Adv in Health Sci Educ DOI 10.1007/s10459-014-9574-9
The mediating effect of context variation in mixed practice for transfer of basic science Kulamakan Kulasegaram • Cynthia Min • Elizabeth Howey • Alan Neville • Nicole Woods • Kelly Dore • Geoffrey Norman
Received: 25 April 2014 / Accepted: 22 November 2014 Springer Science+Business Media Dordrecht 2014
Abstract Applying a previously learned concept to a novel problem is an important but difficult process called transfer. Practicing multiple concepts together (mixed practice mode) has been shown superior to practicing concepts separately (blocked practice mode) for transfer. This study examined the effect of single and multiple practice contexts for both mixed and blocked practice modalities on transfer performance. We looked at performance on near transfer (familiar contexts) cases and far transfer (unfamiliar contexts) cases. First year psychology students (n = 42) learned three physiological concepts in a 2 9 2 factorial study (one or two practice contexts and blocked or mixed practice). Each concept was practiced with two clinical cases; practice context was defined as the number of organ systems used (one system per concept vs. two systems). In blocked practice, two practice cases followed each concept; in mixed practice, students learned all concepts before seeing six practice cases. Transfer testing consisted of correctly classifying and explaining 15 clinical cases involving near and far transfer. The outcome was ratings of
K. Kulasegaram (&) Department of Family & Community Medicine & The Wilson Centre, University of Toronto, 200 Elizabeth St 1ES-565, Toronto, ON M5G 2C4, Canada e-mail: [email protected] C. Min Centre for Health Education Scholarship, University of British Columbia, Vancouver, BC, Canada E. Howey K. Dore G. Norman Programme for Education Research and Development, Faculty of Health Sciences, McMaster University, Hamilton, ON, Canada A. Neville Department of Oncology, Faculty of Health Sciences, McMaster University, Hamilton, ON, Canada N. Woods Department of Surgery & The Wilson Centre, Faculty of Medicine, University of Toronto, Toronto, ON, Canada
123
K. Kulasegaram et al.
quality of explanations on a 0–3 scale. The repeated measures analysis showed a significant near versus far by organ system interaction [F(1,38) = 3.4, p \ 0.002] with practice with a single context showing lower far transfer scores than near transfer [0.58 (0.37)–0.83 (0.37)] compared to the two contexts which had similar far and near transfer scores [1.19 (0.50)–1.01 (0.38)]. Practicing with two organ contexts had a significant benefit for far transfer regardless of mixed or blocked practice; the single context mixed practice group had the lowest far transfer performance; this was a large effect size (Cohen’s d = 0.81). Using only one practice context during practice significantly lowers performance even with the usually superior mixed practice mode. Novices should be exposed to multiple contexts and mixed practice to facilitate transfer. Keywords design
Basic science Transfer Cognition Teaching strategies Instructional
Introduction A central role of preclinical medical education is to provide students with the baseline knowledge they will use to manage clinical issues during clinical training. However, there is an abundance of evidence from other domains that even if students possess the knowledge needed to solve a problem, they may have difficulty accessing that knowledge to solve a new, unfamiliar version of the problem (Ross 1987). This phenomenon, accessing old knowledge to resolve new problems, is called transfer. In undergraduate medical training, the issue of transfer is particularly serious as many different domains of knowledge such as basic sciences are taught with the expectation of future utility for problem solving. Several studies have established (Goldszmidt et al. 2012; Baghdady et al. 2009, 2013; Woods 2007; Woods et al. 2005, 2006, 2007a) that use of basic sciences knowledge enables students to more accurately and efficiently grasp clinical skills such as diagnosis and interpreting examination findings. However, the usefulness of basic science is not simply dependent on retention (Baghdady et al. 2009). Learners must recognize that their previous knowledge from basic science can be used to understand the clinical skill or task they are learning, and must be able to retrieve the particular knowledge they require to understand the clinical situation. In other words, the transfer of basic science knowledge to the clinical task is essential if learners are to derive benefit from basic science teaching. Over a century of research supports the view that transfer is a difficult cognitive process for learners (Eva et al. 1998). This is not a novel problem in medical education, but rather a near universal finding in many domains (Salomon and Perkins 1989). Despite this, curriculum planners and instructors are rarely able to identify transfer as an explicit curriculum goal and devise strategies to facilitate it (Laksov et al. 2008). Developing strategies to teach learners to transfer concepts requires an understanding of why students find spontaneous transfer difficult. One of the most significant reasons for failure to transfer is the crucial impact of context specificity (Eva et al. 1998). For example, illustrating the application of fluid flow dynamics (the concept) with an example from the respiratory system (e.g. asthma or bronchitis) may enable learners to solve similar problems arising in the respiratory system. Fluid flow dynamics is relevant to many body systems, but when learners are presented with novel problems situated in a new organ system context (e.g. cardiovascular), evidence suggests that learners fail to recognize the usefulness of the previous concept in the new context (Kulasegaram et al. 2012). Problems for which the context of learning and that of transfer have similar, familiar, surface details
123
Effect of context variation on mixed practice
