Yrjö Engeström[*]
Introduction
Recent theorizing on organizational change puts emphasis on notions of paradox, spirals, pluralism, ambivalence, and chaos. Organizations have to deal with multiple, partially conflicting values, tasks and goals, multiple parallel lines of change. There is no single, predetermined linear path forward. Change often takes the form of patchwork or bricolage, partly improvised, partly strategic and visionary. This patchwork differs from externalist models in which change is seen primarily as adoption of innovations coming from outside. The patchwork of transformation is actively constructed through problem solving and action within the organization. These actions and solutions proceed in spirals of multiple scales and durations.
“Positive feedback loops emerge to drive people, groups, and organizations into spirals of increasing or decreasing pluralism and change. These loops can have consequences that occur at different points in time, and so intersect in unpredictable ways.” (Eisenhardt, 2000, p. 703)
The idea of spirals is central to the activity-theoretical concept of expansive learning (Engeström, 1987). In an expansive cycle or spiral, externalist and internalist explanations of change are brought together. Expansive learning may be seen as a cycle or spiral that begins with internal turmoil nurtured by seeds of discoordination planted by external contacts at the end of the previous cycle. As the inner unrest grows, more actors are drawn into active engagement with the internal contradictions. This gradually leads to a reformulation of relations to external partner systems. In this phase, lateral transitions and boundary crossing between the organizations or units involved becomes crucial.
Ever since Vygotsky’s pioneering work (see Vygotsky, 1978), activity theory has seen learning as potential trailblazer for developmental transformations. This implies that the current dichotomy between organizational learning and organizational transformation needs to be overcome.Thus, it is appropriate to begin this treatise with the issue of levels of learning.
Levels of learning
The anthropologist Gregory Bateson (1972) suggested a hierarchy of learning processes that can help us overcome the dichotomies that plaque current attempts at theorizing organizational learning. According to Bateson, at the first level (Learning I), the subject learns a behavior within a given situation in which the goal and means for reaching it are given and the appropriate reaction is learned by habituation. At the second level (Learning II), the goal is given and the learner learns by trial and error or experimentation to select an appropriate tool. In other words, in Learning II the learner learns the context and rules of Learning I. The learner learns how the right answers are constructed, what is their underlying logic in the given context. In a work setting, Learning II means that the learner is learning how to be a worker. This may take place largely as tacit socialization (Learning IIa) or more in the form of conscious strategizing (Learning IIb, see Engeström, 1987, p. 145-148).
However, problems and goals do not always present themselves to the practitioner as givens; often “they must be constructed from the materials of the problematic situations which are puzzling, troubling and uncertain” (Schön, 1983, p. 40). In other words, the underlying logic of the given context is contradictory and unstable. Occasionally this uncertainty and instability becomes so pressing that people engage in a major effort to make sense of it. Identifying the right problems to solve is an essential character of the third level of learning (Learning III) described by Bateson. The learner not only learns to solve problems of a certain kind - he or she transforms and expands the context so that the problems appear and become manageable in a radically new light.
In Bateson’s Learning III, or expansive learning (Engeström, 1987), we meet a kind of learning that goes beyond the dichotomies between formal and informal learning, between individual and organizational learning, and between learning and developmental transformation. To construct an expanded context, individuals have to face and articulate the inner contradictions of their organizations and institutions. This requires that they seek and form alliances and initiate joint efforts at analysis, design and experimentation. Such learning is not anymore satisfied with finding the right answers but aimed at grasping why the institution functions as it does and how to go beyond it. Moreover, such expansive learning efforts make use of diverse tools and resources, including informally gained experiences and observations as well as appropriate formal learning opportunities.
According to Bateson (1972), impulse to move to Learning III emerges when individuals acting in a social context are repeatedly exposed to contradictions, or double bind situations, that seem unsolvable. The only way out from such a situation is either to leave the context or to change it radically. In a social practice, the latter can only be done in collaboration with other individuals in the same social system. But what actually is the social system, the context to be changed, and what in the system should be changed?
Activity system as unit of analysis
The cultural-historical activity theory initiated by Vygotsky (1978) and Leont’ev (1978) offers powerful tools for the analysis of contexts (see also Engeström, Miettinen & Punamäki, 1999; Chaiklin, Hedegard & Jensen, 1999). The distinction between short-lived goal-directed actions and durable, object-oriented activity is of central importance. A historically evolving collective activity system is taken as the prime unit of analysis (Figure 1). Goal-directed actions, as well as automatic operations, are relatively independent but subordinate units of analysis, eventually understandable only when interpreted against the background of entire activity systems. Activity systems realize and reproduce themselves by generating actions and operations.
Figure 1 calls attention to both the personal perspective of the subject - any given subject involved in the collaborative activity may be selected - and its relationship to the systems perspective that views the activity from the outside. An individual artifact-mediated action may be depicted as the uppermost sub-triangle in Figure 1.
Figure 1. The general model of an activity system (Engeström, 1987, p. 78)
Such individual actions are embedded in collective activity systems. This collective aspect is represented by the bottom part of the diagram, consisting of the community, its division of labor, and its rules.
