Organizational Improvisation and Organizational Memory Author(S): Christine Moorman and Anne S

Organizational Improvisation and Organizational Memory Author(s): Christine Moorman and Anne S. Miner Source: The Academy of Management Review, Vol. 23, No. 4 (Oct., 1998), pp. 698-723 Published by: Academy of Management Stable URL: http://www.jstor.org/stable/259058 . Accessed: 23/09/2013 16:39

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to The Academy of Management Review.

http://www.jstor.org

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions t Academy of Management Review 1998, Vol. 23, No. 4, 698-723.

ORGANIZATIONALIMPROVISATION AND ORGANIZATIONALMEMORY

CHRISTINEMOORMAN ANNE S. MINER University of Wisconsin at Madison

We define organizational improvisation as the degree to which the composition and execution of an action converge in time, and we examine the theoretical potential of this definition. We then propose that both organizational procedural memory (skill knowledge) and declarative memory (fact knowledge) moderate improvisation's impact on organizational outcomes in distinct ways. We also suggest that improvisation influences organizational memory by (1) generating experiments and (2) permitting the development of higher-level competency in improvisation. Contemporary technological changes related to the nature of organizational memory intensify the salience of these issues.

Observers long have noted the presence and and skills (Crossan & Sorrenti, 1997; Eisenhardt promise of improvisation in the arts, teaching, & Tabrizi, 1995; Hatch, 1997b; Weick, 1993b,c). therapy, and athletics. For example, one can In this article we draw on work from several find detailed data on differences between im- disciplines to generate specific propositions provisational performance by expert and novice about how organizational memory will influ- mathematics teachers (Borko & Livingston, 1989), ence improvisation's impact. We argue that two models of musical improvisation (Pressing, 1984, different types of organizational memory-(1) 1988), and guidelines for encouraging effective procedural (skill memory) and (2) declarative improvisation by therapists (Embrey, Guthrie, (fact memory)-moderate the impact of organi- White, & Dietz, 1996; Gardner & Rogoff, 1990). zational improvisation in different ways. In par- Organizational scholars often seek to provide ticular, we suggest these memory types enhance strategic planning and project planning tools different organizational outcomes and can com- that, presumably, suppress the level of action pensate for each other's drawbacks. We also occurring without prior design. Nonetheless, formalize the insight that improvisation, in turn, those in an important contrasting branch of affects organizational memory and represents work have argued that improvisation not only an identifiable organizational competency occurs frequently in organizations but may also (Eisenhardt & Tabrizi, 1995; Weick, 1993b). have value for organizations (Preston, 1991a; To support these testable propositions, we Weick, 1979, 1987, 1993a,b,c, 1996). Researchers draw on prior work to define improvisation as have also observed, however, that fruitful im- the degree to which composition and execution provisation may require important resources converge in time. Therefore, the more proximate the design and implementation of an activity in The authors contributed equally to this article. Our re- time, the more that activity is improvisational. search benefited from the support of Paula Bassoff, This view of improvisation is primarily tempo- Stephanie Dixon, Bob Drane, Gabor Kemeny, Scott Little, ral, with a focus on the degree of simultaneity of John Miner, Jr., Aric Rindfleisch, David Robinson, Rona Velte, composition and implementation. It is consis- and two anonymous organizations, as well as the comments tent with prior work (Crossan & Sorrenti, 1997; of Jim Burroughs, Mary Crossan, Peter Dickson, Kathy Eisen- hardt, Mary Jo Hatch, Jan Heide, David Mick, Jeff Pressing, Weick, 1993a) but focuses on one key dimension: Thekla Rura-Polley, Bob Sutton, Mary Waller, Jim Walsh, the temporal order of two specific activities. This Karl Weick, and five anonymous AMR reviewers on previous enhances theory development by encouraging versions of this manuscript. National Science Foundation distinctions between the improvisation process Grant SBR-9410419, the Marketing Science Institute, and the itself and important potential correlates or out- University of Wisconsin School of Business Sabbatical Fund and Research Fund have generously supported this re- comes, such as intuition, adaptation, innova- search. tion, and learning.

698

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 1998 Moorman and Miner 699

The importance of predictions about organiza- bear more consistently on action. At the same tional improvisation and memory goes beyond time, organizations increasingly post organiza- refinements to existing frameworks and theo- tional product specifications, policies, and notes ries, however. First, awareness of the changed from group meetings and budget information on pace in competition for many organizations has firm intranets, as well as design internal search created a fadlike press on managers to abandon engines that invisibly guide employees' access traditional planning. Managers face broad ad- to different parts and types of organizational monishments to be nimble and flexible (e.g., memory (e.g., Diamond, 1997; Knott, 1997). Imai, Nonaka, & Takeuchi, 1985) but often lack Our conceptual framework draws on this in- balanced information on the tradeoffs between creasingly powerful role of organizational mem- improvisation and composing well in advance ory by suggesting that firms will need to design of action, and on the broader competencies re- systems that enhance access to the right type of quired to improvise effectively. Such knowledge memory while undertaking improvisational ac- is increasingly useful as contemporary organi- tion. It also implies that organizations could zations face competitive situations, in which benefit from deliberate efforts to accumulate exogenous change may outpace traditional memory that will permit them to improvise more planning cycles (Burgelman, Maidique, & effectively over time, and to take into account Wheelwright, 1996). This shift makes improvisa- the impact of different memory types on differ- tion important to more than startup or creative ent outcomes, such as speed or effectiveness of organizations. A highly successful special ses- actions. Taken together, then, our propositions sion involving jazz improvisation at the 1995 an- imply that developing organizational improvi- nual meeting of the Academy of Management- sational competencies involves long-term atten- documented and analyzed in a special issue of tion. Organization Science (Meyer, Frost, & Weick, in To develop and consider our propositions, we press)-also points to the more general impor- have organized the article as follows. We first tance of improvisation for many types of organi- describe several types of improvisation and dis- zations. cuss the nature of collective improvisation. Next, Second, work on improvisation is made sub- we examine the theoretical potential of this def- stantially more urgent by the impact of techno- inition of improvisation by assessing whether it logical change, not only on the tendency to im- extends the existing menu of organizational provise but on an organization's ability to variables in a meaningful way. Following this, deliberately manage the nature of and access to in the main body of the article, we present prop- organizational memory. Regarding the tendency ositions linking organizational memory and or- to improvise, technological change affects the ganizational improvisation. Taken as a whole, chances an organization may execute an act these propositions suggest that, while improvi- while designing it because traditional buffers sation may represent a potentially powerful or- between choice and action have been reduced ganizational tool, its effective deployment is drastically. For example, a CEO could improvise very demanding and may well require substan- a corporate response to a crisis by electronically tial time and investments by organizations. We mailing comments to thousands of employees conclude by considering the operationalization and customers. This level of simultaneity simply' of key constructs, boundary conditions, and se- would not have been possible if the CEO had lected contributions of this work. developed a strategy in conjunction with a public relations staff and then created written mem- oranda and organized meetings to communicate IMPROVISATION the organization's position. Definition Technological change also influences the way memory can guide action in organizations. For In Table 1 we suggest that the notion of im- example, in technically sophisticated firms, de- provisation arises in varied contexts and that sign engineers, who used to work from princi- the term improvisation has been defined in var- ples and prior training, now work on computers ied ways, even in the organizational context. We with corporate design rules encoded into design suggest that whether or not improvisation oc- constraints so that prior memory is brought to curs depends not only on what happens but also

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 700 Academy of Management Review October

TABLE 1 Improvisation Across Disciplines

Definition Author(s) Domain

A. Organizational perspectives on improvisation

"the acts of composing and performing are Bastien & Hostager (1992) Organizational inseparable, and each composition/ communication performance is different from all previous compositions/performances" (p. 95) "Improvisation in the present .. . to stay focused Brown & Eisenhardt (1997) Product development on current conditions,... while maintaining project schedules" (p. 9) "Intuition guiding action in a spontaneous way" Crossan & Sorrenti (1997) Management (p. 1) "Intuition guiding action upon something in a Hatch (1997b) Management spontaneous but historically contextualized way" (p. 5) an activity which requires no preparation and Mangham (1986) Management obeys no rules" (p. 65) "Improvisation-the casting around for a Mangham & Pye (1991) Management precedent or referent that will enable someone to deal with a circumstance for which no script appears to be immediately to hand" (p. 41) "To be composed while performed" (p. 51) Perry (1991) Management "Improvisation is to be distinguished from Preston (199la) Organizational rewriting a musical in that changes are development introduced during the performance by the performing individuals and not by the composer before the event" (p. 84) on the spot surfacing, criticizing, restructuring, Schon (1983) Management/education and testing of intuitive understandings of experienced phenomena" (p. 147); "Knowing- in-action" and "Reflection-in-action" (p. 276) "a just-in-time strategy" (p. 229) Weick (1987) Management "There is no split between the composition and Weick (1993a) Management performance; no split between creator and interpreter; and no split between design and production" (p. 6) "Improvisation implies attention rather than Weick (1993b) Management intention drives the process of designing" (p. 351) "Thinking and doing unfold simultaneously" Weick (1996) Firefighting management (p. 19); "Retrospective sensemaking" (p. 19) B. Musical perspectives on improvisation

"Imagination guiding action in an unplanned Chase (1988) Music way, allowing for multitude of split second adjustments" (p. 3) "Improvisation follows not the blueprint method Gioia (1988) Jazz/management but this second approach. The improviser may be unable to look ahead at what he is going to play, but he can look behind at what he had just played" (p. 61) "The spontaneous creation of music" (p. 119) Kernfeld (1995) Music "Free from the effects of previous training" (p. Pressing (1984) Music 345); the opposite of pure composition (p. 346) "Real-time composition" (p. 142) Pressing (1988) Music "Unlike compositional creativity, which involves Sawyer (1992) Music a long period of creative work leading up to a creative product, in improvisational creativity, the creative process and the resulting product are co-occurring" (p. 1)

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 1998 Moorman and Miner 701

TABLE1 Continued

Definition Author(s) Domain

"playing extemporaneously, i.e., without the Schuller (1968) Music benefit of written music" (p. 378) "Improvisation involves making decisions Solomon (1986) Music affecting the composition of music during its performance. The fundamental ideal of improvisation is the discovery and invention of original music spontaneously, while performing it" (p. 226) "The art of spontaneously creating music while Toiviainen (1995) Neural network theory playing or singing" (p. 399) applied to music "The spontaneous act of constructing or Zinn (1981) Music reconstructing; using any immediate or available properties (material or immaterial) into either material or nonmaterial forms used for a specific purpose (function) or need" (p. xii) C. Theater perspectives on improvisation

"to substitute ... staid and preconceived notions Knapp (1989) Theater for the unforeseen, the improvised, the unknown, the world of imponderables" (p. 59) "Playing the game; setting to solve a problem Spolin (1963) Theater with no preconception as to how you will do it; permitting everything in the environment to work for you in solving a problem" (p. 383) D. Therapy perspectives on improvisation

"Practice without prior planning" (p. 22) Embrey, Guthrie, White, & Dietz (1996) Pediatric physical therapy "Complete freedom" (p. 221) Esman (1951) Psychoanalysis "In clinical improvisation, therapists experience Forinash (1992) Clinical music therapy moments of spontaneity, creativity, and intuition" (p. 130) Improvisation occurs when there is "evidence Gardner & Rogoff (1990) Developmental that the child used some lookahead psychology [strategies] but little evidence that the child's drawn solution was based on a completely deliberated plan" (p. 482) "to be highly respectful of form even as it Lichtenstein (1993) Psychoanalysis reworks its previous instantiations" (p. 229) "Being, acting, creating in the moment without Nachmanovitch (1990) Therapy props and supports, without security, can be supreme play, and can be very frightening, the very opposite of play" (p. 23) "Spontaneous music behavior" Orsmond & Miller (1995) Music therapy "Juggling with ideas and feelings (in one's Towse & Flower (1993) Music psychiatric therapy head) and trying out silent interventions" (p. 81) E. Teaching perspectives on improvisation

