FROM LEARNING ENVIRONMENTS AND IMPLEMENTATION TO
ACTIVITY SYSTEMS AND EXPANSIVE LEARNING
Yrjö Engeström
University of Helsinki
INTRODUCTION
In studies of computer-supported collaborative learning and related fields of research, the
notion of learning environment has gained a central status. The notion is widely used to
indicate that learning is somehow situated or distributed within a setting which minimally
includes individual learners and digital technologies of some kind. While perhaps useful as a
catchword, the notion of learning environment has become a virtual substitute for, if not an
impediment to, serious theoretical conceptualizations. I will question the explanatory potential
of the construct of learning environment. I will suggest that research may benefit from going
beyond the notion of learning environment, to such concepts as activity system and network
of activity systems.
The implementation of technologically advanced computer-supported learning environments
in educational practices is notoriously difficult. As Larry Cuban has repeatedly shown,
instructional technologies, in particular computer tools for learning, are oversold and
underused (Cuban, 1986, 2001). It is not an exaggeration to say that most of the research and
development on digital learning environments leads to prototypes and isolated demonstrationtype
implementations at best. I will argue that efforts at improving implementation may be
largely misguided to begin with. It may be more fruitful to frame the issue in terms of
expansive learning in collective activity systems, including schools and other educational
institutions.
To concretize these two conceptual shifts (from learning environments to activity systems,
and from implementation to expansive learning), I will present a case analysis based on data
from the Jakomäki middle school in Helsinki, Finland, where my research group conducted
two cycles of intervention studies, in the school years 1998-99 and 2000-01 (see Engeström,
Engeström & Suntio, 2002.
LEARNING ENVIRONMENT AS A WISHFUL IDEOLOGICAL ABSTRACTION
The notion of learning environment is usually presented with an attribute. We have dynamic
learning environments (e.g., Barab & Kirshner, 2001), innovative learning environments (e.g.,
Kirshner, 2005), powerful learning environments (e.g., De Corte, Verschaffel, Entwistle &
van Merriënboer, 2003), collaborative learning environments (e.g., Beers, Boshuizen,
Kirshner & Gijselaers, 2005), networked learning environments (e.g., Wasson, Ludvigsen &
Hoppe, 2003), smart learning environments (e.g., Dodds & Fletcher, 2004), real-life learning
envirnoments (e.g., Järvelä & Volet, 2004), authentic learning environments (e.g., Herrington
& Oliver, 2000), and many many more.
Common to this plethora of attributes is that they are positive, optimistic, promising, and
promotional. They seem to be designed to serve the selling of a wishful image of future
learning in which all good qualities of human interaction come true. In this sense, they are
2
thoroughly and blatantly ideological. The ideology behind them is to a large extent the
familiar belief in technology as the solution to social problems and human limitations
(Feenberg, 1999, Pippin, 1995). In this case, digital information and communication
technology is conveniently embedded within the softer notion of learning environment (de
Castell, Bryson & Jenson, 2002).
It seems that the proliferation of positive promotional attributes is a logical counterpart of the
scarcity of substantive models and critical theoretical examinations of the notion of learning
environment. When you don’t have a theory, you might as well replace it with a positive
slogan. Judging from the flow of funding to research on digital learning environments, the
strategy has worked pretty well thus far.
LEARNING ENVIRONMENT AS A STATIC AND HIERARCHICAL
ABSTRACTION
The notion of learning environment is commonly taken for granted as a self-explanatory
starting point in studies of computer-supported collaborative learning. Attempts at modeling
the anatomy of the notion are relatively rare. Many of those attempts are based on the idea of
concentric circles that represent multiple embedded layers or scales of a learning
environment. A recent model presented by Kirshner serves as a case in point (Figure 1).
Figure 1. The structure of a learning environment according to Kirshner (2005, p. 548)
Models such as this take the context as an envelope or container that surrounds human beings
engaged in action. The general structure of concentric circles dates back at least to
Bronfenbrenner’s (1979) attempt at modeling the nested systemic layers of the ecology of
human development. Such models tend to be inherently static and closed, like Russian dolls
lying still one within another. It is very difficult to depict and analyze movement, interaction,
contradiction, and construction of the context itself with the help of such models. Moreover,
concentric circles commonly imply that the smaller circles are hierarchically controlled and
constrained by the bigger ones.