A collective activity is driven by a deeply communal motive. The motive is formed when a collective need meets an object that has the potential to fulfill the need. The motive is thus embedded in the object of the activity. The object, in turn, is to be understood as a project under construction, moving from potential ‘raw material’ to a meaningful shape and to a result or an outcome. In this sense, the object determines the horizon of possible goals and actions. But it is truly a horizon: as soon as an intermediate goal is reached, the object escapes and must be reconstructed by means of new intermediate goals and actions.
It is important to realize that one and the same goal-directed action may accomplish various different activities and transfer from one activity to another. On the other hand, the object and motive of a collective activity may typically be sought after my means of multiple alternative goals and actions. The object and motive give actions their ultimate continuity, coherence and meaning, even when the ostensible object of many actions does not coincide with the object of the overall activity.
The dynamic and systemic character of activity means that its elements are not mechanically assembled together but develop and co-evolve through reciprocal interplay. The activity system is much more robust and competent than any of its individual members or specific components. The cultural mediators of the activity - tools, rules, division of labor - stabilize the system. On the other hand, the different elements are in constant interaction and keep transforming into each other. What is now an object - for instance, a new information system to be acquired and appropriated - may soon become a tool and eventuallya rule dictating standards and routines. The activity system incessantly reconstructs itself.
Recent developments in activity theory emphasize dialogue, multiple perspectives, and networks of interacting activity systems. Wertsch (1991) introduced Bakhtin’s (1982; 1986) ideas on dialogicality as a way to expand the Vygotskian framework. Ritva Engeström (1995) went a step further by pulling together Bakhtin’s ideas and Leont’ev’s concept of activity. Notions of activity networks (e.g., Russell, 1997) are being developed, and a discussion between activity theory and Latour’s (e.g., 1993) actor-network theory have been launched (Engeström & Escalante, 1996; Miettinen, 1999). These developments indicate that the basic model needs to be expanded to include minimally two interacting activity systems (Figure 2).
In Figure 2, the object moves from an initial state of unreflected, situationally given ‘raw material’ (object 1; e.g., a specific patient entering a physician’s office) to a collectively meaningful object constructed by the activity system (object 2, e.g., the patient constructed as a specimen of a biomedical disease category and thus as an instantiation of the general object of illness/health), and to a potentially shared or jointly constructed object (object 3; e.g., a collaboratively constructed understanding of the patient’s life situation and care plan). The object of activity is a moving target, not reducible to conscious short-term goals.
Figure 2. Two interacting activity systems
Internallyandexternally inducedchange
There are two main ways of explaining how qualitative change and innovation are initiated. One is the structuralist explanation using binary models which depict two cultural systems external to each other coming together or colliding in space, thus leading to translation and hybridization (e.g., Lotman, 1990). The other one is the dialectical explanation that uses triadic models which depict a cultural system struggling with an internal contradiction between its components, leading in time to a re-mediation and to a qualitatively new self-organized pattern (e.g., Ilyenkov, 1977).
These two modes of explaining change are usually considered incompatible. The first one emphasizes lateral external movement in space. The second one emphasizes vertical internal movement in time. In Figure 3, the first three models represent externalist explanations while the fourth one represents internalist explanations of change.
The center-periphery model (e.g., Leblebici & al., 1991) suggests that organizational change is triggered by peripheral actors. They typically survive by being different and innovative, until at a certain point one or more of the big central actors in a field will begin to imitate and copy the innovative solutions of the peripheral actors.
Figure 3. Four models of organizational change
The mosaic model (e.g., Cole, 1999) suggests that an organizational field is changed by a gradual saturation of a set of innovative ideas and practices coming from the outside. The prime example is the impact of the Japanese quality movement on American industry. No single breakthrough is needed. A superficial observer may indeed erroneously conclude that the whole change effort has evaporated -- when in fact the whole constellation of standards and practices has gradually and irrevocably been transformed.
The transplantation model (e.g., Sevon, 1999) argues that organizational change is initiated by importing a new, innovative model into an organization. The transplant is then multiplied and diffused throughout the organization until it becomes dominant.
The problem with all the three externalist models is that they take it for granted that there is a model somewhere outside that is somehow superior and worth adopting. While this is often true, externalist models fail to ask how this superior model emerged in the first place, and correspondingly, what makes the receiving organization ripe and ready to adopt a certain innovative model and not some others from the outside.
The punctuated equilibrium model (Gersick, 1991, Romanelli & Tushman, 1994, Loch & Huberman, 1999) basically suggests that qualitative transformations in organizations emerge out of their inner ripening -- accumulation of small incremental improvements. Major breakthroughs are preceded by lengthy periods of stable routine functioning - equilibrium. At a certain point, the accumulation of incremental changes reaches a point of eruption, and intense reorganization occurs over a relatively short period of turmoil.
