Portal de Periódicos da CAPES

Bounded Rationality, Satisficing, Artificial Intelligence, and Decision‐Making in Public Organizations: The Contributions of Herbert Simon

2022; Wiley; Volume: 82; Issue: 5 Linguagem: Inglês

10.1111/puar.13540

1540-6210

Gary Schwarz, Tom Christensen, Xufeng Zhu,

Complex Systems and Decision Making

In honor of the 75th anniversary of the publication of Herbert Simon's seminal book Administrative Behavior: A Study of Decision-Making Processes in Administrative Organizations, first published in 1947, PAR held a special symposium (Simon 1997). The Royal Swedish Academy of Sciences, which awarded Herbert Simon the Nobel Prize in Economics in 1978, considered Administrative Behavior to be "epoch-making." The book has a special relationship with PAR, as two of its chapters appeared in PAR prior to its publication (Simon 1944, 1946). Moreover, when PAR selected its half-century's "Great Books" in public administration for its own 50th anniversary, Administrative Behavior was selected as the overwhelming winner by a panel of 20 experts (Sherwood 1990). The recent resurgence in interest in the behavioral sciences in public administration has reintroduced the book to a new audience (Battaglio and Hall 2019). Administrative Behavior places decision-making at the center of analysis and examines how individuals make decisions within certain organizational frames or contexts. Whereas standard economic theory assumed that individuals are perfectly rational decision makers, Simon emphasized the limits to rationality that real-life administrators—such as the employees in the municipal governments that he observed first-hand in projects for the International City Management Association—face with regard to memory, attention, and capacity (Augier and March 2001; Rainey 2001). To this end, he developed a theory of bounded rationality (Simon 1955), according to which individuals satisfice (an amalgamation of the words satisfy and suffice) rather than maximize because they cannot evaluate all potential alternatives and their consequences due to their limited cognitive and information-processing abilities, time constraints, and incomplete knowledge (Simon 1956). Thus, rather than considering all relevant factors and alternatives to make optimal decisions, individuals are portrayed as limiting their search and focusing on only a few options to make decisions that are "good enough" to meet their aspiration levels. Simon's concepts of bounded rationality and satisficing heavily influenced classic public administration work on the "science of muddling through" (Lindblom 1959) and on the budgeting process (Wildavsky 1964). Simon's oeuvre is unique for its combination of insights from social psychology and structural theory. He suggested that the formal structure of organizations channels the thoughts and actions of actors, that is, they are not free to pursue self-interests but are parts of a collective endeavor to achieve organizational goals (Egeberg 2014). As a result of structural design, the actors are assigned specific roles, and the formal structure helps in modifying their cognitive limitations. Simon challenged the notion of the single decision maker and demonstrated how organizations influence individual decision-making by, for example, developing decision premises, assigning roles, and establishing operating procedures and communication mechanisms (March and Olsen 1983). Simon cautioned that too much identification with an organizational sub-unit may lead to decisions inconsistent with larger organizational purposes and may necessitate the centralization of decision-making (Miao et al. 2019). Arguing that organizations can attenuate limits to human rationality and nudge boundedly rational individuals to make better choices (Thaler and Sunstein 2008), Administrative Behavior is widely considered to be a foundation of the burgeoning behavioral public administration literature that emphasizes the importance of the behavioral sciences in the study of public organizations (Battaglio et al. 2019; Grimmelikhuijsen et al. 2017). Simon was one of the first to suggest closer integration of the study of public administration and psychology. He believed that such interaction would be mutually beneficial as both sides have much to offer to each other, resulting in more rigorous, higher-quality research (Olsen 2015; Simon 1955). Simon also pioneered research on artificial intelligence and computer science, perceiving computers as instruments to improve human decision-making, for example, by serving as expert systems (Augier and March 2001; Rainey 2001). In The Sciences of the Artificial, Simon (1996) distinguishes between empirical phenomena that are "artificial" and those that are "natural." Here, "artificial" alludes to systems that obtain their form and behavior by adapting to the outer environment. The interface between the outer and inner environments characterizes an artificial system. The inner environment becomes important for behavior when a system reaches its limits of rationality and adaptability. The complexity of an actor's behavior largely reflects the complexity of the environmental context. Based on this basic thesis, and leaning on cognitive psychology and linguistics, Simon developed an information-processing theory of an actor's thinking processes that provides an empirically based alternative to behaviorism. Simon called for empirical research and experimentation on the concepts that he developed, and this PAR symposium issue includes five articles that do so and examine the concepts in a novel context. To overcome bounded rationality, Simon suggested that organizations establish procedural rationality, for example, by ensuring that formal processes are followed to collect, analyze, and use relevant information and ensure "appropriate deliberation" (Simon 