Editor's note: This article is part of a series in Medical Education entitled ‘Dialogue’. Each publication in the series will be a transcription of an e-mail discussion about a current issue in the field held by two scholars who have approached the issue from different perspectives. For further details, see the editorial published in Med Educ 2012;46(9):826–7. In this issue, Douglas P Larsen, Director for Medical Student Education for the Division of Paediatric Neurology, Washington University in St Louis, and Tim Dornan, Professor of Medical Education, Faculty of Health, Medicine and Life Sciences, Maastricht University, discuss the roles that testing, broadly defined, and social interaction play in the acquisition and retention of medical knowledge. In medical education research, we ask a lot of questions about how people learn. However, it seems as if we focus much less attention on what makes that learning durable. In many ways I find this to be just as important a question (if not more so) than that of how the initial knowledge or skills are acquired. Tests have been shown to be one mechanism to promote long-term retention1 (this is known as test-enhanced learning2). This line of research has shown it is the act of retrieval that seems to be key to retention (not just the increased motivation to study that comes with testing).3 When retrieval becomes the focus, testing can be seen much more broadly than as the administration of written questions and scoring of their answers. Really, tests include any activity that promotes retrieval. Tests could include the discussions that occur during hospital rounds; even the actual act of seeing a patient could be considered a test. This idea has led me to think more about the mechanisms for learning from experience. As any doctor can tell you, he or she learned (and retained) much more from seeing patients than from sitting in lecture halls! I know that you have written extensively about experiential learning.4 Why do you think that learning from experience is so durable? Every day we are flooded with experiences, but what differentiates the experiences we learn from and those from which we don't? What role does retrieval play (if any) in learning from experience? Retention of learning is, as you say, a most important issue. You write about retrieval and acquisition of learning. Testing, according to your argument, sits between the two because retrieval under test conditions supports long-term acquisition. I'm reminded of a seminal paper by Anna Sfard about the influence of metaphors on the way we think.5 Your argument sits nicely within her acquisition metaphor, which she contrasts with a participation metaphor. Each, she argues, has its place. As to why experience with real patients is better retained, our own work (albeit qualitative6) also sits within the acquisition metaphor. Medical students were more likely to use the word ‘remember’ than any other word when they gave metacognitively rich accounts of their learning from real patients. Experience helped them link their learning into more durable knowledge networks. I have tended to assume that identifying with patients caused positive emotions, which made their learning durable. But I have two questions for you. How comfortable are you with the notion of knowledge as something that is stored and retrieved like a commodity? And what place do you think emotions have in test-enhanced learning or, come to that, learning on rounds? You're right. The discourse surrounding test-enhanced learning to date does fit squarely in the acquisitionist camp. In many ways the metaphor of acquisition has substantial power in helping us to understand the phenomena observed in experiments about learning and testing. Individual learners do recall more facts (or perform more skills on a checklist) when they are repeatedly tested. Learners also do a better job solving application problems when they learn by repeated testing.7 These results imply that the individual has gained (and retained) more knowledge when compared to other methods of acquisition (e.g. repeated studying). In that sense, the analogy of a commodity that is stored and retrieved by an individual would seem to work to an extent (in the spirit of Sfard's description5). However, I would argue that the concept of a ‘commodity’ and its ‘storage’ is too passive. The process of testing and retrieval is more of an active process of repeated construction and re-construction of knowledge. (Sfard keeps this metaphor in the acquisitionist column,5 but Hager identifies construction as its own metaphor.8) It is important to recognise that tests don't just measure memory, but they actually change memory. For evidence of this effect, we can consider the fact that when errors are repeatedly retrieved on tests, the student actually ‘learns’ these errors9 (see Goff and Roediger10 for a laboratory example of how false memories can be created). Because of this direct effect on memory, testing that requires the construction of an entire network rather than an isolated fact leads to more robust learning and retention (e.g. free-recall tests versus short-answer or fill-in-the-blank tests).11-13 I believe this concept is one of the reasons why retrieval in the context of experiential learning is so durable–in a real-life experience, the learner