"To move away from scripted lesson plans" (p. Borko & Livingston (1989) Teaching 473); "An improvisational actor enters the stage with a definition of the general situation and a set of guidelines for performing his or her role, rather than working from a detailed written script" (p. 475) "[to] process information continuously and make Cleary & Groer (1994) Health education numerous interactive or 'inflight' decisions" (p. 110)

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 702 Academy of Management Review October

TABLE1 Continued

Definition Author(s) Domain

"Situational decision making" (p. 200), "To do Graham, Manross, Hopple, & Sitzman Physical education things in lessons they hadn't planned for-to (1993) discover whether they [teachers] actually made up activities on the spot" (p. 205) "Thinking in the midst of action" (p. 630); Irby (1992) Physician training "Responding to the unknown without advance preparation" (p. 630) The creation of (algebra) problems that fit the Yinger (1986) Teaching immediate context of the students' misunderstanding in a way that preplanned problems could not. The fact that the problems were new for both student and teacher meant that the problems were being composed by the teacher on the spot. (p. 6) F. Other perspectives on improvisation

"Reading and reacting in parallel" (p. 1383); Bjurwill (1993) Sports "Dual tasks" (p. 1384); "Perception-in-action" and "thinking-in-action" (p. 1386) no agreed mechanism for changing the British The Economist (1995) Politics constitution or even any agreement about what the constitution actually contains" (p. 18) "Role improvisation is defined as the extent to Powers (1981) Sociology which the organization and meaning of roles are invented by the people immediately involved in a relationship" .... "actors re- interpret, redefine, and re-structure their relationships during the on-going process of interaction" (p. 289) "Rapid, unplanned change" (p. 565) Volkman (1994) Anthropology

on the temporal order in which things happen. tive activity (Weick, 1993a). This definition In ordinary discourse we usually assume that also describes another feature of improvisa- composition of an activity occurs first and is tion noted by many observers: improvisation followed later by implementation or execution. involves a semi-ordered activity. Specifically, In improvisation the time gap between these improvisation typically has a referent or "an events narrows so that in the limit composition underlying formal scheme or guiding image" converges with execution. The more improvisa- (Pressing, 1984: 346). Yet, the order is not fully tional an act, the narrower the time gap between predesigned and, in that sense, is partial. If composing and performing, designing and pro- the order were derived solely from following ducing, or conceptualizing and implementing. standard procedures, we would not consider it One advantage of the proposed definition is improvisation. that it is consistent with several features sug- This definition also follows in a tradition of gested by prior observers of improvisational research on processes focusing on the simul- activity (see Table 1). "Composition" implies taneity of events (e.g., Cohen, March, & Olson, that the improvisational activity involves 1972; Ginsberg & Baum, 1994; Van de Ven, 1986, some degree of innovation, because it goes 1993), of which improvisation is a particular beyond automatically repeating a pre-existing type. By focusing on when rather than whether routine (Crossan & Sorrenti, 1997; Weick, 1996). the design of action occurs, the improvisation Some improvisational actions may represent construct invites attention to a new set of re- only modest shifts from prior behavior, search issues, such as the advantages and whereas others may involve radical innova- disadvantages of the timing of actions. Like-

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 1998 Moorman and Miner 703 wise, some evolutionary theories have con- nal referent and composes new patterns (Berlin- trasted the potential value of fully random and er, 1994; Hatch, 1997b; Weick, 1993b). This form of accidental actions with preplanned actions improvisation-termed "motive" or "theme" im- (Aldrich, 1979; Burgelman, 1983; Miner, 1987; provisation by Weick (1996)-is often associated Weick, 1979). This improvisation definition of- with "free jazz." There, the improviser may begin fers a third mode of change, which may not be with a standard head but moves on to melodic random, but does not necessarily reflect prior improvisations that have internal patterns unre- planning or even stable goals (Follett, 1930; lated to the original harmonic, rhythmic, or me- March, 1976). lodic structures. In organizations, subgroups may create a new product not only outside of, but actually inconsistent with, existing firm Levels of Improvisation strategy (Burgelman, 1983; Hutt, Reingen, & Observers of improvisation agree that there Ronchetto, 1988). A skunkworks team may impro- are, generally, three distinct levels of improvi- vise a new product, sometimes piecing together sation. We describe each level with examples parts, machines, and people from other projects from the arts-especially music-and studies of (Peters, 1988; Sutton & Hargadon, 1997). When organizations. the composition and execution of such projects The first level of improvisation involves mod- tend to converge in time, these activities repre- est adjustments to a pre-existing piece or pro- sent radical improvisation. From popular cul- cess. In jazz, for example, musicians often begin ture, the work of the NASA team that rescued with "the head" of a piece, playing the song and Apollo XIII represented a radical, improvisa- its standard chords, but making slight modifica- tional use of objects outside prior themes or tions in style and emphasis (Bailey, 1980; Ber- structures (Lovell & Kluger, 1995). These forms liner, 1994; Hatch, 1997b). Slightly more aggres- represent a continuum, rather than a set of sive interpretation of a melody, termed sharp categories. embellishment or ornamentation, also occurs (Hatch, 1997b; Preston, 1991a; Weick, 1996). Ob- Organizational Improvisation serving similar levels of improvisation within organizations, Preston argues, "Many courses of Improvisation can be, and often is, executed action within a factory, for example, the re- by an individual. Weick's (1993a) descriptions of scheduling of production to meet customer de- improvisation by firefighters, for example, fre- mands, were really only of a paraphrase variety. quently detail improvisations by specific indi- In these situations the managers sought to make viduals, although he also considers team phe- minor adjustments to the production plan" nomena. Similarly, in much of the literature, (1991a: 93). researchers describe improvisation by individ- A second level of musical improvisation in- ual actors, athletes, therapists, musicians, and volves stronger departures from the referent or teachers. In one standard form of jazz improvi- underlying song. Preston refers to this as "cho- sation, for example, a rhythm section will main- rus phrasing" (1991a: 84), and Weick labels this tain rhythmic order and underlying harmonic "formulaic improvisation" (1996: 21). In this mode structure, while an individual soloist improvises the listener may hear few if any specific phrases in various modes. from the original melody, but at least one ele- Observers of both artistic and organizational ment of the original referent-chord structure, improvisation also have emphasized that collec- rhythm, or style-remains a template around tive improvisation occurs (Crossan & Sorrenti, which the musician improvises (see Berliner, 1997; Preston, 1991a; Weick, 1993a,b,c). For exam- 1994, and, Kernfeld, 1995). Organizational example, Hutchins (1991) describes a collection of peo- ples of this level of improvisation include impro- ple on a ship, on which the navigational system vised new products that represent variations on has failed. As they try to help navigate, calling existing products (Miner, Moorman, & Bassoff, out estimates of coordinates and calculating 1996) and production processes (Stoner, Tice, & subparts of data needed to make navigational Ashton, 1989). choices, they eventually develop a "system" of In the most extreme forms of improvisation, interaction that permits them to get successfully the improviser discards clear links to the origi- to port. However, although the group improvised

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 704 Academy of Management Review October a solution to the instrumentation crisis, no indi- In some cases an individual's behavior can vidual member of the group fully grasped the trigger collective activities that are improvisa- system he or she was creating or why it was tional in nature (Hutt et al., 1988). The individual working. behavior may itself be extemporaneous, as Some scholars reject such ideas as improper when artists created new projects during the reification and argue that such apparently "col- history of the Canadian Film Board (Mintzberg & lective" order should be treated as merely the McHugh, 1985). In other cases the stimulus for aggregate impact of actions by individual hu- improvisation at the organizational level may man beings (Argyris & Schdn, 1978; see discus- actually be planned or deliberate at the individ- sions in Walsh, 1995, and Walsh & Ungson, 1991). ual level of analysis. Eisenhardt and Tabrizi We believe, in contrast, it is meaningful to see (1995), for example, found that aspects of leader such processes as collective improvisation. This behavior played an important role in what ap- approach follows in the tradition of other work, peared to be partially improvisational group in which researchers argue for the value in con- product development (see also Quinn, 1986). sidering other organizational features, such as Likewise, Miner (1987) reports the creation of memory (Cohen, 1991; Feldman, 1989; Hedberg, new jobs that can be seen as organizational involved in- 1981; Huber, 1991; Walsh, 1995; Walsh & Ungson, improvisation but that sometimes 1991), culture (Martin, 1992; Schein, 1985), and tentional individual action. In this article we sometimes draw on observa- routines (Cohen & Bacdayan, 1994; Nelson & tions made of improvisation by groups and Winter, 1982; Winter, 1987). Although studying teams for insight and ideas concerning improvi- such organizational-level phenomena can raise sation. Is it reasonable to apply such ideas to thorny definition and measurement problems entire organizations? We suggest that any entity (Cohen et al., 1995), these problems do not pre- that can reasonably be thought of as planning clude the existence of these phenomena. We or executing action can also be thought of as argue that organizational improvisation can oc- improvising. This logic implies that problems cur if organizational design and implementa- and processes that occur between five improvis- tion can occur. ing musicians can provide a useful lens for con- then, does collective improvisation oc- How, sidering problems and processes affecting five most common metaphors for cur? One of the departments seeking to improvise a joint group improvisation is the notion of "conversa- project. In each case the crucial question is the tion," in which at least two agents interact degree to which design and execution converge around a theme or referent to compose while in time. executing. Using this metaphor, we argue that This is not to say there are not potentially collective improvisation may be produced by complex interrelationships between improvisa- the joint activities of individuals, who are them- tions at different levels of analysis. The mi- selves improvising. The joint action of the indi- crosteps behind a particular organizational im- viduals produces a "system" that we label a provisation may involve improvisational collective improvisation. Theatrical improvisa- actions by small groups, for example. However, tion offers an obvious example here, when the because improvisation can occur at any level of ultimate shape of a scene arises not from any analysis, it is also the case that not all organi- prior plan but from what unfolds after the first zational improvisation is strictly a process of actor generates lines and movement, a second small group interactions (Burgelman, 1984). For responds to that, and the group continues to example, Mintzberg and McHugh (1985) describe interact (Crossan & Sorrenti, 1997; Mangham, a complex interweaving of improvisation at the 1986; Spolin, 1963). Likewise, the improvisation individual and organizational levels in the evo- of technicians and radiologists in Barley's work lution of organizational priorities at the Cana- (1986) was collective, because they jointly devel- dian Film Board. oped routines through their ongoing interaction. Therefore, if the crew members on Hutchins' Constructs Related to Improvisation ship each individually improvised in isolation of one another to save the ship, we would term In this section we assess whether the con- this individual, not collective, improvisation. struct of improvisation adds to the existing