1976, 67) before a decision is reached. In their symposium article, Al-Hashimi et al. (2022) develop an integrative model for strategic decision-making and implementation that includes procedural rationality. Based on multi-source data on 170 strategic decisions in the public sector in Qatar, a context in which studies on strategic decision-making are rare, they find a positive relationship between procedural rationality, participation, and constructive politics and the successful implementation of strategic decisions. Their study demonstrates that good decision-making alone is not sufficient to ensure that a decision attains its intended objectives, but that decision-making needs to be complemented with effective implementation processes: a topic that has received significantly less attention than it deserves in the literature. Dowding and Taflaga (2022) also start from Simon's concept of procedural rationality. Their article "Procedural Rationality in Westminster Systems: How De-Separation Affects the Decision Premise" deals with the ideal of a clear separation of the roles of political executives and career civil servants, which provide expert advice and support implementation. In studying what they call the de-separation of political careers, they show that new groups of politically appointed advisors participate in undermining procedural rationality. These advisors increasingly provide advice to political executives instead of administrative leaders, which is also the basis for recruitment to the political elite (Craft and Halligan 2020). External groups are also increasingly giving advice to political executives. In this way, politicization of the central civil service is rising; traditional expertise in the public apparatus, based on formal and informal procedures, is weakened (de-skilling); and organizational memory is decreasing. In their article "Decision-Makers' Generation of Policy Solutions amidst Negative Performance," van der Voet and Lems (2022) deal with two important constructs in Simon's decision-making theory, aspiration levels and search. The authors argue that for complex problems, required solutions cannot be identified merely through search, but must be created and designed. The article tests the behavioral framework against expectations based on threat-rigidity theory. Through a survey-experimental research design in the context of youth care policy in Dutch municipalities, they find that negative performance information decreases the creativity of policy solutions that decision makers put forward. Thus, their analysis reveals that the mechanisms of threat-rigidity theory underlie decreased creativity in response to negative budgetary feedback. The article provides an empirical test of solution generation as a central yet empirically undocumented mechanism of behavioral public administration theory. How do managers learn from performance information? Hansen and Nielsen (2022) examine the core implications of the behavioral model of performance-based learning in a public-sector setting. Based on Simon's notion of bounded rationality, this model suggests that performance information holds the potential to generate organizational learning and inform public managers' decisions about important problems, priorities, and organizational change. The authors conduct two experiments among managers working in public education in Denmark, finding that low performance affects problem focus but not managers' intention to engage in innovative search. They also find that low-performance signals enhance preferences for performance-related changes but not for different forms of changes. Owing to their bounded rationality, public sector organizations face uncertainty and, increasingly, need to deal with unknowns and "Black Swans," which are consequential shocks that cannot be anticipated based on past experience (Taleb 2010). Feduzi, Runde, and Schwarz (2022) develop a framework that distinguishes four varieties of Black Swan: (1) events that were unimaged by a public organization, or events that were imagined but (2) considered impossible, (3) sufficiently unlikely to happen to ignore, or (4) lost within the organization. In their article, they use the Challenger space shuttle explosion, 9/11, the international withdrawal from Afghanistan, and the Danish cartoon controversy to illustrate the four Black Swan types. They also provide recommendations on how public organizations can augment their traditional risk identification and assessment systems and reduce their exposure to Black Swans by mitigating the psychological and organizational barriers to uncovering unknowns. These symposium articles demonstrate the continued relevance of Simon's concepts for the study of public organizations. Simon's seminal work on bounded rationality has inspired much research on (public) decision-making (Christensen, Lægreid, and Røvik 2020), although there are still challenges in applying Simon's concepts. One strain of study that is rather lacking concerns how increasingly complex contextual factors are developing, enabling, and constraining decisions. Regarding cultural theory, Selznick (1957) provided insights into the overrepresentation of the past in the presence of path dependency. However, there is a lack of similar structurally-oriented studies building on Simon's work. Simon's focus on information and decisions is currently reflected in numerous studies of advanced uses of large quantities of information in public decisions, such as the use of artificial intelligence, but these studies are only partially relevant for structural design and often underestimate related challenges (Wirtz, Weyerer, and Sturm 2020). Queen Mary University of London, Research Environment Fund. Tsinghua University, High Level Project in Arts, Humanities, and Social Sciences (2021TSG08101).