7 | Theories of Teaching

Pedagogical Families

Terry Anderson and I (2011, 2012) divided the field of distance learning into three distinct (but persistent and nowadays coexistent) generations defined by the dominant pedagogies of successive historical periods. The first we originally described as behaviourist/cognitivist, though recently we have preferred the term “objectivist” (Dron & Anderson, 2022) because what binds such pedagogies is not a set of methods or common theories but the assumption that there is an independent true body of knowledge to be learned and an optimal set of methods to learn it, whatever that might be. Some prefer the term “instructionist” (e.g., Johnson, 2009), which effectively captures the pedagogical emphasis of such approaches, but I am reluctant to fully endorse the negative attitude that it implies, and it is equally applicable to learners who apply such methods themselves.

The second generation we originally described as social-constructivist but now prefer to label as “subjectivist” (Dron & Anderson, 2022), reflecting its epistemological underpinnings that focus on how subjects construct knowledge (not normally that reality is subjective). Subjectivist pedagogies assume that knowledge is individually and socially constructed rather than (or, more often, in addition to) being independently true. Typically, subjectivist teachers adopt a softer set of pedagogies such as problem-based, inquiry-based, and other more student-directed active learning techniques. Some, such as those of Papert (Papert & Harel, 1991) and Piaget (1952), focus mainly on individuals’ learning, whereas others, such as those of Dewey (1916) and Vygotsky (1978), treat learning as fundamentally social in process and substance.

The third generation we (Dron & Anderson, 2011) initially labelled as connectivist (with a small c to distinguish it from the specific theory of that name) to describe models of learning developed in an age of information plenty. We now (Dron & Anderson, 2022) follow Davis and Sumara (2006) in describing this generation of models as “complexivist” because all share the common feature of seeing learning and the processes of learning as complex adaptive systems, many predate Connectivism, and the term better reflects the diversity of the field. Complexivist models treat knowledge as distributed, situated, complex, emergent, as much embedded in the networks of people and stuff that surround us as in our own brains.

Although our model was used to examine the history of distance education, it has broader applicability as a means to distinguish all families of pedagogical theory. In brief, objectivism is concerned with theories of teaching, subjectivism with theories of learning, and complexivism with theories of knowledge. Thus, they can be seen as orthogonal views of the same basic phenomena, and, as I hope to show, they are not mutually exclusive. Each can play a complementary role in the educational technology assembly.

Assemblies of Pedagogies

I have presented these paradigms of pedagogical theory in order of harder to softer: objectivist, subjectivist, and complexivist. In distance education, this is also the order in which they emerged, largely because of the constraints imposed by the communications technologies with which they were assembled, path dependencies caused by what was inherited from in-person education, and the affordances of new inventions such as the internet that created adjacent possibles into which they could evolve. However, though often presented as competing models or successive generations, the reality is far more complex. Far from competing, it is normal to find all coexisting in any given learning trajectory, each playing a different but complementary role in the process. Indeed, there are arguments to be made that this should be so, because each speaks to different aspects of the educational assembly.

Objectivist (behaviourist and cognitivist) pedagogies are usually fairly hard, from a learner’s perspective, and often prescriptive. However, bearing in mind the inherently distributed nature of teaching, pedagogies are soft for learners as well as their ostensible teachers, so the reality has always been that learners under such conditions seldom follow the rigid paths determined for them by instructional designers (Haughey & Muirhead, 2005). Although the norm in distance learning, this is even true, to an extent, in tightly controlled traditional classrooms. In a fascinating in-depth study of a small selection of learners in a conventional classroom, Nuthall (2005) found that they had critical learning experiences because of their own self-designed experiences and resources (from 6.5% of the time for the lowest achiever to 13.1% for the highest), and all learned through interactions with others (from 6.5% to 14.8% of the time, similarly related to lower and higher achievement), and that was not counting occasions when the teacher gave them choices that made them partly autonomous by design.