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 1998 Moorman and Miner 705

menu of variables for understanding organiza- vidual using intuition (see, for example, tions. We explore this by arguing that improvi- Hutchins 1991) and that intuition should not be sation does not describe a new empirical phe- used to define whether action is improvisa- nomenon-clearly, it has been observed in tional. organizations for decades-but that, as a formal Process and outcomes associated with impro- construct, it delineates a theoretically rich and visation. In this section we describe several measurable construct that can be distinguished broad processes in which improvisation may from other, related phenomena. play a role. Correlates of improvisation. In this section we Adaptation represents a fundamental con- describe several features of action that may or struct for scholars in many fields (Durham, 1978; may not be present when improvisation occurs. Gould, 1977; Holland, 1975) and often raises On the one hand, our view of improvisation does problems of potentially tautological definitions. not require the presence of these features for Here, we suggest that adaptation involves ad- defining an activity as improvisation. On the justment of a system to external conditions other, these features may tend to occur when (Campbell, 1989; Stein, 1989). Using this view, design and composition merge in time, or they adaptation is a much broader and more general may facilitate improvisation. construct than improvisation and does not nec- Bricolage is defined as "making do with the essarily invoke the same issues of temporal or- materials at hand" (Levi-Strauss, 1967: 17). der. Adaptation can be achieved when an organ- Weick (1993a) suggests that bricolage may be an ization plans in advance to adapt to a change. important part of improvisation in his descrip- An organization can also adapt by making con- tion of firefighters at Mann Gulch. By putting tingency plans for different competitive situa- temporal order at the heart of improvisation, our tions that may unfold. Finally, an organization definition positions bricolage as an important can adapt by deploying standard response rou- skill that can occur during improvisation, rather tines-as when a fire department follows a rule, than as part of its formal definition. Moreover, such as "adapt to the size of the fire by increas- we believe that the more improvisational an act, ing the number of firefighters in the first two the more likely bricolage is to occur, because hours." Therefore, not only are there many forms there is less time to obtain appropriate re- of adaptation not involving improvisation as sources in advance. Finally, we argue that be- listed above, but, even more important, the con- ing skillful at bricolage may actually help pro- struct of improvisation does not imply that all duce valued improvisation. improvisation is adaptive. Thus, our definition Although creativity has been defined in a va- of improvisation focuses on the process of im- riety of ways, a core aspect of most definitions is provisation and not its potential outcomes. By that creativity involves a degree of novelty or doing so, we confirm the need to examine the deviation from standard practice (Amabile, 1983; question of when and how improvisation is or is Barron & Harrington, 1981; Sternberg & Lubart, not adaptive for organizations. 1996). Creativity may involve absolutely no im- We view learning as a process that involves provisation. For example, an organization could the discovery, retention, and exploitation of design a creative marketing plan without using stored knowledge (Epple, Argote, & Devadas, improvisation. However, creativity may repre- 1991; Huber, 1991; Levitt & March, 1988). If an sent an unusually valuable competence for im- organization improvises, assesses outcomes, provising organizations or individuals. and then acts again, this process can be seen as Crossan and Sorrenti define intuition as oper- trial-and-error learning. However, many other ating when "choices" [are] made without formal kinds of learning are not improvisational. For analysis" (1997: 3) and describe it as central to example, an organization may execute a improvisation (Chase, 1988). We agree that intu- planned set of experiments in new ways to or- ition may be part of some improvisation. For ganize itself, may conduct research and devel- example, a firm improvising a marketing strat- opment to learn about the properties of a key egy may skip market research and follow the material, or may acquire knowledge by hiring intuition of a project leader to guide a product people from outside (Huber, 1991). None of these introduction. However, we also argue that col- activities necessarily involves improvisation at lective improvisation can occur without an indi- all. We argue that automatically equating im-

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 706 Academy of Management Review October provisation with learning again confounds a improvisation is the impact of prior routines and process with one of its potential outcomes. The knowledge of the improvising units (Eisenhardt definition here leaves completely open the & Tabrizi, 1995; Hatch, 1997b; Weick, 1993a, 1996). question of whether improvisation produces For instance, Berliner (1994) describes how mu- learning. Later, we argue that improvisation sicians draw on pre-existing skills to append may, indeed, represent one avenue of organiza- grace notes and produce chromatic fills. Barley's tional learning, and may even be learned itself. (1983) account of the creation of new interaction It is important that these possibilities not be left routines between radiologists and technicians untested, which could happen if we equate the suggests that their new routines drew on previ- construct of learning directly with the definition ous interaction routines. Metcalf's (1986) account of improvisation itself. of the Grenada rescue operation describes the Scholars have defined innovation as a devia- military units involved as having certain impor- tion from existing practices or knowledge (Rog- tant mental skills useful in improvising the res- ers, 1983; Zaltman, Duncan, & Holbek, 1973). Be- cue of an official during the operation. These cause improvisation involves some level of and other examples suggest that prior knowl- creation or design, we believe it is a form of edge and routines are important to improvisa- innovation. However, it is just one kind of inno- tion, but that the presence of stable competen- vation. For example, if an organization inno- cies is not inconsistent with improvising. vates a new way to store chemicals by analyz- Building on this insight, we explicate specific ing its needs, gathering facts, designing the propositions about how the effectiveness of or- new storage facility, getting bids, and finally ganizational improvisation depends upon the building it, the organization has innovated, but processing of stored knowledge, termed organi- it has not improvised. Furthermore, improvisa- zational memory. In Figure 1 we summarize the tional activities may involve very different lev- proposed relationships. Specifically, the figure els of innovation; a very large proportion or rel- suggests that improvisation has a neutral rela- atively small proportion of the activities may be tionship with organizational outcomes, but that novel. Some level of innovation is required for procedural memory (skill or action knowledge) an activity to meet the temporal order definition and declarative memory (fact knowledge) mod- because of its emphasis on design or creation of erate this relationship. In addition to showing action. These considerations suggest a subtle how memory influences the impact of improvi- link between innovation and improvisation. sation, the figure shows that improvisation can, They also imply, however, that treating them as over time, shape the development of organiza- synonymous could reduce our ability to under- tional memory. stand the specific issues of temporal order by confounding degree of innovation with degree of improvisation. Focal Organizational Outcomes Clarifying these distinctions underscores the idea that our proposed definition makes no pre- Organizational scholars writing on improvi- dictions or assumptions in and of itself regard- sation typically have emphasized instrumen- ing the consequences of improvisation. This tal outcomes (Brown & Eisenhardt, 1995; Eisen- makes improvisation more useful as a research hardt & Tabrizi, 1995; Preston, 1991a; Weick, area, since it is then feasible to theorize and test 1993a,c, 1996). Two specific criteria appear to ideas about the conditions under which impro- dominate such outcomes. First, does the im- visation is and is not valuable. We argue here provisation solve a problem? For example, im- that organizational memory represents one of provisation saved the lives of firefighters at the key determinants of the nature of improvisa- Mann Gulch (Weick, 1993a) and helped subvert tional outcomes. a lorry strike (Preston, 1991a). Second, does improvisation permit the organization to har- vest unanticipated opportunities? Brown and IMPROVISATION,MEMORY, AND Eisenhardt for describe how ORGANIZATIONALOUTCOMES (1995), instance, product development teams improvised to One thing that stands out in careful accounts meet customer needs and introduce timely of organizational, artistic, and other forms of new products.

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 1998 Moorman and Miner 707

FIGURE 1 The Relationship Between Organizational Memory and Organizational Improvisation

Organizational declarative memory Action novelty

Organizational Organizational Action c memory improvisation } K \* Declarative * Procedural

Organizational procedural memory Action speed

Time Time1 Time2 The interaction of memory and improvisation Action outcomes Selective retention of action outcomes

Scholars focusing primarily on musical, the- product development team to improvise in a rela- atrical, and other artistic improvisation typi- tively unoriginal manner, producing something cally emphasize aesthetic outcomes (Berliner, that is not very different from prior products. In 1994; Hatch, 1997b; Spolin, 1963). Although it is contrast, the team could improvise and produce possible to analyze both instrumental and an exceedingly unusual and novel product. aesthetic outcomes, we have chosen to focus Third, speed refers to the time to plan and on three instrumental outcomes, given the or- execute an action. Some observers have argued ganizational context. First, action can vary in that executing well-made plans speeds up ac- the degree to which it displays coherence. Co- tion by eliminating unnecessary steps, prevent- herence refers to the degree to which an action ing time-consuming mistakes and midcourse displays internal fit (components of an action changes, and building a -common understand- have an internal consistency or order and ing that reduces coordination problems (Cooper work together well) and external fit (compo- & Kleinschmidt, 1986). Others have suggested nents of an action produce effects that fit the that improvisation may speed up action be- performance context). cause it supplants lengthy planning (Eisenhardt Second, whereas all improvisation involves a & Tabrizi, 1995). This equivocality makes our threshold degree of deviation from prior routines, investigation into the impact of improvisation improvisation can produce varied levels of nov- on speed not as straightforward as it might oth- elty in action. For example, it is possible for a new erwise seem.

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 708 Academy of Management Review October

The Nature of Organizational Memory individual or organization is operating- whether it be jazz (Berliner, 1994; Hatch, 1997b), As with organizational improvisation, scholars physical therapy (Embrey et al., 1996), physical disagree about whether organizations store infor- education instruction (Graham, Manross, mation in memory as individuals do. However, Hopple, & Sitzman, 1993), psychiatry (Esman, there is a growing sense that organizations do 1951; Lichtenstein, 1993), sports (Bjurwill, 1993), have frames of reference, routines, structures, and new product development (Brown & Eisenhardt, other physical artifacts that reflect the presence of 1995; Sutton & Hargadon, 1997), or fighting wild- stored knowledge (see Moorman & Miner, 1997; land fires (Weick, 1993a, 1996). Walsh, 1995; and Walsh & Ungson, 1991, for re- In one study of jazz, Berliner provides dense views of this literature). We adopt this perspective descriptions of the critical role that procedural but do not limit our analysis of organizational memory, which he calls "aural memory," plays memory to the "storage bins" often associated in improvisation. He describes good improvisers with it, nor to its associated acquisition or reten- as having large "vocabularies," "repertory store- tion processes or consequences (see Moorman, houses," "musical biographies," or a "reservoir 1995, and Walsh & Ungson, 1991).Instead, we focus of technique" of "little sections or melodic frag- on the content and level of organizational memory ments" that "provide readily accessible mate- (Moorman & Miner, 1997, 1998). rial that meets the demands of the composing The content of memory refers to what Walsh music in performance" (1994: 102). and Ungson (1991) describe as the "what" of or- A key characteristic of procedural memory is ganizational memory (Walsh, 1995). In this arti- that it becomes automatic or accessible uncon- cle we focus on two types of memory often as- sciously. Often referred to as "motor memory" sociated with individuals-(l) procedural and (2) (Pressing, 1988), procedural memory includes declarative (Anderson, 1983)-that scholars re- the skills needed to ride a bike or use a type- cently have found to be associated with collec- writer (Cohen, 1991; Pressing, 1984; Singley & tives as well (Cohen, 1991; Cohen & Bacdayan, Anderson, 1989). Procedural memory, therefore, 1994). The level of memory, which refers to the often represents tacit knowledge for individuals amount of stored knowledge and experience, is and organizations (Cohen, 1991; Cohen & Bac- a well-known concern among cognitive psychol- dayan, 1994; Nonaka, 1990; Winter, 1987). ogists who study expert-novice differences As a result of its automaticity, we argue that (Chase & Simon, 1973; Chi, Glaser, & Rees, 1981; procedural memory is likely to have contrasting Chiesi, Spilich, & Voss 1979), and is increasingly effects on improvisation. First, by providing a accepted by scholars as a trait of organizations rich vocabulary of action from which to choose, as well (Cohen & Levinthal, 1990; Moorman & it can improve the likelihood that improvisation Miner, 1997, 1998; Walsh & Ungson, 1991). As an will produce coherent action. In a musical set- example of the content and level of memory, an ting, for example, high levels of procedural organization that has been working in a partic- memory provide the jazz artist with a large rep- ular industry for an extended period of time will ertoire of potential actions. As a result, the artist likely accumulate a high level of declarative is able to select the most effective action, de- memory about the competitive structure and de- pending on the context and the actions of other tailed traits of this industry. It might also accu- band members. Likewise, Weick's (1993a) de- mulate a number of standard practices for deal- scription of the bricoleur as a master improviser ing with others in the industry, representing a stems largely from the observation that the bri- high level of procedural memory. coleur draws on a repertoire of pre-existing routines, allowing for the creation of a tool that solves the problem at hand. Procedural Memory in Improvisation In the organizational context, Brown and Procedural memory is memory "for how things Eisenhardt (1995) suggest that firms with deep are done" (Cohen & Bacdayan, 1994: 404) or technological routines will be more likely to memory for "things you can do" (Berliner, 1994: generate improvisations in new product devel- 102). Therefore, procedural memory involves opment. Although they do not use "improvisa- skills or routines. The nature of these skills will tion," Mintzberg and McHugh (1985) describe depend on the particular domain in which the how the Canadian Film Board continuously gen-