are near transfer (e.g. respiratory to respiratory) and are invariably easier than far transfer, which occurs when the surface details are different between the context of learning and the context of transfer (e.g. transfer from respiratory to cardiovascular). Transfer theory suggests that learners tend to classify and retrieve concepts based on the superficial or surface level similarity based on their evaluation of the context of the problem (e.g. organ system or case setting) instead of conceptual similarity (Gick and Holyoak 1980; Holyoak and Koh 1987; Ross and Kennedy 1990). Our previous work (Kulasegaram et al. 2012) has shown that use of contextual familiarity to cue recognition of concepts is predictive of transfer success for near transfer but is inaccurate for far transfer. Moving learners from over-reliance on superficial contextual similarity and appropriate integration of contextual information is a focus of many transfer interventions. Context variation The recognition that the context of learning is meaningful to learners has led to practice strategies that have focused on manipulating the relationship between conceptual knowledge and the context in which concepts are practiced (Kaminski et al. 2013, 2009). One approach is structuring practice for context variation, i.e. multiple examples with dissimilar surface details (Ross and Kennedy 1990). To illustrate, simultaneously practicing the application of a basic science concept such as fluid dynamics in the cardiovascular and respiratory organ system is practice with contextual variation. Viewing conceptual knowledge in multiple contexts forces learners to discount superficial features and focus on conserved elements of the concept present in both examples. Practice in multiple contexts has shown dramatic improvement in transfer performance for novices and is a common strategy in medical education (Norman 2009). Mixed practice Another innovation has been to redesign practice to promote mixed or interleaved practice. Traditional practice involves learning a concept followed by practice with multiple examples of that concept. This strategy is known as blocked practice. Mixed practices involves practicing examples of multiple concepts together. Originally developed and studied in motor skills learning (Lee et al. 1985; Magill and Hall 1990) mixed practice has been extensively investigated for concept and procedural skills learning (Helsdingen et al. 2010; Roher and Taylor 2007; Hatala 2011). Head to head studies of transfer performance for mixed and blocked practice have consistently demonstrated an advantage for mixed practice (Roher and Taylor 2007). The reason for the advantage of mixed practice is that in traditional blocked practice, learners do not have to identify the type of concept to complete the practice problems; it is already known that the concept is the one from the preceding lesson (Rohrer and Pashler 2010). In contrast, mixed practice requires that learners must actively learn to distinguish which concept correctly applies to each problem. As Roher and Taylor (2007) notes: whereas blocked practice requires students to know how to perform a procedure, it does not require them to know which procedure is appropriate. Evidence for the benefit of mixed practice has been demonstrated for clinical learning. Hatala et al. (2003) studied diagnosis of ECGs by novices who were asked to learn to interpret ECGS of three different types of disorders and then given practice in either blocked or mixed fashion. Transfer testing showed participants in the mixed condition successfully diagnosed 46 % cases while in the blocked condition accuracy was 30 %.
123
K. Kulasegaram et al.
Contextual variation in optimizing mixed practice While mixed practice exposes learners to different conceptual structures, context variation exposes learners to the variability of how a single concept may present during future problem solving. Both strategies emphasize variability but at different levels in order for learners to develop the appropriate recognition of conceptual knowledge necessary for future transfer and problem solving. As such, it is possible that these interventions may have similar mechanisms of action and adopting one or the other maybe sufficient to support concept transfer by learners. Given the increasing evidence for the benefits of mixed practice, simply reorganizing existing basic science practice material may be sufficient to promote transfer. On the other hand, novice learners may still intuitively make close associations with contextual surface features if practice does present variations of context and surface features for basic science concepts. To our knowledge, few studies have examined context variation and its role in impacting of mixed practice of conceptual knowledge. The studies to do so, (Simon 2008) have not examined transfer but focused on outcomes more closely related to recall or memory for taught concepts. Furthermore, these studies did not examine the impact on near and far transfer. The central question of this study was whether the efficacy of mixed practice of multiple concepts when it occurs with and without context variation on near and far transfer of basic science concepts. We hypothesized that context variation is essential for the advantages of mixed practice and that without it, mixed practice may lose its efficacy. Specifically, if context variation plays a role in mediating the effects of practice strategy, the effect of multiple contexts should be to encourage participants to discount context as a cue to categorization, and focus on the underlying concept. Consequently, we expected to see a significant reduction in the classic difference between near and far transfer performance for participants who practiced with context variation regardless of practice strategy. On the other hand, if context variation did not play a role, we expected to see a reduction in the difference between near and far transfer for participants with mixed practice compared to blocked practice regardless of context variation.
