Building on Simon’s Bounded Rationality

As noted earlier, one traditional theme in behavioral strategy research hinges on premises that date back to Richard Cyert and James March’s book A Behavioral Theory of the Firm (Reference Cyert and March1963) – a truly foundational contribution in terms of the inspiration it has provided behavioral strategy scholars (and strategy scholars in general). In their behavioral theory of the firm (BTOF), Cyert and March (Reference Cyert and March1963) follow Simon in explicitly criticizing neoclassical economics, specifically the neoclassical theory of the firm (i.e., the basic iso-cost and iso-quant curve apparatus taught in basic microeconomics). The inspiration for key parts of their theory is explicitly drawn from Simon. Rationality in BTOF is best understood in terms of satisficing aspiration levels, which may be thought of as the smallest acceptable outcomes for a goal, the line between perceived successes and failures, or as rules for simplifying evaluations (Greve, Reference Greve1998). Moreover, quasi-resolution of conflict and priority rules govern which goal is addressed. Furthermore, firms are assumed to (only) act when they face a problem – when performance falls below an aspiration level – by searching for a solution. Additionally, similar to goal selection and performance assessment, this search is governed by rules (March & Simon, Reference March and Simon1958), specifying the proximity of search to the problem, the current state of the organization, and vulnerable areas inside of it. Thus, search is problemistic, “simple-minded” (governed by simple models of causality), path-dependent (based on organizational history and accumulated knowledge embedded in the standard operating procedures and routines of the organization), and reflective of managerial preferences and employees’ training and experience. From a methodological standpoint, A Behavioral Theory of the Firm commits to developing process-oriented models of the firm and linking them to the empirical world through decision-making simulations. As a result, this simulation approach has become a key tool for understanding and deriving nonobvious implications in subsequent behavioral theory of the firm research (Gavetti et al., Reference Gavetti, Greve, Levinthal and Ocasio2012; Puranam et al., Reference Puranam, Stieglitz and Pillutla2015).

Another traditional behavioral strategy research area directly descending from Cyert and March’s theory (Reference Cyert and March1963) is organizational learning research (Argote, Reference Argote1999; Levit & March, Reference Levitt and March1988; Argote & Greve, Reference Argote and Greve2007) which examines how organizations, seen as complex adaptive systems (Denrell & March, Reference Denrell and March2001), learn and draw inferences from experience (Levitt & March, Reference Levitt and March1988). Local (myopic) search is assumed as the main mechanism driving learning, resulting in suboptimal outcomes (Levinthal & March, Reference Levinthal and March1993). These studies typically rely on agent-based simulation models, including NK models and bandit models, characterizing choices of action as existing on performance landscapes (Levinthal, Reference Levinthal1997).

Another closely related traditional research stream builds on Cyert and March’s rule-based view of organizational decision-making to examine the implications of decision rules for a series of organizational processes and outcomes, including learning (Lant & Mezias, Reference Lant and Mezias1992), adaptation (Lant, Reference Lant1992), innovation (Mezias & Glynn, Reference Mezias and Glynn1993), and performance. Adjacent to these contributions, is the view that organizational routines perform the double role of (1) providing stability and encapsulating firm-specific knowledge (Nelson & Winter, Reference Nelson and Winter1982), and (2) generating organizational change (Feldman & Pentland, Reference Feldman and Pentland2003). This work augments March and Simon’s (Reference March and Simon1958) and Cyert and March’s conceptualization of routines as abstract and stable patterns shaping and guiding organizational behavior with human agency – organizational members’ doings and acts – to introduce changes to routines, and therefore organizational change and performance.

While we will refer to these foundational contributions and their direct descendants several times in the subsequent sections, because of their centrality and extraordinary influence in the organization sciences, they have already been the subject of excellent reviews by Argote and Greve (2006) (Carnegie’s School tradition), Baumann et al. (Reference Baumann, Schmidt and Stieglitz2019) (search in performance landscapes), Gavetti et al. (Reference Gavetti, Levinthal and Ocasio2007) (Carnegie’s School tradition), Gavetti et al. (Reference Gavetti, Greve, Levinthal and Ocasio2012) (behavioral theory of the firm), Posen et al. (Reference Posen, Keil, Kim and Meissner2018) (problemistic search), and Puranam et al. (Reference Puranam, Stieglitz and Pillutla2015) (bounded rationality).

However, as Figure 1 illustrates, behavioral strategy research encompasses more than these traditional streams, so we focus on the many other parts of behavioral strategy that explore topics beyond search and organizational learning, including, for example, Kahneman and Tversky’s (Reference Tversky1974, Reference Kahneman and Tversky1982) research on heuristics and biases (Section 2), the distinction between “hot” and “cold” cognition (e.g., Hodgkinson & Healey, Reference Hodgkinson and Healey2011) (Section 2), Weick’s thinking on managerial and organizational cognition (Weick, Reference Weick1979) (Section 3), and work based on cognitive or motivational psychological theories, such as regulatory focus theory (e.g., Weber & Mayer, Reference Weber and Mayer2011) (Section 4). We delve into each of these sections in more detail below.

Information Processing, Heuristics, and Biases

In the second section, we present work on microlevel biases and heuristics that build on the notion of bounded rationality (continuous-line-patterned areas in Figure 1). It originates from two major individual-level research traditions that predict organizational decision-making: information-processing research (Simon & Newell, Reference Simon and Newell1971) and behavioral decision research (Tversky & Kahneman, Reference Tversky and Kahneman1974; Kahneman & Tversky, Reference Kahneman and Tversky1979). Information processing research concerns human problem-solving and heuristics and views the decision-making process as a key unit of analysis (Simon & Newell, Reference Simon and Newell1971). In particular, organizational behavior is explained by the discrepancy between the real world and the simplified model of reality in the human mind (Simon, Reference Simon1955).

In behavioral decision research, bounded rationality leads humans to make fast decisions that produce systematic biases (or cognitive illusions). This research stream aimed to create an evidence-based map of bounded rationality “by exploring the systematic biases that separate the beliefs that people have and the choices they make from the optimal beliefs and choices assumed in rational-agent models” (Kahneman, Reference Kahneman2003: 1449). According to the two main exponents of this stream, Amos Tversky and Daniel Kahneman, fast thinking manifests itself in decision-makers relying on rules of thumb (heuristics) – such as representativeness, availability, and anchoring and adjustment – which lead to systematic divergences from the normative models of behavior in classical economic theories. Still lying within the behavioral decision research tradition but promoting a fundamentally different view of heuristics is the work on fast-and-frugal heuristics (Gigerenzer et al., Reference Gigerenzer and Todd1999; Gigerenzer, Hertwig, & Pachur Reference Gigerenzer, Hertwig and Pachur2011). Here, humans are seen as intuitive statisticians, who can adopt heuristics in a way that increases accuracy and minimizes effort, achieving what is defined as ecological rationality. Another important extension of behavioral decision research is represented by the dual-systems theory of cognition (Chaiken & Trope, Reference Chaiken, Trope, Press and handler1999). This theory builds on Kahneman and Tversky view’s of decision-making as dominated by two systems: a system that operates through largely automatic, preconscious processes involving the development of heuristics; and one that operates via more effortful, deeper, and analytical processes. Yet, in contrast to Kahneman and Tversky’s view of automatic thinking to be seen as the default mode and the main source of systematic biases, researchers in this stream view the two systems as operating simultaneously and emphasize the importance of humans’ capability to switch back and forth between these two processes depending on situational circumstances (Hodgkinson & Healey, Reference Hodgkinson and Healey2011).

Interpretive Perspectives

The third section addresses the stream of work exploring enactment and the related notion of sensemaking (dashed-dotted-patterned areas in Figure 1) – which has received substantial attention (e.g., Hodgkinson, Reference Hodgkinson1997; Walsh, Reference Walsh1995). This stream challenges the notion of bounded rationality and its positivistic view of the world as an objective entity. In contrast, scholars belonging to this stream maintain a view of the environment as enacted – that is, socially constructed via a process of sensemaking through which individuals develop meaning from ambiguous environmental cues via local action, in turn, giving rise to even radically different perceptions of the business environment. Research in this interpretive stream can further be grouped into three traditions: micro- and meso-level work on schema and mental maps (Huff, Reference Huff1982, Reference Huff1990; Porac et al., Reference Porac, Thomas and Baden‐Fuller1989), micro- and meso-level work on social identity and identification (Ashforth & Mael, Reference Ashforth and Mael1989; Albert, Ashforth, & Dutton, Reference Albert, Ashforth and Dutton2000); and meso- and macro-level work on enactment and sensemaking (Daft & Weick, Reference Daft and Weick1984). The first sheds light on managers’ cognitive mapping process (Hodgkinson et al., Reference Hodgkinson, Maule and Bown2004), the content of these maps (Porac et al., Reference Porac, Thomas, Wilson, Paton and Kanfer1995; Reger & Huff, Reference Reger and Huff1993), and their implications for strategic decisions (e.g., Barr et al., Reference Barr, Stimpert and Huff1992). It also provides insight into divergences between managers’ cognitive maps and aggregation issues. The second tradition illuminates organizational members’ social identity construction and its impact on the firm (Ashforth & Mael, Reference Ashforth and Mael1989). The third stream focuses on the sensemaking process through which organizational members, when faced with disruptive events, collectively construct meaning and continue to enact the environment (Weick, Reference Weick1995). Therefore, in contrast to the first stream, which assumes that a frame of meaning (or cognitive map) is already in place, and one needs to connect a new cue to the existing frame (through the act of mapping), the second stream is more concerned with the “invention” of a new frame. In this section, we shed insight into the distinctive facets of this stream’s constitutive research traditions. We also discuss the difficulties of studying the mental representations and sensemaking processes of the organization’s top management team members.

Behavioral Governance

In the fourth section of the Element, we focus on the motivation-related part of the behavioral strategy landscape in Figure 1 (dashed-patterned areas), and, specifically, on the behavioral governance traditions. Work in this tradition aims at augmenting corporate governance theories (i.e., agency theory; Fama & Jensen, Reference Fama and Jensen1983; Jensen & Meckling, Reference Jensen and Meckling1976) and transaction cost economics (TCE; Williamson, Reference Williamson1975, Reference Williamson1985) with psychological theories and concepts to predict organizational outcomes, with a special focus on the role of governance – that is, “the formal structures, informal structures, and processes that exist in oversight roles and responsibilities in a corporate context” (Hambrick, Werder, & Zajac, Reference Hambrick, Werder and Zajac2008: 381) – in supporting cooperation among firms’ internal and external constituents (Westphal & Zajac, Reference Westphal and Zajac2013). In particular, work in this stream can be organized into three separate streams differing on the context of application, in terms of the “main agent(s)” to which they apply: (1) corporate leaders; (2) organizational members and employees; and (3) external stakeholders. Work focusing on corporate leaders examines executives’ behaviors, as guided by self-interest and personal risk preferences and characterized by bounded rationality, and how certain incentives (e.g., equitable compensation), structures, and systems, by acting on executives’ intrinsic, extrinsic motivation, and prosocial motivations can channel executives’ efforts in directions that contribute positively to organizational and/or societal outcomes (Devers, Cannella, Reilly, & Yoder, Reference Devers, Cannella, Reilly and Yoder2007).

Research in the second stream explores how different organizational designs induce different motivations, leading to specific types of opportunism or cooperation (Weber, Foss, & Lindenberg, Reference Weber, Foss and Lindenberg2023), as well as specific transaction costs (Foss & Weber, Reference Foss and Weber2016). Finally, work in the third stream examines how contract frames impact employees’ regulatory focus, emotions, perceptions, and behaviors (Weber & Mayer, Reference Weber and Mayer2011), inducing desired exchange behaviors (Weber, Mayer, & Macher, Reference Weber, Mayer and Macher2011) and impacting the relationship between firms. This section also builds on the second and third sections to examine efficient transaction governance decisions that take different organizational cognitive frames into account (Weber & Mayer, Reference Weber and Mayer2014).