typically doesn't retrieve a fact in isolation, but retrieves an entire schema of knowledge.14 With regard to emotion, I think that it certainly does play a large role in what we remember. Most of us would say that almost all of our enduring memories have some sort of emotional component (this makes a lot of sense when you think about the physical relationship of the limbic system with the hippocampus and other brain structures heavily involved with the creation of memories). However, the increased retention that occurs with retrieval practice through testing would seem to be independent of the emotional context of learning. In the laboratory studies of test-enhanced learning, the materials did not have much emotional significance (word pairs,3 short scientific texts,15 etc.) and there was little emotional relevance to the tests (the participants were student volunteers in a laboratory and there was no identifiable consequence to the testing). With that said, in real-life educational settings, emotion is present in most testing situations–whether the test be a written test, an oral test on rounds, or a patient interaction. In fact, emotions are likely to be stronger in those situations where the learner interacts with another person. Emotion typically makes events more memorable; therefore, in those settings, emotion may accentuate the effects of retrieval practice (see McConnell and Eva for a full review of emotion and learning16). Because emotions are typically reactions to a social context, they take us back to consideration of participatory metaphors of learning and the view that learning cannot be extricated from its context. I am curious about your feelings in this regard. Do you think that retrieval practice through testing has a place in participatory learning or is it simply a tool of acquisitionists? What you have written about emotions is reassuring. Testing does not seem to enhance learning by causing negative emotions, a topic that McConnell and Eva16 discussed in relation to ‘pimping’. You have also said that, if the learning that results from testing is a commodity, it is a commodity that is enhanced rather than just embedded by testing. Returning to the acquisition and participation metaphors, they represent two different views of competence: something that is replicable from one situation to another (acquisition), or something that is unique to each new situation (participation). Neither is right for all situations. My practice as an internist was founded on the latter premise. It would be irresponsible, in contrast, for a paramedic to premise emergency life support on anything but the former. I see dialogue as central to learning. If a ‘test’ is a dialogue which helps a learner articulate his or her stance towards a situation that is in at least some regards unique, then yes, retrieval practice has a central place in participatory learning. But Sinclair's ethnography of UK medical student education17 found that medical students were often ‘tested’ in a way that had more to do with maintaining a power differential than with fostering learning – like pimping. What we are exploring together touches on something crucial: competency (outcome)-based education. According to the Flexner centenary recommendations, this should underpin medical student and residency education.18 Testing has a central place in it and all the more so, one might say, if it enhances learning. Something I'd appreciate your view on, however, is how much of a doctor's real competence can be enhanced by testing or, indeed, is testable? Testing, moreover, is a dialogue that constructs a position of power for the tester over the tested. Our research shows that workplace learning takes place through a plethora of other types of dialogue and discourse theory indicates that verbalisation is an act of learning.19, 20 Could it be that, for example, the type of mutual co-exploration captured by the term ‘cognitive scaffolding’ could be as educative as a ‘test’? Of course, that takes us away from test-enhanced learning, but maybe it helps us define the boundaries of the phenomenon. As I think about the various ideas you raise, an issue that comes to mind is whether tests are used as learning tools versus whether they are used as assessment tools. Tests most clearly create a power differential when they are used to assess and categorise students. However, when the main purpose of testing is to provide students with a tool for learning, retaining and applying knowledge, the test becomes empowering. When tests are used as learning tools, their objective is to help all learners to retain all of the knowledge or skills that are targeted, rather than to separate out well-performing from poorly performing learners. In order for tests to help with retention, they must be repeated and spaced.21, 22 When a test is used for learning there is no real need for a grade because the objective is for the learner to master all of the material, as much as possible. Ideally, tests used in this way are repeated until that goal is accomplished.23 A single test at the end of a class or the end of a course of study is not likely to produce the retention effects we are talking about. Test-enhanced learning conceptualises tests in a way that is fundamentally different from that in which they are typically used in