Methods Participants This study was conducted with forty-four (N = 44) first year undergraduate students in the Department of Psychology at McMaster University during 2010–2011. Participants received either $10.00 in compensation for participation or course credit. The study received institutional ethics approval from the McMaster University Research Ethics Board. Procedures, materials, and other aspects of the design are similar to those used in previous studies (Kulasegaram et al. 2012; Norman et al. 2007). Materials We used materials previously used for previous transfer studies (see Norman et al. 2007; Kulasegaram et al. 2012). Participants in this study were asked to read explanations written by an expert clinician (AN) for the three physiology related principles.: Fluid dynamics, Laplace’s law, and Starling’s law. Fluid dynamics illustrates the principles of laminar and turbulent flow, which has application in multiple organ systems including the respiratory, cardiovascular, gastrointestinal, and genitourinary tract. Laplace’s law describes
123
Effect of context variation on mixed practice Table 1 Example of explanation: Laplace’s law and teaching analogy Analogy supplemented clinical explanation Laplace’s law applies in many situations, particularly in the gastrointestinal tract. The intestines are tubes with a radius, intestinal wall tension, and a pressure exerted by the tube’s contents: digesting food. The tension in the intestinal wall is proportional to the radius of the intestine and the pressure of the contents. This relationship can be explained by: T = PR where T is the wall tension, P is the pressure, and R is the radius. Thus, having a small radius decreases the tension in the wall for the same internal pressure. Laplace’s law explains why in normal circumstances intestines don’t rupture their contents and why in some conditions, serious complications can follow dilatation or an increase in the radius of the bowel, as occurs in inflammatory conditions of the bowel A diagrammatic way of explaining this relationship of wall vessel tension and the pressures across the vessel wall is shown in the left diagram. In order to maintain the same downward force (the pressure), the wall tension T must increase as the wall becomes less curved. This is analogous to a weight hanging on a string as in the diagram at the right. When a weight is suspended from a string, the tension in the two ends of the string increases as the string becomes closer and closer to horizontal, because the vertical component of the tension must equal the weight
relationship between tension in the wall of a cylindrical vessel and the radius of the vessel and pressure across the wall; it applies to the same organ systems as the principles of laminar and turbulent flow. Starling’s law describes the elastic behavior of the heart in response to filling of the ventricles and volume of the ejected blood. Because it only applies to the heart, Starling’s law was used to increase the number of concepts participants had to learn and was tested only in near transfer cases. All explanations included a detailed outline of the concept, its relevance to physiology and a brief example of its application, and emphasized that the concept applied in multiple organ systems (except for Starling’s law). Each explanation was also augmented by the provision of a teaching analogy, which illustrated the basic principles of the concept (Kulasegaram et al. 2012). For example, the relationship of tension, pressure, and vessel radius described by Laplace’s law was illustrated using the analogy of a weight suspended in the middle of a string. The increase in vessel wall tension due to increasing the radius of a cylindrical vessel is akin to the increase in the tension of the string as the ends of the string are pulled apart. We have previously used teaching analogies to study transfer performance (Kulasegaram et al. 2012; Norman et al. 2007) and all of the participants in the study were given teaching analogies for each concept (see Table 1). Practice for each concept consisted of two clinical vignettes written by an expert clinician (AN). Participants were asked to provide an explanation of how the concept related
123
K. Kulasegaram et al. Table 2 Example of Laplace’s law’s practice vignette and feedback The patient complains of increasing abdominal pain. On examination, his abdomen is distended and an abdominal X-ray shows a dilated colon. The patient complains of being unable to move his bowels for the past four days and of previously severe pain the area. The doctor is concerned that the patient might suffer a bowel perforation and prescribes anti-inflammatory drugs. He recommends surgery to constrict the size of the stomach. Explain why these measures may help and what the risk is to the patient? The dilation of the colon is an increase in the radius of the gastric tube. While the pressure of the contents may stay the same, the increase in the radius will increase the tension of the colon (T = PR). If the tension continues to increase the colon’s walls might actually rupture from the excess stress. While this is a rare possibility, it is severe. Similarly, the increased size of the bowel reduces the ability of the muscles to generate enough force to push through the contents. Reducing the radius or size should eliminate this difficulty
to the signs and symptoms—in other words, how the concept applied to the problem. They were then provided with the correct answer. See Table 2 for an example. Design In a 2 9 2 design, students were randomized to learn and practice three physiology concepts with either mixed or blocked practice as well as practice with either one or two organ system contexts for two of the concepts (see Fig. 1). Manipulation for mixed and blocked practice In the blocked practice conditions, practice cases followed each explanation so that a participant starting with Laplace’s law would read the explanation for the concept and would then complete the two practice cases before moving onto the next concept. Practice cases in blocked conditions required students to explain how the concept explained the features of the case; participants were then given feedback in the form of the correct answer as written by the expert clinician. In the mixed practice condition, participants read all three explanations prior to any practice; in the practice phase, participants were presented with six practice cases in random order and told to identify which law applied and how it explained the features. As in the blocked group, participants were then given the correct answer for each case.