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 1998 Moorman and Miner 709 erated projects without a clear plan or goal, found that novice teachers took more time and embodying what they define as an "adhocracy." were less efficient than experienced teachers in Their description of the organizational features responding to students who led them away from that supported this strategy clearly includes a scripted lesson plans (1987: 473). rich density of routines embodied in both people Accounts of improvisation in war likewise and technology within the Film Board. The now- suggest that rich repertoires of procedural rou- famous account of Honda's U.S. introduction of tines play a role in fast improvisation. As Vice the Supercub motorcycle provides another or- Admiral Joseph Metcalf observed during the ganizational example of the impact of proce- 1983 U.S. invasion of Grenada, "The rescue of the dural memory. Honda's planned introduction of governor-general had not been included in any large motorcycles through traditional methods of my earlier instructions. But it soon became experienced difficulties, but managers noticed apparent ... that his rescue was of paramount that nontraditional motorcycle customers tried importance" (1986: 288). He goes on to describe a to buy the Supercubs ridden by Honda's repre- rapidly improvised set of troop movements and sentatives. The Honda team successfully re- a landing on an unsurveyed beach. Metcalf at- sponded to the apparent demand in this context tributes the success of the rescue to the pre- by improvising a new strategy to sell Supercubs paredness of the troops, which-in our terms- through sports stores, discarding their original included a diverse and highly practiced plan. Clearly, the effectiveness of this improvi- repertoire of routines concerning movement of sation depended heavily on the fact that the ships, movement of landing vehicles, and land- Honda team had a repertoire of marketing, ing actions of Marines. This pre-existing proce- sales, financial management, and technical rou- dural memory was recombined and redeployed tines (Mintzberg, Pascale, Goold, & Rumelt, 1996; for a purpose and in a setting not part of the Pascale, 1984). original mission. Thus, we propose: Although our argument about the importance Proposition 2: The greater the proce- of procedural memory applies to many forms of dural memory level, the greater the innovation, it carries special strength with re- likelihood that improvisation will pro- spect to improvisation. In planned innovation, duce speedy action. organizations can gather, in advance, tools needed to implement change. They can acquire, A high level of procedural memory can also for example, both physical tools as well as ideas have a third and contrasting effect on improvi- from sources outside the organization. However, sation, which is to constrain novelty. As Berliner given that there is little or no time between notes, "In one of the greatest ironies associated conceiving of and executing an action in impro- with improvisation, as soon as artists complete visation, whether improvisation produces coher- the rigorous practice required to place a vocab- ent action depends fundamentally on the exis- ulary pattern into their larger store, they must tence of a large number and variety of guard against its habituated and uninspired procedural routines that can be recombined to use" (1994: 206). Lonnie Hillyer, a famous jazz fit in a given context. Therefore, we propose the musician, has been quoted as saying, "Just try- following: ing to make phrases come out differently is hard at times, very hard, because we are pro- Proposition 1: The greater the proce- grammed" (Berliner, 1994: 206). This constraining dural memory level, the greater the effect, to some degree, may be true of all types of likelihood that improvisation will pro- information stored in a memory (Leonard- duce coherent action. Barton, 1992). However, its consequences are In addition to increasing the likelihood that more problematic in the case of procedural improvisation will produce coherent action, pro- memory, because procedural memory tends to cedural memory is also likely to improve the be accessed automatically (Cohen, 1991). speed of improvisation. Increased speed is due Consistent with this viewpoint, scholars have to the automatic or tacit quality of procedural referred to procedural memory at the individual memory, thereby producing an "economy of ac- level as "use specific" (Singley & Anderson, tion" (Pressing, 1984: 355). In support of this, 1989), which means that its accessibility and Carter, Sabers, Cushing, Pinnegar, & Berliner utilization are tied to particular issues or con-

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 710 Academy of Management Review October texts present when learning a skill. For exam- noting, "One cannot stress too much the impor- ple, Singley and Anderson (1989) demonstrate tance of total mastery of the old disciplines of that differentiation and integration skills in cal- harmony, counterpoint, all types of canon and culus generally do not facilitate learning of fugues" (1980: 50). Finally, Spolin (1963) de- each other, nor do language generation and scribes several important "principles" important comprehension skills. Carraher, Carraher, and to successful theatrical improvisation, including Schliemann (1985) have found that Brazilian good group dynamics and an effective relation- child street vendors' knowledge of math compu- ship with the audience. tations is limited to use in the marketplace-not A key characteristic of declarative memory is in a laboratory setting. Furthermore, sports com- the variety of uses to which it can be put. As mentators have noted that a player with strong Anderson has suggested, "[D]eclarative knowl- procedural skills often will be "blinded" by the edge provides a basis for transfer between dif- ball and that this "narrowing of attention referent uses of the same knowledge" (1983: 220). duces the visibility of important things that go This characteristic leads to contrasting effects on around him" (Bjurwill, 1993: 1384-1385). on improvisation. First, by providing the basis to Groups with strong procedural memories transfer to a number of new areas, declarative have also been found to be restricted. For exam- knowledge should improve the likelihood that ple, Dougherty (1992) found product develop- improvisation will produce coherent action. Spe- ment teams with well-established procedures cifically, when improvisers have rich stores of least likely to deviate from pre-existing action declarative memory, they are able to recognize Weick how patterns. Similarly, describes expe- various patterns in external events (i.e., see the rienced firefighters were unwilling to "drop higher-order principle) and to select actions that their packs and tools" to "run faster" to escape link their actions to these events so that a coher- catastrophic death (1996: 1). He speculates that ent whole is achieved, both within the action this tendency was due to firefighters "over- itself and within the context. learning" certain skills-a characteristic of pro- The importance of declarative knowledge in cedural memory (Neustadt & May, 1986; Walsh, making sense out of new situations, deriving 1995). Thus, we propose: meaning from unstructured situations, or using Proposition 3: The greater the proce- principles to predict outcomes has been noted in dural memory level, the greater the descriptions of scientific discoveries, of course. likelihood that improvisation will pro- In a famous example, historians of science often duce action low in novelty. assume that Fleming's extended declarative knowledge of facts and theories about bacteri- ology allowed him to see the important implica- Declarative Memory in Improvisation tions when he noticed that bacteria did not grow Declarative memory is "memory for facts, around an accidental spore of the mold Penicil- events, or propositions" (Anderson, 1983; Cohen, lium notatum (Jewkes, Sawers, & Stillerman, 1991: 137). Therefore, unlike procedural memory, 1969). At the firm level, Cohen and Levinthal which involves routine or skill memory, declar- argue that "fortune favors the prepared firm" ative memory can be more general. For exam- (1990, 1994), referring to their finding that firms ple, although the procedural memory of riding a investing in R&D (which, by nature, is highly bike may be useful in some situations, declara- declarative) tend to be more effective at recog- tive memory of the mechanics principles under- nizing, interpreting, and using knowledge cre- lying riding a bike may have more general ap- ated outside the firm. At the group level, declar- plications. ative memory may reside both in collective The importance of declarative memory to im- knowledge structures, such as shared informa- provisation is found in a number of domains. In tion (Walsh, 1995), and also in material forms, jazz, for instance, Kernfeld (1995) describes the such as blueprints, reports, summaries of prior acquisition of musical theory about chord pro- research results, and the like. We propose: gressions or rhythmic patterns as critical to improvisation. For organ improvisation, Bailey Proposition 4: The greater the declar- points to the importance of declarative memory, ative memory level, the greater the

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 1998 Moorman and Miner 711

likelihood that improvisation will pro- Although declarative memory can enhance duce coherent action. the chances of coherent and novel improvisation, it has one distinct drawback. Specifically, In addition to assisting with pattern recogni- as Singley and Anderson note, because "declar- tion, declarative memory does not carry as much ative knowledge is not committed to a particular risk of falling into pre-existing or "use-specific" use, vast amounts of it are potentially relevant (Singley & Anderson, 1989) patterns as does pro- in any problem-solving situation, and this leads cedural memory. Because declarative memory to serious problems of search" (1989: 220). For typically is more theoretical and abstract, it may example, deductive reasoning (a declarative be applied to numerous situations in countless stock) can be retrieved in learning math, playing ways, thereby encouraging greater novelty. For games that rely on logic, or in solving commonly example, in cognitive research, Singley and encountered questions in life (Anderson, 1983). Anderson (1989) have shown knowledge of alge- In the organizational setting we can see evi- bra to be utilized in understanding some calcu- dence of this drawback to declarative memory lus problems. Berliner describes the impact of in improvisation in numerous reports of organi- declarative memory on jazz improvisation when zations designing comprehensive information he notes, "For many, the ability to interpret an systems but finding them disappointing in times appealing idea in theoretical terms facilitates of crisis (Feldman & March, 1981; Goodhue, exploration of the idea's implications, generat- Wybo, & Kirsch, 1992; Laudon & Laudon, 1996). ing new phrases in the process of experimenta- For example, a firm may need to improvise a tion" (1994: 168). Finally, Weick (1996) suggests modified manufacturing process in one plant, that a long, detailed list of firefighting proce- but it may fail to draw on knowledge of similar dures may be harder to adapt to unusual condi- situations in other parts of the firm, because it is tions. Therefore, general knowledge may be too hard to find out rapidly if and where such more useful in situations requiring improvisa- knowledge exists (Goodhue et al., 1992; Laudon tion. & Laudon, 1996). In another example, during the This role of declarative memory influences in- Cuban missile crisis, Kennedy and his advisors novation of many types, but it has particular had to create and execute responses within a importance for improvisation because of the very tight time frame. During this process, it took rapid creation of rich meaning in novel actions. an extensive "search" through different govern- If there is substantial time to plan in advance of and individuals access knowl- implementation, one can seek declarative mem- ment agencies to in ory during the conceptualization or design pro- edge about the true state of events several cess. This might occur, for example, if an organ- important domains. As a consequence, although ization were planning in advance to move into a declarative memory was present, in principle, new technology and sought knowledge from some of it was not used in improvisational as- outside the organization. In improvisation, how- pects of the crisis (Allison, 1971). ever, the organization designs action while im- This difficulty of declarative memory often plementing it, so the strength of its own declar- has been seen by scholars as part of the divide ative memory has a crucial impact on the between engineers and production profession- degree to which improvisation produces novel als. In some settings production professionals actions. This observation informs the standard believe engineers and others with strong theo- insight that to sustain a "first mover" strategy in retical knowledge create problems by searching a fast-moving product market, organizations through far too many forms of abstract knowl- typically need stable access to cutting-edge sci- edge, instead of drawing on heuristics that can ence, whether through their own research or be deployed rapidly (Dean & Susman, 1989). through links with sources of basic science Given these search problems, we expect that (Maidique & Patch, 1988). We propose, then: improvisation will produce slow actions when declarative stocks are accessed. We predict: Proposition 5: The greater the declarative memory level, the greater the Proposition 6: The greater the declar- likelihood that improvisation will pro- ative memory level, the greater the duce novel action. likelihood that improvisation will pro-

