p.6 Much of the motivation for this book arises from relatively recent developments in game theory, economics, and political theory. These developments... permit the formalization and modeling of principles that previously had been too complex or considered too indeterminate - and were thus relegated to the sidelines of analysis until recently.
p.6-7 Strategic interaction occurs whenever agents share strategic interdependence - meaning that their actions affect outcomes for others and that agents typically understand such interdependence... By contrast, the constrained-maximization approach of traditional economics and rational choice theory often fails to address strategic interaction... In pure competition... Others' actions merely become indistinguishable elements of an environment (a field) that constrains individual activity and affects outcomes.
p.7 Strategic interaction is more complicated and intricate than constrained maximization because other parties matter... Such strategic interaction constitutes a game, not just an individual decision... even in competitive markets, game theory can represent transactions that involve any type of commitment... Conditions of this sort are ubiquitous in complex economies... economic choices remain profoundly strategic because outcomes depend on numerous interactions among participants.
p.8 In short, game theory offers an extraordinarily flexible and widely applicable set of modeling techniques. Its use informs... business strategizing, and policymaking... Game-theoretic modeling permeates this book.
p.12 In a non-ergodic environment - one that changes in ways that defy statistical regularity - agents do not know the underlying probability distributions for important social phenomena (North 2005). Accordingly, outside the confines of a narrow problem-complexity boundary - within which problems are simple - agents cannot maximize (Arthur 1992, 1994). They face a gap between their ability to understand and the complexity of problems they face (Heiner 1983). When confronted with such uncertainty, agents resort to heuristics [JLJ - yes, and to meta-heuristics - the problem turns to one of understanding cause and effect and to "what do I do now?". The agent must explore and look for clues in a scheme that positions the agent to adapt in an uncertain world where cause and effect are often not what they seem, and where one must doubt what one finds.]
p.12 Mental models are conceptual representations that agents use for identifying and interpreting categories, patterns, and cause-and-effect relationships. Agents use mental models to produce conclusions or judgements concerning the consequences of various activities or interactions.
p.12 Evolutionary game theory can specify how boundedly rational agents select and discard strategies, ideas, traits, or even social norms, by adjusting individual and shared mental models. Epistemic game theory can represent agents' cognitive limitations and shared understandings (Gintis 2009b).
p.13 For Williamson, the simultaneous presence of bounded rationality, incomplete information, and self-interest generates "strategic opportunism." ...Williamson's core assertion is that a firm's choice between contracts and various forms of hierarchical governance reflects condition-specific and adaptive efforts to minimize transaction costs associated with holdup and related problems.
p.16 Evolutionary game theory can model adaptive learning processes of boundedly rational agents; their tendencies to acquire or modify heuristics, mental models, and behavioral dispositions; and the larger-scale evolution of conventions, norms, formal institutions, policies, and development (or lack thereof). Epistemic game theory can model the individual and shared understandings that frame and motivate strategic behavior.
p.65 Webster's dictionary defines power as the "capability or ability to direct or influence the behavior of others or the course of events." ...Economist Kaushik Basu (2000) defines power... "All agree that power is, broadly speaking, the ability of one person to get another person to do something that is of advantage to the former but not in the latter's interest" (134).
p.65-66 A related conception of power appears in Bowles and Gintis (2008). They identify four characteristics: (i) power involves relationships between or among people (and so is not a property of an individual); (ii) power involves the use of threat of sanctions; (iii) power "should be normatively indeterminate, allowing for Pareto-improving outcomes... but also susceptible to abuse in ways that harm others in violation of ethical principles"; and (iv) to be enduring, power must be "sustainable as a Nash equilibrium of an appropriately defined game" (Vol. 6, 566).
p.66 we define power as
the ability of an individual or group to deliberately employ sanctions
or manipulation in order to induce other individuals or groups to take, alter, or avoid specific actions in a manner that the former... believes is in its own interest and that the latter would not otherwise pursue.
p.66-67 Power relationships permeate political economy... Why, then, is power so rarely discussed in traditional economics? Basu provides a simple historical answer. Until recently, economists simply lacked the conceptual tools to formally address power: "given the importance attached to formalization in economics, the only effective option has been to ignore them [power and coercion]" (2000,132). Fortunately, game theory with information economics now offers formal methodologies for analyzing power.