Tesng
Learning One organ system
Mixed Pracce Two organ systems
Blocked Pracce
One organ system
Two organ systems
Fig. 1 Summary of manipulation and general design
123
10 T/F
6 MCQ, 4 short answer
15 Transfer cases (near and far)
Effect of context variation on mixed practice
Manipulation of context variation Practice contexts were manipulated using organ systems involved in the vignettes depicted in the practice cases, which could involve the same or different organ systems. In the single practice context condition, fluid dynamics cases involved only respiratory disorders and Laplace’s law cases involved only gastrointestinal disorders. In the multiple practice context condition, fluid dynamics had one respiratory and one cardiovascular case while Laplace’s law was practiced with a case from the gastrointestinal and one from the cardiovascular system. Starling’s law only applied to the cardiovascular system and was included to increase the difficulty of the learning and transfer tasks. Procedure This study was conducted in two phases, learning–practice and testing, which occurred in the same session. Phase 1: Learning and practice In phase 1, participants were exposed to the principles and their assigned practice condition. Order of presentation of principles and thus practice of principles was randomized. After viewing the explanations and completing practice with feedback, all participants were asked to complete a ten question True/False quiz on the concepts. Participants were required to obtain a minimum score of 8/10 to move on to a knowledge test or were asked to repeat the quiz before moving onto phase 2: Testing. To avoid test-enhanced learning, participants were not given feedback. Phase 2: Testing Recall and knowledge test Phase 2 began with a knowledge test of the concepts in the form of six multiple choice and four short answer questions for a total score out of 10. Questions directly elicited information presented about the concepts in the explanations. Participants only completed the test once and were not given feedback. Transfer testing All participants were presented with 15 transfer cases in the form of clinical vignettes with similar structure to those in the practice session. Participants were asked to identify the concept that best applied to the vignette, then provide an explanation for how the concept accounted for the clinical signs and symptoms. Responses were scored on a 0–3 scale with one point awarded for correctly identifying the concept involved and additional points for accuracy and depth of explanation. This scoring system was used in previous transfer studies (Norman et al. 2007; Kulasegaram et al. 2012). A sub-sample of cases was scored by two raters to determine inter-rater reliability. See Table 3 for an example of the test cases and scoring. To determine if differences between groups were due increases in correctly recognizing concepts or increasing depth of explanations, we created a secondary measure on the proportion of cases correctly classified by concept. As in our previous study (Kulasegaram et al. 2012) this allowed us to examine if gains in transfer were purely due to participants developing more complex representations of the concepts as expressed through better
123
K. Kulasegaram et al. Table 3 Example of Laplace’s law testing case, response and appropriate scoring A patient complained of vomiting copious amounts at the end of each day. Investigations revealed he had a lax, poorly contracting stomach. He was given a drug that constricted the size of the stomach, resulting in more efficient emptying. Explain why constricting the size impacts stomach emptying Assigned score
Response
Response
Rationale
0
Goethe/ starling
1
Laplace
Not provided or incorrect
Correct law but no explanation or wrong explanation
2
Laplace
This can be due to a decreased radius of the stomach. The drug must have dilated the pathway of the stomach, allowing the food to stay in the stomach for further digestion
Correct law but superficial explanation
3
Laplace
The drug might have decreased tension within the stomach walls. From the knowledge of Laplace’s law, we know that as the tension decreases, so does the pressure. The drop in the pressure from the outside walls, stomach’s radius increases and thus, resulting in more efficient space
Correct response, response captures the deep structure of Laplace’s law
Wrong response
scoring on the explanations or if the mechanism of the interventions worked at the purely better recognition of concepts. To create near and far transfer cases, vignettes involved disorders of organ systems that were familiar (near transfer) and unfamiliar (far transfer) for participants. Near transfer for all participants in the study were Flow dynamics cases involving the respiratory organ system, Laplace’s law cases involving the gastrointestinal system, and Starling’s law cases involving the cardiovascular organ system. Flow dynamics and Laplace’s law cases involving the cardiovascular organ system were considered near transfer for half of the participants who saw this organ system and concept pairing in practice as part of the multiple contexts of practice manipulation. These cases were far transfer for the half that did not. Flow dynamics cases involving the urinary tract and Laplace’s law involving the reproductive system were consider common far transfer for all participants. Outcomes and analysis To control for time on task and as a proxy for effort, we measured time spent during the learning phase and analyzed average time spent per case using 2 9 2 9 2 repeated measures ANOVA with average Near and Far transfer case time as the repeated measures factor, practice strategy (mixed vs. blocked) and practice contexts (one vs. two) as the between subject factors. The outcome for the knowledge test was a total score out of 10; a 2 9 2 ANOVA was conducted on the knowledge test with practice strategy and number of practice contexts as the between subjects factors. The main analysis of the hypothesis involved outcomes on the transfer test analyzed using average score on case (0–3) and using 2 9 2 9 2 repeated measures ANOVA with average Near and Far transfer performance as the repeated measures factor, practice strategy (mixed vs. blocked) and practice contexts (one vs. two) as the between subject
123
Effect of context variation on mixed practice
factors. The secondary analysis was of the recognition or classification score (percent of cases correctly classified) using a 2 9 2 9 2 repeated measures ANOVA with the same design as the main analysis. We determined a priori to use the knowledge test and time spent per case as covariates in the analysis if there were significantly associated with the outcome.
