I went through my draft blog entries and I found this one from 2013. It was supposed to accompany my chapter in Ally & Tsinakos' Book on Mobile Learning. Recently, Marion and I worked on a different book chapter, in which we built on the same concepts. This post discusses why I found it necessary to extend Engestöm's Activity Theory model and what insights on learning design re-arranging the activity components yielded.

This post started from Quinn's decision tree model to educational design. I don't find this decision tree particularly usefull to explain design decisions for the educational designs I work on and it does not represent the flow of the typical educational design processes I encounter. This can be due to the fact that most educational design work actually happens after the decision of the educational format. Therefore, I am particularly interested in designing experiences on the activity level and in arranging and orchestrating activities as processes.

My educational designs I ground of established and well-grounded frameworks, such as 4C/ID, Bloom's Taxonomy together with the EQF, the conversational model, 8-Learning Events, and others. I find them useful, but together these frameworks and models form a complex relationship that educational designers need to balance. Consequently, one must know what to balance for creating meaningful learning experiences. I found that Engeström's Activity Theory model suited this task quite well, so the above figure shows basically my operational version of that model (in the 2013 and the 2017 version). For those who are familiar with Engeström's model you find, that my illustration actually does not match the original model, as you can also see below.

There are three main differences.

1. Engeström's Community is called Context in my model
2. The contexts were moved a bit to highlight the relation of Rules and Division of Labour (which I relabelled as learning and supporting tasks for clarification)
3. Engeström's (learning) Resources are included into the tools and the former Resources are now merely the Topic of an activity

Contexts are more than just Communities

The first change was a result from my work on mobile learning at the OUNL. While many mobile learning projects either ground on Activity Theory or on Laurillards Conversational Theory (Laurillard, 2012), I found it very irritating that neither theory actually considers context beyond the level of social interaction. I wondered, what about location-based learning, where the location plays a cruicial role for the learning activities, even if not social interaction takes place? What about time constraints such as the time-of-day, duration, rhythm, time available etc.? What about surrounding or embedding activities that yield noise or provide information that influence a learning activity? What about relations between objects, information, history? Trying to answer these kinds of questions on the grounds of Activity Theory will force you to work outside of the model, because non-animated aspects of an activity are either tools, objects or irrelevant.

Engeström got it right that there are external factors influence the learning activites, but stuck with the idea that it is only human interaction that can spark or constrain learning. With this idea Engeström is not alone, since many theories and models rely on the idea that only social interactions can effectively frame learning (Merienboer & Kirschner, 2013; Dillenbourg, 2015; Laurillard, 2012; Reigeluth, 1983, 2009; Lave, 1993; Wenger, 1998). I hear the same line of reasoning when I talk with organisational stakeholders that claim "our students learn only on campus and at home" or "our students have better things to do than playing with mobile devices". When I look at smart technologies and further at smart environments formed by the internet of things, then I start wondering whether this assumption is still valid.

I chose to work with the model for context-aware applications of Zimmermann, Lorenz & Oppermann (2007), in order to get a different perspective on other factors that can influence learning activities. This model distinguishes five types of context descriptors: location, time, relations, activity, and identity. Specht (2009) builds his AICHE model on top of this framework. The AICHE model describes the context-senitive adaptation and injection of learning opportunities into a context based on matching the context factors. This allows to describe the availability and flow of digitally enhanced learning experiences in changing environments and settings. This model is particularly useful when experiences in and for multi-device ecologies should be modelled. In my 2009 book I outlined, how the integration of the AICHE model and the context dimensions introduced by Lave (1993) and Wenger (1998) help to design and situate learning experiences for informal learning.

When we look at Engeström's model again, we find that the "community" component refers to a special set of social relations that match the definition of what Zimmermann, Lorenz and Oppermann call "relations". On these grounds I replaced "Community" with "Context", because of the community being defined by a special set of relations so all statements regarding the community remain unaffected by this change. On the other hand this change gains us new contextual aspects such as locations, time, activity and other kinds of relations that have similar influence on activites than Engeström's community concept. On the grounds of the broader definition of the context component, we can now start tackling the questions on locations, on time and on surrounding and embedding activities, as well as on a broader range of relations without leaving the conceptual framework of the model.

Rules, Tasks, and Context

The second change in my interpretation of the Activity Theory is to break its beautiful triangle structure. The main reason for this was that I could not link the rules of an activity with its tasks. Engeström calls the tasks devision of labour, in educational design models this typically refers to role specific tasks. For example, the 4C/ID modell label this component "part-task practice", which links learning activities and supportive activities. In IMS Learning Design there is a clear link between the learning tasks and the rules of an activity. IMS Learning Design goes further and highlights three different types of rules: prerequisites, completion and role assignment. This allows to model processes that link various kinds of activities.