Research Challenges in Behavioral Strategy

In the final section of the Element (Section 5), we present the major research challenges in behavioral strategy research and propose a roadmap to advance research in this field. We focus on definitional and methodological issues. From a definitional standpoint, we describe how existing behavioral strategy research has largely overlooked possible implications of branches of psychology beyond cognition (cf. Gigerenzer & Goldstein, Reference Gigerenzer and Goldstein1996), such as emotions, affect, intuition, and even motivation. Because of these tendencies, we note that there is a huge potential for future behavioral strategy research that has not yet been tapped, for example, examining the impact of strategic beliefs on firm performance, or the psychology of goals (Lindenberg & Foss, Reference Lindenberg and Foss2011), or motivational psychology, and how these aspects influence firm value creation. We also present potential issues with the widespread approach of anthropomorphizing organizations. From a methodological standpoint, we discuss the potential benefits of methodological pluralism to capture and account for the complex dynamics of real organizational settings.

From “Perfect” to “Bounded” Rationality

As its name suggests, bounded rationality was coined as a critique of the assumption of “perfect” rationality. Extreme assumptions often serve science well, and, arguably, the assumption of perfect rationality has, in many ways, served economics well, as it furthers tractability (i.e., decision problems can be represented as maximization problems), tends to lead to unique predictions (given certain assumptions on the maximization problem) and is not completely unrealistic (e.g., when a decision situation is sufficiently simple). And yet, the assumption has also been criticized for as long as it has been articulated (e.g., Veblen, Reference Veblen1898).

To understand what the critique is about, it is important to understand what the assumption entails and what it does not. Perfect rationality does not assume omniscience or perfect foresight. Economics has long since dispensed with those assumptions in favor of assumptions of “asymmetric and imperfect information.” Thus, information and uncertainty or risk are included in the set of constraints on the decision problem, but decision-makers are nevertheless assumed to be capable of maximizing and arriving at the best possible solution. This view – further refined in the 40s and 50s in the works of von Neumann and Morgenstern (Reference Neumann and Morgenstern1944), Friedman and Savage (Reference Friedman and Savage1948), and Luce and Raiffa (Reference Luce and Raiffa1957) – implicitly assumes that decision-makers have an unlimited capacity for processing the information they possess or acquire. Accordingly, any decision problem can potentially be addressed and handled immediately and perfectly.

If taken literally, it is easy to see how extreme this view is, as “there is a complete lack of evidence that, in actual human choice situations … these computations are in fact performed” (Simon, Reference Simon1955: 104). Providing this evidence, and, more generally, understanding how decision-makers actually make decisions and if general principles can be constructed about such decision-making, was a significant part of the multi- and interdisciplinary oeuvre of Herbert A Simon.

Simon (Reference Simon1947) points out that although we may intend to be rational, we do not have the wits to calculate the optimal solution for many decision situations. In Simon’s often-quoted phrase, we are boundedly rational: “intendedly rational, but limitedly so” (1947: 88). Of course, this is a negative definition in the sense that it tells us what bounded rationality is not rather than what it actually is (Foss, Reference Foss and Rizzello2003). However, exploring what bounded rationality entails for individual decision-making and, in particular, in an organizational context, has constituted a central scholarly effort in a number of related but distinct research streams over the last seven decades. This includes work in information processing (Simon & Newell, Reference Simon and Newell1971), within the broad Carnegie School (March & Simon, Reference March and Simon1958) – including, the behavioral theory of the firm (Cyert & March, Reference Cyert and March1963), the attention-based view of the firm (Ocasio, Reference Ocasio1997), and modern computational approaches that highlight the search and adaptation aspects of bounded rationality (Levinthal, Reference Levinthal1997; Puranam et al., Reference Puranam, Stieglitz and Pillutla2015) – as well as work on behavioral decision research from behavioral economics (Tversky & Kahneman, Reference Tversky and Kahneman1974; Kahneman & Tversky, Reference Kahneman and Tversky1979), top management teams (Ha0mbrick & Mason, Reference Hambrick and Mason1984), and even TCE (Williamson, Reference Williamson1996; see Section 4). The first major advance in the bounded rationality program is the expansion to consider satisficing and search.

From Cyert and March to Computational Models

Although Simon’s (1947) concern with bounded rationality emerged from his close study (for his Ph.D. thesis) of decision-making in a public bureau, much of his early work on the topic, including his multiple collaborations with Alan Newell, mainly considers individual decision-making. Instead, the concept of satisficing search was extended to the organizational level by Cyert and March in their landmark Reference Cyert and March1963 book, A Behavioral Model of the Firm. As mentioned in Section 1, the most important elaboration of the book is the notion of problemistic search, that is search triggered by negative divergences between realized performance and aspiration levels (acceptable performance thresholds set by a dominant coalition of boundedly rational decision-makers through quasi-resolution of conflict (see Posen et al., 2020 for a review)). In doing so, Cyert and March emphasized the importance of feedback in guiding search (which had also appeared in Simon 1966).

Since Cyert and March (Reference Cyert and March1963), a substantial number of papers have been published following a basic scheme in which: (1) agents are equipped with models of their environments, (2) representations inform the choices agents make, (3) these choices are “tested” in the relevant environment generating feedback, and (4) the representations in the environmental model are adapted, resulting in novel choices (see Puranam et al., Reference Puranam, Stieglitz and Pillutla2015). This scheme underpins research on the behavioral theory of the firm, including research on the “exploitation-exploration tradeoff” (March, Reference March1991) and work that uses “NK models,” originally developed by theoretical biologist Stuart Kauffman (1993) and introduced to management research by Levinthal (Reference Levinthal1997).

Behavioral strategy using an NK model depicts organizations as moving on a “fitness landscape,” an N-dimensional space that maps each attribute (characteristic or policy dimension of an organization) and its performance consequences. Crucially, the topology of the landscape is determined by the degree of interdependence (i.e., complementarity) between the attributes, K. That is, K measures how many other attributes influence a given attribute. While a small value of K implies few peaks in the landscape or just a single one (a “smooth landscape”), a high value indicates many peaks (a “rugged landscape”). Even though the landscape exists in objective reality, the organization does not know its topology and therefore has to engage in search, either local (i.e., only changing one attribute) or global (“long jumps”; changing all attributes). It is intuitive that if the landscape has just one peak, this peak will be reached through local search. In contrast, in a rugged landscape, organizations that engage in only local search may get stuck on local peaks. While this seems to make global search generally attractive, such search is more costly to engage in. Thus, organizational search behavior results from a tradeoff between the benefits and costs of the two types of search. This line of research has been thriving since the late 1990s (Gavetti & Levinthal, Reference Gavetti and Levinthal2000; Puranam et al., Reference Puranam, Stieglitz and Pillutla2015), percolating widely beyond the community of modelers and resulting in empirical work investigating the impact of negative performance feedback on the likelihood and intensity of organizational and strategic changes (e.g., Audia & Greve, Reference Audia and Greve2006; Greve Reference Greve1998). This stream of empirical research has also shown that, especially under conditions of uncertainty, decision-makers selectively imitate the actions of other firms (Rhee et al., Reference Rhee, Kim and Han2006). In particular, to improve firm performance while avoiding the risks and costs of online search and experimentation, they learn vicariously by absorbing knowledge produced by other firms’ explorations (Baum et al., Reference Baum, Li and Usher2000; Levitt & March, Reference Levitt and March1988).

From March and Simon to the Attention-Based View

Further implications of bounded rationality at the organizational level were added through work from Simon’s collaboration with the political scientist James G. March and, more recently, by Ocasio’s (Reference Ocasio1997) attention-based view of the firm. In Simon’s (1948) original view, the main role of organizations was to structure the attention of individual decision-makers by offering both a physical and cognitive division of labor (member A works on problems of type Z, while B works on type Y problems, etc.) (Dearborn & Simon, Reference Dearborn and Simon1958). In the book Organizations, March and Simon explicitly focus on providing a theory for when, where, and how organizations search for information about urgent problems, alternatives, and their consequences. More specifically, they illustrate that organizations structure and channel the allocation of cognitive effort and attention through the division and spatial allocation of labor, and decide which problems should be worked on based on “decision premises” (March & Simon, Reference March and Simon1958). We will return to some of these aspects in Section 4.

Building on March and Simon (Reference March and Simon1958), Ocasio (Reference Ocasio1997) formulated the attention-based view of the firm in an article published almost four decades after their classic work. Ocasio interprets bounded rationality as individuals’ limited attentional capacity. Attention is the “noticing, encoding, interpreting, and focusing of time and effort on issues (problems, opportunities, threats …) and answers (alternatives, projects, procedures, …)” (Ocasio, Reference Ocasio1997: 188). A key purpose of organizations is therefore to direct attention in the best possible manner (Joseph & Gaba, Reference Joseph and Gaba2020). While the main emphasis is on the attention of the firm’s top decision-makers, Ocasio develops a more general theory that picks up on Simon (Reference Simon1947) and March and Simon (Reference March and Simon1958) in showing how organizations sculpt decisions by defining and allocating the stimuli that channel attention (cf. Ocasio, 2011). Thus, organizational problems (“Hey, demand dropped in market x,” “Ugh, a machine broke down”) in a specific unit selectively trigger the attention of that unit’s employees.

The Heuristics and Biases Program

A key theme in Simon’s idea of bounded rationality is that decision-makers often rely on cognitive shortcuts because they lack the processing capacity to make fully informed decisions (Schwenk, Reference Schwenk1984; Simon, Reference Simon1956). Satisficing is one such heuristic, enabling decision-makers to economize on cognitive resources as compared to computationally more demanding approaches, such as maximizing. The downside is, of course, that decision-makers may choose and stick to inferior alternatives. This basic duality is central in the behavioral decision research initiated by the behavioral economists Tversky and Kahneman (Reference Tversky and Kahneman1974) (see Kahneman (Reference Kahneman2003) for a summary), and often referred to as the heuristics and biases program (Kahneman, Reference Kahneman2003, 2011; Kahneman & Tversky, Reference Kahneman and Tversky1979). Additionally, it is now seen as an important instance of “dual processing theory” which stresses the “twin imperative of having to process information deliberatively and in detail, but also being able to cut through such detail with minimal cognitive effort to perform tasks more efficiently” (Hodgkinson et al., Reference Hodgkinson, Burkhard and Foss2023: 1042). Notably, the heuristics and biases program, albeit originating directly from Simon’s bounded rationality notion, was framed as a new paradigm departing from “bounded rationality, satisficing, and simulations” through the reliance on empirical evidence – mainly of an experimental nature (Sent, Reference Sent2004: 742).

The key idea in the heuristics and biases program is that the manifold heuristics that decision-makers utilizeFootnote ² are likely to result in a rich set of cognitive biases, rendering the decision-making process systematically flawed – at least if evaluated related to a perfect rationality ideal. The search for such biases has become a major industry, with the number of biases at 188 at last count.Footnote ³ While it is debatable how distinct many of these biases truly are (Oeberst & Imhoff, Reference Oeberst and Imhoff2023), representativeness and availability biases, both stemming from the tendency to focus attention on supposedly typical features of a decision situation, are among the more well-known ones (Hodgkinson et al., Reference Hodgkinson, Burkhard and Foss2023; Kiesler & Sproull, Reference Kiesler and Sproull1982). Other well-known biases (probably so well-known that an explanation is not necessary) are hindsight, self-serving, overconfidence, anchoring, and confirmation biases.

One of the most influential theories in this program is Kahneman and Tversky’s (Reference Kahneman and Tversky1979) prospect theory. Conceived as an alternative to expected utility theory, it provides a model for predicting individual decision-making under risk in contexts of isolated gambles characterized by simple choices with two clear prospects or alternatives and stated probabilities (Kahneman, Reference Kahneman2003). The theory is based on the following three core assumptions: (1) decision-makers weigh assured outcomes more than probable outcomes, (2) they simplify decision-making by focusing on differences between choices rather than their commonalities, and (3) they tend to be risk-averse in the domain of (small) gains and risk-seeking in the domain of (small) losses. As a result, decision-makers prefer an assured gain over an uncertain gain with a marginally higher expected value and prefer an uncertain loss over an assured loss with a marginally lower expected value (Bromiley & Rau, Reference Bromiley and Rau2022; Holmes et al., Reference Holmes, Bromiley, Devers, Holcomb and McGuire2011).