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 712 Academy of Management Review October

duce action that proceeds at a slower (1989) describe expert teachers as effective im- pace. provisers, because their theoretical knowledge can assist in choosing quickly those pre- existing routines to deploy in a particular set- Combining Procedural and Declarative ting. Hutchins (1991) likewise describes how de- Memory in Improvisation clarative memory of how key navigational Our propositions imply that both procedural calculations influenced action made clear the memory and declarative memory involve need for someone to retrieve a navigation cal- tradeoffs for improvisation. Procedural memory culator to save the ship. In all of these cases, increases the likelihood that improvisation will declarative memory appeared to guide the cre- produce coherent and rapid action, but it ative deployment of procedural memory. strengthens the danger of automatic behavior. This combination is facilitated by organiza- In contrast, declarative memory permits richer, tional structures and routines that funnel expert more complex meanings and connections (thus declarative memory directly into procedural enhancing potential novelty and coherence), but routines. For example, anticipating the need for it also makes timely improvisation less likely periodic improvisational troubleshooting to since it demands substantial search time. handle manufacturing crises, a chip manufac- Therefore, declarative and procedural mem- turer may require that a scientist with theoreti- ory can be viewed as complementary competen- cal knowledge be on call 24 hours a day. This cies that offset one another's weaknesses. As a means that the procedural routines followed by result, their joint deployment may be especially technicians would be less likely to restrict inno- effective in ensuring that improvisation results vativeness, because they would be comple- in coherent, novel, and speedy action. Bjurwill mented by more formal declarative memory. (1993), in the sports literature, describes the crit- Therefore, the degree to which improvisation icality of both mental skills (declarative mem- produces novel action may depend on whether ory) and motor skills (procedural memory) for the improviser is able to use declarative memory thinking in action or improvisation. Likewise, to make creative use of procedural memory. Alinksy (1969) describes how community groups Such skills might involve applying pre-existing improvised the tactic of using stock shares to routines to new contexts, recombining subunits gain access to shareholder meetings in order to within pre-existing routines, and recombining influence local firms. The community groups entire routines in new ways. Hutchins (1991) de- used many of their pre-existing organizational scribes how the navigation team improvised by routines, such as protest signs and the willing- engaging in all three of these skills. First, they ness to confront authority. However, these rou- applied typical computation sequences to the tines were only effective in combination with new context of trying to save the ship. Second, the declarative memory of the firms they sought this approach evolved into a mediating struc- to influence and of social mores in corporate ture, in which the team used a calculator that boardrooms. changed the relation of the workers to the task Considering the specific mechanisms under- and resulted in the team performing a key rou- lying this complementary relationship, we have tine in a nonstandard sequence (i.e., they recom- emphasized declarative memory's generative bined units within the pre-existing routine). Fi- potential for suggesting new meanings, new in- nally, the team added a person to the terpretations, or new linkages between concepts computation process, thereby changing the and action. There is also some evidence, how- availability of data and the accuracy of team ever, that declarative memory plays an impor- understanding and predictions. tant role in the deployment of procedural mem- Considering the limitations of declarative ory. Specifically, Anderson suggests that memory, we suggest that the degree to which subjects who have a great deal of declarative improvisation positively impacts speed may de- memory are also more likely to perform well in pend on whether the improviser has developed the "abstracted planning space of operator se- procedural skills that allow for rapid access to lection" (1983: 212), which is knowing which be- declarative memory stores. Such skills require havior to employ given certain problem condi- the ability to find and incorporate declarative tions. In support of this, Borko and Livingston memory into action as improvisation occurs. At

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 1998 Moorman and Miner 713 the individual level of analysis in artistic impro- sation suggests that it is dependent on the or- visation, field reports suggest that individuals ganization's (1) ability to use declarative require time to develop procedural skills that memory to make creative use of procedural allow them rapid access to declarative memory. memory and (2) development of procedural In jazz, for instance, Berliner (1994) notes that skills that allow for rapid access to declarative musicians must not only know their chord theory memory stores. This strategic deployment of but also learn to quickly access that knowledge memory requires the organization to be very as- during performance. In theater, improvisational tute regarding how and when to utilize currently actors may draw on declarative memory of liter- held knowledge and routines. The implication, ature, current events, or science, but they must therefore, is to reemphasize the importance of practice to develop procedural skills in being how memory is used by the organization, over able to "find" and quickly incorporate such the mere accumulation of memory. We propose knowledge into action as they improvise (Spolin, the following: 1963). Within organizations, scholars continue to Proposition 7: Combining procedural wrestle with the complex issue of how informa- and declarative memory can enhance tion and knowledge can be stored and accessed the probability that improvisation will formally (Feldman & March, 1981; Lee, Barua, & produce coherent, novel, and speedy Whinston, 1997). Moreover, research for several action. decades has suggested that social ties constitute important avenues for storing and access- The Impact of Improvisation on Organizational ing collective knowledge (Scott, 1992). Consis- Memory tent with this, organizations that deliberately seek to merge creation with execution, such as In the previous sections we explored how or- high-tech startups or firms seeking rapid prod- ganizational memory influences organizational uct development, sometimes maintain multiple improvisation. In this section we consider how mechanisms for rapid sharing of declarative improvisation impacts organizational memory. memory-through designing buildings with Improvisations as experiments that influence open office settings, enhancing interactions in memory. Observation of improvisation in the lunch rooms and informal settings, or creating arts suggests that improvisational actions may norms that permit multiple contacts with varied serve as experiments that shape future behavior sources (Bowen, Clark, Holloway, & Wheel- and the memory of the actors. Preston notes, "In wright, 1994; Brown & Eisenhardt, 1996; Moor- jazz, collective improvisation has much to do man, 1995; Moorman & Miner, 1998). with rehearsal. Much collective improvisation Quick access to declarative memory might takes place in small ensembles during jam ses- also include routinizing search processes (creat- sions. The fruits of the collective improvisation ing procedural memory about how to find are then incorporated into subsequent public things). For example, some organizations have performances" (1991a: 84). invested in dynamic internal electronic commu- Two theoretical frameworks imply that a sim- nication systems involving rather sophisticated ilar process occurs in organizations. Theories of search engines that allow employees to access organizational evolution and trial-and-error specific organizational declarative memory learning emphasize that random variations in quickly (Laudon & Laudon, 1996; Schatz, 1992). organizational activity may produce unexpected Specifically, network designers in large organi- benefits. Firms then observe the beneficial con- zations, with highly developed intranets for sequences of these random variations, and they their firms, can specifically configure search en- repeat and even institutionalize these acciden- gines to help employees find certain product tally discovered actions over time so that they specifications, safety procedures, and other de- become part of the organization's memory (Al- clarative memory likely to be needed during drich, 1979; Miner, 1987; Weick, 1979). Cyert and emergency improvisational activity. March describe a similar process and suggest Our discussion of the impact of the potential this can constitute a process of (unplanned) ad- complementary impact of declarative and pro- aptation (1992: 117). In these models the initial cedural knowledge on organizational improvi- activity-later repeated-may be nonrandom,

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 714 Academy of Management Review October possibly resulting from efforts to solve an imme- Proposition 8: Improvisational actions diate problem. In addition to simply repeating can serve as unplanned experiments successful actions, organizations may draw in- that generate changes in an organiza- ferences beyond simple trial-and-error learning tion's procedural and declarative that guide future action as well. Finally, some memory. observers argue that the process of acting and observing its consequences may produce deep- Improvisation as higher-order procedural er-level learning in terms of goals and interpre- memory. Finally, we suggest that it is possible tive schemes (Follett, 1930; March, 1976; Weick, for an effective improviser to acquire-through 1993c). practice-a general skill or competency in im- Upon considering empirical evidence related provisation itself. This skill, while procedural, to this claim, we note that in some cases organ- represents a metaroutine, as opposed to a single izational improvisation appears to serve local routine that the improviser can access during purposes and leaves undisturbed long-term or- improvisation. Our review of the literature sug- ganizational memory. For example, the improvi- gests that this skill can be learned and that it sation of the new navigation system described can, in turn, influence a variety of improvisa- by Hutchins (1991) permitted the large vessel to tional outcomes (see Borko & Livingston, 1989, avoid disaster at a particular point in time, but it and Pressing, 1984). presumably had no lasting influence on organ- One can find interesting anecdotal support for izational processes, once the regular equipment the idea that improvisation represents a distinct was running again. However, in other cases im- organizational competency in Metcalf's (1986) provisational action can and does influence discussion of the Grenada invasion, in which he organizational memory. Miner (1991) describes argues that the Navy needed to practice impro- how such activities as solving employment visation itself. Simply having good routines and problems, analyzing financial data, and smart actors was not sufficient to produce effec- organizing training meetings moved from being tive improvisation. Similarly, a qualitative improvisational actions to solve a local problem study of new product development activities into new formal jobs, some of which were still in showed two firms that had both developed tra- place 6 years later. In a manufacturing context, ditions that actively encouraged some improvi- Stoner et al. (1989) describe how a firm not only sation within formalized programs, and per- improvised a series of cellular manufacturing ceived these traditions as strengths (Miner et al., constellations but retained those that seemed 1996). effective and kept experimenting with others. In other work Brown and Eisenhardt (1995) The procedural memory used in production identify three firms that exhibited the skill of within these cells, therefore, was shaped managing improvisation. They found that new through this process. product development at these firms relied on Improvisational actions also can contribute to real-time interaction and a targeted structure. the organization's declarative memory. For ex- From the perspective of our framework, the real- ample, Preston (1991a) describes how managers time interaction among team members repre- responding to a lorry driver strike improvised a sents a pooling of procedural and declarative new meaning to their firm identity, moving from memory that increases the amount of knowl- a "plastics" company to a "food" company, edge available for recombination in the firm. which exempted them from the strike. This new Brown and Eisenhardt (1995) also argue that identity, in turn, became a part of the organiza- structured managerial roles and clear project tion's long-term identity. Similarly, Miner et al. priorities are essential to effective improvisa- (1996) describe how design engineers impro- tion. In our framework this structure is likely to vised by rewiring part of a product when it evoke routines (procedural memory) at the ap- failed to work as expected. When the improvi- propriate time and place. Finally, although we sation failed, they then investigated why it see the organizational improvisational compe- failed and discovered new theoretical knowl- tency as largely procedural, it also appears that edge about the product's technological base, some organizations are becoming increasingly which was recorded and became part of the aware of their own improvisational activities. organization's declarative memory. We predict: When they do, they sometimes develop declar-

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 1998 Moorman and Miner 715 ative memory about the process of improvisa- of organizations (Meyer et al., in press). We sug- tion itself (e.g., Sutton & Hargadon, 1997). gested that this is because advances in technology make improvisation not only more likely but Proposition 9: The greater the organi- also more valuable (Preston, 1991b). Our work zational improvisation level, the more provides a sense of the tradeoffs associated likely an organization will be to de- with improvisation, as well as the investments velop a higher-order organizational required for effective improvisation. We sug- in competency improvisation over the gested that a key issue in ensuring that impro- long run. vised action is effective is the development and deployment of organizational memory, includ- DISCUSSIONAND IMPLICATIONS ing the systems that drive its accessibility and utilization during improvisational events (see In the preceding sections we first built on Moorman & Miner, 1998). However, because or- prior research to distill a formal definition of ganizational memory is both costly and difficult improvisation-as the degree to which design to develop, the frequency of improvisation and and execution converge in time-to illustrate the importance of improvising well should, in degrees of improvisation in various settings and the end, drive the degree of investment that the to clarify what collective improvisation is and organization makes in memory systems. when it occurs. Drawing on research suggesting Our definition of improvisation and the frame- that memory may deeply affect improvisational work we have proposed also provide some res- outcomes, we then proposed that procedural olution to a number of theoretical issues. In and declarative organizational memory influ- terms of the construct, our definition offers a ence the degree to which improvisation will pro- more specific and distinct view of improvisation duce coherent, rapid, and novel action and that than generally has been used by scholars in the improvisation can, in turn, influence memory. organizational literature. It also suggests that In this closing section we reconsider our ini- the appropriate focus should be on the simulta- tial theory-based and practice-based problems neity of design and action-not on whether and challenges in initiating this work. In doing planning should or can be eliminated from or- so, we seek to demonstrate that our ideas about ganizations. This approach promotes a different improvisation and the proposed framework con- set of research questions concerning the advan- tribute to resolving these challenges. We accom- tages and disadvantages of various timing op- plish this in several ways. First, we examine tions. For example, designing nearly simulta- issues specifically related to the main foci of our neously with action may enhance success by propositions-memory, improvisation, and or- reducing the time for opposition to organize col- ganizational outcomes-and conclude that our lective resistance (Pfeffer, 1981, 1982). propositions imply a set of rather demanding In this article we focused on one research is- conditions within which improvisation may sue: the impact of organizational memory. We prove effective. Second, we offer concrete advice argued that organizational memory is important to consider in the operationalization of organi- to all forms of organizational innovation, but of zational memory and organizational improvisa- particular importance to organizational impro- tion. Third, we offer insight into selected bound- visation, because of the convergence of compos- ary conditions of both our view of organizational ing and acting. Specifically, as composing and improvisation and its relationship to organiza- acting converge, there is less time for organizational memory. tions to purchase or develop knowledge and skills necessary to complete action. This approach supports work that has emphasized the Improvisation, Memory, and Organizational identification of factors Outcomes influencing improvisational effectiveness (Crossan & Sorrenti, 1997; In opening our article, we argued that impro- Eisenhardt & Tabrizi, 1995; Weick, 1993b), by ex- visation long has characterized actions in se- plicating the moderating impact of two different lected organizations, such as startups or cre- memory types on three different organizational ative agencies, but now represents a crucial outcomes. This contrasts with work by authors factor in the prosperity of many different types who have actively criticized acting without