This relationship appears to be missing in Engeström's model because there is no direct link between the rules and the tasks. The relations between the two elements are either mediated through the subject, the object, or the community. Given that I relabelled community already into context, the authority of the community wasn't as central as to the original model. By moving the context down the direct relation between rules and tasks could be integrated in a way as we find in other educational design concepts.

With this new arrangement the rules and tasks looked out of place due to the broken symmetry. By moving rules and tasks into the new symmetry of the hexagon, two more relations emerged: A relation between Tools and Rules as well as a relation between Tools and Tasks.

The relation between Tools and Rules appears to be obvious. Most tools were made for a specific purpose and therefore implement the rules that are related to that purpose. If we design technology enhanced learning activities we look for or design tools for a specific purpose in order to integrate them fully or partially into the learning activities and experiences we design. At this level the relation between tools and rules is by design and therefore the tools will enforce the rules they incorporate. Besides this designed relation many generic purpose tools have a more subtle relation to rules, which is typically referred to as "affordances". Affordances describe characteristics that suggest or encourage interactions with a tool or instrument. While affordances are typically linked to design elements, their characteristics are not always part of the immediate considerations and decisions of tool developers ad designers.

The second new relation is the relation between Tools and Tasks. In the original model tasks refer to the devision of labour and thus related to (typically) human activity. Engeström refers to tasks often in terms of production processes. The new relation in the model highlights that tools can be responsible for certain tasks. In the relation to learning activities these tasks will supportive tasks. Some of these tasks could be attributed to activities that would be otherwise performed by a learning facilitator, so we could attribute them to the subject. However, other tasks are not related to normal educational activities performed by humans. Like in many educational design models this would be ignored. Most prominently, IMS Learning Design does not event allow modelling tasks performed by a tool. Instead these tasks are considered as intrinsic to the tools. By making the tool-task-relation visible, the related decisions and implications for the overall activity become accessible to the educational design.

Smart Technology and Digitalisation

By breaking the original symmetry three new relations became visible that were not considered by Engeström. These new relations form a previously hidden relational triangle. I call this the smart technology or digitalisation triangle. From the perspective of the Activity Theory, these relations provide an important insight: Smart Technologies can take responsiblity over tasks without becoming a subject. Yet, they support the activity and adhere to the rules related to their tasks. Instead of attributing characteristics of "subjects" to tools, this model allows smart technologies to occupy a distinct role and function that goes beyond traditional tools. More importantly, these aspects are also bound to the decisions we make as educational designers.

Resources are Tools not Objectives

The third change to the model I included just recently. The change is related to a flaw in the original model, which was really hard to detect. I came across it, when trying to explain the difference of learning resources, learning objectives and learning outcomes to practitioners. Learning objectives are typically translated into German as "Lernziele" (learning goals). Consequently, many educational practitioners find it hard to understand the difference between learning outcomes and learning objectives and to apply this appropriately. However, there is another less common German translation of learning objectives: "Lerngegenstand" (Herbart, 1841, §105). A "Lerngegenstand" describes the topic that is taught, which is independent from the resources used for teaching it.

In Engeström's model there is nothing like "objectives", just objects. In the context of learning Engeström explicitly labels "learning resources" as the objects of learning activities. While the classic German educational theories use the term "Lerngegenstand" interchangingly for learning objectives and learning resources, the meaning of term it is always clear in the context of use. So in the classic literature it quite clear if an author means an objective or a resource. This is clearly not the case in Engeström's Activity Theory model.

The flaw in the model becomes clear with the help of the smart technology triangle. A very typical learning resource is a book or a script, while a learning tool would be a online discussion forum or a database. In Engeström's model the difference between a resource and a tool is the interactivity and the capability for manipulation. A traditional book or script does not have these capabilities. If we simply digitise a book, these capabilities remain absent. What happens, however, if we make one visualisation interactive in a way that readers can change parameters and test how different elements influence the outcomes in a small simulation? Does the book become a tool or does it remain a resource? If it remains a resource, what is the difference to digital tools? There are no easy answers to these questions and this is why I consider this a flaw in Engeström's model.

The easiest way out of this dilemma is to utilise the concept of the objective or -- given to the problematic translation of objective -- "topic". If we use topic instead of object in the model, everything that is used to teach a topic in a learning activity becomes a tool. This resolves nicely the inconsistencies when differentiating between learning tools and learning resources and allows learning resources to become subject of digitalisation processes. More importantly, the concept of learning objectives is now represented in the model and it is clearly separated from the learning outcomes.