Prospect theory has been very popular in strategy research (see e.g., Holmes et al., Reference Holmes, Bromiley, Devers, Holcomb and McGuire2011; Hoskisson et al., Reference Hoskisson, Chirico, Zyung and Gambeta2017). It is typically applied to the top management team or the organization to explain the relationship between a firm’s negative performance and its risk-taking (e.g., risky actions such as acquisitions, divestitures, and new product introductions). The application of prospect theory to firm strategic decisions has not come without criticism (Bromiley, Reference Bromiley2010; Bromiley & Rau, Reference Bromiley and Rau2022). While prospect theory examines the behavior of an individual decision-maker confronted with a simple well-structured problem with clear prospects or alternatives and known probabilities, the problems organizations are confronted with are generally complex, collective, and involve a substantial level of uncertainty. Explicit examinations of the validity of prospect theory to predict individual choices in more complex contexts similar to organizational ones have provided far from reassuring results (e.g., Bromiley, Reference Bromiley2010; Spiliopoulos & Hertwig, Reference Spiliopoulos and Hertwig2023).

This aggregation issue emerges in most strategy research building on the heuristics and biases program. Another example is the upper echelons research area founded by Hambrick and Mason (Reference Hambrick and Mason1984), who were early adopters of the heuristics and biases program (e.g., Barnes, Reference Barnes1984; Das & Teng, Reference Das and Teng1999) (see the review and discussion in Hodgkinson et al., Reference Hodgkinson, Burkhard and Foss2023). In their foundational work and that built on it, executives’ and senior managers’ characteristics and traits were grafted onto behavioral theories of decision-making (cf. Hambrick, Reference Hambrick2007) to show the heuristic nature of their decision-making and the possibly biased nature of the resulting firm-level strategic decisions and outcomes (e.g., Chen et al., Reference Chen, Crossland and Luo2015; Li & Tang, Reference Li and Tang2010; Wesley et al., Reference Wesley, Martin, Rice and Lubojacky2022). As suggested, these studies tend to take an aggregate view without delving into how biases and heuristics may be either mitigated or accentuated by social interactions and dynamics between firm leaders. Moreover, the bulk of this work focuses primarily on top-manager overconfidence (Burkhardt et al., Reference Burkhard, Sirén, van Essen, Grichnik and Shepherd2022), with Hodgkinson et al. (Reference Hodgkinson, Burkhard and Foss2023) identifying approximately 50 percent of published papers in this area addressing this particular bias.

In contrast to the work on decision-making in top management, even less attention has been paid to how heuristics and biases at lower organizational levels impact value creation. In one such application, Foss and Weber (Reference Foss and Weber2016) consider framing and social comparison biases that may arise in exchanges between members and units in firms. They argue that different hierarchical structures induce different biases, giving rise to predictable frictions in internal exchanges. Given that this line of research is very recent, this area represents a significant research opportunity in behavioral strategy.

Finally, Kahneman and Tversky’s original goal “to obtain a map of bounded rationality” recently re-gained momentum in work examining individual-level beliefs and strategic choices (Kahneman, Reference Kahneman2003) using neurocognitive experiments. To do so, a group of behavioral strategy scholars directly examine decision-makers’ mental processes via brain imaging techniques (e.g., Laureiro-Martínez, Brusoni, & Zollo, Reference Laureiro-Martínez, Brusoni and Zollo2010; Laureiro-Martínez, Brusoni, Canessa, & Zollo, Reference Laureiro‐Martínez, Brusoni, Canessa and Zollo2015). While holding some promise, this approach may risk excessively detaching decision-making from its organizational context, thereby losing external validity (cf. Felin et al., Reference Felin, Foss and Ployhart2015). Moreover, the nature of brain imaging methodology may change the nature of the decision-making task, perhaps reducing its internal validity. We will return to some of these issues in the final section of the Element.

Ecological Rationality

Whereas most research in the heuristics and biases tradition emphasizes the downside of heuristics (i.e., biases), the ecological rationality program accentuates their upside. Picking up on Simon’s (Reference Simon1955) idea, this approach focuses on the skilled application of heuristics by decision-makers. So in this research, a heuristic is seen as “a strategy with the goal of making decisions more quickly, frugally, and/or accurately than more complex methods” (Gigerenzer & Gassmeier, Reference Gigerenzer and Gaissmaier2011). Strikingly, Gigerenzer and his co-authors show that heuristics are not second-best approximations to maximizing rationality, but may often lead to faster and better outcomes, particularly in uncertain, ambiguous environments with few useful data points (e.g., Gigerenzer & Gaissmaier, Reference Gigerenzer and Gaissmaier2011). As with the neuroscience approach to behavioral strategy, the study of ecologically rational, fast, and frugal heuristics in organizations in general and in strategy in particular is still in its infancy. Pioneer work has been done by Bingham and Eisenhardt (Reference Bingham and Eisenhardt2011), who studied the formation of heuristics during the internationalization processes of six technology-based ventures. However, except for Luan, Reb, and Gigerenzer (Reference Luan, Reb and Gigerenzer2019), there is still very little research examining the formation and use of heuristics at the firm level and their implications for organizational decision-making and strategic adaptation. Again, the dearth of research on this important topic represents an important opportunity for work in behavioral strategy.

Schema Theory, Mental Representations, and Frames

Schemas are cognitive structures through which individuals recognize, process, organize, and interpret information from their environment. Rooted in cognitive psychology, schema theory posits these mental frameworks guide humans’ perceptions, beliefs, and actions by providing a structured means of understanding. In essence, schemas are the “mental templates” that individuals use to categorize the world (Walsh, Reference Walsh1995).

When applying this theory, strategy scholars tend to assume information converges and is interpreted at the top manager level (Fahey & Narayanan, Reference Fahey and Narayanan1989; Lyles & Schwenk, Reference Lyles and Schwenk1992). Thus, researchers examine how managers’ schemas influence their interpretations of complex business environments, their expectations of how these may evolve (Gick & Holyoak, Reference Gick and Holyoak1983), and their subsequent strategic decisions (Kaplan, Reference Kaplan2011). To do so, these scholars generate managers’ mental maps or representations, which are “tangible” expressions of schemas, by outlining managerial perceptions and beliefs of the internal and external environment of the firm. These mental maps may include firm structure, technology, industry landscapes, stakeholder interplay, and competitive dynamics (Kaplan, Reference Kaplan2011). For instance, prominent work by Huff (Reference Huff1982), Porac et al. (Reference Porac, Thomas and Baden‐Fuller1989), and Reger and Huff (Reference Reger and Huff1993) shed light on the role mental maps play in shaping managerial perceptions of strategic groupings and competitive positioning. While some of these studies indicate strategists within a given sector tend to hold highly similar representations of the competitive landscape (e.g., Porac et al., Reference Porac, Thomas and Baden‐Fuller1989; Reger & Huff, Reference Reger and Huff1993), others suggest these representations reflect subjective perceptions (Rindova & Fomburn, Reference Rindova and Fombrun1999), prompting radically different strategic groupings across firms in the same industry (e.g., Hodgkinson & Johnson, Reference Hodgkinson and Johnson1994) within the same timeframe (e.g., Fahey & Narayanan, Reference Fahey and Narayanan1989). One such example is from Kilduff (Reference Kilduff2019), in which he argues that managers may interpret certain firms as relational rivals (similar to long-standing university rivals), increasing the likelihood of nonrational competitive actions and reactions, exceeding those predicted from market conditions.

Closely related to schema theory is framing theory (Goffman, Reference Goffman1974), where frames define the way experience is organized, described, and presented (Cornelissen & Wener, Reference Cornelissen and Werner2014; Fiss & Zajac, Reference Fiss and Zajac2004; Hodgkinson et al., Reference Hodgkinson, Bown, Maule, Glaister and Pearman1999). By influencing the categorization process (Starbuck & Milliken, Reference Starbuck, Milliken and Hambrick1988), frames determine the perceived importance, urgency, or relevance that managers attribute to issues and events (Jackson & Dutton, Reference Jackson and Dutton1988), activate expectations about potential outcomes (Goffman, Reference Goffman1974), shape how attention is allocated (Benner & Tripsas, Reference Benner and Tripsas2012; Ocasio & Joseph, Reference Ocasio, Joseph, Szulanski, Porac and Doz2005), and influence how strategy is formulated (Dutton, Fahey, & Narayanan, Reference Dutton, Fahey and Narayanan1983; Kaplan, Reference Kaplan2008) and implemented (Thomas, Clark, & Gioia, Reference Thomas, Clark and Gioia1993).

One prolific stream of this strategy research examines how frames impact decision-makers’ categorizations and responses to stakeholder evaluations (Bitektine, Reference Bitektine2011; Elsbach, Reference Elsbach2003). For example, some studies show that the way decision-makers frame stakeholders’ feedback results in more or less symbolic responses (e.g., Bass et al., Reference Bass, Pfarrer, Milosevic and Titus2023; Nason, Bacq, & Gras, Reference Nason, Bacq and Gras2018; Wang et al., Reference Wang, Jia, Xiang and Lan2022). In contrast, another stream highlights how mental representations and frames, despite their advantages for information processing and meaning construction, lead to rigidities (Benner & Tripsas, Reference Benner and Tripsas2012; Gilbert, Reference Gilbert2006; Lovallo, Clarke, & Camerer, Reference Lovallo, Clarke and Camerer2012) that can, in turn, introduce biases or blind spots in the strategic decision-making process (e.g., Barr et al., Reference Barr, Stimpert and Huff1992; Tripsas & Gavetti, Reference Tripsas and Gavetti2000). This work suggests that socialization with other relevant actors, including social class peers, business advisors, and family members may alter preexisting schemas, potentially reducing this negative effect (Bartunek & Moch, Reference Bartunek and Moch1987; Nason, Mazzelli, & Carney, Reference Nason, Mazzelli and Carney2019; Strike & Rerup, Reference Strike and Rerup2016).

The conceptual apparatus provided by schema and framing theories is cross-level, and microfoundational, offering individual-level explanations for higher-level organizational phenomena. Nevertheless, the field faces challenges due to the proliferation and inconsistent application of related constructs. This issue is particularly evident in the overlapping and sometimes interchangeable use of terms such as mental representations, cognitive frames, frameworks, schemas, maps, and knowledge structures (Walsh, Reference Walsh1995). For behavioral strategy research to progress, it is crucial to achieve greater conceptual clarity. This entails precisely defining each construct and delineating their specific roles and interactions in the context of interpretation and strategic decision-making. Furthermore, despite notable attempts (e.g., Kaplan, Reference Kaplan2008), further empirical evidence is needed to examine the mechanisms through which individual-level cognitions translate into collective cognition or shared mental models, especially in the presence of spatial and temporal distances between organizational members.

Social Identity Theory, Self-Categorization, and Organizational Identification

Social psychology theories based on identity and self-categorization suggest individuals delineate groups based on perceived features and then affiliate themselves with those satisfying their personal needs for self-definition and meaning (Tajfel, Reference Tajfel1978; Hogg & Terry, Reference Hogg and Terry2000; Tajfel & Turner, Reference Tajfel, Turner and Austin1986). This social categorization process shapes member behaviors. That is, individuals incorporate distinctive group features into their own identity (Tajfel & Turner, Reference Tajfel, Turner and Austin1986; Terry & Hogg, Reference Terry and Hogg1996) and embrace a shared set of values and beliefs, fostering ingroup cohesion and normative behavior (Brewer & Gardner, Reference Brewer and Gardner1996; Tajfel & Turner, Reference Tajfel, Turner and Austin1986). Moreover, this delineation is preserved, as ingroup members craft their attributes, norms, and values relative to outgroups to achieve “optimal distinctiveness” (Brewer, Reference Brewer1991).