medical education. In our studies, most students welcome tests used as learning tools.1, 14 In my opinion, tests used in the manner outlined above have the potential to make a much larger contribution to the development of competence than the summative tests used to measure that concept. However, competence is a much larger construct than knowledge or even application of knowledge. Competence includes judgement, compassion, management, communication, etc. I have yet to see a test that I feel truly captures competence. Knowledge and application of knowledge can be accentuated by testing (when implemented as described by test-enhanced learning), but they represent only a piece of the pie. With regard to dialogue and learning, I think that the contrasts you create between pimping and cognitive scaffolding fit in many ways with the framing of assessment and learning tools that I discuss above. Pimping is a ‘dialogue’ meant to assess the student and often leads to humiliation: students ‘fail’ if they are not able to answer the question correctly. Cognitive scaffolding is a conversation or interaction used as a learning tool to guide students through a complex process to help them discover, learn and retain the knowledge or skills that are being explored. Often when I am working with students in a clinical setting, I will explain the ground rules for the questions I ask. I tell them that I will likely ask questions they won't be able to answer and that is okay because I am not trying to make an assessment but, rather, am trying to engage them in a conversation. I explain that my questions are meant to point them in the direction of the thought processes and knowledge that are key to the work we are doing. Interestingly, there is laboratory evidence that asking students questions they are unable to answer does lead to better retention once they do learn the answers to the questions.24 In my mind, it is all about how the dialogue is framed and what the intention is. I consider cognitive scaffolding and testing (as described by test-enhanced learning) as converging education techniques based on similar cognitive principles: retrieval and analysis. Learners retrieve what they already know and seek to expand on it. As I mentioned above, though, repetition and spacing are crucial elements that lead to retention in test-enhanced learning. However, they are not typically part of the discussion in relation to cognitive scaffolding. Do you think they could play a role in the learning that occurs as part of this more participatory method? That is an interesting question. To answer it, let me tug you back from the cognitive into the socio-cultural world. The concept of test-enhanced learning creates two roles: the tester and the person being tested. Cognitive scaffolding creates the role of a person with knowledge, which allows that individual to scaffold, and a person who lacks such knowledge. Of course, relationships between teachers and learners are inherently asymmetrical and, as you have pointed out, experts can use their mastery in kind as well as unkind ways. Nevertheless, those two terms reinforce the novice status of learners. Social learning theory views interaction in a more communal way and recognises that teachers stand to gain from learners, as well as the reverse. Speech is the currency of their social interaction. There is ‘inner speech’, which is the foundation of learning and identity development, as well as outwardly visible and audible social speech. Speech binds the teacher and learner together into the social act of learning. They both story their lives.19 If you think of testing from that perspective, it is a type of discourse, which helps people incorporate the speech of practice into their inner speech. My research with Etienne Wenger and others found some evidence for that. We rediscovered what Wenger and Jean Lave had said: that novices like medical students learn ‘from talk and to talk’25 and testing is one of many ways of fostering talk.20 I can see interesting parallels, then, between the repetition and spacing that are crucial to test-enhanced learning and the engagement of learners into the talk of communities, through which they learn. Whatever one's theoretical stance, keeping learners engaged in conversations that continue over a period of time seems a really important way of fostering durable learning. I don't know if you use the term, but, to my mind, you have added ‘constructive testing’ (This test is helping me learn) to the more familiar concepts of formative (How am I doing?) and summative (Is this learner competent enough to progress?) testing. Van der Vleuten et al.'s concept of programmatic assessment26–regular, low-stakes testing to steer and support learning–seems a very natural bedfellow to test-enhanced learning. I feel more comfortable with your empowering approach to testing than with the judgemental approach from which, in your last contribution, you moved our dialogue away. As we come towards the end of this dialogue, I would appreciate any comments that might allay a concern of mine. I think of the talk of learning communities as being co-exploratory and stimulating divergent thought and creativity. Do you have any concerns that test-enhanced learning might thwart exploration and be convergent and normative? Do you have any