Asynchronous learners who work independently, of necessity have more choices than synchronous learners in a classroom—because a teacher does not directly control any moment of the activity—so the overall technology of learning is therefore, at least for the duration of the formal teaching process, almost invariably softer, regardless of the teacher’s intentions. For students trapped in a physical classroom, though they will construct their knowledge differently and play some of the teaching role for themselves, the pedagogy used by the teacher, especially when following a typical lecture format, may be considerably harder than it would be for students watching (for example) a videotaped lecture online, which they can pause and rewind or play at a different speed as needed. Notwithstanding a reduced capacity to interrupt to seek a different explanation (something that normally demands assembly with other technologies, such as discussion forums in a distance setting), the online learner usually controls the pace, the place, and the time of learning and typically is more able than an in-person counterpart to take divergent paths not planned by the learning designer. There are distinct limits to this autonomy. The fact that the almost ubiquitous focus on set outcomes and the assessment of those outcomes makes the distance teacher’s control strong and places fixed limits on how far a learner can diverge, so detracting significantly from learner autonomy, but (compared with an in-person classroom context) there is always greater freedom to follow alternative paths to achieve the same goals. Objectivist assessments have a tendency to focus on those fixed outcomes, and it is not uncommon to find the use of objective tests, the hardest of all assessment technologies, playing a significant role as well as other mostly hard assessment tools such as written exams and quizzes. For distance learners, given the absence of classroom roles that emphasize the dominance of the teacher, these are often the primary means by which teachers assert control over the learning process.

Whereas objectivist models are explicitly concerned with teaching or training, subjectivist theories are primarily concerned with learning. This means that teaching processes do not arise directly from theory but are developed with a learner model in mind and assembled responsively as that model changes. Subjectivist pedagogies thus tend to be softer than their objectivist counterparts, requiring learners to engage creatively with problems, discussions, arguments, and constructions, each learning in unique ways because part of the process is enacted explicitly by the learner in interaction with others. This has led to some criticisms. Some have noted that subjectivist approaches tend to be time consuming, inefficient, and expensive (Annand, 1999, 2019), and this is indeed what lies at the root of what Daniel et al. (2009) describe as the “iron triangle” of access, cost, and quality—the need for skillful technique and constant adaptation makes subjectivist approaches expensive and unreliable. Others object to the softness itself.

It is precisely that softer lack of prescribed process that Mayer (2004) finds objectionable about pedagogies based upon subjectivist principles. If learners are left entirely to their own devices and have insufficient skills to add their own pedagogical processes, then the results might be (and often are) relatively poor when measured by predetermined outcomes. Equally, if unskilled teachers fail to provide the necessary scaffolding and support, then there is a good chance that their efforts will fail. However, though such instances can and do occur frequently, Mayer is wrong to dismiss subjectivist methods altogether. First, the notion of the unguided learner who makes discoveries alone is a myth: we have already seen that there are always other teachers, and this is explicit in social constructivist pedagogies. Second, a lack of prescribed process does not mean a lack of process: it is just a softer technology that leaves plentiful gaps to be filled, and therefore demands active creation and refined technique by its instantiators, who might do it well or not. Because they are soft, subjectivist methods usually require responsiveness, skill, and talent from a teacher (especially when learners are inexperienced in the method and/or subject area), who must responsively adapt and invent pedagogies to address changing needs and concerns as individual and group problems are addressed. This is both the biggest weakness and the greatest strength of such pedagogies, inasmuch as a poor (or time-poor) teacher will do much worse and a good (or time-rich) teacher will do much better. In social constructivist approaches, there are likely many teachers, further softening the overall assembly.

Effective subjectivist methods are not free, however, of cognitivist or even behaviourist pedagogies. Every participant in every group of learners has a model of how others learn that is brought into play during interactions with them, and this is especially true of one playing an explicit teacher role, who may use any number of different pedagogies along the way to help support learners in their discoveries. Simply explaining the process, even if it is thereafter very hands off, demands at least a rough model of how best to impart information in a manner that will be understood, remembered, and utilized by the learner (and is often one of the ways that it goes wrong). Objectivist pedagogies are therefore unavoidable. However, the difference between subjectivist and objectivist models tends to be not that subjectivist teachers avoid objectivist approaches altogether (which would be absurd and inefficient) but that, they are assembled as needed, often on demand, rather than being dictated in advance by an outcomes-focused teacher.

Complexivist models, notably in their most archetypal form of Connectivism, are not quite theories of learning nor theories of teaching but theories of knowledge. Their various forms explain how knowledge comes to emerge in individuals and in networks or groups of connected individuals and the artifacts that they create. Common to this idea is that bounded systems—from cities to ecologies to termite colonies to human brains—learn in analogous ways, adapting and accommodating change through similar processes. Therefore, the boundaries of the “system that learns” are not necessarily drawn around the learner. The learner is one significant bounded element of the process, recursively a part of a whole taught by the whole. As Davis and Sumara (2006, p. 15) put it, “the physical or conceptual boundaries of a complex/open system are always contingent on the criteria used to define or distinguish the system from its backdrop.” They explain that, “in complexity terms, learners can include social and classroom groupings, schools, communities, bodies of knowledge, languages, cultures, species—among other possibilities” (p. 14). In such a system, technologies are not just reflections or products of cognition but also active participants in it.

