Citekey: @Engestrom2009-fq

Engeström, Y. (2009). The future of activity theory: A rough draft. Learning and Expanding with Activity Theory, 303–328. Retrieved from


Summarize: A thoughtful piece that explains AT deeply and also explores its neighboring concepts.




Recently Aaro Toomela has published two papers in which he critically assesses the contents and potential of activity theory. The first one is titled ’Activity theory is a dead end for cultural-historical psychology’ (Toomela, 2000), the second one ’Activity theory is a dead end for methodological thinking in cultural psychology too’ (Toomela, 2008). According to Toomela, there are five fatal faults in activity theory: 1. It relies on unidirectional instead of a dialectical view of culture- individual relationships. 2. It focuses on analyses of activities without taking into account the individual involved in the activity at the same time. 3. It underestimates the role of signs and the importance of focusing on sign meaning. 4. It approaches mind fragmentally, without understanding the holistic nature of mind. 5. It is fundamentally adevelopmental and therefore not appropriate for understanding emerging phenomena, including mind. (p. 1)

the term ’activity theory’, the two books of Leont’ev available in English (although sold out long ago), my book Learning by Expanding (p. 1)

article from 1993 by Cole and Engeström. (p. 2)


Activity theory is a theory of object-driven activity. Objects are concerns, they are generators and foci of attention, motivation, effort and meaning. Through their activities people constantly change and create new objects. The new objects are often not intentional products of a single activity but unintended consequences of multiple activities. (p. 3)

compatible with Popperian worldview (p. 3)

In the present era, we need to understand and deal with what I have called ‘runaway objects’ (Engeström, 2008). (p. 3)

Runaway objects have the potential to escalate and expand up to a global scale of influence. They are objects that are poorly under anybody’s control and have far-reaching, unexpected effects. Such objects are often monsters: They seem to have a life of their own that threatens our security and safety in many ways. (p. 3)

Runaway objects are contested objects that generate opposition and controversy. They can also be powerfully emancipatory objects that open up radically new possibilities of development and well-being. The Linux operating system is a well-known example. (p. 3)

Leont’ev’s (1978) well-known dictum was that there is no activity without an object. With runaway objects, we may ask: Are there objects without an activity? Whose object is the global warming, for example? (p. 4)

But which activities take responsibility for such a huge object as global warming? (p. 4)

I have often used the representation depicted in Figure 2 to capture the challenge of constructing a shared object between two or more activity systems. (p. 4)

Figure 2. Two activity systems and a potentially shared object (p. 4)

However, with large runaway objects, the challenge would look more like Figure 3. The are typically numerous activity systems focused on or affiliated with the object. But the object is pervasive and its boundaries are hard to draw. Thus, the positions of the activity systems are ambiguous and they often seem to be subsumed to the object rather than in control of it. (p. 5)

Figure 3. Large runaway object and activity systems (p. 5)

Various social movements try to do just that. Organic farming, Wikipedia, open models of scientific research and publishing are examples. Most such attempts fail or remain marginal. A crucial question is: What gives some objects inherent drawing power? (p. 6)

this is aligned with promising ideas (p. 6)

First of all, a benign runaway object must have intrinsic properties that transcend the limits of utilitarian profit motive. In this sense, a benign runaway object is at the boundary between legitimate and illegitimate, sensible and crazy, work and leisure, technology and art. These properties are experienced in acting on and with the object over a long haul, with persistence and patience, oscillating between intensity and withdrawal. The object must yield useful intermediate products, yet remain an incomplete project. The object must be visible, accessible and cumulable - allowing participants to return time and again. There must be effective feedback from and exchange among the participants acting on the object. (p. 6)

In the following sections, I will discuss the five themes of this book in the light of the challenges posed by the emergence of runaway objects. (p. 6)


(Engeström, 1996). The first generation built on Vygotsky’s notion of mediated action. The second generation built on Leont’ev’s notion of activity system. The third generation, emerging in the past 15 years or so, built on the idea of multiple interacting activity systems focused on a partially shared object. (p. 6)

Figure 4. A possible unit of analysis for examining power relations at work (p. 7)