New Insights

The transformed model yielded new insights for modelling learning activities, besindes the smart technology triangle. The rearranged model shows three pairs of components. The first pair (subject/actors-topic) reflects the classic educational notion of the interaction of teachers, students and the topic. Therefore, I call this the "interaction dimension" of an activity. The second pair (roles-tasks) refers to the internal and external procedures of an activity. I call this the "procedural dimension" of the model. The third pair (tools-context) refers to the enabling aspects of an activity. I call this the "environmental dimension" of an activity.

When we combine the components of the interation and the procedural dimensions, the model yields almost naturally the key principles of 4C/ID. If we do the same for the interaction and the environmental dimensions, we will get the key concepts of mobile learning.

Another insight came from looking at the environmental dimension. The relation between the tools and the context is marked as a bi-directional relation in the model. However, in the educational design process there is often a starting point for the design. Depending on our starting point, we apply different reasoning for working through our educational ideas. Therefore, this bi-directional relation should be memorised as two directional relations during the design phase. This means for the environmental dimension, that if we start from a tool-centric perspective, the possible learning contexts are constrained by our tool choice. If we start from the contexts that we have to support or to bridge, these contexts will limit our choice for tools. While this sounds similar, in practice this will lead educational designers to very different results.

The kind of design decisions depending on the starting point on the environmental dimension are typical for two educational design concepts. The tool-centric perspective is very typical to blended learning concepts, while the context-centric perspective is typical for seamless learning concepts.

Conclusions

I started from working with Engeström's Activity Theory, but I was not really satisfied with what I could explain and express with it. Initially, I had a only rough ideas about the underexpressed context in the model. Over the time this grew into a better understanding and reflection of my own educational design practice. Eventually this lead me to leaving the original model and extend it, which I now find more consistent with the design decisions as well as with the recent developments in the field of smart technologies and digitalisation. With the extended model I can focus more clearly on the different aspects of learning activities. I find it very useful to have a broad representation of contexts in the model, because it helps exploring new kinds of educational designs from starting with rather conventional ideas.

References

• Dillenbourg, P. (2015). Orchestration Graphs?: Modeling Scalable Education. Lausanne: EPFL Press.
• Engeström, Y. (1987, 2015). Learning by expanding?: an activity-theoretical approach to developmental research (Second Edition). New York: Cambridge University Press.
• Glahn, C. (2009). Contextual Support of Social Engagement and Reflection on the Web. Heerlen: Open University in The Netherlands.
• Herbart, J. F. (1841). Umriss pädagogischer Vorlesungen. Göttingen 1835; 2., vermehrte Ausgabe 1841
• Laurillard, D. (2012). Teaching as a Design Science: Building Pedagogical Patterns for Learning and Technology. New York and London: Routledge.
• Lave, J. (1993). The practice of learning. In S. Chaikin, & J. Lave (Eds.), Understanding practice, perspectives on activity and context (pp. 3-33). Cambridge: Cambridge University Press.
• Merriënboer, J. J. G. van, & Kirschner, P. A. (2013). Ten steps to complex learning: A systematic approach to four-component Instructional Design (Second edition). New York: Routledge.
• Koper, R., Olivier, B., & Anderson, T. (Eds.) (2003). IMS Learning Design Information Model. IMS Global Learning Consortium.
• Koper, R., & Tattersall, C. (Eds.) (2005). Learning design: A handbook on modelling and delivering networked education and training. New York. Berlin/Heidelberg: Springer.
• Reigeluth, C.M. (1983). Instructional Design: What is it and Why is it? In C.M. Reigeluth (Ed.): Instructional-Design Theories and Models: An Overview of their Current Status (pp. 3-36). New York: LEA.
• Reigeluth, C.M. & Keller, J.B. (2009). Understanding Instruction. In C. Reigeluth & A.A. Carr-Chellman (Eds.) Instructional-Design Theories and Models: Building a Common Knowledge Base, Vol. III (pp. 27-40). New York: Routledge.
• Reigeluth, C.M. & Carr-Chellman, A.A. (2009). Situational Principles of Instruction. In C. Reigeluth & A.A. Carr-Chellman (Eds.) Instructional-Design Theories and Models: Building a Common Knowledge Base, Vol. III (pp. 57-72). New York: Routledge.
• Specht, M. (2009). Learning in a Technology Enhanced World: Context in Ubiquitous Learning Support. Inaugural Address. September, 11, 2009, Heerlen, The Netherlands: Open University in The Netherlands.
• Wenger, E. (1998). Communities of practice: learning, meaning, and identity. Cambridge, New York: Cambridge University Press.
• Zimmermann, A., Lorenz, A., & Oppermann, R. (2007). An Operational Definition of Context. Modeling and Using Context, 558-571.