2 | A Handful of Observations about Elephants

People Must Be Made to Learn

We are all born with an insatiable thirst for learning. When we want to learn something, as children or as adults, we need no special encouragement. We do it as naturally as we breathe, drink, or eat. It is intrinsically fulfilling to overcome challenges, to become more competent (Ryan & Deci, 2017), in almost anything, including many things that others (and perhaps we too on reflection) would find pointless or trivial. There is an intense joy in learning that almost everyone has felt at some point. Learning is its own reward, yet, in formal teaching and training, we usually force people to do it with rewards and punishments. In doing so, we could not send a stronger message that the activity itself is undesirable (Kohn, 1999). Even when apparently successful—usually measured by grades—a remarkable amount of what is intentionally learned under such conditions is forgotten once the rewards or threats are removed, though we might remember how we felt about it. This is perfectly natural. Just as we usually remember the number of our hotel room while we are staying there but forget it almost immediately when we no longer need to know it, so too, once the grade or credential has been achieved, or the punishment has been avoided, we tend to forget much of the information that we needed at the time. It can also teach us limited, dependent, and often ineffective ways of learning that rarely serve us well throughout our lives outside a few atypical contexts. It is also probable, worryingly, that we learned little or nothing in the first place. I am sure that neither you nor I ever cheated on a test, but we are in a minority. In some cultures, over 80% of the population admit (anonymously) to cheating on assessments (Ma et al., 2013). In the United States and Canada, over half of students admit to it (Jurdi et al., 2011; McCabe & Trevino, 1996).

At least some of our dominant teaching methods, whether or not they result in the intended learning in the short term, are counterproductive in the long term, too often resulting in learners with no desire to learn more, little recollection of whatever they were supposed to learn in the first place, and little passion for what they have learned. Given our natural love of learning, this is more than a little strange.

Online Learning Dominates in-Person Learning—Except in Formal Education

What do you do first when you need to discover some information or learn something? The odds are that your answer will be an internet search, most likely through Google or one of its close competitors. If not, or maybe as a result, depending on your field of interest and learning needs, you might visit Wikipedia, or Stack Exchange, or a bulletin board or Q&A site for your field, or YouTube, or ChatGPT. You might send a message—an email, a direct message, a tweet—to someone whom you believe can help you. You might even ask someone nearby. The chances, though, that your first port of call is a library, let alone a course of some sort, are fairly slim, even though you might use internet technologies to find one and it might help to guide a good part of your learning journey later. For those of us with access to the internet, it has fundamentally changed how we acquire knowledge and skills. We are now used to being able to discover more or less anything almost wherever and whenever we need it. Whether the knowledge that we gain in the process is reliable, sufficient for our needs, appropriate to our understanding, or relevant to our wishes might remain a hit-and-miss affair. Nonetheless, we are all learning, most of the time, anywhere, anytime, anyplace, and we are doing so online. Moreover, we do so without coercion, without the threat or reward of being graded for it. If we are talking about intentional learning, then it appears that online learning dwarfs its in-person counterpart in quantity if not necessarily in quality.

So why is it that, in formal education, online learning is often considered to be a poor second cousin to in-person learning (Protopsaltis & Baum, 2019), and, when given the choice, many people prefer to avoid it? And who is teaching us when we learn this way?

No Significant Difference in Learning Outcomes No Matter Which Media or Tools You Choose

Common sense suggests that, if people are taught in different ways, the results should be different. For instance, as alluded to in the previous section, many people believe that online learning is inferior to in-person learning. However, a large body of research over many decades has shown fairly definitively that, on average, this belief is false. The no-significant-difference phenomenon has been observed for a long time in the case of different learning media. Russell (1999) catalogued 355 explicit examples from 1928 to 1999 that illustrate the phenomenon that the mode of delivery appears to be insignificant (on average) in achieving effective learning. The same is true of online learning. In what is likely the largest metastudy to date, conducted by the US Department of Education, researchers looked at over 1,000 comparative studies and revealed no significant difference (Means et al., 2009). Indeed, metastudies of such metastudies equally reveal no significant difference between learning outcomes for online and face-to-face learning (Allen & Seaman, 2013; Pei & Wu, 2019). Similarly, large-scale individual studies tend to show little or no difference in outcomes (Cavanaugh & Jacquemin, 2015). A few metastudies reveal a slight tendency toward better outcomes for online and distance learning, and slightly better still for blended approaches, but that can usually be explained by demographics of students, competence of early adopting teachers, publication bias, or other methodological flaws (Chen et al., 2010; Tamim et al., 2011). This is not to suggest that there are no consistent and important differences in the online learner experience between in-person and online modalities. There are huge differences. But the learning results, as they are normally measured (which I will question in Chapter 10) tend, on average, to be pretty much the same for online learners as those for in-person learners.

Perhaps even more surprisingly, not even considering media or tools, it makes little difference how one teaches. In what must surely be among the most rigorous and influential metastudies of metastudies in the field of education, John Hattie (2013) synthesizes over 800 metastudies, relating to millions of learners, drawing from this vast catalogue those strategies, techniques, and methods that research shows to be most effective. Perhaps the most central message that we can draw from all this is that, as Hattie (pp. 34—35) himself puts it, “almost everything works. Ninety percent of all effect sizes in education are positive. Of the ten percent that are negative, about half are ‘expected’ (e.g., effects of disruptive students); thus, about 95 percent of all things we do have a positive influence on achievement.”

How is this possible? Surely the ways in which we teach must make some difference. Yet, as most teachers know, not only do different methods appear to work equally well, but also we can use what appear to be the same methods repeatedly over multiple iterations of a course yet achieve utterly different results.