Wolf-Michael Roth and his co-authors (2008) call for the inclusion of sensuous aspects of work into the unit of analysis. They name emotions, identity, and ethico-moral dimensions of action as salient sensuous aspects. (p. 7)

Analyzing actions together with their social and material consequences is indeed a promising way to approach emotions and other sensuous aspects of activity empirically. (p. 7)

Third generation activity theory expands the analysis both up and down, outward and inward. Moving up and outward, it tackles multiple interconnected activity systems with their partially shared and often fragmented objects. Moving down and inward, it tackles issues of subjectivity, experiencing, personal sense, emotion, embodiment, identity, and moral committment. (p. 8)

there is a risk that activity theory is split into the study of activity systems, organizations and history on the one hand and subjects, actions and situations on the other hand. This is exactly the kind of split the founders of activity theory set out to overcome. To bridge and integrate the two directions, serious theoretical and empirical efforts are needed. (p. 8)

isn’t this also a reality in CSCL research? different work at diff levels that rarely meet together? (p. 8)

David Russell (in press) suggests ‘genre as social action’ as a unit of analysis complementary to the unit of activity system. For Russell, genres are classifications of artifacts-plus-intentions. They are links between subjects, tools and objects. (p. 8)

Activities are mediated by multiple modalities, from bodily movements and gestures to pictures, sounds, tools, and all kinds of signs. Written text is but one of the mediational modalities. (p. 8)

useful; quotable (p. 8)

What is particularly interesting about genres as systems of typified written communication is their mobility and ability to cross organizational boundaries. (p. 8)

Third generation activity theory still treats activity systems as reasonably well-bounded, although interlocking and networked, structured units. What goes on between activity systems is processes, such as the flow of rules from management to workers depicted in Figure 4. Processes are commonly assumed to be relatively straightforward, stepwise movements from point A to point B. (p. 9)

make sense. Highlight the importance of the networked viewpoint. (p. 9)

In social production or peer production, the boundaries and structures of activity systems seem to fade away. Processes become simultaneous, multi-directional and often reciprocal. The density and criss-crossing of processes makes the distinction between process and structure somewhat obsolete. The movements of information create textures that are constantly changing but not arbitrary or momentary. The textures are made up of traces or trails which are both cognitive, ‘in the mind’, and material, ‘in the world’ (Cussins, 1992). Wikipedia is a good example in that every alteration of an entry is automatically stored and retrievable for anyone as a cumulative record of previous versions and alterations. So the constantly moving texture is also multi-layered and historically durable. (p. 9)

quotable Combining AT with trace data to uncover complex processes. (p. 9)

I have characterized these new forms of activity as ‘wildfire activities’ and ‘mycorrhizae activities’ in which interaction takes the shape of knotworking without a single stable center (Engeström, 2006, 2007a, 2008). (p. 9)

Activity system models are very appropriate for the analysis of such hubs. The challenge is to integrate such analytical tools with new concepts appropriate for the analysis of trails and mycorrhizae. Perhaps this implies a need for a fourth generation of activity theory. (p. 10)

nice. Analyze hubs vs. trails. Pointing to future direction of AT. (p. 10)


Georg Rückriem (in press) argues that activity theory as it presently exists is captive of the historically passing medium of print and writing. For Rückriem, the whole idea of mediation of specific activities by specific tools and signs misses the point of the ongoing societal and cultural transformation engendered by digital media, especially by Web 2.0. Mediation is an issue of the historically leading or dominant media. (p. 10)

Here is disagree with Rückriem. I see Rückriem’s insistence on the decisive role of media as a particular form of technological determinism. (p. 10)

Sweeping technological determinism leaves little room for human agency in concrete activities. Focusing on contradictory objects in specific activities calls for new forms of agency. (p. 11)

We do also need new concepts to make sense of Web 2.0. For example the notions of ‘open’ and ‘closed’ have great potential although they remain theoretically underdeveloped for the time being. Perhaps more importantly, digital media make very problematic the Vygotskian distinction between tool and sign. (p. 11)

? (p. 11)