Ashforth and Mael’s (Reference Ashforth and Mael1989) work illuminates the role of organizations (and their subgroups) in organizational members’ social identity construction and its impact on the firm (see Ashforth, Harrison, & Corley, Reference Ashforth, Harrison and Corley2008; Ashforth & Johnson, Reference Ashforth, Johnson, Hogg and Terry2001; Dutton, Dukerich, & Harquail, Reference Dutton, Dukerich and Harquail1994; Liu, Fisher, & Chen, Reference Liu, Fisher and Chen2018).Footnote ⁴ For instance, some scholars illustrate how CEO organizational identification – defined as a CEO’s sense of oneness or belongingness with an organization (Lange, Boivie, & Westphal, Reference Lange, Boivie and Westphal2015; Mael & Ashforth, Reference Mael and Ashforth1992) – can foster goal alignment (Davis, Schoorman, & Donaldson, Reference Davis, Schoorman and Donaldson1997; Kogut & Zander, Reference Kogut and Zander1996), reducing agency costs and, in turn, increasing firm performance (Boivie et al., Reference Boivie, Lange, McDonald and Westphal2011). Relatedly, other scholarly work examines how a CEO’s identification within a particular group in the organization, or adoption of a specific role, shapes their priorities and conduct (e.g., Miller & Breton-Miller, Reference Miller and Le Breton–Miller2011; Miller, Breton-Miller, & Lester, Reference Miller, Le Breton‐Miller and Lester2011).

This work on social identity branches out beyond CEOs. Literature on group faultlines – referred to as virtual dividing lines that split a group into relatively homogeneous subgroups based on group members’ alignment and identification with the subgroup prototype (Lau & Murnighan, Reference Lau and Murnighan1998; Thatcher & Patel, Reference Thatcher and Patel2012) investigates how organizational members’ multiple identities and their relationships shape firm strategy and performance (Lawrence & Zyphur, Reference Lawrence and Zyphur2011). For example, Leicht-Deobalda et al. (Reference Leicht‐Deobald, Huettermann, Bruch and Lawrence2021) show that organizational identification among firm employees can be hampered by demographic faultlines, especially when functional heterogeneity within demographic subgroups is low, in turn, generating negative consequences for firm performance and innovation. Kaczmarek, Kimino, and Pye (Reference Kaczmarek, Kimino and Pye2012) article illustrates how task-related faultlines within corporate boards can have value-destroying effects. Faultlines also have a similar effect in top management teams. Although TMT heterogeneity positively impacts the formulation of deviant competitive actions (Ndofor, Sirmon, & He, Reference Ndofor, Sirmon and He2015), this benefit disappears when strong faultlines are present. This represents a more nuanced view of the impact of demographic characteristics on firm performance in the upper echelon’s research stream (Hambrick & Mason, Reference Hambrick and Mason1984).

Despite these important contributions, the literature on how social identification impacts firm behavior and outcomes is still in its infancy, especially when examining the dark side of overidentification in organizations (Bouchikhi & Kimberly, Reference Bouchikhi and Kimberly2003; Fiol, Reference Fiol1991, Reference Fiol2001). Furthermore, significant empirical challenges exist in reliably assessing whether and to what extent social identification affects distal firm-level variables (cf. Ashforth et al., Reference Ashforth, Harrison and Corley2008).

The Attention-Based View of the Firm

Ocasio (Reference Ocasio1997) expanded Simon’s (1961) bounded rationality into the attention-based view of the firm, which addresses coordination inside the firm. He specifically suggested that firm behavior results from the way in which firms direct the attention of the individuals within it. As such, the organizational actors’ information is both limited and distorted by the firm in which they work. This approach is frequently used in both theoretical work (Ocasio & Joseph, Reference Ocasio, Joseph, Szulanski, Porac and Doz2005) and empirical studies (e.g., Bouquet & Birkinshaw, Reference Bouquet and Birkinshaw2008; Bouquet, et al., Reference Bouquet, Morrison and Birkinshaw2009; Hung, Reference Hung2005; Ocasio & Joseph, Reference Ocasio and Joseph2008). For example, Joseph and Wilson (Reference Joseph and Wilson2018) used the attention-based view to predict when the firm will expand to include new subunits to address intrafirm attentional conflicts. This type of work explores using organizational design to direct employee attention to increase coordination.

Organization Design

There is also a large body of work on organization design (e.g., Galbraith Reference Galbraith1973; Burton & Obel, Reference Burton and Obel2018). This research stream addresses coordination and cooperation inside the firm. That is, organizational design research “investigate[s] the information flows essential for accomplishing the organization’s objectives, then examine[s] what these information patterns imply for organization structure” (Simon, Reference Simon1967: 1). This work relies on the simplistic mental representations proposed by March and Simon (Reference March and Simon1958), and the influence that organizational structure has on them. Thus, organizational actors in this theory have both limits on the amount of information processed, as well as distortions of this information due to the organization’s design. There are many studies in this area that examine how these information processing needs lead to centralization and decentralization in the organization (e.g., Obel & Burton, Reference Obel and Burton1984; Galbraith, Reference Galbraith1973, Reference Galbraith1977; Thompson, Reference Thompson1967; Tushman & Nadler, Reference Tushman and Nadler1978).

Misaligned Behavioral Assumptions

However, the behavioral assumptions underlying these two economic-based theories do not match those of the attention-based view and organization design, making it difficult to combine them. In particular, agency theory is firmly based on the traditional economics model of maximization. Transaction cost economics, which creates a discriminating alignment between exchange characteristics and efficient governance to mitigate transaction costs, assumes boundedly rational actors. Even with these assumptions, the theories narrowly define actors’ cognitive limitations, in the case of TCE including bounds on the amount of information they process (Foss & Weber, Reference Foss and Weber2016) and in the case of principal-agent theory recognizing no such bounds but postulating asymmetric information conditions. The thin bounded rationality assumption of TCE (Foss, Reference Foss2001) allows all contracts to be unavoidably incomplete, which is necessary for agency and transaction costs to arise ex post (Eisenhardt, Reference Eisenhardt1989; Williamson, Reference Williamson1985). However, it ignores the distortions due to cognitive biases and heuristics (e.g., Tversky and Kahneman, Reference Tversky and Kahneman1974) that Simon intended. Thus, as conceived, it is incompatible with the full form of bounded rationality assumed in both the organizational design and the attention-based view literatures.

Moreover, both theories focus solely on negative motivations. Agency theory only considers self-interest, suggesting agents act in their own interest, unless these are aligned with those of the agent (Eisenhardt, Reference Eisenhardt1989). TCE takes this self-interested view further by assuming an even stronger form of self-interest, opportunism. Opportunism is self-interest seeking “with guile” (Williamson, Reference Williamson1985), which permits actors to lie, cheat, and steal in pursuit of their own interests, as they are not constrained by laws and rules. Again, the motivational assumptions of these economic-based governance theories are incompatible with those of organizational design research and the attention-based view, which both allow for more expansive positive (e.g., reciprocity) or negative (e.g., internal politics) motivations.

Given these extensive differences in both cognitive and motivational assumptions of economic-based governance theories and the attention-based view and organizational design literatures, there is a need to augment the less complex behavioral assumptions of the former to bring them in line with those of the latter. Some work has started to do just that. However, this work has only begun to scratch the surface of expanding these behavioral assumptions for governance.

Behavioral Agency Theory

Scholars expanded the behavioral assumptions of agency theory in two ways. First, they put bounded rationality assumptions in the form of cognitive distortions explicitly into the foundations of the theory. Second, they increased the precision of negative motivations, as well as added positive motivations into agency theory.

Cognitive Expansion. Behavioral agency theory augments agency theory with a variety of cognitive constraints that distort the information processes. Specifically, problem framing, drawing on prospect theory (Kahneman & Tversky, Reference Kahneman and Tversky1979), was added to address agent risk-taking behavior (Wiseman & Gomez-Mejia, Reference Wiseman and Gomez-Mejia1998), with greater risk-taking linked to positive versus negative framing. Further, Foss and Stea (Reference Foss and Stea2014) augmented agency theory with boundedly rational sensemaking, allowing the principal to learn the agent’s effort and type. This addition allowed for the reduction of information asymmetry, leading to greater value creation in the exchange than predicted by traditional agency theory.

Motivational Expansion. More recent work elaborated on the motivational assumptions of agency theory. The broad concept of self-interest was replaced with the negative emotions of envy, guilt, and greed, and augmented with the positive motivation of fairness (Pepper, Gosling, & Gore, Reference Pepper, Gosling and Gore2015). This greater specificity was better able to explain the results found in empirical data than the rational choice model. Further, Bosse and Phillips (Reference Bosse and Phillips2016) combined traditional agency theory with the positive motivations of reciprocity and fairness to limit self-interest, resulting in the expansion of the theory from simply limiting self-serving agent behaviors to expanding social welfare. Further, Pepper and Gore (Reference Pepper and Gore2015) centered the agents in the theory, suggesting motivated managers tended to perform better than those with incentives aligned with the principal. Finally, Cuevas-Rodríguez, Gomez-Mejia, and Wiseman (Reference Cuevas‐Rodríguez, Gomez‐Mejia and Wiseman2012) added trust to suggest that agents can behave honestly and loyally, leading to cooperative relationships with the principal.

Behavioral Transaction Cost Economics

Additionally, since the inception of TCE, scholars continued to offer expansions of the behavioral assumptions underlying the theory (e.g., Ouchi, Reference Ouchi1980; Ghoshal & Moran, Reference Ghoshal and Moran1996). As with behavioral agency theory, this augmentation resulted in different predictions than traditional TCE.

Cognitive Expansion. Like agency theory, TCE’s bounded rationality assumption was also expanded. In the context of contractual governance, Weber and Mayer (Reference Weber and Mayer2011) examined how contract frames impact views of the exchange, emotions, behaviors, and exchange relationships. Further, Weber and Bauman (Reference Weber and Bauman2019) empirically extended this theory by uncovering attributions of benevolence as the underlying psychological mechanism. Additional research in this area examined how regulatory focus impacts contract design (Weber, Mayer, & Macher, Reference Weber, Mayer and Macher2011), as well as how contract design influences trust development and learning in repeated exchanges (Weber Reference Weber2017), and innovation (Mayer, Xing, & Mondal, Reference Mayer, Xing and Mondal2022). Finally, Weber and Coff (Reference Weber and Coff2023) elaborated on the idea of cognitive biases to suggest asset specificity perceptions drive most governance decisions, and these perceptions are likely to be biased and even manipulated.

Foss and Weber (Reference Foss and Weber2016) moved the discussion of augmented bounded rationality away from contract framing. Instead, they focused on how bounded rationality alone may lead to specific transaction costs inside the firm. Moreover, Nickerson and co-authors drew on an expanded bounded rationality assumption to identify antecedents to governance decisions, creating the problem formulation-problem solving perspectives (e.g., Nickerson, Silverman, & Zenger, Reference Nickerson, Silverman and Zenger2007; Baer, Dirks, & Nickerson, 2009).

Motivational Expansion. Additionally, due to the concern over Williamson’s negative motivational assumption in TCE (Ghoshal & Moran, Reference Ghoshal and Moran1996), scholars expanded this behavioral assumption. Ouchi (Reference Ouchi1980) brought social comparison into TCE, introducing clan governance to address it. Moreover, Ouchi and Barney (Reference Ouchi and Barney1980) suggested goal alignment influences the efficient governance form, expanding the options from market and hierarchy to include clans, bureaucracies, quasi-markets, and their intermediate forms. Further, Husted and Folger (Reference Husted and Folger2004) elaborated on this idea, incorporating organizational justice into the theory and arguing that transaction costs typically occur because it is difficult to assess fairness in an exchange. Nickerson and Zenger (Reference Nickerson and Zenger2008) further expanded on the incorporation of social comparison in TCE, showing that these comparisons and their resulting negative emotions can lead to hierarchical failure, as they and their associated costs increase with the scope and scale of the firm. More recently, Weber, Foss, and Lindenberg (Reference Weber, Foss and Lindenberg2023) expanded Williamson’s opportunism assumption into different types of opportunisms, as well as added the positive motivation of cooperation. Specifically, they suggested that different hierarchical forms tend to invoke distinct motivations, which lead to specific behaviors.