p.69 Robert Dahl (1957) identifies three fundamental elements of power relations: sources, instruments, and domains of power. The base or source of power indicates where an agent's power originates. Fundamentally, power arises from three related sources: position (social, political, or economic), access to resources, and an ability to resolve CAPs associated with mobilizing support.
p.69 The means or instruments of power indicate the manner through which one party can exercise power over another. There are two basic types of power instruments:
- actual, threatened, or promised sanctions; recall that we use the general term sanction to include both punishment and reward (respectively, negative and positive sanctions); and
- the use of communication as a means of manipulation independent of sanctions.
p.69 Dahl's third element of power, the domain of influence... Roughly speaking, power has three fundamental domains of influence (faces): (i) behavior, (ii) conditions of engagement or rules of access, and (iii) players' inclinations.
p.76 With respect to collective action, Lukes (1974) argues that power1 arises from the ability of agents to mobilize support. Mobilized support... can enhance one party's ability to pressure another. [JLJ - ok, now I see where Nick Crossley formed his opinion of power as 'mobilized sanction']
p.81 Arguably, threats are the most important type of strategic move. One party may preclude or reduce another's ability to participate in various arenas by credibly threatening some form of punishment.
p.88 Analyzing power involves identifying its three basic sources (position, access to resources, and an ability to resolve organizational CAPs), its two instruments (sanction and manipulative communication), and its three domains or faces. Power1 directly affects behavior, power2 affects underlying rules and expectations, and power3 affects deeper-level preferences and beliefs. Power1 manifests itself in contests... that arise in given, mutually understood contexts or arenas.
p.125 We now turn to modeling bounded rationality. Our purpose is to develop what Ostrom calls a minimalist conception of rationality - one that uses the principle of vicarious problem solving in a manner that incorporates both potential influences of social context on understandings and the cognitive limits that accompany uncertainty in non-ergodic environments.
p.125 Kahneman (2003) asserts that people use two mental systems to process information: a reactive, habitual, impressionable, doubt-suppressing, and essentially effortless intuitive system (S1); and a slow, deliberate, reflective, and effortful reasoning system (S2). Despite its grounding in intuition, S1 employs previously stored knowledge, responds to language... and incorporates learning with repetition over time... The outcomes of S1 processes are impressions. By contrast, S2 outcomes are reasoned deductions, such as solutions to differential equations.
p.126-127 we define heuristics as mental procedures that readily combine various inputs from current and prior experience to produce impressionistic judgments. Specifically, heuristics incorporate accessible cognitive inputs, which arise from observation and/or memory... to generate intuitive (primarily S1) impressions.
p.127 Judgements mix S1 intuition and heuristics with S2 deliberation to generate conclusions. [JLJ - which might be a recommendation for further deliberation]
p.127 Mental models are "the internal representations that individual cognitive systems create to interpret the environment" (Denzau and North 1994, 4). Mental models combine S1 and S2 evaluations of prior experience into understandings of key categories, patterns, and cause-and-effect relationships.[JLJ - corrected quote from Denzau and North - Ferguson mistakenly had the word 'created' instead of the correct 'create']
p.128 Mental models not only reflect previous learning, they also inform future learning. Such learning occurs at two levels: internal hypothesis testing and reevaluative learning. Both levels embody evolutionary, adaptive processes. Internal hypothesis testing employs existing mental models to run thought experiments. People evaluate plausible hypotheses that arise from their mental models as they strive to predict outcomes of strategic interactions and other processes... Hypotheses that survive this internal (evolutionary) selection process tend to solidify into expectations or judgments... The considerably less frequent reevaluative learning adjusts mental models themselves in response to significant adverse feedback. Because mental models economize on cognitive effort and adjusting them is costly, agents have an incentive to retain their models... Mental models persist with small adjustment until a sufficient volume of salient information feedback (e.g., a dramatic surprise) demonstrates large enough inconsistencies to undermine them (Denzau and North 1994).
p.132 Boundedly rational agents learn adaptively. They follow one or more of their heuristics for some time (following temporarily fulfilled expectations). They observe outcomes and then occasionally use S2 monitoring to reevaluate... Evolutionary game-theoretic models can represent this sequential adaptive learning.
We now construct a model, based largely on Heiner (1983), that uses evolutionary logic to show why heuristics improve outcomes for boundedly rational agents facing uncertainty.
p.132 The extent of the CD gap encountered by boundedly rational agents depends on relationships between environmental complexity and agents' cognitive abilities.
p.136 Evolutionary game theory was initially developed by biologists to illustrate processes of natural selection.
p.136 There are five fundamental differences between classical game theory (CGT) and evolutionary game theory (EGT)...