When categories are imposed upon people they often become iron cages which reduce and rule out possibilities. Such closed ‘stabilization knowledge’ (Engeström, 2007b) is commonly the result of exclusively empirical generalizations taught in schools as authorized ‘correct knowledge’ (Davydov, 1990). On the other hand, existing social categories can also be turned into discursive tools that generate new, emancipatory meanings when blended with new contents and new categories. (p. 11)

power (p. 11)

Such transitions from ‘stabilization knowledge’ to ‘possibility knowledge’ are at the core of zones of proximal development. The zone is never an empty space to be (p. 11)

simply filled with the new. It is inhabited by previous categories that need to be opened up, challenged and transformed. (p. 12)

In radical transformations aimed at the creation of qualitatively new patterns of activity, opening up and blending existing categories are not enough. What is needed is re-mediation by new theoretical concepts that serve as ‘germ cells’ for expanded horizons of possibilities. (p. 12)


I will suggest a set of potential mechanisms that may stimulate further work in activity- theoretical studies of development. These mechanisms are (1) living movement, (2) breaking away, (3) double stimulation, (4) stabilization, and (5) boundary crossing. (p. 12)

(1) In the tradition of activity theory, a key metaphor for development is that of a zone. Often the zone of proximal development is interpreted as a vertical step which leads to a higher stage or level. I find it more useful to think of the a zone as a terrain of activity to be dwelled in and (p. 12)

explored, not just a stage to be achieved or even a space to be crossed. The zone is explored by moving in it. The movement may take various directions and patterns. (p. 13)

fits well with KB – ‘not a vertical step.. but a terrain of activity to be dwelled in and explored…’ (p. 13)

(2) The dwellers create trails and the intersecting trails gradually lead to an increased capability to move in the zone effectively, independently of the particular location or destination of the subjects. However, the zone is never an empty space to begin with. It has pre- existing dominant trails and boundaries made by others, often with heavy histories and power invested in them. (p. 13)

When new dwellers enter the zone, they both adapt to the dominant trails and struggle to break away from them. (p. 13)

(3) Breaking away from a pre-existing trail or terrain requires expansive agency. This can be achieved by employing external cultural artifacts that are invested with meaning and thus become powerful mediating signs that enable the human being to control his or her behavior from the outside. This is the mechanism of double stimulation. (p. 13)

(4) New trails and intersections are marked, stabilized and made durable mainly in three ways, namely by means of critical conflicts, by means of authority, and by means of reification into artifacts and conceptualizations. (p. 13)

Formation and execution of authority is an obvious source of stability, yet it is an issue barely touched by activity theorists thus far (p. 13)

(5) Boundary crossing occurs because human beings are involved in multiple activities and have to move between them. A school student must move from home to school to peer culture and back home. Boundary crossing also happens between collective activity systems and organizations, in partnerships and mergers but also in espionage and hostile takeovers. Boundary crossing provides material for double stimulation. It requires negotiation and re-orchestration. It is the most obvious aspect of the horizontal or sideways dimension of development. (p. 14)

better conceptualize boundary crossing? (p. 14)

These five mechanisms partly overlap the conceptual framework of expansive learning (Engeström, 1987, 2001). Obviously breaking away is closely connected to facing and resolving contradictions in the different steps of a cycle of expansive learning. (p. 14)

There is much research and theorizing in developmental psychology that is compatible with the idea of development as ’breaking away and opening up’. What is missing is sustained research programs that would integrate the psychological, institutional and societal aspects of development, not only observationally and retrospectively but also proactively and by means of interventions. (p. 14)

GAP (p. 14)

Development happens – and should be studied - in the forging of the future in politically and affectively loaded everyday discursive actions, decisions and change efforts. (p. 14)


Authority is foundational for the sustained existence of a community – yet, as Taylor (in press) points out (p. 14)

In my recent book From Teams to Knots (Engeström, 2008), I try to capture something of the historical evolution of authority by means of a condensed table (Table 1). (p. 15)

Table 1. A historical sketch for conceptualizing authority, agency, and community (p. 15)