Given that the self-organized emergence of order in richly connected systems is central to all variants of the complexivist model, and that intentional design tends to take a more reactive, partial, and (if visible at all) structural role, complexivist pedagogies, such as they are, are thus so soft that they might provide little or no process guidance beyond dictating a theme or general principles of assembly. This is particularly significant when we remember the mantra that soft is hard, and hard is easy. As we move away from objectivist teaching methods, learners have to make more and more creative decisions, to be active creators, not just users, of the pedagogy. Because pedagogies (to the creator) are soft technologies, it is possible to create them and use them with greater or lesser expertise. In complexivist accounts, this relates not only to individuals but also to the social networks, groups, and sets of which they are parts, to which it is rarely easy to ascribe volition, let alone intention or design. Nonetheless, collectives (emergent entities formed from local interactions of independent agents), as well as designed technologies, more formal groups, and their processes, play important roles in shaping the behaviour of self-organizing systems. Locally, at the level of an individual learner, hardness can start to creep in through both intentional design and unintended emergent structure.

Achieving the right balance between soft and hard is important, and it is not enough simply to assume that the right help is a click away. Learners must be able to choose when to choose, because choice alone does not give them control (Dron, 2007). Once again complexivist accounts might explain how knowledge emerges, but, beyond some broad patterns of role modelling and engagement, they do not predict in any detail how to make learning happen, let alone what will be learned. It is not a totally self-organizing free-for-all but a richly connected assembly of both intentional and unintentional interactions and processes. This itself is in keeping with a complexivist account given that complexity-driven explanations invariably recognize the emergence of different levels of explanation according to the boundaries that matter at any given time. Although there might be important and interesting commonalities (e.g., that they obey laws common to all scale-free networks), there is a need for different kinds of explanation of, say, the structure and dynamics of social groups or organizations than of, say, the exchange of chemical messengers in cells, neural networks, the operation of the endocrine system, or the formation of network cliques.

In both complexivist and subjectivist accounts, guidance can come from any of the many teachers in a learning transaction, but especially and most effectively it is best when it comes from the learner. Thus, for subjectivist and, especially, complexivist models of learning to work well, support is needed to allow learners to gain expertise in learning itself, to become effective users of pedagogies, not just to become proficient in the subject of what is being learned. This is an assembly that grows more by accretion than replacement. The assembly leads to greater overall softness, but the parts themselves can be hard. We still need to learn, for instance, hard, human-enacted technologies such as spelling or the actions needed to submit a blog post. One interesting feature of such phenomena is that the pedagogies used by learners typically, at the finest granularity, tend to be objectivist. It would make little sense for them to be anything else, though multiple scales of assembly can make it fairly common for subjectivist pedagogies to be assembled from other subjectivist pedagogies too.

Subjectivist and complexivist learning technologies are therefore assemblies constituted largely by objectivist pedagogies. Where they differ from purer objectivist pedagogies is that, to a greater or lesser extent, learners themselves perform much of the assembly, rather than their teachers, and (at all scales) the various participants have greater freedom in their choices of pedagogy than those following a more objectivist approach. In the case of complexivist models, it is assumed that there will be many teachers, including those that are non-human (software, texts, and emergent collective entities), and that learning will occur at a system-wide level as well as in the individuals of which the system is composed. Subjectivist and complexivist approaches have many advantages, not the least of which are the fact that learning is tailored to the learner and integrated with a learner’s existing knowledge and the fact that hard prescriptive pedagogies forcefully applied by someone else are demotivating for the same reason that all prescriptive technologies sap motivation—they reduce control. However, their major disadvantage lies in the expertise needed to instantiate those pedagogies effectively. The enactment of an objectivist method demands relatively little skill, once it has been expertly designed. Objectivist pedagogies can be designed largely in advance, whereas subjectivist and especially complexivist pedagogies must be developed, or emerge, on the fly.