Psychology-Augmented Governance Theory

First, the work on augmenting economic-based governance theories with psychology is in its infancy. While work on contract framing has gained some traction, there are still ample opportunities to examine the impact of other bounds on rationality in these theories. Moreover, the studies on expanding motivations and adding the impact of external environments on governance are just starting to emerge. Finally, the idea of bringing in informational, attentional, and governance issues together to design high-performance transactions has not yet been done. Thus, this area of research is wide-open, offering extensive opportunities for future work.

Opportunities to Expand Bounded Rationality. There are several areas for additional research in expanding the bounded rationality assumption in economic-based governance theories. First, behavioral agency theory is much less developed than behavioral TCE. As a result, research on the impact of framing tasks and incentives, beyond those using prospect theory (Kahneman & Tversky, Reference Kahneman and Tversky1979), should be considered, as the prior work using prospect theory has been criticized as improperly applying base predictions from the lens to the firm (Bromiley, Reference Bromiley2010). For example, considering the impact of prevention and promotion framing in agency theory could create interesting prescriptions for managers who are looking to induce vigilant versus eager behavior in their agents.

Additionally, even in TCE, the work on framing only considered regulatory focus theory. However, there are other ways to frame the same contract. For example, the contract’s time horizon could be framed as short term versus long term. Additionally, contract detail could be framed at a high-level or extremely detailed, based on construal theory. Moreover, the exchange relationship between the partners can be framed as competitive or cooperative. Thus, there are many different possibilities for framing the exact same contract that have not yet been explored.

Another distortion of human cognition emerges from stereotypes. While organizational behavior research examines how gender and racial stereotypes impact individual career paths and job satisfaction, this work does not examine how these cognitive elements influence transactions within and between firms. Yet, these stereotypes can influence organizational actors’ perceptions of opportunistic behavior or the attributions of benevolence, directly impacting transaction performance. Further, while some studies explore attributions of performance in exchanges, no one explicitly examines when the fundamental attribution error may occur, which may impact the formation of trust, as well as the perception of opportunistic behavior in the exchange. As such, both aspects of governance are virtually unexplored at this time.

In addition, only one study (Weber & Coff, Reference Weber and Coff2023) examines the biases and malleability of perceptions of key TCE constructs, such as asset specificity, uncertainty, and frequency. Additionally, the concept of perceptions has not been brought into agency theory at all. Yet, in both cases, it is likely that perceptions of these key constructs drive both choices and behaviors in these theories. Thus, it is imperative to understand how incorporating perceptions into these economic-based governance theories may change their predictions and may explain empirical evidence that traditional theories cannot address.

Finally, most psychological theories address individual behaviors, which is appropriate for its application to agency theory. However, TCE discusses firm-level outcomes. Thus, when augmenting TCE with psychology, there is a need to either work with transactions involving a small group of managers or to examine how these effects aggregate to the firm level. That is, the cognitive impact on a single person does not matter unless that impact is likely to dominate the individuals in the transaction. Thus, there is a need for a deeper examination of this topic.

Opportunities to Expand Motivations. In addition to these opportunities to extend the cognitive assumptions in agency theory and TCE, there is also a need for research that expands the motivational assumptions of these theories. For example, nonconscious affect could impact contract design. That is, witnessing a supplier manager berating their assistant during a negotiation, may unconsciously activate negative exchange attitudes, leading to the inclusion of more safeguards in the contract.

Moreover, while emotions are examined as an outcome variable in behavioral governance, their influence on transaction costs or governance is not yet explored. Emotions were shown to impact both cognition and motivation. Thus, they are good candidates for future exploration in economic-based governance theories.

Additionally, most motivations examined in agency theory and TCE are broad. That is, self-interest and opportunism are both negative, but not precise motivations. Thus, there is also a need to expand the types of opportunisms that may be motivating specific behaviors. So far, only one paper explores a more nuanced look at opportunism (Weber, Foss, & Lindenberg, Reference Weber, Foss and Lindenberg2023). Further, while a few positive motivations are incorporated into agency theory (e.g., Devers, Cannella, Reilly, & Yoder, Reference Devers, Cannella, Reilly and Yoder2007), fewer are incorporated into TCE. Thus, there is a broad area of motivational expansion that can occur in economic-based governance theory.

Finally, while the expansion of TCE’s motivations started with social comparison, little empirical research examines envy costs. In addition, there has not been an application of this concept to alliances or contracts. As a result, there is a significant opportunity to explore this extension as well.

Holistic Behavioral Governance Theories

There are also several opportunities to bring attention, knowledge flows, and governance together to design optimal transactions within and between firms. For example, incentives can be used to increase knowledge flows. Further, organizational design can impact perceptions of transaction hazards, impacting governance choice. To understand the potential for this combination to expand our understanding of transaction performance, we explore a few of these potential areas of interest in detail.

Situated Cognition and Motivation in Governance. First, there is a chance to bring in the attention-based view of the firm to understand how perceptions of governance variables are formed, and how they may be biased. Additionally, by bringing concepts of organizational design into governance, Williamson’s concept of administrative mechanisms can be expanded to include centralization and decentralization, as well as other aspects of structure. Further, organizational structures are likely to impact managers’ and employees’ motivations, which also can be used to inform contracts in agency theory. As there may be opportunities to shape agent behavior without aligning their incentives in an employment contract.

Design to Optimize Organization Performance. In addition, organizational design may increase or reduce the need for governance, while potentially also allowing for optimal information flow. Thus, rather than focusing on one aspect of the other, the field can return to multi-theoretical perspectives, such as those that dominated economic-based strategy in the early 2000s (transaction cost plus the resource-based view, for example), but in a behavioral strategy realm. Neither information flows nor governance are mutually exclusive in an organization. Instead, each aspect likely influences the other, so it is important to understand how these elements work together or in opposition when designing the organization.

Design to Optimize Transaction Performance. Moreover, the only research examining how transactions are designed is within the context of organizations (i.e., the organizational design literature). While this is an interesting topic, the governance literature is also interested in how transactions between firms (e.g., alliances) or within markets are designed. Given that not all markets, contracts, or alliances are the same, but instead may have many different characteristics, these may be used as design levers to both increase information flows and minimize transaction costs. Also, these different governance forms shape actors’ attention, which further impacts their perceptions in these transactions, bringing in the attention-based view (Ocasio, Reference Ocasio1997).

Tradeoffs between Informational Benefits and Transaction Costs. Finally, there are likely tradeoffs between optimizing information flow and minimizing transaction costs. There is a huge opportunity to examine when these would occur and the most effective solutions to address them. There is also a significant prospect to examine how to potentially shape managers’ or employees’ attention to avoid these tradeoffs.

The Who?

Most work in behavioral strategy focuses on top managers’ decision-making, whereas applications to decision-making at lower levels are less common. However, decision-making at lower levels may impact the phenomena that behavioral strategy researchers are investigating, notably because such decisions may aggregate to firm-level outcomes (e.g., Foss & Weber). Such aggregation may happen through social interaction. Relatedly, a nascent microfoundational stream in strategy research encourages a focus on interfaces – that is, “instances in which individuals’ attributes, aspirations, and activities influence one another” – as germane for explicitly considering “both sides of the ledger” in terms of the actors involved” and their proximate and distal organization-level outcomes (Simsek, Heavey, & Fox, Reference Simsek, Heavey and Fox2018: 284). Indeed, interfaces are not only the primary means for strategic leaders to carry out their daily work but also the main conduit for bottom-up influence. For example, Dutton and Ashford (Reference Dutton and Ashford1993) show how middle managers’ issue-selling efforts focus senior managers’ attention and shape their understanding of strategic issues. and teams outside of the TMT, but catalyzed by the interactions of managers at different levels. Huy, Corley, and Kraatz (Reference Huy, Corley and Kraatz2014) shed light on how the emotions of middle-level managers can play a significant role in fostering or hampering the implementation of strategic changes, thus potentially affecting firm adaptation and performance. Sonenshein (Reference Sonenshein2010) highlights how managers’ and employees’ narratives about strategic change interact to either facilitate or hinder implementation.

In sum, framing the “who” from an interface perspective is likely to trigger questions about what intersects at the interface (e.g., goals, cognitions, emotions). It also prompts investigation into the type of influence at the interface (motivational, cognitive, behavioral), its direction (up, down, lateral), and its pattern (constraining, enabling) (Simsek et al., Reference Simsek, Heavey and Fox2018). This examination offers opportunities to understand which interactions, among whom, shape firm strategies, influence decision-making processes, and ultimately drive firm performance.

The What?

Our discussion of the behavioral strategy landscape so far suggests that existing perspectives largely explore the cognitive side of bounded rationality (Gigerenzer & Goldstein, Reference Gigerenzer and Goldstein1996). Yet, even this topic has not been fully examined, suggesting that there is still significant potential for research in this area. Moreover, noncognitive elements including motivations, goals, emotions, imagination, and predispositions, have not been incorporated into behavioral strategy research. We examine the potential for incorporating these aspects into behavioral strategy research below.

Strategic Beliefs. In traditional decision-making models, beliefs are estimates of the payoffs associated with the decision alternatives. Some scholars suggest that explicitly integrating them into the existing theories may lead to new firm-level predictions, such as search in response to increasing performance (rather than decreasing, as traditionally posited) (e.g., Keil, Posen, & Workiewicz, Reference Keil, Posen and Workiewicz2023). Belief updating is another area for potential work in behavioral strategy research. While research on Bayesian approaches to belief updating (e.g., Cyert & DeGroot, Reference Cyert and DeGroot1974; Daw et al., Reference Daw, Gershman, Seymour, Dayan and Dolan2011) could be extended to predict firm behaviors and outcomes, it would not solve the basic problem that Bayesianism cannot address unforeseen events, surprises, and novelty (Ehrig & Foss, Reference Ehrig and Foss2022). Moreover, as March and Olsen point out, individual beliefs and models of the world are not always tied to experience and objective reality – “individuals, as well as organizations or nations, develop myths, fictions, legends, and illusions.” (1975: 154). In particular, ideologies often serve as lenses through which information is interpreted, guiding how it is assimilated and beliefs are updated (Semadeni, Chin, & Krause, Reference Semadeni, Chin and Krause2022). This interpretation process can, therefore, lead to belief updating that is less about Bayesian adjustments and more about reinforcing existing frameworks and worldviews (Staw & Ross, Reference Staw and Ross1980). Further exploration into the nuances of belief formation and updating in organizations, especially those not rooted in direct experience, may offer valuable insights for behavioral strategy. Indeed, such an exploration would allow for a more dynamic and comprehensive understanding of experience- vs. non-experience-based belief updating. This distinction may make beliefs more or less likely to change, and therefore, have a greater or lesser influence on strategic decisions and competitive outcomes.

Motivations. Motivational issues in general, and employee motivation in particular, are under-explored research avenues in behavioral strategy (with some exceptions, such as Bridoux, Coeurderoy, & Durand, Reference Bridoux, Coeurderoy and Durand2011; Lindenberg & Foss Reference Lindenberg and Foss2011). However, there are many opportunities for linking motivations at various levels in the firm to key strategic issues. The current dominant motivational theory is self-determination theory (Deci & Ryan, Reference Deci and Ryan2012: 416) which theorizes a spectrum of motivational drivers. Future behavioral strategy research may explore how this fine-grained theory of motivation can be linked to organizational characteristics, different governance instruments and ways of managing, and different aggregate outcomes (e.g., competitive advantage).

Complementing intrinsic and prosocial motivations is “idealism,” that is, the pursuit of lofty, often unbounded goals, unconstrained by conventional notions of practicality. This form of idealism, as Schilling (Reference Schilling2018: 338) notes, can lead to the formation of superordinate goals, driving individuals towards higher aspirations. Such an idealistic pursuit within organizational contexts can, for example, lead to employee resilience, a relatively unexplored but potentially important source of sustained competitive advantage.

Goals. Combining motivation and cognition, goal frames are an important construct for understanding the microfoundations of value creation, a key concept in strategy research. The basic notion is that behavior is influenced by higher-level goals that, when salient, steer an individual’s motivation and cognition. Goal-framing theory proposes different goal frames that are differentially salient to individuals. These goal frames are linked to specific individual behaviors, and also shape firm behaviors in the aggregate (Foss & Lindenberg, Reference Foss and Lindenberg2013; Lindenberg & Foss, Reference Lindenberg and Foss2011).

New research on goal framing may provide more detail on why, how, and when specific goal frames become dominant at the organizational level, fostering collective commitment toward certain choices or actions. Expanding this line of inquiry could involve further exploring the interactions between goal frames and other organizational elements, such as leadership, organizational design, and culture. In particular, in the formation and reinforcement of goal frames that are beneficial to an organization’s performance, the interplay of organizational culture, leadership styles, and broader cultural norms is likely to be crucial. A culture that embodies core values such as justice, equity, and fairness may increase the salience of goal frames conducive to value creation. Leadership styles, particularly those exemplifying servant leadership which focuses on the needs and emotional well-being of employees (cf. Wu et al., Reference Wu, Liden, Liao and Wayne2021) may lead to the same outcome. Yet, there is also a possibility that lower-level employees may adopt and champion beneficial goal frames, independent of direct managerial influence. Here, organizational identification (Section 3) might play a role. Future research should also examine the individual personality traits and attributes that encourage organizational members to support and spread specific goal frames.

Additionally, there has been work on dispositional and situational regulatory focus (Higgins, Reference Higgins1997, Reference Higgins1998) and its impact on interfirm relationships (Weber & Mayer, Reference Weber and Mayer2011), learning (Weber, Reference Weber2017), acquisition decisions (Gamache et al., Reference Gamache, McNamara, Mannor and Johnson2015), and firm stakeholder strategies (Gamache et al., Reference Gamache, Neville, Bundy and Short2020). This research examines how prevention- versus promotion-framed goals impact top management and lower-level managers’ attention, emotions, behaviors, and strategic decisions, which influence firm-level outcomes. Again, this work has only begun to examine the impact of regulatory focus theory on firm performance, creating an opportunity for future research in this area.

Emotions. Despite emotions driving motivation, cognition, and behavior, and acting as a distinct influence in social interaction (Huy, Reference Huy2012; van Knippenberg & van Kleef, Reference Van Knippenberg and Van Kleef2016) that may affect organizational outcomes (Healey & Hodgkinson, Reference Healey and Hodgkinson2017), behavioral strategy research disproportionally focuses on cognition. If emotions are considered, they are typically incorporated as impediments to effective decision-making (cf. Damasio, Reference Damasio1994; Rafaeli & Worline, Reference Rafaeli and Worline2001). However, more recent work showed that emotions may facilitate (interactive) decision-making in some contexts (e.g., Ehrig et al., Reference Ehrig, Manjaly, Singh and Sunder2022; Healey & Hodgkinson, Reference Healey and Hodgkinson2017; Healey et al., 2018; Vuori & Huy, Reference Vuori and Huy2020).

Leadership research shows how leaders’ affective displays – that is, expressions of leader mood or emotion observable to followers – motivate and mobile organizational members through emotional contagion (Elfenbein, Reference Elfenbein2014; Visser et al., Reference Visser, van Knippenberg, Van Kleef and Wisse2013. Besides leaders’ emotions, organizational members’ emotions also play a crucial role in organizational change (Bartunek, Balogun, & Do, Reference Bartunek, Balogun and Do2011). While most of these studies focus on employees’ resistance to change and thus on managing and mitigating such resistance (Ford, Ford, & D’Amelio, Reference Ford, Ford and D’Amelio2008; Kiefer, Reference Kiefer2005; Reger et al., Reference Reger, Mullane and Gustafson1994), recent work provides evidence that more positive emotions (including happiness, enjoyment, and enthusiasm) play a catalytic role in accelerating change (Bartunek et al., Reference Bartunek, Balogun and Do2011; Kiefer, Reference Kiefer2002; Liu & Perrewé, Reference Liu and Perrewé2005).

Imagination. Although behavioral strategy research overlaps with fields like entrepreneurship and innovation, forward-looking, creative, or imaginative thinking tends to lie outside of the main research program (see Gavetti, Reference Gavetti2012 and Winter, Reference Winter2012). The reason arguably is that dominant approaches in behavioral strategy, such as the behavioral theory of the firm (Cyert & March, Reference Cyert and March1963), focus on the role of the past in shaping the future. Thus, the emphasis has been primarily on standard operating procedures, routines, organizational memory, and myopic, local search to explain why past firm behavior is the best predictor of future firm behavior. Another major approach to behavioral strategy, behavioral decision theory, assumes (like standard decision theory) that choice alternatives are essentially given. How actors generate (new) solution spaces under conditions of deficient knowledge and experience has generated less interest (e.g., Feduzi & Runde, Reference Feduzi and Runde2014). This is surprising given that the construction of (new) decision situations is a key aspect of decision-making in March and Simon (Reference March and Simon1958).

The imaginative aspect of strategic decision-making is starting to garner renewed interest (e.g., Alvarez & Porac, Reference Alvarez and Porac2020: 739). However, how decision-makers extract relevant information from the past and use it to predict and evaluate possible futures is not very clear. The emerging theory-based view in behavioral strategy (e.g., Felin & Zenger, Reference Felin and Zenger2017; Rindova & Martins, Reference Rindova and Martins2021) posits that decision-makers form theories much like scientists, which is anticipated in the psychology literature in George Kelly’s “personal construct” theory (Kelly, Reference Kelly1970). The theory-based view characterizes valuable theories as novel, simple, falsifiable, and generalizable (Felin & Zenger, Reference Felin and Zenger2017), but does not offer further insight into how they are generated. The “conviction narrative theory” associated with David Tuckett (e.g., Johnson, Bilovich, & Tuckett, 2023) goes somewhat further. It suggests the psychological foundations of theories are “conviction narratives” that identify beliefs about relevant future variables and the mechanisms that link them. This perspective also shows how such narratives are subject to simulation, affective evaluation, and communication (cf. Rindova & Martins, Reference Rindova and Martins2022). As such, both the theory-based view and conviction narrative theory align with an older “scenario” tradition in strategy (see also Feduzi & Runde, Reference Feduzi and Runde2014), which has made limited contact with behavioral strategy.

Predispositions. The notion of predisposition emphasizes the nonreflective aspect of human agency (Cardinale, Reference Cardinale2018), and refers to the impromptu mobilization of skills in response to evolving situations, which is typically transmitted unconsciously via social practices. Nayak and colleagues further suggest predispositions are microfoundations of firm dynamic capabilities, as they provide a “tacitly honed capacity for improvisatory adaptive action that is unconsciously acquired in situ through extensive immersion in changing environmental conditions” (2020: 282).

The notion of predisposition helps in better understanding how firm constituents perceive and respond to environmental cues, processes that reflect perceptual sensitivities honed through prolonged exposure to both internal organizational dynamics and external environmental factors (Nayak et al., Reference Nayak, Chia and Canales2020). This view also underscores the importance of focusing on interfaces as opportunities for the unconscious transfer and expansion of predispositions. Shedding light on how interfaces facilitate the transfer and adaptation of predispositions can offer profound insights into how firms develop their dynamic capabilities and, thus, sustained competitive advantage.

The Where and When?

Routines. In Section 1, we mentioned a shift in the understanding of organizational routines, moving away from a view of them as static entities to dynamic, evolving processes (e.g., Feldman & Pentland, Reference Feldman and Pentland2003). This evolution underscores the importance of human agency in renewing and adapting routines, challenging earlier notions of routines as mere repetition (Feldman, Reference Feldman, Howard-Grenville, Rerup, Langley and Tsoukas2016; Parmigiani & Howard-Grenville, Reference Parmigiani and Howard-Grenville2011).

While the study of routine dynamics traditionally focused on actions and their associations (Pentland et al., Reference Pentland, Feldman, Becker and Liu2012), actions are inseparable from the agents performing them (D’Adderio & Pollock, Reference D’adderio and Pollock2014). As such, routines and routine dynamics provide fertile ground for behavioral strategy research, as they represent the main locus of interfaces. In this sense, behavioral strategy could both draw from and augment routine dynamics research by delving into routines as points of confluence for multiple agents’ predispositions, cognitions, motivations, and actions (Weick & Roberts, Reference Weick and Roberts1993). Investigating routines as conduits of influence among various agents can clarify the relation between the action networks within routines and the individual actors involved. Furthermore, examining the intrinsic and extrinsic motivational factors, (beyond mere feedback mechanisms) that drive routine (re)production could illuminate when routines are more susceptible to change, which in turn affects organizational performance and adaptability (Abell, Felin, & Foss, Reference Abell, Felin and Foss2008).

Artificial Intelligence. Simon (1995) not only developed the concept of bounded rationality but was also one of the pioneers in AI research in the 1950s. He believed that human behavior should be studied not just with psychology, but also modeled using computers (cf. also Newell & Simon, Reference Simon and Newell1976). Today firms are increasingly using artificial intelligence (AI) – machines that perform cognitive functions typically associated with human minds (Nilsson, Reference Nilsson1971) – across a range of operational and managerial tasks. The integration of AI is ushering in new forms of agency and new interfaces, redefining the boundaries of bounded rationality and transforming firms’ experiential, structural, and relational systems in unprecedented ways (Kemp, Reference Kemp2023; Murray, Rhymer, & Sirmon, Reference Murray, Rhymer and Sirmon2021).

From a cognitive perspective, AI-based algorithms are supercarriers of formal rationality because they can both optimize and improve themselves at an unprecedented speed (Lindebaum, Vesa, & den Hond, Reference Lindebaum, Vesa and Den Hond2020). Delegating decision-making authority to AI can thus reduce firms’ information processing costs resulting from the creation of vertical and horizontal information structures and the incentives to diminish information asymmetries. Moreover, AI has a significantly greater attentional capacity than humans, allowing for the simultaneous pursuit of multiple goals and, thereby, reducing the need to rely on priority rules for attention allocation (von Krogh, Reference Von Krogh2018). On the other hand, AI may exacerbate certain cognitive biases such as myopia (Balasubramanian, Ye, & Xu, Reference Balasubramanian, Ye and Xu2022), and illusory causation (cf. Kiesler & Sproull, Reference Kiesler and Sproull1982). Delegating decision-making authority to AI can also hinder humans’ reliance on perceptual sensitivity for discriminating between environmental stimuli and, thus, their capacity for improvisatory adaptive action (Daugherty & Wilson, Reference Daugherty and Wilson2018).

The evolving role of AI in mediating human interactions within organizations offers a novel area of investigation. AI systems, acting as intermediaries in communication processes, have the potential to significantly influence the flow and interpretation of information. They can also play a crucial role in governing sensemaking and social influence processes. For instance, AI systems can promote high-quality, “heedful” interrelations among firm constituents, fostering a shared understanding and facilitating concerted action (cf. McKee et al., Reference McKee, Tacchetti and Bakker2023; Shirado & Christakis, Reference Shirado and Christakis2020). Additionally, even though AI systems do not possess emotions, their ability to emulate human emotional responses can be instrumental in spreading positive emotions during crisis situations (cf. Han, Yin, & Zang, Reference Han, Yin and Zhang2023; Prinz, Reference Prinz2022). However, there is a risk that AI might lead to increasingly uniform mental representations that suppress the requisite variety needed for adaptive shifts in information processing styles (cf. Steinbach, Gamache, & Johnson, Reference Steinbach, Gamache and Johnson2019), potentially hindering firms’ adaptability and eroding competitive advantage (Kemp, Reference Kemp2023).

Beyond their cognitive influence, AI and related technologies can also serve as motivational drivers. For instance, AI-driven performance monitoring systems can influence employee behavior by providing real-time feedback, thus fostering a more dynamic and responsive work environment (Tong et al., Reference Tong, Jia, Luo and Fang2021). This aspect raises important questions about the potential of AI to facilitate both individual and organizational learning. However, complete delegation to AI systems might lead to reduced employee engagement and participation as well as encourage opportunistic behaviors, as employees could exploit the automated systems for personal gain, potentially at the expense of organizational goals. Such scenarios underscore the importance of maintaining a balance between AI automation and human involvement, ensuring that employees’ accountability is preserved through some level of involvement with and control over relevant tasks (Raisch & Krakowski, Reference Raisch and Krakowski2021).

Network and Connections beyond the Organizational Boundaries. There is increasing recognition of the significant role played by social relationships and interfaces beyond firm boundaries in shaping strategists’ decision-making and actions, and, as a consequence, firm behavior and outcomes (Tasselli et al., Reference Tasselli, Kilduff and Menges2015; Westphal & Zajac, Reference Westphal and Zajac2013). The extensive literature on board interlock ties, where board members concurrently serve on multiple corporate boards, shows that such ties act as conduits for social learning and the diffusion of strategies, practices, and norms between firms (e.g., Haunschild, Reference Haunschild1993; McDonald & Westphal, Reference McDonald and Westphal2003;). Relationships between firm leaders also were found to increase the likelihood of cooperative strategies (Gulati & Westphal, Reference Gulati and Westphal1999).

Furthermore, social networks play a crucial role in shaping individual motivations and emotions (Burkhardt, Reference Burkhardt1994). For instance, individuals who are surrounded by pro-social (self-interested) others are more likely to become pro-social (self-interested) (Christakis & Fowler, Reference Christakis and Fowler2013; Tsvetkova & Macy, Reference Tsvetkova and Macy2015); and individuals who are surrounded by happy others are more likely to become happy (Fowler & Christakis, Reference Fowler and Christakis2008). These networks also contribute to social identity formation, fulfilling fundamental human needs for belonging and serving as a foundation for identification (Walker & Lynn, Reference Walker and Lynn2013).

Recent research also focused on the psychological processes underpinning tie formation and their effective exploitation. Active pursuit or avoidance of certain relationships by individuals (Tasselli et al., Reference Tasselli, Kilduff and Menges2015), and the interplay between network position and motivation, significantly impact both individual and firm-level outcomes (Tasselli et al., Reference Tasselli, Kilduff and Menges2015). For instance, Reinholt, Pedersen, and Foss (Reference Reinholt, Pedersen and Foss2011) found that individuals’ knowledge acquisition and provision were highest when both network centrality and autonomous motivation were high. This indicates that while structural position in a network is crucial, it is motivation that allows for leveraging network positions effectively (cf. Vissa, Reference Vissa2010). Shedding further light on the motivational drivers leading specific firm participants to coalesce may also help to enhance our understanding of whether and how distributed and shared cognition may coexist and interact (Stevenson & Greenberg, Reference Stevenson and Greenberg2000). While we know shared cognitions may not be a necessary condition for coalescence (e.g., Fiol, Reference Fiol1994) (see Section 3), the extent to which mental models need to be shared among participants to allow them to form a coalition has not been examined.

Furthermore, trust, as an output of microlevel interactions within these networks (cf. Emerson, Reference Emerson1976), acts as a resource that can be exchanged to build and modify higher-level structures. Within firms, trust can serve as a complementary or substitute governance mechanism to traditional authority (Adler, Reference Adler2001; Westphal & Zajac, Reference Westphal and Zajac2013). At the same time, it may also undermine authority and reduce monitoring and supervision, particularly when arising from informal ties between superiors and subordinates (e.g., CEO and board members) (Fredrickson, Hambrick, & Baumrin, Reference Fredrickson, Hambrick and Baumrin1988). Overall, by situating interfaces into networks of structural ties beyond firm boundaries, behavioral scholars may gain a better understanding of the origins of strategists’ emotions, motivations, and cognitions and formulate better predictions regarding their subsequent impact on firm behavior and outcomes.

Uncertainty. Much work on behavioral decision theory is predicated on the existence of “uncertainty,” and this is largely true of behavioral strategy. Thus, individuals and firms have to make choices among multiple alternatives with different outcomes (Tversky & Kahneman, Reference Tversky and Kahneman1974; Simon, Reference Simon1955), search “landscapes” they may know little about (Levinthal, Reference Levinthal1997), and construct representations of those landscapes and their decision situations more generally (Levinthal, Reference Levinthal2011). However, it is not always clear what “uncertainty” means in behavioral strategy. One reason is that scholars are often not forthcoming about the kind of uncertainty, or, more broadly, the lack of knowledge assumed. For example, a definition of uncertainty is seldom explicitly mentioned in the search literature derived from the behavioral theory of the firm (e.g., Greve, Reference Greve1998), although models that invoke “novelty” (e.g., Gavetti, Levinthal, & Rivkin, Reference Gavetti, Levinthal and Rivkin2005) seem to implicitly suggest that such search takes place under conditions of partial ignorance about the future as well the present “landscape.”

Indeed, as one surveys the behavioral strategy literature, everything from subjective probability (in Bayesian as well as non-Bayesian forms) to ambiguity to sheer ignorance has been labeled as “uncertainty.” For example, some key contributions to behavioral decision theory (Tversky & Fox, Reference Tversky and Fox1995) explicitly assume or posit conditions of risk. For example, much of the prospect theory literature shows how framing gambles differently (notably, loss versus gain) influences how risks are assessed, impacting choice behavior. Indeed, Gerd Gigerenzer argued that much of the heuristics and biases literature is fundamentally based on standard rational decision-making models, albeit with tweaks and add-ons (e.g., Gigerenzer, et al., Reference Gigerenzer, Hertwig and Pachur2011) Still, many others have used reasoning derived from such models to approach behavior under even radical uncertainty (i.e., decision-makers cannot describe future state spaces) (e.g., Feduzi & Runde, Reference Feduzi and Runde2014; Mousavi & Gigerenzer, Reference Mousavi and Gigerenzer2014).

Perhaps the reason behavioral strategy research has not typically clearly defined uncertainty is that it focuses on behaviors. Thus, it may loosely, but plausibly, assume that ambiguity leads to behaviors that reflect ambiguity, and that uncertainty leads to analogical reasoning in strategy-making (Gavetti et al., Reference Gavetti, Levinthal and Rivkin2005). However, the microfoundations for such assumptions are not always clear. For example, there is much more to know about the psychological mechanisms producing analogies under different circumstances, or the extent to which cognitive bias findings derived in probabilistic settings are transferable to non-probabilistic situations (i.e., identifying decision alternatives and their outcomes is difficult).

These problems in behavioral strategy seem to stem from the lack of a clear understanding of decisions and behaviors under what is often called “Knightian uncertainty.” Of course, behavioral strategy is not alone in lacking a clear understanding of Knightian uncertainty; the problem is a more general one in social science where multiple interpretations of such uncertainty have been proffered (Foss, Reference Foss2023). However, behavioral strategy is championed because it attempts to grapple with real behaviors in realistic decision-making situations in a strategic context. If “Knightian uncertainty” means that our knowledge about the future often, perhaps usually, eludes quantification, it is incumbent upon behavioral strategy scholars to address such uncertainty.

While decision science and economics have developed several models of Knightian uncertainty, many of these are essentially tweaks of the basic Bayesian model of decision-making (Foss, Reference Foss2023). Game theorists are also experimenting with modeling “unawareness” (i.e., there are parts of the state space that decision-makers are initially unaware of). However, none of these approaches make room for the role of imagination, stressed above as an important future research avenue in behavioral strategy. As argued in the section on Imagination (page 63), when the future does not exist, decision-makers exercise their imagination to be able to evaluate the desirability of potential outcomes. Such imagination often emerges as socially embedded narratives that respond to the existence of uncertainty in terms of including sense-making, outlining possible actions, and criteria for evaluating the future (Johnson, Bilovich, & Tuckett, Reference Johnson and Tuckett2023).

The Why and How?

As argued, most research in behavioral strategy is influenced heavily by Simon’s conception of individuals as information processors, emphasizing organizational rules and routines to reduce complexity and treating organizations as unitary entities (e.g., Levinthal, Reference Levinthal1997). This has resulted in a general lack of clarity with respect to the mechanisms that connect microlevel psychological aspects to firm-level behaviors and outcomes (cf. Christensen et al., Reference Christensen, Dahl, Knudsen and Warglien2021). Still, recent work has made significant strides in understanding the aggregation of individual choices and their impact on firm behavior and performance (see Cialdini & Goldstein, Reference Cialdini and Goldstein2004).

Such research largely focuses on intended aggregation mechanisms, explicitly designed to align individual motivations, cognitions, and actions to enhance firm performance and competitive advantage, and including compliance via formal governance structures, social influence via power and communication, and political processes (e.g., Christensen & Knudsen, Reference Christensen and Knudsen2010; Csaszar & Eggers, Reference Csaszar and Eggers2013; Piezunka, Aggarwal, & Posen, Reference Piezunka, Aggarwal and Posen2022). “The basic premise in this line of research is that the organization has a fundamental role in aggregating choice functions, which characterize the organizational members’ ability to pass judgment.” (Christensen et al., Reference Christensen, Dahl, Knudsen and Warglien2021: 2). For instance, some work examines the effectiveness of different governance forms to mitigate the transaction costs ensuing from differences in participants’ mental models as well as participants’ self-interest (e.g., Foss & Weber, Reference Foss and Weber2016) (Section 4). Studies also explore decision structures that promote endogenous adaptation and desired aggregate outcomes (e.g., Piezunka & Schilke, Reference Piezunka and Schilke2023), revealing that decentralized decision structures may engender unintended firm-level consequences and lead to suboptimal firm outcomes, even when individuals are unbiased (Christensen et al., Reference Christensen, Dahl, Knudsen and Warglien2021; Piezunka & Schilke, Reference Piezunka and Schilke2023). Furthermore, certain decision rules, such as averaging participants beliefs, albeit being particularly effective at aggregating individual knowledge into firm decisions, may ultimately be ineffective at fostering individual learning (Piezunka et al., Reference Piezunka, Aggarwal and Posen2022) and participation (Piezunka & Schilke, Reference Piezunka and Schilke2023).

Social norms may also have a role in connecting the micro and macro levels. Individuals often rely on social norms for guidance, especially in ambiguous situations where formal rules are not applicable (Cialdini, Reference Cialdini2001). The influence of norms, ranging from subtly shaping behaviors to driving overt compliance, hinges on two factors: the prominence of these norms in an individual’s consciousness and the alignment between different types of norms (Cialdini, Reference Cialdini2003). Existing studies suggest that norms influence individual behavior when they are the focus of attention (cf. Hargadon, & Bechky, Reference Hargadon and Bechky2006). This happens when individuals are exposed to a certain stimulus for prolonged periods of time (Keizer, Lindenberg, & Steg, Reference Keizer, Lindenberg and Steg2013) and intermittently (Bernstein, Shore, & Lazer, Reference Bernstein, Shore and Lazer2018). For instance, Weber and Murnighan (Reference Weber and Murnighan2008) demonstrate that when a single individual consistently exhibited prosocial behaviors, members came to perceive the norm as cooperative, becoming more willing to contribute themselves. This, however, implies that norms may explain “aggregation” only when they are consciously and prominently considered by individuals (Cialdini & Goldstein, Reference Cialdini and Goldstein2004).