- ...boundedly rational EGT agents face a CD [competence-difficulty] gap in which the complexity of encountered problems exceeds their cognitive capabilities. Consequently, they resort to heuristics and mental models that are inherited from previous learning.
- ...EGT agents... inherit strategies as elements of heuristics or mental models, where such inheritance reflects accumulated judgements derived from prior experience
- ...EGT payoffs represent evolutionary fitness... As in any strategic encounter, received payoffs depend upon the player/strategy combinations actually encountered... hence, a game.
- ...In EGT, strategies, not individuals, are the "dramatis personae of the social dynamic"
- ...EGT (as its name suggests) is inherently dynamic. Equilibria signify stable trajectories over time
p.141 Overall, standard EGT models can represent the adaptive learning dynamics of heuristics and mental models as well as behavioral patterns or strategies that emerge from them, such as approaches to bargaining... Closely related phenomena that are also amenable to evolutionary modeling include the development of general preference orientations
p.141 We may employ indirect evolutionary modeling to illustrate interactions between immediate choices and the longer-term evolution of preference orientations. Indirect evolutionary game theory (IEGT) combines principles from classical and evolutionary game theory. It mixes CGT evaluation of utility payoffs in given strategic contexts on the basis of existing preference orientations with EGT modeling of the adaptive processes that generate or alter such orientations over time.
p.144 A bounded rationality approach to cognition, when combined with evolutionary game-theoretic modeling, can represent relevant dynamics and outcomes.
p.144 In moderately complex situations, boundedly rational agents face a [competence-difficulty] gap [the gap between an agent's competence and the difficulty of the decision problem to be solved] that arises from the costs and limitations of cognition. Agents adapt by mixing minimal-effort S1 intuition with high-effort S2 deliberation - in order to produce impressions and judgements... Over time, impressions congeal into heuristics that rule out consideration of multiple strategies - those that fail to meet a reliability criterion based on their potential for successful use, given a [competence-difficulty] gap. At a deeper level, mental models are cognitive frameworks that combine S1 intuition, associated heuristics, and S2 reasoning to establish categories, patterns, and conceptions of cause-and-effect relationships. Like heuristics, mental models tend to emerge from context-dependent selection based upon perceived reliability in uncertain environments.
p.145 Evolutionary game theory facilitates modeling the dynamics and outcomes of learning processes associated with adoption of strategies, heuristics, and mental models, as well as behavioral predispositions. Relatively simple EGT models can, for example... [account] for influences of behavioral dispositions or social norms. Indirect evolutionary game theory allows for combined modeling of substantively rational CGT responses within established stable contexts, based on material and social preference, along with evolutionary development of basic predispositions, such as intrinsic reciprocity.
p.149 Creating the foundations for complex exchange requires resolving collective-action problems.
p.150 Ultimately, a game-theoretic approach enables systematic representation of complex exchange and collective action.
p.152 The term social means created by humans and beyond the control of a single actor. [JLJ - I disagree - the term social means jointly-interactive. Animals can be social. In defense, Greif is defining the term 'institution', and Ferguson is breaking down the words in the definition.]
p.152 Beliefs have two components: (i) expectations concerning the behavior of others... and (ii) mental models - cognitive frameworks that indicate categories, patterns, and cause-and-effect relationships. Greif asserts: "Together these components motivate, enable, and guide individuals to follow one behavior among the many that are technically feasible in social situations" (2006a, 30).
p.157 To understand economic, political, and social choice, therefore, we must consider such institutionally conditioned framing and motivation. Game-theoretic modeling offers a methodological foundation for such inquiry.
p.160-161 Broadly speaking, individuals or coherent groups who confront problems related to the design, functioning, or maintenance of elements of institutional systems - or the myriad exchanges that operate within them - may either try to escape such problems or endeavor to address them... Exit is precisely what it sounds like: a decision to leave an economic, political, or social relationship... Hirschman defines voice as "any attempt at all to change, rather than to escape from, an objectionable state of affairs" (1970, 30). Voice is the archetypical mechanism of politics; it emerges within myriad efforts to reform institutions or alter organizational behavior... Vertical exercises of voice tend to reflect actual or attempted exercises of power... Agents may regard exit and voice as alternatives... For voice, costs and benefits depend on both the prospects for success and its relative value. Prospects for successful voice, in turn, depend on the nature of the relevant problem and on agents' ability to mobilize bargaining power.
