Cognitive Science

  1. Broaden the Tasks Computers Can Address: In 1957, Simon and Newell forecast that heuristic programs would extend computers beyond well-structured calculations into increasingly varied ill-structured problems. Their published examples were still limited, so this was a research forecast rather than a demonstrated ability to perform every human job. — Simon and Newell — Heuristic Problem Solving.
  2. An Overoptimistic Chess Prediction: In a paper Simon presented with Allen Newell, they predicted—based on the state of AI in 1957—that a digital computer would become the world's chess champion within ten years, unless barred from competition. The deadline was a prediction, not a result. — Simon and Newell — Heuristic Problem Solving.
  3. Use Computer Models to Study Thinking: Simon distinguished using AI to make computers perform useful tasks from using computer simulations to understand human learning and thought. A successful task program need not mimic the way a person learned the task. — Simon — Why Should Machines Learn?.
  4. Treat Symbolic Intelligence as a Testable Hypothesis: Simon proposed that a physical symbol system could supply the capabilities needed for intelligent behavior. He presented this as an empirical hypothesis to be tested against human and computer problem solving, not as a settled theory of every aspect of mind. — Simon — Physical Symbol System Lecture.
  5. Study Design as a Human Activity: Simon argued that design—the search for courses of action that change existing situations into preferred ones—belongs at the center of professional education and is, in large part, a way to study human thought and action. — Simon — The Science of Design.
  6. Make Computers Decision Aids: Simon expected interactive expert systems to advise managers, with knowledge and intelligence shared between the human decision maker and the automated component. — Simon — Making Management Decisions.
  7. Build a Large Store of Recognizable Patterns: Simon's account of expertise emphasized many learned patterns in long-term memory. Those patterns help experts recognize a situation and retrieve relevant knowledge quickly; the source does not establish a universal 10,000-hour threshold. — Simon — Making Management Decisions.
  8. Use Chess as a Testbed for Intelligence: Simon and Jonathan Schaeffer described chess as a standard testbed for studying both computer search and human expert thinking. They noted that it had sometimes been called the “Drosophila” of AI and cognitive science, rather than claiming the phrase as Simon's invention. — Simon and Schaeffer — The Game of Chess.
  9. Test Cognitive Models as Programs: Simon used computer programs to express candidate models of human problem-solving strategies. The program's behavior, and the conditions under which people use a strategy, could then be compared with observations. — Simon — What Is an Explanation of Behavior?.

On Science, Learning, and Knowledge

  1. "Learning is any change in a system that allows it to perform better the second time on repetition of the same task or on another task drawn from the same population." — Simon — Why Should Machines Learn?.
  2. Build Models to Test Discovery Processes: Simon proposed making a scientific-discovery process precise enough to implement as a computer program, then testing whether it can reproduce discoveries from the information available to historical scientists. — Simon — Studying Scientific Discovery.
  3. Separate Description from Explanation: Simon treated scientific discovery as involving several activities: collecting data, finding compact descriptions of it, and formulating and testing explanatory theories. One activity should not be mistaken for the whole of science. — Simon — Studying Scientific Discovery.
  4. Test Patterns Beyond the Data That Revealed Them: Simon warned that extracting a pattern from a finite data set does not by itself establish that the pattern will predict new observations. Discovery and testing require different questions. — Simon — Does Scientific Discovery Have a Logic?.

On Society and Economics

  1. Know When Price Signals Lose Clarity: Simon argued that prices simplify coordination when they are known or reasonably predictable. Under rapid change or uncertainty, organizational procedures may help people coordinate where price signals alone are less useful. — Simon — Organizations and Markets.
  2. Use More Than Price for Coordination: Simon argued that prices are only one way to coordinate economic activity. Standards, authority and adjustments in quantities can also help firms and their partners align their actions. — Simon — Organizations and Markets.
  3. Put Organizations Into Economic Theory: Simon argued that modern economies cannot be understood as transactions among isolated market actors alone. Much economic decision-making happens within firms, where authority, identification and coordination matter. — Simon — Organizations and Markets.
  4. Account for Group-Regarding Motives: Simon observed that people in a cooperation experiment sometimes stayed with a group even when a narrowly self-interested model predicted exit. He treated group-regarding motives as part of explaining that behavior, not as proof that every interaction benefits everyone. — Simon — Human Nature in Politics.

Selected original works and publisher editions for further reading: