This review of cognition in strategic management research takes as its starting point the appreciation of the seminal paper, ‘Competitive groups as cognitive communities: the case of Scottish knitwear manufacturers’, by Porac, Thomas and Baden-Fuller on cognitive categorization of competition, published in the Journal of Management Studies only 20 years ago. In this paper, I reflect on the context in which their paper emerged, the impact it has had, and the future paths that research on cognition in strategy might take. In doing so, I highlight the challenges associated with establishing cognition as a legitimate factor in strategic management (alongside the traditional explanations of capabilities and incentives) and of showing the causal relationship between cognition and strategic outcomes. Subsequent work in cognition explored the dynamic relationship between cognition, capabilities, and incentives, and, in process models of framing, linked cognition with political action. Rather than managerial cognition becoming its own independent field, cognitive concepts have diffused throughout work in many different managerial fields, leading to a proliferation of terms, concepts, and approaches. I conclude by exploring some of the paths that research in cognition and strategy is taking in the present day – particularly those involving studies of the construction of markets and categories, each of which are themes that the work by Porac, Thomas and Baden-Fuller brought to our attention.
Just over 20 years ago, Joe Porac, Howard Thomas and Charles Baden-Fuller published their article on ‘Competitive groups as cognitive communities: the case of Scottish knitwear manufacturers’ (Porac et al., 1989) in a special issue of the Journal of Management Studies focused on ‘Managerial Thinking in Business Environments’. This paper proved to be an important contributor to the movement towards acknowledging the role of managerial cognition in shaping strategic outcomes and other organizational processes. The authors showed in their analysis of Scottish knitwear firms that rivalry was a cognitive construction based on the set of firms identified by managers as competitors.Their work broke with the strategic management field by suggesting that categorization processes played an important role in competition and that the environments in which firms operated were endogenous to the interpretations and actions of the managers in firms themselves.
In this paper, I analyse the impact of their publication on the field of strategic management and explore some paths that research in this arena has taken since the paper was published. Note that this article is accompanied in this issue of JMS by a piece by Porac et al. (2011), who reflect on the conversations in which their work has been engaged and provide a nice complement to the analysis I provide here. For my part, I highlight the challenges in establishing the legitimacy of cognitive explanations – both in creating valid measures and in showing the power of these theories alongside the traditional explanations in strategy associated with capabilities and incentives. In discussing the subsequent extensions of their ideas over the past 20 years, I show that the most powerful works sought to integrate cognition with these other explanations into contingent models of strategic management. Throughout the discussion I interrogate the methods used in cognitive research on strategy, showing what kinds of insights can be achieved using various methodological approaches. I close with some thoughts about promising future paths of research that build on the primary insights of Porac, Thomas and Baden-Fuller, focusing specifically on the resurgence of interest in categorization and the emergence of new fields.
COGNITIVE COMMUNITIES IN SCOTTISH KNITWEAR PRODUCTION
This section explores the Porac, Thomas and Baden-Fuller paper in context. I identify the broad trends in which it fell in order to situate its impact subsequent to publication. I also highlight not only the stream of papers in which the Porac et al. paper moved but also its contribution to the broader development of the field through, for example, the creation of an interest group in the Academy of Management and book series on the topic. An examination of the numerous citations to the paper shows that its reach has been very broad to the point that it has been transformed into a pro forma citation in the area of managerial cognition.
Antecedents to Porac et al. and the JMS Special Issue on Managerial Thinking
The emerging research on cognition in organizations made two related moves. The first was to address the question of organizational response to their environments, which had been dominant in the discourse of organizations as least since Lawrence and Lorsch's (1967) classic. The cognitive perspective suggested that the environment is not purely exogenous and therefore organizational response to the environment is mediated by the interpretations made of that environment by managers. Cognitive researchers pointed out that this is particularly true because of the uncertainties and complexities of the environment that decision-makers faced. Some argued, drawing on the work of Simon and March (March and Simon, 1958; Simon, 1947), that the need for interpretation comes from the cognitive limits of managers who therefore use simplifying heuristics to make sense of the world. Others suggested that the environment is fraught with Knightian (Knight, 1921/1965) uncertainties, that could not be resolved, even with more analytical capacity, because the range of outcomes is unknown and probabilities could not be assigned to them. In either case, the effect was to endogenize the environment by suggesting that cognitive frames are the means by which managers could make sense of the environment, and such sensemaking shapes strategic choice and action (Daft and Weick, 1984).
The second move was to bring the focus back to the actions of managers after the late 1970s when theories that privileged position and situation – such as population ecology (Hannan and Freeman, 1977), transaction cost economics (Williamson, 1975), and resource dependence (Pfeffer and Salancik, 1978) – were introduced. The cognitive approach suggested that structural features do not determine outcomes but rather that organizations only ‘act’ through the choices and actions of the managers within them. These two moves were tied together in a reaction to the ‘rational actor’ model of utility maximization in explaining strategic choice and action. As Charles Stubbart (1989, p. 326) argued in his piece in the same JMS special issue, ‘In economics, axioms provide scant comfort to managers who are uncertain about their preferences, unsure about their economic alternatives, and baffled as to which strategy represents the marginal utility-maximizing choice.’
Reflecting this increasing interest in cognition, a group of scholars organized a conference at Boston University in 1987 on managerial thinking, and it is from these contributions that Porac and Thomas, as editors of the special issue of JMS, drew representative papers. Stubbart (1989) provided an overview of cognitive theories, suggesting that managerial interpretation is implicit in any strategic management theory that is based in analysis of internal or external factors. The remaining papers were empirical and descriptive, examining the differences in competitors' perceptions following environmental jolts (Gronhaug and Falkenberg, 1989), the changing cognitive maps of managers in one firm over time (Fahey and Narayanan, 1989), dimensions managers use to sort issues (Dutton et al., 1989), and the mental maps of managers regarding diversification opportunities (Ginsberg, 1989), in addition to the analysis of cognitive strategic groups within an industry by Porac, Thomas and Baden-Fuller.
The Porac et al. paper mobilized these emerging ideas about organizational cognition to address a challenge in the strategic management literature posed by asymmetries within industries. There was an emerging critique (driven in part by Howard Thomas) of the tendency (escalated by the publication of Michael Porter's 1980 book on competitive strategy and industry analysis) to focus strategic analysis at the industry level. Thomas and others suggested that such a high-level grouping omitted asymmetries within industries that could explain outcomes, and they defined strategic groups as an intermediate structure between firms and industries (McGee and Thomas, 1986). The goal was to explain strategic behaviour better by focusing on intra-industry structures. Much of the subsequent research was aimed at developing better analytical techniques for identifying these groups within industries. Porac, Thomas and Baden-Fuller (1989, p. 413) turned away from this task, stating that such definitions were ‘analytical abstractions’. Instead, they based their theorizing and analysis on the idea that managers' categorizations of their competition enacted the strategic grouping. Rather than being purely material, the groups were constructed of both material and cognitive dynamics.
To develop these ideas, the authors drew on data from their examination of Scottish knitwear manufacturers. They identified 17 such firms, which appeared from the outside to be following similar ‘focused’ strategies –Porter's (1980) term – by selling ‘premium quality, expensive garments through specialist distribution channels to a limited number of high income consumers’ (Porac et al., 1989, p. 404). No other firms outside of this predetermined set were examined. Of these 17 firms, the authors interviewed managers from 35 per cent (six firms) and complemented this primary data collection with archival information about the types of products, production processes, sales volume, location, and beliefs of each.
Using methods to elicit ‘cognitive taxonomies’ of the industry, they found that the managers they interviewed had a hierarchical categorization of their competitive space, for example, starting broadly with ‘textiles’, then narrowing to ‘knitwear’ (vs. ‘hosiery’ and ‘lace’), then ‘full-fashioned classic’ (vs. ‘fashion’ which had its own subsets of ‘fully fashioned’ and ‘cut and sew’), and finally ‘high quality’ (vs. ‘the rest’). They found that most managers thought only of other Scottish knitwear manufacturers as their competition. Firms from other parts of the UK and other countries, even if they produced fully-fashioned knitwear at similar price points, were seen as being in different businesses or only ‘somewhat’ competitors (Porac et al., 1989, pp. 407–8). The authors concluded that the set of beliefs about which firms are capable of producing a certain kind of sweater defined the strategic group because managers in these manufacturers only saw these related firms as potentially posing competitive threats.
They further argued that the definition of the competitive group was self-reinforcing because the managers drew on data from a fairly limited set of sources. These managers tended not to do systematic market research but instead consulted their existing agents, design consultants, and the shop owners who sold their products. As a result, their own network ‘acts as an information filter by narrowing the range of market data to that which is of immediate concern’ to the core participants (Porac et al., 1989, p. 409). Porac et al. further argued that not only do ‘transactional networks’ filter the information that manufacturers receive, but also beliefs about the various network elements shape strategic choices. They outlined a set of causal beliefs about the customer (e.g. ‘the top 5 per cent of wage earners are unaffected by economic trends’), competitors (e.g. ‘Scottish companies are not good a fashion design’), suppliers (‘Scottish spinners are the finest in the world’), and retailers (‘minimum order quantity is one’) that influenced a general strategy pursued by the firms in this competitive group.
These causal beliefs – similar to Spender's (1980, 1989) notion of industry recipe or the concept of industry macrocultures developed slightly later by Abrahamson and Fombrun (1994) – are associated with a set of actions such as purchasing from a certain set of suppliers, using a specific type of equipment, working with certain designers, distributing through agents, selling at flexible quantities, and targeting a narrow set of consumers. The pattern of transactions in the market was therefore, they claimed, stabilized by the interplay between mental models and strategic choices. However, because Porac et al.'s data is essentially cross-sectional, such causal inferences can only be seen as suggestive.
The most important idea that emerged from their analysis involved the connection between competitive groups and individual firm strategic behaviour. They suggest that the link between group-level and firm-level dynamics is the shared mental model of strategy makers. More importantly, they argued that such mental models are the basis of coordinated behaviour in some competitive settings. ‘Cognitive oligopolies’ exist because competitors define each other as such. Managers do this because of their cognitive limits, which create the need for a definition of the competitive set that identifies a small enough set of firms to be monitored. What is powerful about this argument is that it reverses the causality proposed by economists, suggesting that ‘interfirm monitoring and co-ordination create rather than result from oligopolistic situations’ (Porac et al., 1989, p. 413). Their model endogenizes rivalry rather than treating it as an immutable external force. More generally, the paper should be seen as a contribution to a growing literature that placed cognitive dynamics at the centre of strategy formulation and strategic outcomes.
Subsequent Impact of Porac et al.
The management field was slow to pick up on the potential of the ideas proposed in the article. As demonstrated in Figure 1, for the first five or six years, Porac et al. did not draw much attention (as was the case with all of the papers from the special issue). However, in the mid-1990s, citations to this work grew quite substantially, achieving 226 in the ISI citation index by the end of 2009 (and emerging as the third most cited paper published by JMS). While all of the other papers in the special issue also received an important increase in citations in 1994–95, attention subsequently waned for the rest and no other article from that issue attracted as many citations.
Not surprisingly, an examination of the pattern of papers that have cited the work by Porac et al. shows that it has been picked up mostly in journals related to strategic management (see Figure 2). On the other hand, general management and organizational theory journals as well as journals specializing in particular domains such as sports management and health care are also represented. We also see substantial representation of journals from both sides of the Atlantic. Conceivably, because this paper is co-authored by top scholars based in both the USA (Porac and Thomas) and the UK (Baden-Fuller), the article has been noticed by authors in a range of geographic locations.
Interestingly, their most powerful insight concerning cognitive oligopolies seems not to have been addressed as prominently. Karl Weick (Weick, 1995; Weick and Roberts, 1993) drew on the idea in some of his works on sensemaking, but otherwise it has appeared in only a few other publications, and often outside of the field of management (e.g. marketing or engineering). Indeed, as shown in Table I, only one third of the papers that cite Porac et al. are even focused on the broader topics of strategic groups or rivalry. Most of these papers were published in the mid- to late-1990s. More recently, the paper seems to be cited in a generic way, as a study that looked at cognition's connection to strategy (including strategy formulation) or even more broadly at cognition in organizations. Smaller groups of citations included in the ‘other’ category are made by papers focusing on topics such as legitimacy and organizational identity. Thus, while Porac, Thomas and Baden-Fuller's paper has indeed been well cited, over time the citations seem to have taken on the character of pro forma recognitions of a foundational paper in the literature on cognition and organizations rather than specific links to ideas in their work. One exception to this conclusion is an emerging trend in the 2000s to cite Porac et al. in studies of business agglomeration, clusters and networks that are not related to cognition per se. This suggests a re-reading of the paper in light of new research agendas in strategic management.
Table I. Themes in citations to Porac et al.
LOOKING BACK AT 20 YEARS OF RESEARCH ON COGNITION AND STRATEGY
The special issue in JMS on managerial thinking was part of a flurry of activity related to cognition in the late 1980s and early 1990s. In 1989, Marlene Fiol and Bill Starbuck founded the Managerial and Organizational Cognition (MOC) Interest Group within the Academy of Management and its membership grew rapidly in the early years, achieving about 800 members by 1994 (and becoming a full division of the Academy in 1999). Anne Huff published her foundational edited volume, Mapping Strategic Thought (Huff, 1990), which included both a series of empirical studies as well as several chapters highlighting specific methods for eliciting and measuring cognition. The JAI Press launched a series on Advances in Managerial Cognition and Organizational Information Processing in 1994, which produced six volumes yearly through 1999, edited by many of the people whose work appeared in the 1989 JMS special issue.
This burst of interest, particularly in the field of strategic management, led to rapid growth in the number of management articles using cognition as a key construct, moving from an average of 6 per year in the late 1980s, to 17 per year in the early 1990s, and 29 per year in the late 1990s (see Figure 3 for a count of the number of articles in a selected list of management journals). However, at this point, growth appears to have stalemated, both in MOC membership as well as in publications focused on ‘cognition’.
In the remainder of this section, I will evaluate the progress made in the study of cognition in strategic management and, in doing so, suggest some explanations for these trends. Rather than provide a comprehensive review of the literature, I will explore a few exemplars of different kinds of empirical research on cognition in strategy to illustrate key moves scholars have made, important progress to date, and future directions for research (see Table II for a classification). In order to avoid engaging in any pro forma citing behaviour myself, I will go into some detail on the papers I cite, describing their empirical contexts, methods, and findings.
Table II. Phases in research on cognition and strategy after 1989 (selected papers)*
Establishing ‘proof of concept’
Establishing the link with outcomes
Looking at organizational dynamics
Only selected empirical and simulation papers included here (not theory or review pieces).
In exploring what each of these papers do, I demonstrate how progress has been achieved in the field. I find that the field emerged and solidified through a process of establishing a ‘proof of concept’, in other words, creating and validating measures of cognition that could be shown to be distinctive from other constructs. Later, research moved to testing the accuracy of managers' cognition using these validated constructs and, more importantly, showing how cognition could matter for strategic outcomes. In the more recent history of the field, scholars have now moved away from looking at cognition in isolation and towards an examination of how cognition interacts with other organizational factors in dynamic models of organizational action.
‘Proof of Concept’
The first phase of work in cognition and strategy aimed to create legitimacy for the line of inquiry and to establish that shared mental models existed. In other words, they sought to generate a ‘proof of concept’ that would demonstrate the usefulness of the domain for research into strategy and organizations. Following the JMS special issue on managerial thinking and other related work in the 1980s, scholars fleshed out methods for representing cognitive frames and demonstrated their presence in multiple empirical settings.
Much of the research in this period examined the content of these frames. Reger and Huff (1993) employed the repertory grid technique (which uses interviews to elicit informants' perspectives on similarities between various entities) to generate perceived groupings of banks in the Chicago area. They found evidence for three different cognitive strategic groups that were widely recognized by actors in the industry and that were also associated with performance differences across groups. Porac and Thomas followed up the Scottish knitwear study with an examination of retailers in the Midwest of the United States (Porac and Thomas, 1994). They showed that managers based their categorization of competitors on a hierarchical understanding of the product offerings (from the general category of retail, to food retailer to grocery stores). Porac and Thomas (along with Fiona Wilson, Douglas Paton, and Alaina Kanfer) also revisited the Scottish knitwear setting in a more systematic study of a larger number of competitors (including interviews at 20 different firms and 89 questionnaires) (Porac et al., 1995). In it, they established, as they did in their JMS paper, that rivalry occurs as managers in firms define other firms as rivals (a socially constructed notion of rivalry). They also showed that an industry recipe of the attributes for categorizing competition developed over time.
Beyond demonstrating the presence of cognitive understandings of rivalry, several studies examined the degree to which these cognitive frames substantively reflected rivalry. For example, Porac et al. (1995) showed that the identification of rivals was often asymmetric primarily due to the high visibility of larger firms (where small firms were more likely to designate large firms as rivals than the reverse). Lant and Baum (1995) used interviews with managers of Manhattan hotels in a competitor group mapping exercise to show that size also mattered for classification. Managers of larger hotels categorized competitors over a wider range of prices than did those of small hotels (perhaps because large hotels can offer a wider range of room rates). Peteraf and Shanley (1997) built on these findings to suggest that only some cognitive strategic groups mattered. Those with ‘strong identities’ were ones that received enough managerial focus to affect organizational action, while groups with ‘weak identities’ did not function in any meaningful way. Reger and Palmer (1996) suggested that managers' categorizations evolved at different rates in dynamic industries, depending on the type of cognitive processing (automatic or controlled) in which they engaged.
Other studies began to expand the question of interpretations of rivalry to cognition about other aspects of the environment. Milliken (1990) used surveys of managers in liberal arts colleges to examine their perceptions of an important environmental change, and concluded that managers' assessments of organizational characteristics affected their perception of uncertainty in the environment. Fiol (1990) used an analysis of letters to shareholders of ten chemical companies to show that their perceptions of the external constraints on their actions varied remarkably despite being in similar market conditions.
One crucial aspect of these studies was that they created and validated direct measures of cognition rather than using proxies such as those based in managers' demographic characteristics. Much work on top management teams had used demographic measures to capture cognitive constructs such as receptivity to change or the willingness to take risks (Finkelstein and Hambrick, 1990; Hambrick and Mason, 1984; Norburn and Birley, 1988; Wiersema and Bantel, 1992). Research in managerial cognition began to suggest that demographics were but weak approximations of cognition, and therefore more direct measures needed to be used (Barr et al., 1992; Markoczy, 1997). The view was that demographics are important indicators of sources of cognition – they can reasonably be seen to represent managers' knowledge accumulations from prior experiences – but they cannot, by definition, capture context-specific interpretations.
The mid-1990s marked a turning point in the managerial cognition field. Among other punctuation points, James Meindl, Charles Stubbart, and Joe Porac edited a special issue on managerial cognition in Organization Science (see their introduction, Meindl et al., 1994); the MOC division hosted a discussion (between Anne Huff, Joe Porac, and Jim Walsh) on the future of Managerial and Organizational Cognition at the Academy of Management Annual Meeting in 1994; Jim Walsh published his exhaustive review of the field, also in Organization Science (Walsh, 1995); and Gerard Hodgkinson (1997) reviewed the cognitive strategic groups literature for Human Relations.
In the main, these interventions came down in the same place. They all acknowledged the tremendous progress in describing different aspects of cognition and in identifying various methods for eliciting cognitive frames. Having done so, they first advocated for a refinement of the constructs and methods to study cognition. Meindl et al. (1994) pushed for a narrowing to the most promising concepts and methodologies. Hodgkinson (1997) promoted a move away from small-scale exploratory investigations towards larger-scale theory testing approaches. The MOC panel discussants noted that there was increasing ‘tolerance’ between those who advocated more focus on qualitative approaches that could add richness and those who wanted the field to move more in the direction of quantitative studies. Walsh (1995) specifically encouraged an assessment of the accuracy of cognitive frames against the environments they represent. The challenge for this, he noted, was in locating the ‘objective’ reality in what many believed was a socially constructed world. As I discuss in the next section, scholars have subsequently addressed this challenge in a few different ways.
However, these scholars ultimately were less interested in accuracy than in the effect of cognitive frames on outcomes. Walsh quoted Weick (1990, pp. 6–87): ‘Accuracy is nice, but not necessary, and the reason is that organizations generate action which creates its own substitutes for accuracy and learning . . . the truth of the map lies in the action and in the conditions of use.’ To pursue this line of thinking, Walsh recommended a ‘moratorium on research whose only aim is to describe the content of a particular knowledge structure’ (p. 303). The more than 25 studies that he identified (a few of which are described above) went a long way to legitimizing the pursuit of studies in managerial cognition, but he (as well as Hodgkinson, Meindl, the MOC panel, and others) urged scholars to take up where these studies had left off and begin to examine the effects of cognitive frames on outcomes. As Meindl et al. (1994, p. 293) prognosticated, ‘In the future, the most important studies will clearly show linkages between cognition, behavior, and organizational outcomes.’
Walsh went further to say that the ultimate objective was to understand ‘the process explanations that undergird these effects’ (p. 306). He noted however that these dynamic analyses could only happen after adequate study of the accuracy of cognitive frames and the correlations between cognition and action. It is therefore to these next phases of managerial cognition research that I turn in the sections that follow.
Assessing the Accuracy of Managers' Cognitive Frames
Research appears to have responded to the agenda articulated by Walsh (1995) and others. One stream of research built from the notion of bounded rationality (March and Simon, 1958), acknowledging that cognitive frames existed but seeking to understand how accurately managers' views represented the underlying competitive landscape. From this standpoint, any inaccuracies were due to the kinds of biases in perception that have been identified by cognitive psychologists (Kahneman and Tversky, 1979; Kahneman et al., 1994). A theory paper by Zajac and Bazerman (1991) suggested that ‘blind spots’ (inaccuracies) in perception could lead to problematic outcomes for firms and industries. Others followed with both empirical studies (using surveys) and analytical models to examine the sources and effects of such inaccuracies.
A main empirical method of studying accuracy was the use of survey data. In early work, Sutcliffe (1994) used surveys to identify the degree to which managers accurately perceived environmental instability and munificence, comparing perceptions to standardized measures of these factors from industry segment financial data. She also identified top team characteristics – such as work history diversity, team tenure, centralization, and scanning behaviour – that were associated with ‘inaccuracy’ of perceptions.
Later, Baum and Lant (2003) compared Manhattan hotel managers' categorization of their competitors (also from survey data) with an analytical model of rivalry based on material factors in the industry (size, location, and price). They solved the problem of identifying the information environment that cognitive frames are meant to represent by using what they call the ‘objective–empirical’ basis of competition derived from detailed analyses of rivalry in the industry (Baum and Mezias, 1992). They did find that the managers' maps of competition were closely associated with the important material factors shaping competition, though managers appeared to underestimate the importance of size similarity and overestimate the importance of location similarity. As a result, they found that managers made both Type I errors (categorized some hotels as competitors that were not predicted by their analytical model of similarity) and Type II errors (failed to categorize a competitor that the model predicted would be a rival). These errors were attributed to experience (the least and most experienced managers made the most errors), and this was compounded for the least experienced managers by information overload and the organization's deviations from performance expectations.
Some simulation models were also developed to address problems of accuracy. Gavetti and Levinthal (2000) used computer simulations to model cognitive frames as low-dimensional representations of more complex ‘landscapes’. They solved the problem of identifying the underlying environment by generating its structure themselves in their computer simulation. They described cognition as a ‘forward-looking’ process based on an actor's beliefs about the links between choices and actions. The assumption was that the more accurate the cognitive frame (or mental model) from which these beliefs were derived, the better the outcomes. In their model, the actor's beliefs were not fully accurate, and therefore different cognitive frames were better or worse at driving the actor to a performance peak in the landscape. They first showed that actors' experiential learning influenced changes in the cognitive frames and that these changes affected further choices and actions to respond to their understanding of the landscape. While they based their construction of the cognitive frames on an objectively known landscape (much as Baum and Lant started from the basis of a known set of dimensions predicting rivalry), what is most interesting about their analysis is the demonstration through computer simulations of the ways that different frames shaped subsequent paths of search and action.
Connecting Cognition to Strategic Outcomes
This links to another stream of research that was less concerned with accuracy and more concerned with the impact that these representations might have on outcomes. It was not enough, researchers noted, to say that managers had cognitive frames (though, of course, the proof of concept was an important first step); the real goal was to show that these interpretations could be connected to organizational actions and outcomes. Much of the challenge here was methodological. Could measures of cognition be found that would allow for longitudinal studies that would enable analysis of causality? And, ultimately, could these be compared to and combined with other factors previously identified as contributing to strategic outcomes?
In Porac et al. and other early work on cognition in strategy, the authors found cross-sectional correlations between the cognitive frames of managers and the patterns of strategic choices and actions in the firms and industries. Another early study (Thomas et al., 1993) used scenario-based surveys of hospital managers to determine their scanning and interpreting propensities and correlated these with their hospital performance measures. The problem with these observations is that it is difficult, especially in the cross-section, to attribute any form of causality. It is possible that cognitive frames direct subsequent strategic choice and action, but based on the evidence in these studies, it is equally plausible that frames simply capture a pattern of actions within the industry and are therefore epiphenomenal.
In order to establish causal relationships, later work moved towards longitudinal studies in which multiple firms could be compared. The challenge of this approach, however, was in finding sources of data that allow these comparisons to take place longitudinally across firms. Interviews were useful in capturing details of cognitive frames at one point in time, but were not as useful in capturing longitudinal perspectives because of the risk of retrospective bias. Another method for gathering data on managers' cognition was to conduct surveys. Such surveys proved useful in understanding such factors as the urgency, understandability, and manageability of environmental change – as Ginsberg and Venkatraman (1995) showed in their analysis of tax preparation firms' response to the introduction of electronic filing of tax returns – but because of their cross-sectional nature suffered from the lack of ability to identify the direction of the relationship between cognition and outcomes. An exception is research by Plambeck and Weber (2009), who conducted a two-stage survey examining the relationship between German CEOs' ambivalence about the enlargement of the European Union and subsequent organizational actions. Yet, getting a reasonable response rate (Plambeck and Weber got about 30 per cent at each stage) is difficult across multiple stages, and therefore surveys are not well suited for examining causality over longer time periods.
Following initial work by Bowman (1982), one solution that many scholars have settled on is the use of the Letter to Shareholders in a company's Annual Report to Shareholders (Abrahamson and Hambrick, 1997; Barr, 1998; Barr et al., 1992; Cho and Hambrick, 2006; Clapham and Schwenk, 1991; Eggers and Kaplan, 2009; Kaplan, 2008b; Kaplan et al., 2003; Nadkarni and Barr, 2008). They are documents that capture managers' views contemporaneously. In this way, they are superior to interviews in which managers likely engage in retrospective reconstruction of events to fit more neatly with the understanding about how things have subsequently evolved. They are also available for all (publicly traded) firms over time. Though many critiques of this source of information on managerial cognition can be (and have been!) made, it remains an important option for scholars who are seeking ways to examine cognition in a consistent manner across firms and over time (Duriau et al., 2007). It was the development of the Letter to Shareholders, and specifically word counts of themes within them, as a legitimated measure of managerial cognition that made it possible to proceed with longitudinal studies connecting cognition to action.
In small sample studies, Barr and colleagues (Barr, 1998; Barr et al., 1992) used Letters to Shareholders to map links between cognition and action using a causal mapping approach (Huff et al., 1990). In an early study, Barr et al. (1992) examined a matched pair of railroad companies and found that simply noticing new environmental conditions was not sufficient for change. Instead, managers had to make explicit links to the organization's strategy and then change those links as environmental conditions changed. Barr's (1998) careful examination of the evolution of managers' maps of the environment in six pharmaceutical firms tracked how each firm's executives interpreted an important change in the environment (increased regulation of pharmaceutical research and production). Through a qualitative comparison, she found that different firms manifested different interpretations of the event (for example, some seeing it as a major shift in the industry, others linking it only to a change in the expense structure of the firm) and that these interpretations changed over time. Importantly, she found that the firms did not undertake strategic responses to the event until interpretations had evolved to the point of recognizing a substantial impact on performance.
To move to larger samples, causal mapping approaches have proven too cumbersome, and scholars began to focus on CEO attention as captured by the count of words in the documents. Osborne et al. (2001) also examined pharmaceutical firms, this time 400 of them. They used words in the Letters to Shareholders to identify different themes (for example, the terms ‘international’, ‘alliances’, ‘foreign exchange’, ‘risk’, and ‘licensing’, defined a theme they labelled ‘overseas expansion’) addressed by the executives of the different firms. They found that various firms grouped together according to the themes they pursued and that these cognitive strategic groups were closely related to the strategic groups identified based on performance measures (which they drew from Cool and Schendel's (1987) analysis of the strategic groups of these same firms). More importantly, they also found statistically significant relationships between the cognitive groups as defined in the initial time period, strategic actions in subsequent periods (e.g. facilities expansions or introduction of new product lines), and later reports of the success or failure of those activities.
In their study of deregulation in the airline industry, Cho and Hambrick (2006) extended this line of thinking by showing how changes in the top management team's focus of attention (from an ‘engineering’ orientation to an ‘entrepreneurial’ one) were associated with subsequent changes in strategic actions (such as choices about the number of cities served, the number of planes in the fleet, the level of passenger service, and expenditures on advertising and sales). Their more unique contribution was in identifying a series of antecedents to changes in top management team attention, such as changes in the average industry tenure of the team, changes in the type of functional experiences, and increases in the heterogeneity of the team.
The challenge of a number of these studies that attempted to associate cognition with strategic outcomes is that they did not control meaningfully for the alternative explanations of strategic action that exist in the strategic management literature. Without these controls, one can raise the reasonable concern that measures of cognition may simply be epiphenomena of actions that the firm might reasonably take based on its interests or its capabilities. To address these concerns, I, along with Rebecca Henderson and Fiona Murray, examined the response of 15 pharmaceutical firms to the emergence of biotechnology over a period of 20 years using biotechnology-related words in the Letter to Shareholders to proxy for managerial attention to this new technology (Kaplan et al., 2003). The study found that efforts to get patents and publish scientific papers in this realm were associated with prior changes in managerial attention. This paper responded to scepticism about the connection between cognition and performance, by first verifying the relationship even when including a series of measures for plausible alternative explanations for differences in firms' responses (e.g. scale, prior related experience (capabilities), market position, competitive actions, isomorphic pressures, and financial performance); and second, ruling out reverse causality between organizational actions and subsequent attention to the issues.
This phase of research in cognition and strategy made strides in connecting cognition to outcomes of interest in strategic management. This progress came through innovations in measurement and through careful attention to alternative explanations of organizational action already well established in the strategic management literature. This research completed the primary research agenda established by the early scholars of cognition and strategy by showing not only that managerial cognitive frames exist (‘proof of concept’) but that they matter.
Building Dynamic Models of Cognition in Strategy Making
What happened next? If one looks at the membership of the Academy of Management's Managerial and Organizational Cognition Division, one would not be particularly hopeful about the subsequent momentum of research on managerial cognition in general nor cognition in strategy specifically. Its membership stalemated in the late 1990s and grew only slightly (and more slowly than the overall Academy's membership growth) in the 2000s. What is apparent, however, is that cognition began to spread throughout the different areas of managerial scholarship. Indeed, in the Academy of Management Programme in 2009, of the 96 sessions including the words ‘cognition’ or ‘cognitive’, only 22 were associated with the MOC division. Interestingly, the next most highly associated division was Business, Policy and Strategy (with 16 sessions) (see Figure 4). This trend was presaged by another series of MOC-sponsored discussions at the 1999 Academy of Management Annual Meeting in which one of the main themes (particularly championed by Anne Huff and William Starbuck) was that of the ‘pluralistic, bridging nature’ of cognition. These scholars recognized that many researchers interested in cognition might have their primary intellectual homes in other divisions, such as strategy, organizational behaviour, or elsewhere.
Thus, contrary to the notion that research in ‘managerial cognition’ may have stalemated after the burst of activity from the late 1980s to the late 1990s, what we see is a diffusion of ideas about cognition throughout strategy and management scholarship. Looking at papers that have cited the Porac et al. article, we see that different aspects of cognition have been picked up by different scholars. The concept of cognitive strategic groups was connected to ideas about mental models, shared knowledge, social learning, technological frames, etc. (see Table III). That the Managerial and Organizational Cognition division in the Academy of Management has not grown substantially may actually be a sign of the success of the pioneers of this field. ‘Cognition’ and all of the concepts related to interpretive processes can be seen as a sort of ‘general purpose technology’ (Bresnahan and Trajtenberg, 1995) that has become an essential part of the theorizing in many different fields of management, including, or especially, strategy.
Table III. Cognition-related, author-supplied keywords for 226 articles citing Porac et al.
Source: ISI Web of Knowledge. Coding by the author based on the author-supplied keywords of 226 papers citing Porac et al. to the end of 2009.
• collective strategy frames • domestic mindsets • egocentrism • environmental interpretation • frames • frames of reference • framing • heuristics and biases • information exchange • information processing • information sharing • insight • issue selling • judgment • knowledge base • knowledge structures • knowledge transfer • managerial cognition • measuring knowledge • mental maps
• mental models • organizational image and identity • organizational knowledge • overconfidence • overlapping classification • perception • personal construct theory • problem representation • repertory grid (technique) • sensemaking • shared cognition • shared understanding • signalling • similarity judgments • social cognition(s) • social learning • sociocognitive resources • strategic cognition • technological frames
It is therefore not possible in the space available to trace the paths that managerial cognition research has taken, even just in the field of strategy. Instead, in this section, I will highlight a few significant directions (admittedly those biased by my own research interests) that researchers took as they more fully integrated cognition into the field of strategic management (with apologies to all those scholars whose work I have omitted for reasons of space).
Having accomplished the objective of finding systematic associations between cognitive frames and strategic outcomes, a set of scholars have turned to what Walsh (1995) had described as the ultimate objective of managerial cognition research, that of understanding the processes and practices underlying these effects. In their 1989 piece, Porac, Thomas and Baden-Fuller took as an assumption a set of processes that connect managers' identification of cognitive strategic groups to strategic outcomes in which ‘individuals attend to cues in the environment, interpret the meaning of such cues, and then externalize these interpretations via concrete activities’ (p. 398). These views built on earlier theoretical work by Daft and Weick (1984) and Kiesler and Sproull (1982) who proposed that strategic action is influenced by how managers notice and interpret change and translate those perspectives into strategic choices.
In the 2000s, scholars returned to these earlier models with an aim to develop a richer empirical understanding of the ways in which cognition shapes action. One fruitful area for these explorations was in understanding how organizations respond to changes in their environments, be they regulatory changes, new technologies, or market crises. While some organizations are able to adapt, many are subject to strong inertial forces, and scholars have found cognitive explanations to be particularly powerful in understanding these differences in outcomes. The research to explore these dynamics took two major forms: the first was to place cognitive explanations in the context of the two other major sets of explanations pursued in research on strategy, those of incentives and capabilities; and the second was to look inside the organization to uncover the daily practices that connect cognitive frames to outcomes.
Cognition, capabilities, and incentives. In the field of strategy, substantial research efforts have been devoted to understanding differences in organizations' responses to change, and the explanations have been grounded in differences in either incentives or capabilities (only a very few citations to which are provided below). Economists have argued that differences in responses to changes such as new technologies can be explained as rational responses to differential incentives (Gans and Stern, 2000; Henderson, 1993; Zucker and Darby, 1996). Researchers drawing on the Resource Based View and behavioural theories of the firm argue that firms often have difficulty adapting to change because of the path dependence associated with initial endowments (Cockburn et al., 2000; Leonard-Barton, 1992), unless the firm's capabilities happen to be ‘pre-adaptive’ (Cattani, 2005; Klepper and Simons, 2000) or firms possess dynamic managerial capabilities that allow resources to be reconfigured (Helfat, 1997; Teece et al., 1997).
A series of recent studies – mainly rich qualitative inquiries – have sought to add cognitive explanations to these discussions. Tripsas and Gavetti (2000) used their analysis of Polaroid's failure to transition from analogue to digital imaging technologies to highlight the paradox of Polaroid's early development of technical capabilities but ultimate failure to be competitive in the digital camera market. They suggested that this paradox can be explained by managerial cognitive frames that placed primacy on technical excellence and modelled economic success on a ‘razor/razor blade’ model for making money that turned out to be inappropriate in the digital world. They demonstrated that the presence or absence of capabilities was not an adequate explanation of organizational inertia and that cognition was a central contributor to outcomes. This analysis amplified earlier findings by Dougherty (1992) – whose study of new product development processes showed that departmental ‘thought worlds’ and organizational product routines interacted to inhibit the acquisition of new technology and market knowledge – and by Garud and Rappa (1994) from their study of the development of cochlear implants – which highlighted the intimate relationship between understandings of a technology, the organizational routines developed for evaluating those technologies, and the strategic choices of firms to invest in one technology or another.
This empirical research connected to a set of theoretical developments about the relationships between capabilities, incentives, and cognition. In outlining his attention-based view of the firm, Ocasio (1997) argued that one set of the contextual factors shaping the allocation of attention in the organization is what he calls the ‘rules of the game’– the routines and incentive systems that structure the process by which interpretations are made. Routines can be understood to have both cognitive and motivational dimensions – they embed both an understanding of how things should be done and what gets rewarded (Cohendet and Llerena, 2003; Coriat and Dosi, 1998; Dosi et al., 2003; Kaplan and Henderson, 2005). As a result, these routines and economic forces may reinforce ‘attention traps’ (Johnson and Hoopes, 2003, p. 1058) that cause managers to miss seeing change. These various streams of work all imply that cognition, capabilities, and incentives are interconnected factors shaping organizational outcomes.
Laamanen and Wallin (2009) built on these insights in a comparative case study analysis of three software firms to show how capability development paths paralleled cognitive paths of managers, where shifts in how managers thought about the business enabled the organization to build needed capabilities. Indeed, it was the ability of managers to decide which capability bottleneck to focus on that made adaptation to changing conditions possible.
My colleague J. P. Eggers and I (Eggers and Kaplan, 2009; Kaplan, 2008b) explored these dynamics in larger samples to examine how CEO cognition interacts with organizational capabilities and incentives in shaping strategic outcomes. These studies of communication technology firms' responses to the emergence of fibre optics showed that the effects of cognition, incentives, and capabilities are related. Regarding choices about the direction of R&D (Kaplan, 2008b), changes in managerial attention had an important main effect and this effect was intensified if firms did not have prior related competences in the optical arena or customer-related incentives to act. This implies that having the appropriate capabilities or strong incentives to act may improve a firm's chance for responding to environmental change, but they alone may not be enough to instigate the change. Echoing Tripsas and Gavetti's (2000) findings, even when a firm has the correct technical capabilities, it may be unable to respond effectively to a new technology if managers' beliefs are not correctly aligned with the opportunity. Similarly, when incentives to act are low, cognition will have an increasingly substantial effect on outcomes.
Regarding subsequent launching of optical products (Eggers and Kaplan, 2009), the results suggest that the effect of the CEOs' cognition differs depending on the focus of their attention – accelerating entry when attention is directed to the emerging technology and slowing response when attention is directed to the existing technology. This implies that firms with the same underlying organizational capabilities and incentives might move in different directions if the different CEOs' attention is either myopic or far-sighted. It is therefore not necessarily organizational capabilities or incentives associated with the existing technologies that hold firms back, but rather greater attention to these technologies by managers that lead to inertia. Thus, this work indicates that understanding the interaction between cognition, capabilities, and incentives ends up being essential to understanding strategic outcomes in organizations.
Internal cognitive dynamics in organizations. Another stream of research that picked up on the process assumptions in Porac et al. has focused on the internal cognitive dynamics in organizations that produce the relationships between cognition and strategic outcomes described above. This followed the advocacy by several managerial cognition scholars to move research beyond purely content issues in order to focus on cognitive processes (Bogner and Barr, 2000; Elsbach et al., 2005; Lant, 2002). These scholars argued that an understanding of cognitive mechanisms in such processes as strategy making requires engagement with the situated and interactive nature of cognition. They suggested that it is risky to reify the organization as a cognizer and that scholars should endeavour to understand how a collective frame and ultimately a collective decision might emerge through actors' interactions (Garud and Karnøe, 2001; Spender, 1998; Weick and Roberts, 1993) (for the most recent update of these ideas, see Garud et al., 2010). Weick (1995) was clearly inspired by the Scottish knitwear study as he devoted several pages to a discussion of it in his 1995 book on sensemaking, and it was this portrayal of the interactive process of interpretation that provoked questions in the field of strategic management about static treatments of cognition to that date.
Some scholars have responded to this challenge with in-depth case studies of strategy making inside organizations. The models developed from these various case studies (a few of which are described below) depart from the way cognition has traditionally been construed in the managerial literature. Instead of a static representation of cognition in which actors are depicted as constrained by rigid mental models, these studies of the process of strategy making portray cognition as a dynamic, purposive, and politically-charged process of meaning construction.
These studies have mainly been published in the past decade, but one early study in this regard was that by Gioia and Chittipeddi (1991) who examined an organizational change process inside a university. Using interviews and participant observations, they found that the ‘CEO’ of the university and other senior managers were engaged in efforts, first, to make sense of the strategic change based on ideas from their backgrounds, the history of the university, and information from key stakeholders; and, second, to provide a vision to others (what they call ‘sensegiving’). This analysis demonstrated that interpretation was both an individual and a social process. Individuals need social interaction in order to understand and interpret their environments but they also use these social interactions to shape the interpretations of others. Thus, sensegiving is intertwined with one's own sensemaking efforts.
Rouleau (2005) explored these dynamics further in her study of middle managers in a top-of-the-line clothing company. She demonstrated that these middle managers had a crucial role as both interpreters and sellers of strategic change within the organization and, crucially, to outsiders such as clients. Similarly, Markoczy (2001) looked at how consensus in beliefs was achieved in three different Hungarian companies recently acquired by Anglo-Saxon firms. Using causal mapping techniques, she showed that consensus building was driven not by top management but by an interest group of organizational members who would benefit from the strategic change.
Szulanski et al. (2004) conducted case studies of the strategic decision-making processes in three firms dealing with another kind of strategic change, the emergence of the internet. By looking inside these firms, they were able to make links between organizational dynamics and strategic outcomes. They found that the type and breadth of cognitive frames as well as the timing of efforts to develop these frames affected both the degree to which each firm responded to the technical change and the success of those strategic choices. They argued that successful managers in organizations take deliberate steps to interpret change (technical or other) and that ‘default’ frames based on past individual or organizational experience can lead to poor outcomes.
Inspired by these ideas, I conducted an ethnographic case study of strategy making inside one (communications technology) organization facing a radical change (the collapse of the fibre optics market in the early 2000s) (Kaplan, 2008a). To interpret these data, I drew on research about social movements to develop a model of ‘framing contests’ inside the organization. This model described the ways that actors translate their own cognitive frames into a predominant collective frame as the process of strategy making unfolds. I showed that when cognitive frames were not congruent, different groups attempted to make their own frame resonate with others and therefore mobilize action in their favour. Those actors who most skilfully engaged in these practices shaped the frame that prevailed in the organization. This paper modelled framing in strategy making as both highly contested and connected to the political pursuit of interests. The focal point of the political process is the collective effort of meaning making, such that the pursuit of interests can only be understood alongside the frames actors use to interpret those interests. Making strategy under uncertainty can therefore be comprehended as a product of contests over which frame should guide the understanding of an ambiguous environment and of choices about how to respond to it.
This series of studies on framing shows that cognition and the political pursuit of interests are intertwined. From this perspective, frames are an ex ante part of the political process that produces decisions, defining the nature of the problems around which political jockeying will take place. As a result, these papers brought framing and interpretation into research on strategy processes, which has traditionally emphasized only the political side of the equation (Eisenhardt and Bourgeois, 1988; Pettigrew, 1973). The papers described above offer a glimmer of the possibilities created by the examination of interpretive processes inside organizations. Much potential remains in this stream of research.
CONCLUSION – LOOKING FORWARD
Where to next? The answer, of course, is many places. As I have sketched out above, cognitive constructs have diffused widely both in strategy and more broadly in organizational analysis. Scholarship on cognition in strategic management over the past 20 years has put behind us questions about whether cognition matters for outcomes. It has also steered us away from the analysis of accuracy and bias in perception – these are not central or even perhaps appropriate questions in a context where interpretative processes have been shown to endogenize organizational and market structures. Instead, I suggest that, in addition to further studies of cognitive dynamics inside organizations as discussed above, there are three avenues that Porac, Thomas and Baden-Fuller opened up which are experiencing a resurgence of interest.
The first stream of research builds on the insight from Porac et al. that the environment is endogenous to the actions of organizations. They showed that rivalry in the Scottish knitwear industry was not preordained but rather shaped by how managers in each of the firms defined who their rivals were and what criteria upon which rivalry would take place. More recently, studies have identified and analysed other dimensions of the environment – such as relative reputations of firms, the legitimacy of new practices, industry velocity, and the understanding of value – that are produced through the interpretative processes of the actors in those environments.
Rindova and Fombrun's (1999) study of IBM's evolution as a company found that the reputation of the firm was constructed through the sociocognitive dynamics of interactions between the firm, its constituents, and its rivals. Furthermore, the changing reputation affected the firm's ability to gain access to key competitive resources. As a result, the interpretations of the various players produced the terrain upon which competition took place. Kennedy and Fiss (2009) found that the diffusion of TQM practices depended not only on desires for economic performance or legitimacy – as classic models of adoption would suggest – but crucially on whether such practices were perceived as posing a threat or an opportunity.
Similarly, Nadkarni and Narayanan (2007) found in their study of aircraft and semiconductor firms that the very pace of industry change was constructed by different firms' interpretations about the degree of industry velocity. High- or low-velocity conditions were shaped by the assumptions of managers and then reinforced through their social networks and the feedback from their own strategic actions. Further, Fiona Murray and I (Kaplan and Murray, 2010) found in our analysis of the evolution of the biotechnology industry that the notion of what was of value, and therefore what technological directions should be pursued, was a product of the interpretations of a wide range of actors in the field. These studies suggest that environments have multiple dimensions – rivals, reputations, velocity, value – and that these dimensions are not given characteristics of the field but rather constructed through the interpretations of the actors who make up the field.
Categories and Their Emergence
The second stream of research returns to the idea of categorization. The Scottish knitwear study showed how the categorization of different competitors as direct rivals affected the strategic choices and actions of firms. They claimed associations between categorization and performance. More recent work has picked up on these ideas and has sought to nail down this relationship. Porac returned to the issue of categorization with co-author José Antonio Rosa (Rosa and Porac, 2002) to demonstrate that in the motorcycle industry categorizations had different levels of persistence depending on the source ideas for the categorizations themselves, with those coming from scientific authority, for example, having greater persistence, and those coming from perceptible properties of the product having lower persistence. Zuckerman (1999) showed that stock prices suffer for firms that are not easily classified into an industry sector by analysts. Hsu et al. (2009) found that products (films or items for sale on eBay) that span many categories are less likely to attract customers than products where the categorization is clear. In addition, they show that firms that span multiple categories have difficulty in developing capabilities to do well in any one specific category.
The logical follow on question is of course how these categories emerge in the first place. Researchers have begun to explore this topic, identifying key antecedents to category creation and the ways in which new categories emerge and are legitimated, with the media and high status or powerful actors playing central roles. In early studies in this arena, Joe Porac, along with a number of authors, found that in the case of the introduction of minivans into the motor vehicle market (Rosa et al., 1999) or different classes of motorcycles (Rosa and Porac, 2002), new categories were negotiated between producers and users through the media. Kennedy's (2005, 2008) studies of the emergence of the computer workstation market showed that new market categories need to be legitimized, and media coverage of new firms is a crucial means for doing so.
Rao et al.'s (2003, 2005) analysis of French chefs showed that, as a new category emerged (nouvelle cuisine relative to classical cuisine), chefs' self-categorizations changed based on the status of the proponents of the new category and the defections by peers into the new category. Furthermore, Rao's related study with Lounsbury (Lounsbury and Rao, 2004) on the mutual fund industry emphasized that these actors can exercise their power to either preserve the existing structure of categories or change them. The research highlighted above delves further into the ideas set up by Porac et al. that the interpretations of actors embedded in these contexts shape the contexts themselves in an iterative process. However, there is much more to learn about the process of emergence, and cognition will certainly have a central role in the theories still to be developed.
Reciprocal Dynamics between Cognition, Materiality, and Interests
Porac et al.'s project in 1989 was to move away from purely material definitions of strategic groups and show that they were constructed of both material and cognitive dynamics. However, because their emphasis was on introducing cognition to models which had previously omitted it, they did less work at the time to show how the cognitive and material interacted. Recent research digs into these reciprocal dynamics between categories, material products, and the interests of actors in the creation and stabilization of categories (Kaplan and Tripsas, 2008).
Pontikes (2008) suggested, in her analysis of new categories in the software industry, that there is an intertwining between the nature of a new technology, the character of existing categories, and the creation of new categories. She found that new technologies are seeds for defining new categories but that a new category will only emerge if existing categories are not flexible enough to accommodate the technology. Weber et al. (2008) showed, in a study of the emergence of a market for grass-fed beef, that a grass roots social movement solved challenges in creating new markets by stimulating production, creating a system of exchange, and crucially, generating a collective identity and shared understanding of what the market would be.