Negotiation is a central coordinating mechanism of the distributed agency required in knotworking within social production. Negotiation is required when the very object of the activity is unstable, resists attempts at control and standardization, and requires rapid integration (p. 15)

of expertise from various locations and traditions. Negotiation is more than an instrumental search for a singular, isolated compromise decision. It is basically a construction of a negotiated order (Strauss, 1978) in which the participants can pursue their intersecting activities. As Firth (1995, p. 7) put it, “in quite implicit ways, negotiation activity implicates the discourse process itself, revolving around such things as acceptability of categories used to describe objects or concepts, and the veridicality of facts, reasons or assessments.” Putnam (1994, p. 339- 340) takes a step further and points out that successful negotiations tend to transform the dispute, not just reach an instrumental end. “By transforming a dispute, I refer to the extent that a conflict has experienced fundamental changes as a result of the negotiation. Fundamental changes might entail transforming the way individuals conceive of the other person, their relationship, the conflict dilemma, or the social-political situation. …In the transformative approach, conflicts are no longer problems to be resolved; rather, they are opportunities to create a new social reality, a new negotiated order, a different definition of a relationship, or a transformed situation.” (p. 16)

important notes about negotiation (p. 16)

Social production, such as the Open Source software movement or Wikipedia, is dependent on constant, publicly accessible critical commentary and peer review. When peer review becomes reciprocal, open and continuous, it actually coincides with Putnam’s notion of transformative negotiation. (p. 16)

Authority and agency are closely related. In agentic actions, we gain authority and become authors of our lives. (p. 16)

hese challenges may be characterized by means of the notion of relational agency (Edwards, in press). It seems clear that individuals engaged in multi-agency collaboration aimed at the creation of a new activity need to nourish and manifest relational agency in order to achieve, as a collective, the expansive agency necessary for the accomplishment of radical transformations. Relational agency and expansive agency are complementary lenses, one focused on the individual, the other focused on the distributed collective. (p. 16)

words about agency (p. 16)

The analysis of agency is still in its infancy. We need to link Il’enkov’s (1977) concept of contradiction with Leont’ev’s (1978) concepts of (p. 16)

need, object and motive, and these further with concrete manifestations of will and agentive action. In between, there is space for intermediate concepts such as conflict, envisioning, identification, responsibility, experiencing, and committment. (p. 17)

Table 1 implies that we are moving toward increasingly open, amoeba-like communities characterized by multi-directional swarming, weak boundaries and no single stable center. If this is the case, authority and agency may also grow in unexpected ways, as multiple simultaenous and interacting ‘minority influences’ (Moscovici, Mugny & van Avermaet, 2008) from the peripheries rather than as a single dominant majority influence from the center. (p. 17)


In the past few years, the United States educational authorities have aggressively launched legislation and national guidelines that define the ‘gold standard’ of educational research. The ‘gold standard’ emphasizes the use of randomized controlled trials, the selection of valid control groups, and ‘scalability’ implying large statistical samples and multiple research sites. (p. 17)

The main difference between ‘gold standard’ interventions and design experiments seems to be that the former expects the design of the intervention to be complete at the outset while the latter, recognizing the complexity of educational settings, expects the design to proceed through multiple iterations of ‘refinement’. But even design experiments aim at closure and control. (p. 18)

interesting attack on DBR. (p. 18)

“Intervention is an on-going transformational process that is constantly re-shaped by its own internal organisational and political dynamic and by the specific conditions it encounters or itself creates, including the responses and strategies of local and regional groups who may struggle to define and defend their own social spaces, cultural boundaries and positions within the wider power field.” (Long, 2001, p. 27) (p. 19)

applies to learning analytics implementation design (p. 19)

“Crucial to understanding processes of intervention is the need to identify and come to grips with the strategies that local actors devise for dealing with their new intervenors so that they might appropriate, manipulate, subvert or dismember particular interventions.” (Long, 2001, p. 233) (p. 19)

In other words, resistance and subversion are not accidental disturbances that need to be eliminated. They are essential core ingredients of interventions, and they need to have a prominent place in a viable intervention methodology. Melucci (1996) extends this point into a threefold methodological guideline for intervention research. (p. 19)