3
The static and hierarchical nature of many models of learning environment corresponds to
what Davydov (1990) calls empirical generalization. Learning environment refers to a set of
empirically observable phenomena, namely to various uses of digital technologies with the
purpose of facilitating learning. The concentric circles actually capture something we often
observe in educational settings where digital technologies are used, namely the encapsulation
of the new digitally mediated environments as bubbles within institutions such as schools.
Since an empirical generalization does not recognize the origins and dynamic inner
contradictions of the phenomena it tries to cover, it becomes an empty shell, a placeholder
which can be filled with any number of dimensions or variables that are then used for
classifying further items that seem to fall into the general category in question. This is in fact
one way the notion of learning environments has been used. Thus, de Kock, Sleegers and
Voeten (2004) produced a classification scheme with no less than 18 types of learning
environments.
THE DILEMMA OF AUTHENTICITY
One of the wishful attributes attached to learning environments is ‘authentic’. A recent paper
by Gulikers, Bastiaens and Martens (2005) illustrates the dilemma of authenticity and, more
generally, the limitations of the notion of learning environment.
According to Gulikers and her co-authors (2005, p. 509), an authentic learning environment
“provides a context that reflects the way knowledge and skills will be used in real life.”
“This includes a physical or virtual environment that resembles the real world with real-world complexity and
limitations, and provides options and possibilities that are also present in real life.” (Gulikers, Bastiaens &
Martens, 2005, p. 509-510)
Reference to ‘real life’ implies that there is something that is not real, that is, artificial. But
what in this world is not artificial, made or modified by humans? There is no such pristine
domain of untouched nature.
Or perhaps ‘real life’ refers to something that is not selected, bounded and controlled by
teachers, curricula and instructional materials? But being subjected to instruction, supervision,
domination and manipulation is by no means unique to schools and educational institutions.
Instruction in the broad sense is a pervasive feature of all walks of life. Paradoxically, it may
also be the necessary precondition of learning (Sutter, 2001).
It seems practically impossible to define what is ‘non-authentic’ or ‘not real’. Would school
be non-authentic or non-real, as compared to work? Anyone familiar with the regimes of
assembly line work, for example, would recognize the absurdity of such a claim. Would
objects, events and symbols represented on a computer screen be non-authentic, as compared
to objects, symbols and events seen in the street? This might seem plausible until one realizes
that the objects, events and symbols in the street are to a large extent staged, purposefully
prepared to influence us. Examples of such staging range from the makeup, clothing and
cosmetic surgeries displayed by passers-by to the logos and advertisements painted on cars
and plastered on buildings.
Gulikers and her co-authors are not bothered by these problems. They simply characterize an
authentic learning environment as “a realistic simulation of the real world” (Gulikers,
4
Bastiaens & Martens, 2005, p. 510). Their empirical study is a comparison of the effects of
two learning environments, an ‘authentic’ and a ‘non-authentic’ one, offering basically the
same contents.
“Buiten Dienst (Dutch for ‘Out of Service’) is an ... authentic learning environment that makes use of a lot of
multimedia features to improve the realistic nature of the simulation. The student is placed in the role of a junior
advisor of a consultancy agency who is given the authentic task to write a report about the causes for the high
sick-rate in a bus company and what can be done about it.” (Gulikers, Bastiaens & Martens, 2005, p. 513)
According to the authors, an ‘authentic’ context was created by simulating a consultancy
agency in a virtual way with the help of numerous multimedia features combining visual,
aural, and written information. These features included virtual employees who answer
questions aurally and a secretary who can assist with administrative issues. Students could
talk to the virtual employees, observe a virtual bus driver in his job, and read articles from an
archive. They received coaching from a virtual senior advisor.
The ‘non-authentic’ condition was also an electronic learning environment, but without the
multimedia features. All the information was provided in written form only, there was less
context information, and there was no virtual senior advisor or secretary. In other words, the
practical criterion for