Results Forty-two students completed the study. One participant did not complete the test phase and another randomized to the Blocked practice with two organ systems scored 0 in all transfer cases; these were excluded from analyses for a final sample size of 11 per group in the mixed with two and blocked with one group and ten per group for the mixed with one and blocked with two. Inter-rater reliability for rating of explanations was high (ICC = 0.85). Effect of time on task and time per case Participants on average took 2,378 s on the learning phase (SD = 202 s); participants in the mixed condition took longer [2,734 s (190 s)] than in the blocked condition [2,023 s (134 s)] and this difference was significant [F(1,38) = 5.7, p \ 0.02, Cohen‘s d = 4.3]. Time spent during learning and practice did not correlate with scores on knowledge test or on transfer cases (r = 023, p \ 0.35). Average time per far transfer case was 152.3 s (61 s) and for near transfer case was 92.2 s (54). Analysis of time per case showed no significant differences due to practice condition, organ system but a main effect of far transfer time being greater than near transfer time [F(1,38) = 15.8, p \ 0.0001, d = 1.03]. There was no significant correlation between average case time and performance for near transfer (r = 0.210, p \ 0.18) or for far transfer (r = -0.19, p \ 0.11). Performance on knowledge test Scores on the multiple choice and short answer knowledge tests are shown in Table 4. Analysis showed no effect of practice strategy [F(1,38) = 0.993, p \ 0.35] or number of organ systems during practice [F(1,38) = 2.3, p = 0.21] as well as no significant interaction [F(1,38) = 2.1, p \ 0.19]. Scores on knowledge testing did not correlate significantly with average performance with near transfer cases (r = 0.19, p \ 0.43) and far transfer cases (r = 0.33, p \ 0.18) separately and thus were not included as covariates in the analysis of transfer performance. Performance at transfer testing The repeated measures analysis showed a significant near versus far by organ system interaction [F(1,38) = 3.4, p \ 0.002, g2p = 0.16] with individuals practicing with a single organ system showing lower average far transfer scores than near transfer; mean (SD) for far 0.58 (0.37) versus 0.83 (0.37) for near (d = 0.68). The two organ systems of practice group had similar far and near transfer scores [1.19 (0.50) for far vs. 1.01 (0.38) for near; d = 0.4]. Post-hoc testing using the Least-Significant-Differences method showed no significant differences between the groups on near transfer performance (see Fig. 2) but a significant main effect for two organ systems over one on far transfer (see Fig. 3)
123
K. Kulasegaram et al. Table 4 Average total score on multiple choice and short answer knowledge test by condition
Practice type Blocked Mixed
No of organ systems
Mean (SD)
1
7.50 (0.32)
2
8.10 (0.45)
1
7.30 (0.44)
2
7.66 (0.56)
[F(1,38) = 20.3, p \ 0.0001, d = 0.81]. There was no significant main effect of practice strategy [F(1,38) = 2.4, p \ 0.128, d = 0.1] or for the practice by organ system by near versus far interaction [F(1,38) = 0.07, p \ 0.78, g2p = 0.03]. Overall, the participants who completed mixed practice with two organ systems had the highest performance for near [1.16 (0.30)] and far transfer [1.24 (0.41)]. Blocked practice participants with two organ systems had similar far transfer scores at mean (SD) of 1.14 (0.32) and outperformed participants in mixed practice and blocked practice with a single organ system as 0.44 (0.51) and 0.73 (0.41) respectively. Classification of transfer cases On average participants classified more near transfer cases correctly at [mean (SD)] 65.35 % (19 %) compared to far transfer cases [55.1 % (22 %); F(1,38) = 7.68, p \ 0.01, d = 0.5]. There was a significant interaction with organ system [F(1,38) = 19.14, p \ 0.0001, g2p = 0.3] but no interaction with practice condition [F(1,38) = 1.5, p \ 0.22, g2p = 0.02]. On average, participants in the two organ systems condition classified 60 % (15 %) of near and 65 % (18 %) of far transfer cases correctly. Participants in the one organ system group classified 70 % (20 %) of near transfer cases but only 43 % (19 %) of far transfer cases. The practice by organ system interaction was marginally non-significant (F1,38) = 3.87, p \ 0.06, g2p = 0.11. The mixed practice with two organ systems group had similar rates of success at classifying near and far transfer cases at 66 % (13 %) versus
Fig. 2 Average score on near transfer cases by condition
123
Effect of context variation on mixed practice
Fig. 3 Average score on far transfer cases by condition
70 % (19 %). The mixed practice with one organ system group classified 72 % (23 %) of near transfer cases correctly but only 37 % (22 %) of far transfer cases.