Social psychology literature offers relevant insights into unintended aggregation mechanisms that operate subconsciously, aligning individuals’ motivations, emotions, and behaviors without deliberate design. Mechanisms like social contagion (i.e., the spontaneous distribution of ideas, emotions, attitudes, and behaviors among larger groups of people (Christakis & Fowler, Reference Christakis and Fowler2013)) automatic activation through priming (i.e., information in the environment that makes related mental content accessible in a way that can potentially be used to guide judgment and behavior (Chartrand & Bargh Reference Chartrand and Bargh1999; Chen et al., Reference Chen, Latham, Piccolo and Itzchakov2021; Loersch & Payne, Reference Loersch and Payne2014)), and belief synchronization (i.e., an inherent tendency to select popular alternatives over unpopular ones (Denrell & Le Mens, Reference Denrell and Le Mens2017)) all subtly shape individuals’ decisions and actions in a way that aggregates to organizational-level outcomes. Future research in behavioral strategy could delve into understanding these mechanisms and their interplay, which has often been overlooked. The next step of the behavioral strategy roadmap will consist of linking strategic choices, such as momentum, exploitative and explorative strategies, imitative strategies, and anticipatory strategies (e.g., forbearance) (cf. Greve, Reference Greve2013; Andrevski & Miller, Reference Andrevski and Miller2022), to their performance outcomes and understanding the role played by the aggregation mechanisms as drivers of such relationships.

Multi-agent Experiments

Behavioral strategy scholars are increasingly adopting experimental methods as primary data-gathering and analysis techniques (Levine et al., Reference Levine, Schilke, Kacperczyk and Zucker2023). Experiments offer multiple advantages, as they (1) are highly complementary to traditional modeling techniques in behavioral strategy (Billinger et al., Reference Billinger, Stieglitz and Schumacher2014); (2) provide evidence of causal relationships (Greenberg & Tomlinson, Reference Greenberg and Tomlinson2004); and (3) explicate materials and instructions, easing replication (Croson, Anand, & Agarwal, Reference Croson, Anand and Agarwal2007). Furthermore, with the growth of online crowdsourcing sites, such as Amazon Mechanical Turk and Prolific, conducting experiments has become more scalable. Many consider online experiments just as valid as lab experiments, even offering advantages such as subject diversity, market-like conditions, and the feasibility of longitudinal studies (Horton, Rand, & Zeckhauser, Reference Horton, Rand and Zeckhauser2011; Paolacci, Chandler, & Ipeirotis, Reference Paolacci, Chandler and Ipeirotis2010; Rand, Reference Rand2012). Yet, most online experiments involve very simple and often noninteractive tasks done independently (Giamattei et al., Reference Giamattei, Yahosseini, Gächter and Molleman2020), raising concerns about the generalizability of findings to real-world settings (Levitt & List, Reference Levitt and List2008).

New platforms allowing for synchronous and longitudinal experiments may address this issue (e.g., Mao et al., Reference Mao, Chen and Gajos2012), particularly when tasks and interactions are designed to enable participants’ socialization, thus mimicking virtual work in actual organizations. Despite some criticism regarding the lack of realism and fidelity, virtual reality also has the potential to augment experimental settings, aligning them more closely with the organizational context under study (Brookes et al., Reference Brookes, Warburton, Alghadier, Mon-Williams and Mushtaq2020; Hubbard & Aguinis, Reference Hubbard and Aguinis2023).

In choosing the experimental approach, behavioral scholars must carefully consider the trade-offs between accuracy and generalizability. This decision is compounded by the need to assess the applicability of findings on bounded rationality in strategic decision-making, particularly outside the controlled confines of laboratory settings (see Jussim, Reference Jussim2012). Field experiments may offer greater generalizability, especially when participant sampling allows for broader extrapolation. On the other hand, complementing online experiments with field-based case studies or subsequent firm-level data can complement the advantages of experiments with the richness of field data, offering realistic insights into the psychological underpinnings of firm competitive advantage.

Conversations, Narratives, and Natural Language Processing

The study of interaction and what we call “interfaces” above in behavioral strategy may also be furthered by conversation analysis, which uses management meeting textual data to analyze the situated processes of decision-making (Cooren, Reference Cooren2013; Cooren et al., Reference Cooren, Vaara, Langley and Tsoukas2014). This methodology hinges on the intertextual relationships between utterances (i.e., who says what, when, and how), in terms of such aspects as topic continuity, argumentative progression, and rhetorical devices, to gain a better understanding of what happens at the interface (e.g., König et al., Reference König, Mammen, Luger, Fehn and Enders2018; Tuggle, Schnatterly, & Johnson, Reference Tuggle, Schnatterly and Johnson2010). For instance, the application of conversation analysis has uncovered multiple discursive tactics that strategists employ to drive decision-making (Wodak et al., Reference Wodak, Kwon and Clarke2011).

Although this method has been largely adopted from a qualitative (ethnographic) perspective, the burgeoning field of computational linguistics – particularly machine learning and deep learning techniques for the analysis of textual data – presents opportunities to extend the scope and enhance the inferential reproducibility of conversation analysis. This expansion into computational techniques allows for scaling up the analysis to include a broader range of textual data beyond management meetings, such as text messages, emails, memos, presentation slides, etc. By incorporating interactions with organizational members at various hierarchical levels, this approach can provide a more comprehensive understanding of interactive processes at all organizational tiers and their cumulative impact on strategic decisions. Moreover, sentiment analysis of these conversations is an essential addition. By evaluating the emotions behind individual utterances, the combination of conversation and sentiment analysis can provide rich insights into firm-level implications of the interplay between participants’ affective states, discursive tactics, and linguistic devices.

Earlier in this section, we pointed to the importance of narratives as key mental representations to imagine plausible futures and, thus, as means for sensemaking and decision-making under radical uncertainty (Chong & Tuckett, Reference Chong and Tuckett2015). Topic modeling is a promising method to augment traditional narrative analysis, treating stories as the main unit of analysis (Riessman, Reference Riessman1993), and aiming at uncovering “semantically cohesive topics and their combination across document collections.” (Evans & Aceves, Reference Evans and Aceves2016: 32). Algorithmically derived topic structures tend to be more transparent, reproducible, and less biased than pure qualitative analyses based on individual researchers’ reading. Nevertheless, the structures generated by topic modeling are not self-explanatory and require careful interpretation. To enhance both reproducibility and interpretability, a promising approach is to present selections of prototypical text from the corpus corresponding to each identified topic, which may allow them to engage directly with the language used by individuals, facilitating a deeper understanding and a form of triangulation (Marchetti & Puranam, Reference Marchetti and Puranam2023). Additionally, integrating sentiment analysis into topic modeling could further refine the understanding of narratives. Sentiment analysis helps detect the emotional underpinnings within the narratives, offering insights into the mood and attitudes of the narrators. This combination can be especially powerful in understanding how strategists’ narratives express more or less positive futures, shape organizational culture, and, in turn, influence decision-making processes.

Finally, algorithmic-aided text analysis is now regularly used in measuring various psychological constructs such as CEO motivation (regulatory focus) (Gamache et al., Reference Gamache, McNamara, Mannor and Johnson2015), temporal focus (Nadkarni & Chen, Reference Nadkarni and Chen2014), justification content (Wade, Porac, & Pollock, Reference Wade, Porac and Pollock1997), and attention (Cho & Hambrick, Reference Cho and Hambrick2006), primarily through the analysis of letters to shareholders in annual reports of publicly traded companies. The advent of sophisticated unsupervised machine learning techniques, including word embeddings and large language models, heralds new possibilities for measuring socio-cognitive constructs with greater reliability and accuracy. Word embeddings, which include methods like bag-of-words models, enable the extraction of semantic patterns from large text corpora, potentially revealing relationships between the cognitive frameworks of different organizational members (cf. Piezunka & Dahlander, Reference Piezunka and Dahlander2015). By quantitatively analyzing word relationships in vector spaces, word embeddings can uncover implicit cognitive maps that guide organizational decision-making and strategy formulation, as well as provide insights into the collective cognitive alignment or divergence within an organization (cf. repertory grid techniques in Hodgkinson, Wright, and Anderson (Reference Hodgkinson, Wright and Anderson2015)). Furthermore, large language models, such as GPT-4, are now capable of generating measures of semantic typicality (i.e., the degree of similarity of a text document to a concept) that closely mirror human judgment, which allows for considerable refinement with respect to measuring concepts like ambiguity, diversity, extremity, and polarization (Le Mens et al., Reference Le Mens, Kovács, Hannan and Pros2023).

Visual Methods and Image Recognition

Nonparticipatory observation has been one of the most widely adopted methods in behavioral and social sciences where systems of categories have been developed to interpret and classify vocal, verbal, and nonverbal aspects of behaviors in social contexts, including static visual characteristics (e.g., race, sex, and facial appearance) and dynamic ones (e.g., gestures, facial expressions, and body language). Research suggests that visual cues significantly influence individual attention allocation, memory encoding, and decision-making processes (Mehta & Zhu, Reference Mehta and Zhu2009; Tsay, Reference Tsay2021; Tversky, Reference Tversky1974), and that affective displays and subsequent emotional contagion were typically more salient through sight than other senses (Jack et al., Reference Jack, Sun, Delis, Garrod and Schyns2016).

Nevertheless, compared to textual data analysis methods, approaches to analyze and interpret visual content are relatively less developed (Bell & Davison, Reference Bell and Davison2013). Traditional visual content analysis relies on manual coding, leading to scalability and reliability issues. The advent of deep learning techniques in image processing and recognition presents opportunities for using visual data in behavioral strategy research. These advanced algorithmic methods offer scalability and reproducibility in analyzing visual content, surpassing the limitations of manual coding. Future studies in behavioral strategy could rely on these techniques to understand the types of visual cues that may provide information to facilitate effective decisions versus those that introduce errors in decision-making, as well as when these effects occur. Also, scholars could combine image and language processing models to investigate the salience of different sensory cues in judgment and decision-making processes under various situational circumstances (e.g., radical uncertainty) and whether primes or formal mechanisms can alter such salience (cf. Tsay, Reference Tsay2021). Another interesting avenue involves exploring how inconsistencies between different verbal and visual cues may influence dynamics at the interfaces. Such discrepancies can provide unique insights into the perceived authenticity and reliability of communicated messages, either facilitating or hindering organizational change. Thus, these methods offer a new approach to examining behavioral strategy questions that address some of the main issues typically faced in these studies.

Cognitive Maps and Geometric Spaces

In the late 1940s, Tolman discovered that animals assign locations to objects of experience in memory creating relationships between them in the form of a spatial maze (Tolman, Reference Tolman1948). Recent advancements show that similar map-like encoding mechanisms apply to nonspatial conceptual relationships, extending to imagination and abstract thinking (Horner et al., Reference Horner, Bisby, Zotow, Bush and Burgess2016; Kaplan, Schuck, & Doeller, Reference Kaplan, Schuck and Doeller2017; Schiller et al., Reference Schiller, Eichenbaum and Buffalo2015). The hippocampal-entorhinal complex plays a key role in the organization of memory and the formation and navigation of such cognitive maps (Stoewer et al., Reference Stoewer, Schilling, Maier and Krauss2023). These recent discoveries underscore the legitimacy and validity of geometric spaces and Euclidean distances to represent individual mental models and relationships between states in both empirical and computational studies in behavioral strategy (Hodgkinson et al., Reference Hodgkinson, Wright and Anderson2015). Intriguingly, this stream of work suggests mental maps are predictive: they represent each state in terms of its successor states (Stachenfeld et al., Reference Stachenfeld, Botvinick and Gershman2017). “Two states that predict similar future states will have similar representations, and two physically adjacent states that predict divergent future states will have dissimilar representations.” (Stachenfeld et al., Reference Stachenfeld, Botvinick and Gershman2017: 1). This raises some crucial questions about how, for instance, decision-makers navigate solution spaces. Indeed, the focus of traditional NK models has been on similarity in terms of spatial proximity, whereby decision-makers progress through solution spaces by making incremental changes to current alternatives, often varying only one dimension at a time (Section 2). The novel perspective provided by contemporary neuroscience suggests a notion of similarity that aligns more closely with the notion of equifinality; with important implications for our traditional understanding of attention allocation and search (cf. Parker & Witteloostuijn, Reference Parker and Witteloostuijn2010).