“What we must recognize is that actors themselves can make sense out of what they are doing, autonomously of any evangelical or manipulative interventions of the researcher. (…) Secondly, we need to recognize that the researcher-actor relation is itself an object of observation, that it is itself part of the field of action, and thus subject to explicit negotiation and to a contract stipulated between the parties. (…) Lastly, we must recognize that every research practice which involves intervention in the field of actioncreates an artificial situation which must be explicitly acknowledged. (…) a capability of metacommunication on the relationship between the observer and the observed must therefore be incorporated into the research framework.” (Melucci, 1996, p. 388-389) (p. 20)

If agency is not a central concern in the methodology, there is something seriously wrong with it. In educational research, one of the few scholars who have taken this seriously is David Olson. “Research in the human sciences, it may be argued, is less designed to dictate what one does than to provide information that agents, both teachers and students, can use in making informed decision s about what to do in the multiple and varied contexts in which they work.” (Olson , 2004, p. 25) (p. 20)

To fully appreciate the radical potential of the methodology of double stimulation, we need to reconstruct Vygotsky’s more general conception of intentionality and agency. Vygotsky described this artifact-mediated nature of intentional action as follows. “The person, using the power of things or stimuli, controls his own behavior through them, grouping them, putting them together, sorting them. In other words, the great uniqueness of the will consists of man having no power over his own behavior other than the power that things have over his behavior. But man subjects to himself the power of things over behavior, makes them serve his own purposes and controls that power as he wants. He changes the environment with the external activity and in this way affects his own behavior, subjecting it to his own authority.” (Vygotsky, 1997b, p. 212) (p. 22)

Classic examples of culturally mediated intentionality include devices we construct and use to wake up early in the morning. Vygotsky’s examples of voluntary action are mostly focused on individual actors. This must not be interpreted as neglect of collective intentionality. (p. 22)

Intervention may be defined simply as “purposeful action by a human agent to create change” (Midgley, 2000, p. 113). This definition makes it clear that the researcher does not have a monopoly over interventions. (p. 24)

quotable (p. 24)

Activity theory takes the subjects, the participants, the local practitoners, very seriously. But it does not assume that the researcher has a magic formula with which he or she can objectively decipher how the participants understand and judge the unfolding events. Instead, the practitioners themselves are asked to look at, comment on and make sense of the researcher’s initial data and provisional analysis. (p. 24)

In the first chapter of Thinking and Speech (1987), Vygotsky presented the famous contrast between analysis into elements and analysis into units. “In contrast to the term ‘element’, the term ‘unit’ designates a product of analysis that possesses all the basic characteristics of the whole. The unit is a vital and irreducible part of the whole.” (Vygotsky, 1987b, p. 46) (p. 25)

Expansive learning is above all stepwise expansion of the object. (p. 27)

The study of expansive learning in complex settings requires a longitudinal intervention approach which may be crystallized in the form of three methodological rules (Engeström, Engeström & Kerosuo, 2003): (1) follow the objects of activity in their temporal and socio-spatial trajectories, (2) give the objects a voice by involving the clients or users in dialogues where the object is made visible, articulated and negotiated, (3) expand the objects by organizing intervention sessions and assignments where the producers and clients construct new shared models, concepts and tools to master their objects. (p. 27)

Learning Analytics Implementation as an expansive learning experience? (p. 27)

The Change Laboratory sessions are a purposeful blend of elements familiar from existing practices and new elements brought in by the researchers. They are designed to serve as microcosms where potentials of co-configuration and knotworking can be experienced and experimented with. “A microcosm is a social testbench and a spearhead of the coming culturally more advanced form of the activity system. …the microcosm is supposed to reach within itself and propagate outwards reflective communication while at the same time expanding and therefore eventually dissolving into the whole community of the activity.” (Engeström, 1987, p. 277-278) (p. 27)

The procedure allows for the collection of rich longitudinal data on the micro-interactions and (p. 27)

cognitive processes involved in expansive learning as the participants make visible their work, moving between actions and activity, between the past, the present, and the envisioned future. (p. 28)

However, my tentative conclusion was: ‘We need intermediate runaway objects which are less spectacular and more inviting.’ This is indeed a task for activity theory: Bring together the big and the small, the impossible and the possible, the future-oriented activity-level vision and the here-and-now consequential action. (p. 28)

Blog Logo

Bodong Chen



Crisscross Landscapes

Bodong Chen, University of Minnesota

Back to Home