Discussion This study examined the impact of varying the number of practice contexts and mixed strategies for transfer of concept knowledge. We found that there was a significant interaction between the number of practice contexts participants were exposed to and the difference between near and far transfer. This suggested that increasing the number of practice contexts in both practice schedules reduced reliance on contextual information. Our results showed that mixed practice performance on far transfer was dependent on the number of practice contexts used during learning thus supporting our initial hypothesis. As the number of practice contexts increased, so did far transfer performance in both practice strategies. While mixed practice with multiple organ systems had the highest overall transfer performance, mixed practice with a single organ system per concept had the lowest far transfer performance. This suggests that without contextual as well as conceptual variability, mixed practice may not always be successful at promoting far transfer. Analysis of accurate recognition or classification indicates the advantage of practice with multiple contexts results from improvement in the ability to recognize concepts in new contexts not with increasing depth of explanation. This is reflected by scores on quality of explanations showed participants in all groups to still have low performance indicating both the difficulty of transfer and lack of strong conceptual understanding. There are several limitations to this study. Studies of mixed and blocked practice typically find an increase in performance for mixed practice with delayed testing (Helsdingen et al. 2010). Thus, the absence of a delayed test limits the inferences from this study to everyday educational practice. However, the purpose of this specific study was to evaluate mechanisms that mediate the impact of practice strategies on transfer
123
K. Kulasegaram et al.
performance. We were not interested in retention of knowledge and thus extrapolation to applied settings maybe limited. Still, comparisons with other studies using these materials and including delayed testing indicated that differences between groups remain robust after delay though there is also significant forgetting (Kulasegaram et al. 2012). A second limitation is that the sample used in this study consisted of psychology students generally naı¨ve to the physiology and clinical conditions that were used in testing. Previous exposure or content knowledge can negate differences and obscure the mechanisms involved in mediating transfer. In our local context, available health sciences trainees are typically knowledgeable in the concepts used to study transfer in this study. Interpretations of these results should once again focus on the mechanisms that impact the efficacy of mixed practice in more complex settings and with more advanced learners. While the sample size is also small, the large effect sizes suggest we were adequately powered to detect differences. A third limitation is that we did not control for prior knowledge using a screening test. Pre-testing could have been a source of bias in this study due to the phenomenon of test-enhanced learning and thus was part of the design. Instead, we chose to control for knowledge gained by students in phase 1. While these limitations as a whole limited the external validity of the results to applied education practice, they do not necessarily impact interpretations of the mechanisms of transfer of conceptual learning in basic sciences. Other studies using these materials and protocol (Norman et al. 2007; Kulasegaram et al. 2012) have investigated transfer mechanisms under similar circumstances. As in those studies, participants had very low performance on our primary outcome. In fact, the highest mean score for any group was still less than half of the possible maximum (1.5/3). While recognition scores were higher, the overall rating of the transfer case explanations suggest some limitations of the design and intervention. In general, transfer is difficult for novice learners and alleviation of this difficulty or building of sophisticated responses to clinical vignettes is unlikely to occur through a brief one-shot learning intervention. Indeed, given the difficulty of the task for the sample, it is remarkable we could detect differences between groups. Furthermore, while raw scores were low, the effects size of differences between comparison groups were moderate to large. Still, a larger study with delayed testing is necessary to address the external validity of the results and the direct applicability to classroom educational practice. There are several explanations for the overall benefits of mixed practice. Generally, these explanations argue that mixed practice poses additional processing requirements for learners in that they must not only process the solution to the practice problems but must also attempt to distinguish the type of problem (Roher and Taylor 2007). For example, this extra processing activity has been characterized as desirable difficulty (Schmidt and Bjork 1992) and is thought to induce task-specific processing for solving transfer problems. A more instrumental explanation is provided by cognitive load theory and its application to instructional design. Cognitive load theory posits that learning tasks draw upon limited mental resources such as working memory and that learning is maximized when this load is optimized to the types of processing necessary for building conceptual schemas for problem solving (Paas and van Merrie¨nboer 1994). In this regard, while extraneous load or distracting processing should be reduced, processing relevant to retention and transfer should be increased according to the learner‘s needs (Sweller 1998). The effects of mixed practice and contextual variation or interference, and more generally, variability during learning can be understood in this framework. Variability increases the mental effort devoted to understanding the differences between concepts (germane load), thus leading to more robust representations of the concept or category (van Merrie¨nboer et al. 2006). Contextual inference in the form of contextual variation (i.e. multiple organ systems for a