The Best Ways to Teach Are Not the Best Ways to Teach

Social constructivist pedagogies, typically drawing inspiration from Dewey (1916) or Vygotsky (1978), are widely taught to teachers around the world, used across the educational spectrum, and often upheld as models of best teaching practice, with good reason: theory suggests that they should be effective, and many published studies seem to indicate that the theory holds in practice. However, the empirical evidence to support such beliefs is not compelling. For example, Andrews et al. (2011) investigated the effectiveness of active learning (broadly covering a range of methods inspired by constructivist epistemologies that demand engagement with, rather than absorption of, knowledge) in a random sample of college biology courses, finding little or no benefit to learners who had been subjected to an active learning pedagogy compared with those who had been taught in a more conventional fashion. In some cases, at least by the crude measures of success employed (mainly terminal grades), things were actually worse.

The researchers surmised that previous research had involved researchers skilled in the use of active learning pedagogies who were enthusiastic about and engaged in what they were teaching. When applied by the rank and file of teachers with limited skill or engagement with the methods, there were no significant benefits to be seen. This is not an unusual finding: Klahr and Nigam (2004) found much the same thing when comparing direct instruction and the active approach of discovery learning, and Mayer (2004) has long made similar claims. De Bruyckere et al.(2015) make the more nuanced assertion that problem-based approaches can be useful to extend existing knowledge, but not to acquire it in the first place, and that discovery learning can sometimes be effective, but only with the right guidance and support. Hattie (2013, p. 331) draws from many metastudies to conclude that minimally guided, facilitative approaches such as problem-based, inquiry-based, and project-based learning tend, on average, to be relatively ineffective.

For most teachers who have undergone any kind of teacher training in the past 50 years or so, this tends to come as a surprise.

No One Has Solved the 2 Sigma Problem

Bloom’s (1984) paper on the 2 sigma problem shows, among other things, that one-to-one tutoring leads, on average, to a 2 sigma improvement in performance compared with “conventional” didactic teaching (again according to the measures used). This has since become the gold standard for educational interventions, especially in online learning. No methods identified so far have consistently achieved that gold standard. Given the tens of thousands of papers written every year since Bloom set his challenge, which represents only a small fraction of millions of teaching interventions, this might seem to be a little surprising. Why is one-to-one tutoring so much better than any other method of teaching?

Matching Teaching Style to Learning Style Offers No Significant Benefit

People are different, and (self-evidently) different people learn better in different ways. Common sense suggests that we should therefore teach them differently. Following this intuition, numerous learning style theories postulate that individuals have one among a fixed range of persistent learning styles or preferences and that they will learn better when teaching is designed to fit with their identified style or preference. The boldest theories claim that learning styles are unalterable traits and field independent, whereas others make the weaker assertion that they are potentially changeable states and/or field dependent (Curry, 1983). The meekest theories claim only to identify learner preferences, rather than fixed styles, though this is a much less useful claim that might have little impact on teaching practice because the most preferred ways of learning might not coincide with the most effective ways of learning. In fact, the odds are against it (Clark, 1982). A vast majority of professional teachers believe that learning style theories—often as expressed in their strongest, trait-like, field-independent form—are valid (e.g., Boser, 2019; Dekker et al., 2012). There are many reasons to challenge this belief, such as that these theories cannot all be true, that misapplication can disadvantage those inaccurately diagnosed, that people seem to learn any which way when they have to, and so on. But perhaps the most obvious reason for doubt is that there is virtually no reliable evidence to support any of them, despite countless studies and thousands of papers published on the subject every year for several decades (Coffield et al., 2004; De Bruyckere et al., 2015; Derribo & Howard, 2007; Hattie, 2013; Husmann & O’Loughlin, 2019; Pashler et al., 2008; Riener & Willingham, 2010). If different individuals do learn better in different ways from one another (and this is undoubtedly true), then why is it that, when we adapt our teaching methods to ways that should suit them better, such adaptation does nothing to improve the desired learning outcomes? Is it just that we do not yet have a good theory, or is there some other reason?

Experimental Educational Research Methods Appear Not to Work Very Well

There have been hundreds of thousands of randomized controlled tests (RCTs) and null-hypothesis significance tests (NHSTs) performed in educational settings over many decades that seek to discover or confirm what causes learning. Yet, apart from in some very limited and rigidly proscribed contexts, we have little proof that any generalizable method is much better than any other (Hattie, 2013), and there is little evidence that education has improved significantly in quality over the past several decades. Makel and Plucker (2014) observe that, in top educational journals, only 0.13% of experimental studies replicated earlier studies and that, of those few that successfully replicated the originals, the vast majority involved some overlap in authorship and thus might be subject to similar errors and biases. Why is it so difficult to find proof of things that work? Given the massive investment in research time spent on education, has there been so little obvious improvement in outcomes? Why is it so hard to replicate success?

Explaining the Elephants

The apparently diverse phenomena described in these two chapters are closely related to one another, in fact, and they all stem from similar, closely entwined causes. Much like the parable of the “Blind Men and an Elephant,” each allows a glimpse of one part of a single phenomenon.1

In the chapters that follow, I will offer some unifying explanations of these phenomena and plenty more. These explanations emerge naturally from the nature of technology and especially the nature of the technologies through which we commonly learn. They are among the consequences of how technologies are designed, how they work with (or sometimes against) one another, and above all how we and our technologies are intimately and irreversibly entwined, as essential parts of one another. Before offering my explanations, though, it is necessary to understand the whole elephant. The starting point for this must be the nature of technologies in general and, later, how they can contribute (positively or negatively) to learning. This is the purpose of the next few chapters.