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 716 Academy of Management Review October planning well in advance, in great detail (Coo- izations, including order planning, order gener- per & Kleinschmidt, 1986), or have emphasized ation, order entry and prioritization, order the potential virtues of improvisation (e.g., Pas- scheduling, order fulfillment, and billing and cale, 1984). payment. Therefore, to measure procedural Our propositions also clearly imply that im- memory related to this domain, one could mea- provisation may be effective only within a very sure the degree to which an organization has narrow window of organizational contexts. They stored routines regarding these processes. This point to boundary conditions that may constrain might include examining the number of formal the short- and long-term value of improvisa- written procedures as well as informal norms tional activity. For example, the propositions im- and accepted practices. ply that improvisation may be ineffective at To measure declarative memory, one could best, and possibly harmful, unless an organiza- examine the degree to which an organization tion has a rich repertoire of procedural memory, has factual information relevant to this process, in the form of organizational routines, and a principles invoked in making decisions in this deep reserve of declarative memory, in the form domain, and conscious representations of the of more abstract or theoretical information. Al- reasons for the procedures used. Factual infor- though not startling, these ideas imply a fairly mation in the example of order fulfillment pro- unusual combination of organizational memory cesses might include knowledge; drawings; for- that must be developed and deployed for impro- mulas; or stored facts concerned with post- visation to represent a fruitful tool in organiza- sales-service activities, with cost-estimation tional effectiveness. pricing activities, and with relevant manufacturing and logistics processes. Such declarative memory might be stored in written documents, Operationalizing Our Propositions databases, group records, individual knowledge Organizational memory. Testing our theoreti- bases, and-in selected organizations-intranet cal propositions requires measuring procedural systems specifically designed to make such de- and declarative memory in organizations. The clarative memory widely available within an measurement of organizational memory is a organization. These intranet systems might con- complex task, but in a growing body of work, tain drawings of parts specifications and stan- scholars have begun to address it (Cohen & Bac- dards that the firm uses, site drawings of build- dayan, 1994; Cohen & Levinthal, 1990, 1994; ings and systems within the buildings, laws of Epple et al., 1991; Moorman & Miner, 1997, 1998; physics or specific formulas driving firm prod- Walsh, 1995; Walsh & Ungson, 1991). Cohen and ucts, specifications for product families, and Bacdayan (1991) provide important evidence company policies. that we can both define and measure memory at Organizational improvisation. Concerning the the collective level, using experiments with measurement of improvisation, we see several small groups. Our definition of these constructs plausible approaches to assessing improvisa- is consistent with their view that declarative tion in field studies of organizations. memory involves the storage of facts, proposi- Drawing on its core definition, the degree of tions, and events, whereas procedural memory improvisation can be operationalized by esti- stores the elements of skills and routines. Fur- mating the length of time between design and ther, we suggest that both procedural and de- execution of action. The smaller the gap, the clarative memory can reside in social structures higher the degree of improvisation. Estimates of and practices (such as in group values or rou- the gap could be gained in a number of ways. tines) and in material structures and practices One could ask knowledgeable informants how (such as in the placement of a building or main- long the time was between conceptualization tenance routines for the building [Moorman & and execution of specific activities at any level Miner, 1997; Walsh & Ungson, 1991]). of analysis. This approach carries the danger of A key to this measurement process is defining demand effects, however. Specifically, firms the boundary of knowledge relevant to a partic- with strong sanctions against "shooting from ular organizational concern-say, "order fulfill- the hip" might underreport improvisation, ment processes." Day (1994) describes several whereas more flamboyant organizational cul- processes relevant to order fulfillment in organ- tures might overreport improvisation in an at-

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 1998 Moorman and Miner 717 tempt to appear creative and flexible. These improvisation will be relative to the annual considerations underscore the potential value of maintenance plan-not to the microlevels of archival and longitudinal measures of improvi- plant activity. A different study of the same sation. Ideally, we recommend obtaining actual plant would make explicit that the measure- records or archival traces of plans or designs of ment applies to the design and execution of action and of execution. In the new product de- action on a daily basis. There, some of the tac- velopment context, for example, it is possible to tical activities might, indeed, be scored as im- get formal plans and dates of execution. Re- provisational. It may also be helpful to focus on searchers could also set up prospective obser- measuring changes in degrees of improvisation, vation or reporting schemes, in which both de- rather than absolute levels. For example, it may sign and action are observed as events unfold. be more meaningful to contrast a particular or- This strategy would offer the advantage of ganization's degree of improvisation at two avoiding the typical normative pressure to re- points in time than to compare levels between port organizational actions as being planned two firms, all else being equal. before execution. In addition to these approaches, we (Moorman Another important measurement complication & Miner, 1998) also experimented with simple is that the degree of improvisation could seem to rating scales that asked informants to assess depend on the "grain" used to partition actions. directly the degree to which specific actions that For example, an observer might start with a occurred in a new product setting fit these de- formal plan specifying that certain machines in scriptions: "improvised in carrying out this ac- a manufacturing plant should be checked and tion," "figured out action as we went along," and maintained each month in a rotating schedule "ad-libbed action." Using these ratings, we over a year. If the observer then checks whether found, in a study of over 100 actions in two or- those actions were followed, and they are, this ganizations, that 47.5 percent of all actions qual- measurement approach would not register as ified as primarily improvisational (scoring 5 and improvisation. However, if the same observer above on a 7-point scale), with a mean of 4.242 proceeds to make fine-grained observations of (s.d. = 1.985). In the same study two independent everyday activities in pursuit of the monthly observers also rated improvisation, and their commitment, she or he might observe many ac- joint ratings (at 70 percent agreement) produced tions more or less composed as the actors go an equivalent mean improvisation rating of along, some of which might not involve prior 4.014 (s.d. = 1.539). In this study (Moorman & routines. Even within prior routines-say, a Miner, 1998) we also measured organizational standard maintenance activity-an operator memory (generally) and organizational innova- might make, for example, adjustments on a day tion and found that they exhibit discriminant when some employees are absent or the validity with measures of organizational impro- weather is unusually hot. Are these tactical ad- visation. justments improvisations? From one viewpoint Operationalizing collective improvisation re- they are, because they represent convergence of quires special care, but it seems to us both fea- design and execution at the level at which they sible and critical. Walsh's (1995) review of the occur. However, in the context of the large-scale debate regarding collective knowledge and of monthly target plans, they are not, because that recent empirical research, we believe, supports plan is followed more or less as written. At this the idea that teams, departments, and whole level of analysis, the activity has all followed a organizations can improvise (Eisenhardt & prior plan and been implemented quite sepa- Tabrizi, 1995; Hutchins, 1991; Mintzberg & rately from its design. McHugh, 1985). Memory embodied in physical How can we resolve this apparent problem in artifacts is easy to distinguish as collective. measurement? We suggest that the solution lies Moreover, we have described how group pro- in making explicit the specific contrast that in- cesses, social interaction, and repeated commu- forms it. In general, we recommend that impro- nications may embody collective memory. An visation should be measured relative to the interesting additional form of collective impro- same level of analysis at which plans are made. visation arises when a small group or a few Thus, in the plant maintenance example, an ob- team members undertake a task extemporane- server should make explicit that the measure of ously. We argue that if they do so when officially

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 718 Academy of Management Review October charged with the task, this constitutes organiza- conditions-including structural and cultural tional improvisation. But if an individual impro- conditions-that further influence the accessi- vises not only alone but solely with his or her bility of memory types. For example, when in- own authority, this does not represent organiza- formation is made available quickly to decision tional improvisation. Overall, we agree with makers as the action unfolds, this facilitates the Walsh, who suggests that the challenges asso- impact of memory on improvisation (Bastien & ciated with measuring improvisation, as with Hostager, 1992, 1988; Imai et al., 1985). Finally, other collective traits, should "stimulate inves- our propositions suggested that improvisation tigation, not serve as a rationale for a research could influence memory. However, that view as- moratorium" (1995: 286). sumes that the organization has adequately developed retention systems (Walsh & Ungson, 1991). Boundary Conditions of Our Theoretical In addition to memory and its associated sys- Approach tems, the organization's environmental context In this article we present improvisation as a is a second factor that may limit the generaliz- special and important case of innovative activ- ability of our propositions (Moorman & Miner, ity, which, in the presence of appropriate levels 1998). Specifically, in very stable environments and types of memory, can be effective. However, the value of speed for its own sake may de- as with all conceptual frameworks and hypoth- crease so that the negative risks of improvisa- esized relationships, our theories may not hold tion begin to outweigh some of its potential under certain conditions. We see two factors, in value. At the other extreme, some environments particular, that may prove key boundaries for may move so quickly that even the recombined the relationships we propose here. or redeployed previous memory may provide in- The first factor deals with various aspects of adequate knowledge for new conditions-as organizational memory. Following Walsh and may occur in some instances of very swift tech- Ungson's (1991) idea that memory contained in nological change (Anderson & Tushman, 1990). storage bins (physical artifacts) or workplace Appropriate retention of inferences from impro- ecology is the hardest to access, we suggest that visational events may be less likely to occur if the form in which procedural and declarative there is a high level of interference created by memory occurs may influence or limit the de- noise in the environment (Baumgardner, Leippe, gree to which our propositions hold. Specifi- Ronis, & Greenwald, 1983). cally, if declarative memory lies in fixed, endur- Finally, by way of future research, the adop- ing objects, rather than in portable information, tion of a focused definition and testable propo- it will be more difficult to recombine and rede- sitions for empirical research leaves many in- ploy knowledge to new ends. For example, a teresting related issues for further work. For hotel cannot improvise a new combination of example, this article sets aside, for now, issues location and operating procedures as easily as of aesthetic versus instrumental organizational a new combination of two different operating standards (Hatch, 1997b), questions of the emo- processes, because the hotel itself cannot easily tional experiences of actors involved in organi- be moved. zational improvisation (Crossan & Sorrenti, In addition to memory form, there are also 1997), and nuances of interpretative activities organizations with extremely low levels of (Weick, 1979, 1993b). Our definitions and propo- stored knowledge. In such conditions we would sitions leave the door open for further elabora- not expect memory to facilitate the effect of im- tion along these lines. For example, one could provisation on outcomes. Other conditions exist distinguish even more precise aspects of the where organizations have memory but their sys- ordering of design and implementation activi- tems for accessing it and transmitting it (Moor- ties by contrasting more or less simultaneous man, 1995; Moorman & Miner, 1997; Walsh & composition and execution from improvisation Ungson, 1991) make the impact of memory more that specifically involves design after ac- limited. We have captured some of this in our tion-as when meaning is found in previous discussion of the differences in procedural and actions, which then guides future activities declarative memory. However, beyond these en- (Weick, 1993b). In addition, although our work during differences, there are certain contextual explicitly seeks to facilitate deductive research