Fleischer (2009) tied the shaping of categories directly to the interests of focal actors by looking at the forces shaping brokerage firm rating systems of different companies. She found that where brokerage firms had conflicts of interest (in that they were engaged in underwriting activities associated with firms they were also rating), the ambiguity of the categorizations increased. The definition of categories was not only related to technical features of the companies themselves, but was shaped by the interests of those constructing the categories. Benner and Tripsas (forthcoming) showed, in the context of digital cameras, that firms' prior industry experience shaped their beliefs about what form the new technology would take, but as they gained experience with particular features of the technology, the influence of those previously shared frames declined. These studies beginto unpack the complex relationship between technologies, categories, and actors' interests as they emerge and co-evolve.
The research highlighted above delves further into the ideas set up by Porac et al. that the interpretations of actors embedded in these contexts shape the contexts themselves in an iterative process. The essential questions that researchers interested in cognition and strategy can pursue today are: Where do categories come from? How can we account for the relationship between cognition, materiality, and actors' interests in understanding how categories emerge? How do organizational strategies shape categories and how do categories shape organizations and their strategies?
It is clear from this review of research in cognition and strategy over the past 20 years that one cannot advocate using a single construct in approaching these questions. I use the language of ‘category’ in the research questions above only because it has become a recently popular label for interpretive constructs, but one could easily substitute in ‘cognitive frame’ or ‘schema’ or ‘technological frame’ or many other terms to those questions and they would be equally valid.
Neither can I advocate a single method. Some of the questions will only be answered with detailed qualitative studies of internal organizational dynamics, but others will be amenable to many different approaches already developed in the field. One of the most recent trends, driven by increased electronic availability of texts and new sophistication in computational methods, is textual analysis of themes and relationships. The field is moving rapidly beyond the simple word counts that made cognitive constructs tractable in the past. Further, while not covered in any detail here, methods developed in cognitive psychology, both behavioural experiments and cutting edge brain-scanning (fMRI) techniques being exploited fairly aggressively in the marketing and decision sciences literature, may also bring new insights to the field. Despite 20 years of scholarship in cognition and strategy, there is still much to know and thus ‘a thousand flowers [of constructs and methods] blooming’ is very much in order.
Joe Porac, Howard Thomas and Charles Baden-Fuller, along with a number of other pioneering scholars in the 1980s, set us on a path to explore the role of cognition in strategy. This path has turned into many paths, only some of which I have highlighted in this appreciation of the impact of their work. Cognitive explanations have gained legitimacy and have been analysed alongside more traditional explanatory factors in strategic management research. Scholars have established and legitimized various measures of cognitive constructs. Managerial cognition as a topic has diffused throughout the Academy, especially in the field of strategy. Thus, the scholars of the 1980s can claim substantial victories, and they can also greet with satisfaction new streams of research that build on the ideas born more than 20 years ago.
The author would like to thank Timothy Vogus for data from the Academy of Management Managerial and Organizational Cognition Division archives; Joel Baum, Joep Cornelissen, Anne Fleischer, Sucheta Nadkarni, and Davide Ravasi for helpful comments; and Keyvan Vakili for research assistance.
The other important link here is to images of institutional isomorphism (DiMaggio and Powell, 1983), which, while not specifically focused on cognition, did suggest that factors other than rational economic choice of actors were at play in shaping firm strategic choices and action.
This is a concept that Porac and Thomas had been developing in a series of conference papers which resulted in a theory building paper in the Academy of Management Review in 1990 (Porac and Thomas, 1990).
These statistics are based on searches of the ISI Web of Knowledge (28 January 2010). According to Google Scholar, the article has been cited 541 times, but these data are difficult to analyse at a granular level because Google does not provide keywords and abstracts systematically. So the analyses presented here draw from the ISI results alone, and are therefore only a partial representation of the intellectual landscape in which the Porac et al. paper sits.
This information is based on searches of the ISI Web of Knowledge and Google Scholar (28 January 2010).
I wish to thank Professor Timothy Vogel, the Archivist of the Managerial and Organizational Cognition Division, for providing membership statistics.
For reasons of both interest and space, I focus this analysis on a selection of empirical papers. However, it should be noted that there have been many interesting theoretical papers proposing new ways of theorizing about cognition and strategy (such as Ginsberg, 1994; Milliken and Lant, 1991; Ocasio, 1997) that should be considered in a fuller appreciation of the development of the field.
As summarized by Frances Milliken in the MOC interest group newsletter, November 1994.
Gavetti and Levinthal, as well as Rivkin pursued this line of thinking in a series of subsequent papers contrasting ‘cognition’ with rational search and analysis (Gavetti and Rivkin, 2007; Gavetti et al., 2005). These perspectives see cognition as an effect of bounded rationality (March and Simon, 1958), treating it mainly as a computational problem.
First, the Letter to Shareholders can meaningfully be interpreted as representing CEO attention to specific issues. Unlike the rest of the Annual Report, which is often written by the investor relations department of a firm, the Letter to Shareholders is signed by the Chief Executive Officer and is at least closely reviewed if not specifically written by him/her (Cho and Hambrick, 2006; Kohut and Segars, 1992). In my own research using Letters to Shareholders (Eggers and Kaplan, 2009), we saw clear signs that changes in CEOs had an impact on the style, length, and content of the Letters. For example, the majority of the CEO changes covered in our study were associated with an at least 20 per cent change in the average number of words used in the Letter. We also noticed significant shifts in tone and approach used by new CEOs, even if the CEO had been a member of the senior leadership team for quite some time prior to assuming the top job. Such changes are signals that each CEO put his or her imprint on the Letter to Shareholders. It is also unlikely that the CEO would engage in an excessive amount of misdirection in the Letter because fiduciary requirements make it hard for the CEO to suppress important discussions (Abrahamson and Park, 1994). Finally, research has shown that the themes emphasized in the Annual Report mirror closely the discussions in internal strategic planning documents (though the positive or negative spin on the issues may differ) (D'Aveni and MacMillan, 1990; Fiol, 1995).
Note that Nadkarni and Barr (2008) expanded the causal mapping approach to 24 firms in their analysis of the effect of perceptions of industry velocity and the speed of response to environmental changes in four industries.
The MOC division grew at about 2.6 per cent per annum from the mid-1990s to 2008 (slightly higher at 4.4 per cent from 2003 to 2008), but the Academy of Management overall grew 4.9 per cent p.a. since the mid-1990s and 5.3 per cent from 2003 to 2008 (data from the MOC Division archivist, Timothy Vogel).
As previewed in the June 1999 newsletter of the MOC division announcing the session.
In his 1995 review of the field, Jim Walsh identified 77 different terms used in the managerial cognition field, from ‘knowledge structures’, to ‘frames’, to ‘beliefs’, etc. This pluralism of terms connects in part to the different disciplines from which people have approached managerial thinking. It also explains a good deal of the diffusion of the field in more recent years: paths pursued by different scholars have been diverse. Advocacy in the 1990s (described above) to narrow the number of concepts used in managerial cognition has not been heeded.
Note that, in research on individual choice and action in behavioural economics and finance, there is a huge literature examining accuracy and bias that is too large to summarize with any subtlety here. This primarily experimental work has made important psychological contributions to understanding why neoclassical economic models of behaviour are not fully representative of the dynamics we observe in markets. However, they are less salient in understanding the organizational dynamics associated with strategic management that is the subject of Porac et al.'s (1989) paper and my own review of the related literature here.