123
Effect of context variation on mixed practice
concept) can teach learners about relevant diagnostic information by thinking about how concepts generalize across contexts (van). In this study, the mixed practice with multiple organ systems had two levels of variation that increase learner’s germane load: concept and context. Learners must distinguish between the representations of the concept within each organ system as well as learn that contextual information is an imperfect guide to concept identification. This group thus had the greatest difficulty during practice. The additional processing or germane load—distinguishing structural and surface features during practice—likely forced them to recognize and use structural features during problem solution. This in turn reduces reliance on surface features and possibly decreases the intrinsic load of future problem solving. Accordingly, this group had the highest near and far transfer performance and the difference between near and far transfer performance was lowest for this group. On the other hand, mixed practice with one organ system has the benefit of task variability without contextual interference. With the constraining of one organ system to one concept, using organ system as a cue for identifying concepts is cognitively less demanding activity. While this approach minimizes overall cognitive load, it also decreases the germane processing necessary for successful future transfer and reinforces use of surface level organ system cues as diagnostic of underlying concepts. A complementary, and more fundamental account of the effects seen here can be found in the transfer appropriate processing framework (TAP; Morris et al. 1977; Needham and Begg 1991). Essentially, TAP suggests that transfer occurs when there is congruence in mental processing at learning and at transfer. Many transfer teaching strategies emphasize the importance of fostering increasing abstraction of concepts as the transfer appropriate process when concepts apply in multiple contexts (Holyoak 1989; Lowenstein et al. 2003; Norman 2009). In this study, processing contextual information could have been accurate method for classifying near transfer problems. But, with problems where concepts are not neatly constrained to solely one context, reliance on contextual cues can be misleading. Processing that reinforces reliance on contextual cues for classification does not align with the processing required for transfer in this case. Previous work suggests that novices tend to naively use contextual information in problem-solving unless taught otherwise (Salomon and Perkins 1989). Learners do not create automatically abstract representations or schemas; instead, associations arise with the example used to teach the concept. It is the example that is called to mind when recalling the concept as it is easy and salient to access (Reeves and Weisberg 1994). Use of a single context in practice reinforces the confounding of concepts and contexts; presenting concepts in multiple contexts serves to deemphasize the importance of the context in recognizing concepts. Examining the recognition scores, it is apparent that the mixed with one organ system group used contextual similarity effectively for near transfer problems but could not use this strategy for far transfer problems with accuracy as context was not diagnostic of concept. Transfer appropriate processing to solve these problems requires using the abstraction of the conceptual features and minimizing reliance on contextual cues. The mixed with two organ systems group displays some features of this transfer appropriate processing as evidenced by the reduced gap between near and far transfer. The germane processing induced by variability in both concepts and contexts likely helped foster processing of conceptual structure and thus led to reduced reliance on contextual cues by participants in the multiple organ systems group. There are other benefits for combining conceptual and contextual variability which were not explored in this study. For example, in motor skills training (Magill and Hall 1990), contextual variation is integrated into mixed practice and is the defining feature of creating practice schedules that support transfer. This contextual interference has the impact of
123
K. Kulasegaram et al.
giving learners a range of possible variations of a task or concept and practice with adapting knowledge to multiple contexts. Mixed practice with multiple contexts thus furthers learning by reinforcing adaptation of previous experiences to new contexts. In general, our results reinforce the continuing difficulty of transfer and support the need for extended practice in multiple contexts and focus on teaching deep structure. Most transfer interventions (Norman 2009) focus on enhancing the abstraction of concepts—for example using teaching analogies, active comparison, mixed practice etc. Exposure to multiple contexts during practice forces learners to rely on more abstract, conserved features of concepts and possibly aids their ability to adapt knowledge to new contexts. A judicious mix of deep structure teaching aids that allow abstraction and practice with multiple contexts can push learners to more readily extract the essential features of concepts. This will likely prepare them to recognize these features in future transfer problems and ease the recall and application of that knowledge. Teaching of basic science concepts in undergraduate medical training should find ways to supplement learning along these axes.