practical tips about how, in our teaching practices, we might make best use of it? I have really enjoyed exploring the interface of cognitive and socio-cultural approaches to learning in our exchange. Learning is based on memory, which includes both conscious and non-conscious memories,27 whether that memory is derived from a test or a social interaction. Mechanisms for creating memories are likely to apply to various settings and methods if they are based on the ways in which our brains learn. Therefore, if retrieval, spacing and repetition are important in creating memories of facts and ideas learned through testing, those same principles may play important roles in creating memories through other means, such as conversations and reflections. At least this may represent a fruitful area to investigate further–I don't know that learning through participation has been thoroughly investigated with that particular frame of reference. For instance, I don't think that reflection has ever been investigated as a form of retrieval practice. From my personal experience, though, I would agree with your assessment that sustained engagement, whatever the method, is key to durable learning. I like your term ‘constructive testing’. I think it gets at the idea that we are building knowledge through retrieval practice, not simply measuring it. The term also emphasises the fact that we cannot simply use our existing testing programmes and re-label them as test-enhanced learning. Educators must plan testing in ways that are likely to actually promote retention for it to really be test-enhanced learning–or constructive testing–using the principles of repetition, spacing and feedback.28 With all of its strengths, though, test-enhanced learning does not encompass every aspect of what we are trying to achieve in medical education. In many ways, I see the role of test-enhanced learning as that of a tool through which an individual can learn the basic building blocks of medical practice. I have found the techniques of test-enhanced learning useful in helping students learn complex sets of facts such as anti-seizure medications, developmental milestones, and the elements of the neurological examination, to name a few examples. I see these efforts as representing the learning of the alphabet and vocabulary of medicine. Once we are able to speak a common language, we are able to explore the rich nuances and variations that come with practice within the community of medicine. The presence of common principles and facts does not force conformity and convergence any more than the common language of William Shakespeare and Mark Twain created uniformity between the two authors. The building blocks form the starting point. Having a common language empowers participation in the conversation and exploration of common experiences. Even though his terminology may be antiquated, I think that John Dewey captured this dynamic when he said: ‘In the recitation the teacher comes into his closest contact with his pupil… [T]he recitation is a place and time for stimulating and directing reflection, and that reproducing memorised matter is only an incident–even though an indispensible incident–in the process of cultivating a thoughtful attitude.'29 In reference to your question about the application of test-enhanced learning, the most practical use of this technique that I see is for educators to use it to plan for retention. So often retention is taken for granted until, when learners struggle, we realise we had assumed that initial learning was sufficient. When educators identify information which they want students to be able to remember and use over long periods of time, they should plan retrieval practice through written, verbal or activity-based (e.g. simulation or actual clinical encounters) methods. While we typically think of test-enhanced learning in classroom settings, as medical educators we need to seek out and create opportunities for retrieval practice in the context of real-life experiences. We have identified many advantages of sharing perspectives on learning. What the educator in me takes from our dialogue is the importance of questioning. You have written about how patients, by the problems they represent, can ‘test’. I have emphasised the importance of engaging learners in conversation, and you have argued from theory and empirical evidence that those conversations should include questions. So, we should help learners identify ways in which patients test them. We can also test learners directly because that will prepare them to list, for example, alternative anticonvulsants to the one a patient is currently taking. I see some dangers in the common ground we have identified. My emphasis on constructive conversation might blur your clear and simple message that we should ask testing questions–and vice versa. I am concerned that testing should not be too reductionist. We both insist, however, that the goal of questioning should not be to ‘show up’ learners’ lack of knowledge in order to humiliate them. If teachers can be really confident that their testing is truly positive for learners, then I think experiential learning and test-enhanced learning complement one another rather nicely. This paper is a transcript of an original e-mail correspondence between DPL and TD. none. none. not applicable.