Obviously such a model focuses on the very issues neglected by externalist models. But versions of the punctuated equilibrium model seem quite helpless in cases where no clear, lengthy periods of overall equilibrium can be detected. Authors such as Quinn & Cameron (1988) and Brown & Eisenhardt (1998) claim that major organizational change is indeed much more ongoing, distributed and multi-faceted than the punctuated equilibrium model would have us believe. As Eisenhardt (2000, p. 703) puts it, “perhaps conceptualizing change as a quantum leap from one frozen state to the next is being superceded by viewing it as having a more complicated, continuous scale distribution.” A more specific critical point is made by Van de Ven and Poole (1988, p. 34) who observe that “in the punctuated equilibrium model, the punctuation process itself remains underdeveloped.”
The ideaof expansive learning cycles differs from punctuated equilibrium models in two respects. First, there are no frozen periods of mere incremental change. Second, any major cycle or spiral of transformation consists of multiple, partly parallel smaller cycles of problem solving, learning and innovation.
Contradictionsandexpansivecycles
For situated learning theory (Lave & Wenger, 1991), motivation to learn stems from participation in culturally valued collaborative practices in which something useful is produced. This seems a satisfactory starting point when we look at novices gradually gaining competence in relatively stable practices. However, motivation for risky expansive learning processes associated with major transformations in activity systems are not well explained by mere participation and gradual acquisition of mastery.
As pointed out earlier, Bateson (1972) suggested that learning III is triggered by double binds generated by contradictory demands imposed on the participants by the context. This is well in line with cultural-historical activity theory which sees contradictions as central sources of change and development (Ilyenkov, 1977, 1982). Contradictions are not the same as problems or conflicts. Being historically accumulated structural tensions within and between activity systems, contradictions generate and manifest themselves in problems, conflicts, and disturbances.
The primary contradiction of activities in capitalism in that between the use value and exchange value of commodities. This primary contradiction pervades all elements of our activity systems. Activities are open systems. When an activity system adopts a new element from the outside (for example, a new technology or a new object), it often leads to an aggravated secondary contradiction where some old element (for example, the rules or the division of labor) collides with the new one. Such contradictions generate disturbances and conflicts, but also innovative attempts to change the activity.
The developmental importance of contradictions becomes understandable when we see the contradictions themselves as emerging and evolving through the steps of a cycle of expansive learning. As new elements from neighboring activity systems enter and accumulate in the central activity system, the central activity system moves from a relatively stable ‘business-as-usual’ state to an unarticulated ‘need state’. Practitioners realize that something is wrong but cannot formulate what the root problem is and what should be done about it. In other words, they are still dealing with a primary contradiction which resides within each component of the activity system and has not yet been aggravated so as to take a clearly identifiable shape.
As failures, conflicts and tensions within the activity system become increasingly frequent and intolerable, attempts at analyzing and articulating their causes and remedies increase, too.Aggravated secondary contradictions between components of the activity system become dominant.In the midst of regressive and evasive attempts to deal with the problems, some practitioners may put forward a novel idea that promises to solve the contradictions and opens up a vision of an expanded object of activity. The idea, or prototypical new solution, gains momentum and is turned into a model. The model is enriched as corresponding tools and patterns of interaction are designed and experimented with.
As the new model is implemented in practice, new conflicts and problems surface between the designed new ways of working and customary old ways of working. By working through these tertiarycontradictions, the designed new model is replaced with a created new model, firmly grounded in practice.
The qualitative transformation in the central activity shakes its external relations. This leads to quaternary contradictions between the central activity and the neighboring activities, manifested in new kinds of trouble and innovation.
These idealized and simplified phases of a cycle of expansive development are depicted in Figure 4.The two-headed arrows signify the iterative, nonlinear character of the process.
Figure 4. Phases of a cycle of expansive learning (Engeström, 1987, p. 189)
Batesons’ third level of learning, learning to transform the context of actions, can thus be seen as a complicated historical process of transforming an institutionalized form of social practice. In this process informal and formal learning, individual and collective learning, cognitive development and development of culturally new artifacts and organizational arrangements are intertwined. The crucial point is expansion. The object of activity is radically broadened.
The expansion of the object takes place along four dimensions: the social-spatial (‘who else should be included?’), the anticipatory-temporal (‘how should the past and the future be included?’), the moral-ideological (‘who is responsible, who decides and who is excluded, who benefits and who suffers?’), and the systemic-developmental (‘how does this shape the future of our activity?’).
Expansive learning proceeds in embedded cycles of different temporal and socio-spatial scales (Figure 5). At the macro level, a cycle of expansive learning is a process of qualitative transformation of an activity system in which all the essential elements of the activity have changed qualitatively. These macro cycles of expansive learning may take several years. They comprise several smaller, meso-level cycles. Typically a cluster of specific learning actions (see below), for example those of questioning the present practice or the design of a first testable idea of the new expanded object, may be analyzed as such meso-level learning cycles, typically requiring some days, weeks, or even months to complete. What makes these clusters cyclic is that they often include some form of encounters with all the four successive types of contradictions, even if these encounters remain partial and unresolved.