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 1998 Moorman and Miner 719 on improvisation, it does not foreclose investi- quired for fruitful research on improvisation gation of fundamental questions of the creation seems, to us, a task whose time has come. of meaning in parallel qualitative or hermeneu- tic research (Barrett & Hatch, 1995; Hatch, 1997a; Meyer et al., in press). REFERENCES Aldrich, H. E. 1979. Organizations and environments. Engle- wood Cliffs, NJ: Prentice-Hall. CONCLUSION Alinsky, S. 1969. Rules for radicals. Chicago: University of In this article we reviewed the literature on Chicago Press. improvisation to distill and explicate the view Allison, G. T. 1971. Essence of decision: Explaining the Cuban that improvisation occurs when composition missile crisis. Boston: Little, Brown. and execution converge in time. We suggested Amabile, T. 1983. The social psychology of creativity. New that this definition of improvisation offers a dis- York: Springer-Verlag. tinct and interesting construct that we hope can Anderson, J. R. 1983. The architecture of cognition. Cam- be used in further theoretical development and bridge, MA: Harvard University Press. empirical research. In addition, by focusing at- Anderson, P., & Tushman, M. L. 1990. Technological discon- tention directly on the temporal order of two tinuities and dominant designs: A cyclical model of technological change. Administrative Science Quar- specific organizational activities, we believe the terly, 35: 604-633. improvisation construct may enhance research Argyris, C., & Schbn, D. 1978. Organizational learning: A on the dynamics of organizational processes. theory of action perspective. Reading, MA: Addison- Stimulated by insights in prior research on Wesley. improvisation and the observation that techno- Bailey, D. 1980. Improvisation: Its nature and practice in logical change appears to be influencing not music. Ashbourne, England: Moorlan Publishing. only the value of organizational improvisation Barley, S. R. 1986. Technology as an occasion for structuring: but also organizations' ability to radically alter Evidence from observations of CT scanners and the so- organizational memory's content and access, we cial order of radiology departments. Administrative Sci- developed here a framework that focuses on the ence Quarterly, 31: 78-108. impact of memory on improvisational outcomes. Barrett, F. J., & Hatch, M. J. 1995. Organizational innovation We then detailed the effects of two types of and jazz improvisation: Exploring a postmodern meta- organizational memory-(1) procedural and (2) phor. Paper presented at the annual meeting of the Academy of Management, Vancouver, BC. declarative-and their varying effects on three improvisational outcomes. In particular, our Barron, F., & Harrington, D. M. 1981. Creativity, intelligence, and personality. Annual Review of Psychology, 32: 439- framework suggests that procedural memory 476. should enhance improvisational effectiveness Bastien, D. T., & Hostager, T. J. 1992. Cooperation as commu- and speed, while reducing its novelty. Declara- nicative accomplishment: A symbolic interaction anal- tive memory, however, should enhance improvi- ysis of an improvised jazz concert. Communication Stud- sational effectiveness and novelty, while reduc- ies, 43: 92-104. ing its speed. Given these tradeoffs, we Bastien, D. T., & Hostager, T. J. 1988. Jazz as a process of proposed that the presence of both procedural organizational innovation. Communication Research, and declarative memory would be especially 15: 582-602. likely to produce valuable improvisation and Baumgardner, M. H., Leippe, M. R., Ronis, D. L., & Greenwald, that organizations could develop specific com- A. G. 1983. In search of reliability persuasion effects: petencies to overcome the negative effects of Associative interference and persistence of persuasion in a message-dense environment. Journal of Personality each individual type of memory. Finally, we ar- and Social Psychology, 45: 524-537. gued that improvisational activities can, in turn, Berliner, P. F. 1994. Thinking in jazz: The infinite art of im- influence the nature of organizational memory, provisation. Chicago: University of Chicago Press. if an organization observes the outcomes of im- Bjurwill, C. 1993. Read and react: The football formula. Per- provisational actions and incorporates new rou- ceptual and Motor Skills, 76: 1383-1386. tines or inferences into its memory. Borko, H., & Livingston, C. 1989. Cognition and improvisa- As a whole, taking into account the theoretical tion: Differences in mathematics instruction by expert promise and potential practical impact, tackling and novice teachers. American Educational Research the tough conceptual and empirical issues re- Journal, 26: 473-498.

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 720 Academy of Management Review October

Bowen, H. K., Clark, K. B., Holloway, C. A., & Wheelwright, Cohen, M., March, J. G., & Olson, J. 1972. A garbage can S. C. 1994. The perpetual enterprise machine. New York: model of organizational choice. Administrative Science Oxford University Press. Quarterly, 17: 1-25. Brown, S. L., & Eisenhardt, K. M. 1995. Product development: Cohen, W. M., & Levinthal, D. A. 1990. Absorptive capacity: A Past research, present findings, and future directions. new perspective on learning and innovation. Adminis- Academy of Management Review, 20: 343-378. trative Science Quarterly, 35: 128-152. Brown, S. L., & Eisenhardt, K. M. 1996. Product innovation as Cohen, W. M., & Levinthal, D. A. 1994. Fortune favors the core capability: The art of dynamic adaptation. Working prepared firm. Management Science, 40: 227-251. paper, Department of Industrial Engineering and Engi- Cooper, R. G., & Kleinschmidt, E. J. 1986. An investigation neering Management, Stanford University, Stanford, into the new product process: Steps, deficiencies, and CA. impact. Journal of Product Innovation Management, 3: Burgelman, R. A. 1983. A process model of internal corporate 71-85. venturing in the diversified major firm. Administrative Crossan, M., & Sorrenti, M. 1997. Making sense of improvisa- Science Quarterly, 28: 223-244. tion. In J. P. Walsh & A. S. Huff (Eds.), Advances in Burgelman, R. A. 1994. Fading memories: A process theory of strategic management, vol. 14: 155-180. Greenwich, CT: strategic business exit in dynamic environments. Ad- JAI Press. ministrative Science Quarterly, 39: 24-56. Cyert, R. M., & March, J. G. 1992. (First published in 1963.) A Burgelman, R. A., Maidique, M., & Wheelwright, S. C. 1996. behavioral theory of the firm (2nd ed.). Oxford, England: Basil Blackwell. Strategic management of technology and innovation (2nd ed.). Chicago: Irwin. Day, G. S. 1994. The capabilities of market-driven organizations. Journal of Marketing, 58: 37-52. Campbell, D. 1989. An introduction to nonlinear dynamics. In D. L. Stein (Ed.), Lectures in the sciences of complexity: Dean, J. W., & Susman, G. I. 1989. Organizing for manufac- 3-105. Redwood, CA: Addison-Wesley. turable design. Harvard Business Review, 67: 28-36. Carraher, T. N., Carraher, D. W., & Schliemann, A. D. 1985. Diamond, S. 1997. Internet's impact on the oil and gas indus- Mathematics in the streets and in the schools. British try: Reality vs. hype. Oil and Gas Journal, 95: 17. Journal of Developmental Psychology, 3: 21-29. Dougherty, D. 1992. Interpretive barriers to successful prod- Carter, K., Sabers, D., Cushing, K., Pinnegar, S., & Berliner, uct innovation in large firms. Organization Science, 3: D. C. 1987. Processing and using information about stu- 179-202. dents: A study of expert, novice, and postulant teachers. Durham, W. H. 1978. The coevolution of human biology and Teaching & Teacher Education, 3: 147-157. culture. In N. Burton-Jones & V. Reynolds (Eds.), Human Chase, M. P. 1988. Improvisation: Music from the inside out. behaviour and adaptation: 11-32. London: Taylor and Berkeley, CA: Creative Arts Book Company. Francis. Chase, W. G., & Simon, H. 1973. Perception in chess. Cogni- The Economist, 1995. Reclaiming Britain's constitution. No- tive Psychology, 4: 55-81. vember 11: 18-19. Chi, M. T. H., Glaser, R., & Rees, E. 1981. Expertise in problem Eisenhardt, K. M., & Tabrizi, B. N. 1995. Accelerating adaptive solving. In R. Sternberg (Ed.), Advances in the psychol- processes: Product innovation in the global computer ogy of human intelligence: 7-15. Hillsdale, NJ: Lawrence industry. Administrative Science Quarterly, 40: 84-110. Erlbaum Associates. Embrey, D. G., Guthrie, M. R., White, 0. R., & Dietz, J. 1996. Clinical decision making by experienced and inexperi- Chiesi, H. L., Spilich, G. J., & Voss, J. T. 1979. Acquisition of enced pediatric physical therapists for children with domain-related information in relation to high and low diplegic cerebral palsy. Physical Therapy, 76: 20-33. domain knowledge. Journal of Verbal Learning and Ver- bal Behavior, 18: 257-273. Epple, D., Argote, L., & Devadas, R. 1991. Organization learning curves: A method for investigating intra-plant trans- Cleary, M. J., & Groer, S. 1994. Inflight decisions of expert and fer of knowledge acquired through learning by doing. novice health teachers. Journal of School Health, 64: Organization Science, 2: 58-70. 110-114. Esman, A. 1951. Jazz: A study in conflict. American Imago, 8: M. D. 1991. Individual and Cohen, learning organizational 219-226. routine: Emerging connections. Organization Science, 2: 135-139. Feldman, M. 1989. Order without design: Information production and policy making. Stanford, CA: Stanford Univer- M. Cohen, D., & Bacdayan, P. 1994. Organizational routines sity Press. are stored as procedural memory: Evidence from a laboratory study. Organization Science, 4: 554-568. Feldman, M., & March, J. G. 1981. Information as signal and symbol. Administrative Science Quarterly, 26: 171-186. Cohen, M. D., Burkhart, R., Dosi, G., Egidi, M., Marengo, L., Lassimo, W., & Winter, S. 1995. Routines and other recur- Follett, M. P. 1930. Creative experience. New York: Long- ring action patterns of organizations: Contemporary re- mans, Green. search issues. Working paper, Santa Fe Institute, Santa Forinash, M. 1992. A phenomenological analysis of Nordoff- Fe, NM. Robbins approach to music therapy: The lived experi-