p.163 For Hirschman, however, exercising voice is an art, a set of techniques that individuals and groups improvise and usually improve with practice. Boundedly rational agents learn to apply voice by trial and error.
p.186 Reciprocal behavior is ubiquitous in social life. It involves exchanges ranging from greetings to transfers of goods and services, along with myriad forms of trading favors or reprisal. Such exchanges occur among... even strangers.
p.213 Unlike CGT models, a bounded rationality framework suggests trial-and-error learning. Using their mental models, agents engage in forward-looking experimentation (hypothesis testing) rather than CGT backwards induction... By anticipating short-term responses from others, agents develop expectations concerning the gains from their own contributions to cooperative endeavors.
p.233 "Working rules are the set of rules to which participants would make reference if asked to explain and justify their actions to fellow participants" (Ostrom 2005, 19).
p.236 Other things equal, as exchange complexity increases, the predictability of individual behavior declines.
p.257 A social network embodies a set of relationships or exchanges among various groupings of individuals, specifying precisely who interacts with whom.
p.284 social networks operate as fundamental components of institutional systems. As such they constitute foundations of complex economic, political, and social exchange. Social networks thus create, exacerbate, and help resolve myriad CAPs of social, political, and economic interaction.
Finally, the potential for networks to occasionally generate largely unpredictable information cascades presents affected parties a particularly challenging type of CAP... policymaking processes exhibit the properties of punctuated equilibria: they usually operate during periods of relative stability, during which institutional/network social choreography generates relatively predictable and slowly adjusting patterns. Yet occasional and largely unpredictable shocks can induce periods of rapid change - periods of dissolution, revolution, or rapid reform - that may be interpreted as information cascades.[JLJ - The I, me, we, you of our internal "self" interact, almost as a primitive social network. Much like children making up imaginary friends when they are alone or bored, perhaps our mind invents itself, in order to have objects which can interact, and determine how to go on. Ultimately, we might be, at our core, an invention which we ourselves create.]
p.288 Stressing the political side of policymaking, Deborah Stone asserts, "The strategies we call policy instruments are all ways of exerting power, of getting people to do what they otherwise might not do" (2002, 260). Stone's term "strategies" invokes game-theoretic interpretation. Thus policy... is a contingent plan that must consider anticipated actions and reactions of multiple parties. Policy remains a decision... it is a strategic decision, one that specifies a set of contingent actions and responses - or lack thereof. Moreover, policy involves exerting power. [JLJ - This is a new idea for me - the concept of a policy - it is a "strategy", a way of exercising power, and is aimed at getting people to do what they otherwise might not do. A policy works on people who go about their daily business, making decisions as they usually do, but with the net result somehow different than before.]
p.291 The IAD [Institutional Analysis and Development] framework offers a comprehensive conceptual structure for analyzing institutions, policy, and behavior. It identifies "major types of structural variables present to some extent in all institutional arrangements, but whose values differ from one type of institutional arrangement to another (Ostrom 2007, 27). The foundational concept of this framework is the action situation. An action situation is any social setting in which two or more individuals or organizations engage in any form of transaction, including exchange, production, conflict, group decisions, problem solving, and so forth.
p.295 The vast majority of policymaking occurs within collective-choice situations.
p.299 Because policymaking involves strategic decision making with intent to alter behavior, it constitutes an exercise of power.
p.301 Collective-action problems can motivate members of affected groups to construct social mechanisms that they believe will modify individual behavior - so as to align it more closely with... interests of certain members. Indeed, the core thesis of this book is that such mechanisms are a necessary, though often flawed, prerequisite for development. To the extent that social mechanisms for resolving CAPs are deliberately crafted human artifacts, they embody policies; they make and apply policy statements.
p.319 Policymaking involves conceptualizing, designing, negotiating, and implementing complex strategic decisions - made at some level of governance - that are intended to influence behavior via exercises of power. Collective-action problems both explain the existence of policy and shape policymaking processes.
p.350 Successful exchange, and indeed development, requires some ability to resolve fundamental collective-action problems. Such problems (CAPs) arise from any conflict
p.351 to represent dynamic learning processes among boundedly rational agents, we use evolutionary game theory (EGT), a modeling framework in which agents inherit strategies from past learning.