Because of their innate softness, to both learner and teacher, a teacher (including a learner) who wishes to use subjectivist or complexivist approaches needs to provide, discover, or invent support for the process. Technique is critical. In a subjectivist approach, this support typically takes the form of scaffolding, of creating tasks that gently lead learners outside their comfort zones while providing feedback, encouragement, and support, answering questions, prompting reflection, and critiquing methods. In a complexivist approach, it typically means modelling good practice, exposing ideas, providing opportunities and support for active creation, discovery, and curation of knowledge artifacts, revealing interesting and diverse resources, and helping to aggregate a strong network of interested people and artifacts they find or create around a topic. Often such support is emergent, for example through the sharing of useful ideas or resources that, if useful to more than one or two people, will be reshared and recommended by enough people to allow learners to assume some value in them. Digital tools for aggregation and discovery usually play a significant role in this process. Despite such possibilities, one of the most common criticisms of subjectivist and especially complexivist models is that learners can receive insufficient guidance. They can be set adrift, fail to notice important facts, take suboptimal learning journeys, or shuffle a limited range of ideas.

The social context often can work against them. As Kay (1996) explains, simply putting a piano in a classroom without support for learning it leads to a chopsticks culture in which little progress is made. The blind lead the blind. Although, especially in a complexivist model in which online resources play a significant role, the social environment can afford both exemplars and direct tuition, this can be haphazard, incomplete, and inadequate for individual needs. Although it is great to be exposed to diverse ideas, and to be afforded the opportunity to discover the best that the world has to offer, it is easy to learn falsehoods, or to learn ineffectively, or to learn too little. Much of my own early research was devoted to finding ways around these problems through software that supported the collective organization of resources, as summarized in Dron (2007). The challenge is to influence the development of the collectively generated, emergent structure so that it is more likely to support learning, with a focus on connections and the signals that pass between them. Analogously, just as the signals that pass between termites and their environment have evolved to support the building of intricate, air-conditioned towers that support the colony’s well-being, so too it is possible to support signals that pass between learners to support effective learning.

Objectivist, subjectivist, and complexivist accounts of learning and teaching differ mainly in the relative softness or hardness of their orchestration. On closer examination, however, and from the perspective of a learner, the lines are blurry, and there are few hard and fast distinctions between them. When successful, objectivist models benefit from softening; likewise, complexivist and social constructivist models benefit from hardening. Although they might be the creations of their participants, the softest Connectivist MOOC (cMOOC) is filled with hard elements, and the hardest objectivist course, at least for mature and wise learners who exercise their autonomy, is as soft as it needs to be. Unsurprisingly, as Hattie (2013) notes, there turns out to be little difference in learning outcomes on average no matter what pedagogical method is used, but of course the devil is in the detail, and virtually no learner is average. For instance, when discussing the surprisingly minimal effects of class size on learning outcomes (measured in achievement), Hattie observes that different approaches are needed for different numbers of students, but it appears that a great many teachers adopt the same approach for all, thus negating any potential advantages or disadvantages of a particular class size.2 Just because a particular set of pedagogies is used does not mean that it is used well. Pedagogies and other technologies have to work together if they are to be effective. Unfortunately, such issues are seldom examined with the care that they deserve in reductionist studies of educational interventions.

Mitra’s Holes in Walls

Sugata Mitra’s (2012) Hole in the Wall project affords a useful example of the interplay between different pedagogical models in a real-life setting. The project provided (and, at the time of writing, in some places continues to provide) internet-connected computers in open spaces, designed (through placement and positioning as well as software and design) so that only children would be likely to access them. In a loosely complexivist account of learning, Mitra writes of the remarkable way that what he describes as “self-organized learning” emerges as small groups of children gather around machines, without apparent guidance, and learn to operate them. As a result, they learn to use the machines to learn more, making and sharing discoveries with one another, co-creating learning strategies that result in all learning together in a virtuous circle of ever-increasing knowledge and understanding.

In keeping with complexivist accounts, Mitra (2012) puts this down to the combination of computers and the emergent processes of groups of children interacting with them and one another. The computers themselves are not the technologies that do the teaching: largely, the software and content that they provide do the work, along with the interactions of the children with one another. The computers are filled with reifications of knowledge and ideas, myriad small, hard pieces that can be assembled together, each containing knowledge and learning, each filled with implicit ways of understanding the world. Some are objectivist tutorials, help files, and other deliberate acts of teaching, others are simply things to interact with and, in so doing, to learn from, in the subjectivist tradition. Attached to the internet, the kids have access to a countless number of teachers, including those who have embedded many intentional teaching processes as well as a vast amount of reified knowledge that, intentionally or not, informs, inspires, influences, and explains. Rather than having a single teacher, these children (potentially) have millions of them as well as (in a somewhat more self-organizing way) one another.

