#18: Transfer: High Roads & Low Roads
...and ping-pong balls
Consider the following conundrum (given to participants in a study conducted by Pamela Ansburg and Roger Dominowski):
Throw a Ping-Pong ball so that it will go a short distance, come to a complete stop and then reverse itself. You can’t bounce it against a wall or any other object and you can’t attach anything to it.
Let’s start by examining what we know about Ping-Pong balls. They are used to play table tennis (or Ping-Ping) where opponents use small bats to hit the balls over a low net attached to a table. Our image of the Ping-Pong ball is of it being hit back and forth horizontally. We could try doing one of those flick things where the ball travels a short distance and then returns, but the ball won’t come to a complete stop; it will keep spinning before heading back to us. Throwing it against the wall will violate the rules, as will pushing it up against any other kind of object.
There is a very simple solution, but what’s stopping most of us from reaching it is bound up in our stored knowledge of Ping-Pong balls. Would it make any difference if I swapped the Ping-Pong ball for a basketball? It might help some of you, but not all.
The problem with a Ping-Pong ball is that our experience informs us that it travels on a horizontal trajectory, that is, back and forth from one player to another, but a basketball is more likely to travel vertically, that is, up and down. What if I throw the Ping-Pong ball up into the air? It will travel a short distance vertically, stop, and then drop back down into my hand. For some people, the introduction of the basketball primes them to think of other possibilities.
Problems such as these are often solved through insight, that ‘aha’ moment when the solution seems to pop into your head.
Having the knowledge of something and being able to retrieve that knowledge is an indication of learning, but there’s more to it than that. Learning often involves similarities between different types of information or the acquisition of general rules that can be applied to other problems or situations. This learning outcome is referred to as transfer, or transfer of learning, and it’s without doubt one of the most important aspects of knowledge acquisition. But it’s also one of the most difficult to get right. Some even argue that transfer isn’t even a thing, but here I’m going to assume it is.
What is transfer?
We can define transfer as the ‘productive application of prior learning and experiences in novel situations’ (Hajian, 2019). By ‘novel situations’ we mean a context in which the original material was different to the current application. We might, therefore, learn a formula in a mathematics lesson and then apply it to a physics problem or a research methods class. Or we might take something learned in the classroom and use it within a real world context.
But we might also apply transfer when our goals change or when what we have learned is being assessed differently to the way it was originally. We might, therefore, use what we have learned in a physics class to design a model aircraft or knowledge about human behaviour in one culture and apply it to a different one or the same culture in a different situation.
Transfer is, therefore, a major component of problem solving and your success on the ping-pong ball conundrum will have, at least in part, been dependent on how well you were able to transfer knowledge you already had to the new, novel situation. The theory of transfer has a long history, at least as far as the early days of psychology.
At the turn of the twentieth century, behaviourist Edward Thorndike found that the courses high school students took had very little influence on their problem solving skills. They may have gained knowledge from their studies in mathematics, physics, chemistry and Latin, but were unable to apply this to real-world problems. As a result of his findings, Thorndike became an advocate of a more active type of learning, having concluded that the traditional teacher-led approach to education wasn’t able to prepare young people for the challenges they would ultimately face later.
The high road and the low road
Of course, not all types of transfer are the same. As an experienced car driver I could, no doubt, successfully drive other types of vehicle (such as a van or truck) even though I never have. The task is different, but the habitual experience, such as working the clutch and changing gears, as well as obeying the rules of the road, are very similar.
This type of transfer is what David Perkins and Gavriel Salamon call low road transfer. There are many features of driving a vehicle that are shared with other types of vehicle. I learned to drive in a car with a standard gearbox and eventually became accustomed to the way I would need to step on the clutch every time a shifted from one gear to another, but many years later I acquired a car with automatic gearbox and the transition from one to the other was barely noticeable.
However, if I were to learn computer coding and my proficiency was then assessed by having to write a game program, simply knowing how to code isn’t going to guarantee I’ll successfully complete the task. To develop a working game, I’m going to have to search for connections between what I know and how I need to apply it. This type of transfer is known as high road transfer. High road transfer is more demanding, requiring exploration, discovery, and the flexible adaptation of skills. It, therefore, represents a deeper form of processing, as the skills we need to apply in order to achieve transfer require elaboration.
According to Perkins and Salamon, transfer also results in two broad instructional strategies: hugging and bridging. Hugging is a low road strategy that directly guides and engages the learner in a particular desired performance. Teachers, therefore, might encourage students to revise for exams by completing mock exam papers. We might also prepare people for job interviews by having them engage in a practice interview, rather than providing them with general advice. In this way, the learning task ‘hugs’ the target of assessment - we teach to the test, if you like. On the other hand, bridging encourages intentional abstraction of the general rules by searching for possible connections among our many varied experiences and applying these examples to new situations. For example, a teacher might suggest to students that they adapt an exam strategy based on their previous experience.
This, then, creates an opportunity to analyse and reflect on the strengths and weaknesses of the strategy and create a general strategic plan for future exams. Bridging would suggest that instructional design should include the creation of generic skills that can be used in different situations.
Cognitive theories of transfer generally refer to the application of abstraction and analogy. According to the late Richard Skemp, an important figure in mathematics education, abstraction is an activity by which we become aware of similarities among our experiences (Skemp, 1986 p21).
In a biology class, for example, we may be taught about the pioneering work of Gregor Mendel who, in the nineteenth century, uncovered the genetic laws behind multiple generations of pea plants. Mendel’s discoveries gave rise to the modern discipline of genetics. What we learn from Mendel’s work goes beyond pea plants, so we would need to generate a general law from his findings to use within different contexts. A learner, therefore, needs to be able to understand that what they learn about pea plants can then be applied to other species and the science of genetics more generally.
Mathematics and science progress through the use of these deep conceptual principles. On the surface, they may look dissimilar, but they are defined in terms of the same categories or models.
Initial learning
How well we apply transfer to what we’ve learned depends partly on how well the initial learning has gone and the depth of this learning. There is a threshold to initial learning, I can’t simply read the Wikipedia entry on quantum physics and claim that I know about quantum physics. Chances are, I’ll forget it quite quickly and even if I do manage to retain some of it, I’m unlikely to be able to apply this information to difference scenarios.
Since the growth of the Internet, there has arisen a view in some circles that learning is less necessary than it once was because we can always google it. Doing so certainly has its uses but is no substitute for deep learning. When once we might learn about a new area of interest by focusing intently on a book or completing a course, the tendency today seems to be more geared towards browsing and skimming. This often results in a more shallow type of learning, and shallow learning certainly makes accurate recall less likely, but it would also make the potential transfer of this learning unlikely.
Deep processing
The accumulation of knowledge requires deep elaborative processing and consolidation into networks of previous learning, and for this to take place we need more than a simple quick flick across a Wikipedia entry. This is, of course, before we even begin to think about the accuracy of what we’re reading. What we then end up with is a loose collection of semi-recalled facts, often isolated from other related learning. In addition, inaccuracies will creep in and be consolidated, creating false memories and, unless these errors are corrected re-consolidating processes, it becomes impossible to distinguish errors from accurate accounts of the information. For obvious reasons, this makes transfer unlikely because we’re unable to recognise general rules and similarities that exist between groups of information.