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 1998 Moorman and Miner 721

ence of clinical improvisation. Music Therapy, 11: 120- Kernfeld, B. 1995. What to listen for in jazz. New Haven and 141. London: Yale University Press. Gardner, W., & Rogoff, B. 1990. Children's deliberateness of Knapp, B. 1989. Machine, metaphor, and the writer. Univer- planning according to task circumstances. Developmen- sity Park, PA: Pennsylvania State University Press. tal Psychology, 26: 480-487. Knott, D. 1997. Get smarter by sharing ideas. Oil and Gas Ginsberg, A., & Baum, J. A. C. 1994. Evolutionary processes Journal, 95: 21. and patterns of core business change. In J. A. C. Baum & Laudon, K. C., & Laudon, J. P. 1996. Management information J. V. Singh (Eds.), Evolutionary dynamics of organiza- systems. Upper Saddle River, NJ: Prentice-Hall. tions: 127-151. New York: Oxford University Press. Lee, B., Barua, A., & Whinston, A. B. 1997. Discovery and Gioia, T. 1988. The imperfect art. New York: Oxford University representation of causal relationships in MIS research: Press. A methodological framework. MIS Quarterly, 21: 109- 136. Goodhue, D. L., Wybo, M. D., & Kirsch, L. J. 1992. The impact of data integration on the costs and benefits of informa- Leonard-Barton, D. 1992. Core capabilities and core rigidi- tion systems. MIS Quarterly, 16: 293-311. ties: A paradox in managing new product development. Strategic Management Journal, 13: 111-125. Gould, S. J. 1977. Ever since Darwin. New York: Norton. Levi-Strauss, C. 1967. The savage mind. Chicago: University Graham, G., Manross, M., Hopple, C., & Sitzman, T. 1993. of Chicago Press. Novice and experienced children's physical education teachers: Insights into their situational decision mak- Levitt, B., & March, J. G. 1988. Organizational learning. An- 14: 319-340. ing. Journal of Teaching in Physical Education, 12: 197- nual Review of Sociology, 214. Lichtenstein, D. 1993. The rhetoric of improvisation spontaneous discourse in jazz and psychoanalysis. American Hatch, M. J. 1997a. Exploring the empty spaces of organizing: Imago, 50: 227-252. How improvisational jazz can help reconceptualize organizational structure. Paper presented at the Annual Lovell, J., & Kluger, J. 1995. Apollo XIII. New York: Simon and Standing Conference on Corporate Symbolism meeting, Schuster. Warsaw, Poland. Maidique, M. A., & Patch, P. 1988. Corporate strategy and Hatch, M. J. 1997b. Jazzing up the theory of organizational technology policy. In M. L. Tushman & W. L. Moore (Eds.), improvisation. In J. P. Walsh & A. S. Huff (Eds.), Advances Readings in the management of innovation: 236-248. in strategic management, vol. 14: 181-191. Greenwich, New York: Harper. CT: JAI Press. Mangham, I. L. 1986. Power and performance in organizations: An exploration of executive process. Oxford, Hedberg, B. 1981. How organizations learn and unlearn. In England: Basil Blackwell. P. C. Nystram & W. H. Starbuck (Eds.), Handbook of organizational design: 3-27. London: Oxford University Mangham, I. L., & Pye, A. 1991. The doing of managing. Press. Cambridge, MA: Basil Blackwell. Holland, J. H. 1975. Adaptation in natural and artificial sys- March, J. G. 1976. The technology of foolishness. In J. G. tems: An introductory analysis with applications to biol- March & J. P. Olsen (Eds.), Ambiguity and choice in ogy, control and artificial intelligence. Ann Arbor: Uni- organizations: 69-81. Bergen, Norway: Universitetsfor- versity of Michigan Press. laget. in Three Huber, G. P. 1991. Organizational learning: The contributing Martin, J. 1992. Cultures organizations: perspectives. New York: Oxford University Press. processes and the literatures. Organizational Science, 2: 88-115. Metcalf, J. 1986. Decision making and the Grenada rescue operation. In J. G. March & R. Weissinger-Baylon (Eds.), Hutchins, E. 1991. Organizing work by adaptation. Organi- Ambiguity and command: Organizational perspectives zation Science, 2: 14-39. on military decision making: 277-297. Marshfield, MA: Hutt, M. D., Reingen, P. H., & Ronchetto, J. R., Jr. 1988. Tracing Pitman Publishing. in marketing strategy formation. emergent processes Meyer, A., Frost, P., & Weick, K. In press. The organization Journal of Marketing, 52: 4-19. science jazz festival: Improvising as a metaphor for or- Imai, K., Nonaka, I., & Takeuchi, H. 1985. Managing the new ganizing. Organization Science. product development process: How Japanese companies Miner, A. S. 1987. Idiosyncratic jobs in formalized organiza- learn and unlearn. In K. B. Clark, R. H. Hayes, & tions. Administrative Science Quarterly, 32: 327-351. C. Lorenz (Eds.), The uneasy alliance: 337-381. Cam- Miner, A. S. 1991. The social ecology of jobs. American So- bridge, MA: Harvard University Press. ciological Review, 56: 772-785. Irby, D. M. 1992. How attending physicians make instruc- Miner, A. S., Moorman, C., & Bassoff, P. 1996. Organizational tional decisions when conducting teaching rounds. Ac- improvisation: How firms "make it up as they go along" ademic Medicine, 67: 630-638. in new product development. Paper presented at the Jewkes, J., Sawers, D., & Stillerman, R. 1969. The sources of annual meeting of the Academy of Management, Cin- inventions (2nd ed.). London: Macmillan. cinnati, OH.

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 722 Academy of Management Review October

Mintzberg, H., & McHugh, A. 1985. Strategic formation in an Preston, A. 1991b. The 'problem' in and of management in- adhocracy. Administrative Science Quarterly, 30: 160- formation systems. Accounting Management and Infor- 197. mation Technology, 1: 43-69. Mintzberg, H., Pascale, R. T., Goold, M., & Rumelt, R. P. 1996. Quinn, J. B. 1986. Innovation and corporate strategy: Man- The "Honda effect" revisited. California Management aged chaos. In M. Horwich (Ed.), Technology in the mod- Review, 38: 78-117. ern corporation: A strategic perspective: 167-183. New Moorman, C. 1995. Organizational market information pro- York: Pergamon Press. cesses: Cultural antecedents and new product out- Rogers, E. 1983. The diffusion of innovations. New York: Free comes. Journal of Marketing Research, 32: 318-335. Press. Moorman, C., & Miner, A. S. 1997. The impact of organiza- Sawyer, K. 1992. Improvisational creativity: An analysis of tional memory on new product performance and creativ- jazz performance. Creativity Research Journal, 5: 253- ity. Journal of Marketing Research, 34: 91-107. 263. Moorman, C., & Miner, A. S. 1998. The convergence of plan- Schatz, B. R. 1992. Building an electronic community system. ning and execution: Improvisation in new product de- Journal of Management Information Systems, 8: 87-109. velopment. Journal of Marketing, 61: 1-20. Schein, E. H. 1985. Organizational culture and leadership. Nachmanovitch, S. 1990. Free play: Improvisation in life and San Francisco: Jossey-Bass. art. New York: Putnam. Schon, D. 1983. The reflective practitioner: How professionals Nelson, R. R., & Winter, S. G. 1982. An evolutionary theory of think in action. New York: Basic Books. economic change. Cambridge, MA: Harvard University Press. Schuller, G. 1968. Early jazz: Its roots and early development. New York: Oxford University Press. Neustadt, R. E., & May, E. R. 1986. Thinking in time: The uses of history for decision makers. New York: Macmillan. Scott, R. 1992. Organizations: Rational, natural, and open systems. Englewood Cliffs, NJ: Prentice-Hall. Nonaka, I. 1990. Redundant, overlapping organization: A Jap- anese approach to managing the innovation process. Singley, M. K., & Anderson, J. R. 1989. The transfer of cogni- California Management Review, 32: 27-38. tive skill. Cambridge, MA: Harvard University Press. Orsmond, G. I., & Miller, L. K. 1995. Correlates of musical Solomon, L. 1986. Improvisation II. Perspectives of New Mu- improvisation in children with disabilities. Journal of sic, 24: 224-235. Music Therapy, 22: 152-166. Spolin, V. 1963. Improvisation for the theater: A handbook of Pascale, R. T. 1984. The Honda effect. Excerpted from Per- teaching and directing techniques. Evanston, IL: North- spectives on strategy: The real story behind Honda's western University Press. success. California Management Review, 26: 47-72. Stein, D. 1989. Preface. In D. L. Stein (Ed.), Lectures in the Perry, L. T. 1991. Strategic improvising: How to formulate and sciences of complexity: XIII-XXII. Redwood, CA: Addi- implement competitive strategies in concert. Organiza- son-Wesley. tion Dynamics, 19: 51-64. Sternberg, R. J., & Lubart, T. I. 1996. Investing in creativity. Peters, T. 1988. The mythology of innovation or a skunkworks American Psychologist, 51: 677-688. tale, Part II. In M. L. Tushman & W. L. Moore (Eds.), D. K. Readings in management of innovation: 138-147. Cam- Stoner, L., Tice, J., & Ashton, J. E. 1989. Simple and bridge, MA: Ballinger. effective cellular approach to a job shop machine shop. Manufacturing Review, 2: 119-128. Pfeffer, J. 1981. Power in organizations. Marshfield, MA: Pit- man Publishing. Sutton, R. I., & Hargadon, A. 1997. Brainstorming groups in context: Effectiveness in a product design firm. Admin- Pfeffer, J. 1982. Organizations and organization theory. New istrative Science Quarterly, 42: 685-718. York: Putnam. Toiviainen, P. 1995. Modeling the target-note technique of Powers, C. 1981. Role-imposition or role-improvisation: Some bebop-style jazz improvisation: An artificial neural net- theoretical principles. The Economic and Social Review, work approach. Music Perception, 12: 399-413. 12: 287-299. Towse, E., & Flower, C. 1993. Levels of interaction in group Pressing, J. 1984. Cognitive processes in improvisation. In improvisation. In M. Heal & T. Wigram (Eds.), Music W. R. Crozier & A. J. Chapman (Eds.), Cognitive processes therapy in health and education: 73-81. London: Kings- in the perception of art: 345-363. Amsterdam: North- ley. Holland. Van de Ven, A. H. 1986. Central problems in the management Pressing, J. 1988. Improvisation: Methods and models. In J. A. of innovation. Management Science, 32: 590-607. Sloboda (Ed.), Generative processes in music: The psychology of performance, improvisation, and composi- Van de Ven, A. H. 1993. Managing the process of organization: 129-178. Oxford, England: Oxford University Press. tional innovation. In G. P. Huber & W. H. Glick (Eds.), Organizational change and 269-294. New Preston, A. 1991a. Improvising order. In I. L. Mangham (Ed.), redesign: York: Oxford Press. Organization analysis and development: 81-102. New University York: Wiley. Volkman, T. A. 1994. Our garden is the sea: Contingency and

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions 1998 Moorman and Miner 723

improvisation in Mandar women's work. American Eth- Weick, K. E. 1993c. Managing as improvisation: Lessons from nologist, 21: 564-585. the world of jazz. Working paper, University of Michigan Graduate School of Business Administration, Ann Arbor, Walsh, J. P. 1995. Managerial and organizational cognition: Notes from a trip down memory lane. Organization Sci- MI. ence, 6: 280-321. Weick, K. E. 1996. Drop your tools: An allegory for organiza- Walsh, J. P., & Ungson, G. R. 1991. Organizational memory. tional studies. Administrative Science Quarterly, 41: Academy of Management Review, 16: 57-91. 301-313. Weick, K. E. 1979. The social psychology of organizing (2nd Winter, S. G. 1987. Knowledge and competence as strategic ed.). Reading, MA: Addison-Wesley. assets. In D. J. Teece (Ed.), The competitive challenge: Strategies for industrial innovation and renewal: 159- In D. J. Teece (Ed.), Weick, K. E. 1987. Substitutes for strategy. 185. New York: Harper & Row. The competitive challenge: Strategies for industrial innovation and renewal: 221-233. New York: Harper & Row. Yinger, R. J. 1986. Examining thought in action: A theoretical and methodological critique of research on interactive Weick, K. E. 1993a. The collapse of sensemaking in organi- teaching. Teaching & Teacher Education, 2: 263-282. zations: The Mann Gulch disaster. Administrative Sci- ence Quarterly, 38: 628-652. Zaltman, G., Duncan, R., & Holbek, J. 1973. Innovations and organizations. New York: Wiley. Weick, K. E. 1993b. Organizational redesign as improvisation. In G. P. Huber & W. H. Glick (Eds.), Organizational Zinn, D. 1981. The structure and analysis of the modern im- change and redesign: 346-379. Cary, NC: Oxford Univer- provised line, vol. 1. New York: Excelsior Music Publish- sity Press. ing.

Christine Moorman received her Ph.D. from the University of Pittsburgh and is currently an associate professor of marketing at the University of Wisconsin at Madison. Her current research interests include the use of market information by individual managers and organizations, as well as the market-level dynamics associated with information flows. She is on the editorial review boards of the Journal of Marketing Research, the Journal of Marketing, and the Journal of Public Policy & Marketing. Anne S. Miner received her Ph.D. from Stanford University and is currently an associate professor of management at the University of Wisconsin at Madison. Her current research interests include population-level learning, emergent organizational processes, and the strategic management of technology. She serves as an Associate Editor for Management Science, as well as an editorial board member for the Amer- ican Sociological Review. Her recent work focuses on international differences in the commercialization of university inventions.

This content downloaded from 152.3.153.148 on Mon, 23 Sep 2013 16:39:34 PM All use subject to JSTOR Terms and Conditions