The internet links and connects them, cross-referenced and infinitely varied, the embodiment of the augmentative, cognitively enhancing Memex imagined in the mid-20th century by Vannevar Bush (1945). Through mistakes and accidental discoveries, amplified through implicit pedagogies employed by the children as they explore together, thought processes shape themselves around and with those offered by the machines. Beyond that, to use a computer means having to think, at some level, like the designer of the hardware and software that it uses. The user has to interpret both the explicit metaphors provided in an interface and the mental models of its designer. We do not necessarily need to be taught this by someone else, or to share the same models, as long as those whom we work with understand our meaning. Mitra notes that children create their own vocabularies for things such as icons and cursors in an act of sense making shaped around and by the machines.

What is less clear about the allegedly successful Hole in the Wall project is whether the pedagogies that emerge as children engage in constructive dialogue with the thought processes embedded in the machines are particularly efficient, effective, or useful: soft is hard. The pedagogies that children encounter are many and varied, and not all are of equal quality. They appear to work sometimes, and sometimes well, but this might well be a result of high motivation (the relatedness, control, and competence aspects that underpin intrinsic motivation are extremely high) and consequent time on task rather than innate value in the pedagogies that children encounter.

There are clues in Mitra’s (2012) work that things are not quite as self-organizing as he suggests. In fact, even in the early interventions, there were pre-installed training programs provided by one of the project sponsors (a commercial learning technology company), each of which strongly embodied intentional teaching, though it is not clear to what extent they were used by the children.

In later studies, Mitra and his team have discovered that the addition of an adult mentor to help children focus more on specific tasks can increase the effectiveness of their learning by a considerable amount. The pedagogies employed by such mentors harden the assembly a little and, in so doing, increase the efficiency of the process, the focus, and the equity of use. The mentors play a guiding and moderating role, helping the children to discover things that might be more useful. Most of the computers provided in recent years have not been out in the open but in controlled spaces such as school playgrounds. In these later iterations of what Mitra (2012) somewhat misleadingly christened as SOLEs (self-organized learning environments), the pedagogies are far closer to subjectivist methods used in much traditional teaching than to a complexivist model, though they do benefit from the vast web of knowledge with which learners are connected.

Indeed, without mentors, things did not go well. Mitra sometimes leaves it unsaid that the original Hole in the Wall experiments were not a long-term success. After the researchers stopped paying attention, the holes in the wall soon succumbed to misuse and abuse, with larger and more assertive children dominating the process, great gender inequalities, vandalism, and lack of educational benefit. There was a great deal of game playing that might not have been particularly educational, notwithstanding whatever the kids learned by playing games and engaging in “negotiations” to use the machines (De Bruyckere et al., 2015, pp. 158–160).

The vast majority of the original holes in the wall are now simply that—holes in the wall. Having millions of teachers might seem to be a wonderful thing, but it is just as important to have the power to choose between them (Garrison & Baynton, 1987) and the support to discover their value. It is not the specific methods that matter as much as their appropriate uses at the right times, and how they are assembled, that make a difference.

Using the Right Technologies

There is a place for more or less any technology, model, principle, or approach in any educational assembly as long as it works with the rest of the parts. If a designated teacher has not considered the process from the multiple perspectives implicit in objectivist, subjectivist, or complexivist models, then chances are that, somewhere in the assembly, they will occur anyway because each represents a meaningful and useful way of coming to know something. Designated teachers do not need to do or enable all of this, but they do need to be aware that it is happening. This is necessary because it is also important to ensure that one model does not inhibit or crowd out the others. Complexivist or subjectivist approaches are virtually useless without at least some objectivist teaching, whether it comes from the designated teacher, one another, the internet, a book, or a friend. Equally, an objectivist approach is virtually useless without careful consideration of the social context and the ways that hard skills and knowledge will be applied: at best it will demotivate, at worst it will be bypassed altogether (e.g., by cheating or dropping out). Without an authentic, meaningful, socially beneficial, and personally relevant context of application, without being put to uses that matter, the machines constructed in learners’ heads will break, be forgotten, or, worse, be instantiated at a cost—in attitudes, values, and beliefs—greater than the benefit derived from them. Similarly, the goal-driven nature of subjectivist and objectivist models can blind both teachers and learners to the many other important effects and learning that result, and emergent effects of complexivist behaviours can create barriers or brakes on intended outcomes. Above all, therefore, there is a need for all teachers in the process—especially the learner and (where applicable) the designated teacher—to be aware of how, why, and for what purposes teaching occurs, from whatever sources it derives. If you learn anything from this book, then at least learn this: teaching must incorporate learning about whoever or whatever is learning. Even when we let go of the learning process, it is important to stay close to the learner.