Conclusion This study demonstrated that the effects of mixed practice for transfer of conceptual knowledge were dependent on the number of practice contexts. More generally, transfer performance is affected by contextual knowledge and ignoring the effect of context specificity can negate teaching interventions. Mixed practice as a tool to promote transfer should include as many forms of variability as appropriate for learners to appreciate the extent to which concepts generalize across problems. Teaching for transfer of basic science should consider that practice of concepts should expose learners to variability in both conceptual and contextual dimensions in order to maximize the possibility of transfer. Acknowledgments We would like to thank Dr. Wei-Zhen Lee and the Program for Educational Research and Development at McMaster University for assistance with data collection.
References Baghdady, M. T., Pharoah, M. J., Regehr, G., Lam, E. W., & Woods, N. N. (2009). The Role of basic sciences in diagnostic oral radiology. Journal of Dental Education, 73(10), 1187–1193. Baghdady, M. T., Carnahan, H., Lam, E. W., & Woods, N. N. (2013). Integration of basic sciences and clinical sciences in oral radiology for dental students. Journal of Dental Education, 77(6), 757–763. Eva, K. W., Neville, A. J., & Norman, G. R. (1998). Exploring the etiology of content specificity: Factors influencing analogical transfer and problem solving. Academic Medicine, 73(10 Suppl), S1–S5. Gick, M. L., & Holyoak, K. J. (1980). Analogical problem solving. Cognitive Psychology, 12(3), 306–355. Goldszmidt, M., Minda, J. P., Devantier, S., Skye, A. L., & Woods, N. N. (2012). Expanding the basic sciences debate: The role of physics knowledge in interpreting clinical findings. Advances in Health Sciences Education: Theory and Practice, 17(4), 547–555. Hatala, R. (2011). Practice makes perfect… sometimes. Medical Education, 45(2), 114–116. Hatala, R. M., Brooks, L. R., & Norman, G. R. (2003). Practice makes perfect: The critical role of mixed practice in the acquisition of ECG interpretation skills. Advances in Health Sciences Education: Theory and Practice, 8(1), 17–26. Helsdingen, A. S., Gog, T., & van Merrie¨nboer, J. J. G. (2010). The effects of practice schedule on learning a complex judgment task. Journal of Educational Psychology, 103(2), 383–398. Holyoak, K. (1989). Analogical mapping by constraint satisfaction. Cognitive Science, 13(3), 295–355.
123
Effect of context variation on mixed practice Holyoak, K. J., & Koh, K. (1987). Surface and structural similarity in analogical transfer. Memory and Cognition, 15(4), 332–340. Kaminski, J. A., Sloutsky, V. M., & Heckler, A. F. (2009). Concrete instantiations of mathematics: A double-edged sword. Journal for Research in Mathematics Education, 40(2), 90–93. Kaminski, J. A., Sloutsky, V. M., & Heckler, A. F. (2013). The Cost of concreteness: The effect of nonessential information on analogical transfer. Journal of Experimental Psychology: Applied, 19(1), 14–29. Kulasegaram, K., Martimianakis, M. A., Mylopoulos, M., Whitehead, C. R., & Woods, N. N. (2013). Cognition before curriculum: Rethinking the integration of basic science and clinical learning. Academic Medicine, 88(10), 1578–85. Kulasegaram, K., Min, C., Ames, K., Howey, E. H., Neville, A. J., & Norman, G. R. (2012). The effect of conceptual and contextual familiarity on transfer performance. Advances in Health Sciences Education: Theory and Practice, 17(4), 489–499. Laksov, K. B., Lonka, K., & Josephson, A. (2008). How do medical teachers address the problem of transfer? Advances in Health Sciences Education : Theory and Practice, 13(3), 345–360. Lee, T. D., Magill, R. A., & Weeks, D. J. (1985). Influence of practice schedule on testing schema theory predictions in adults. Journal of Motor Behaviour, 17(3), 283–299. Lowenstein, J., Thompson, L., & Gentner, D. (2003). Analogical encoding facilitates transfer in negotiation. Psychological Bulleting Review, 6(4), 586–597. Magill, R. A., & Hall, K. G. (1990). A review of the contextual interference effect in motor skill acquisition. Human Movement Science, 9, 241–289. Morris, C. D., Bransford, J. D., & Franks, J. J. (1977). Levels of processing versus transfer appropriate processing. Journal of Verbal Learning and Verbal Behaviour, 16(5), 519–533. Needham, D. R., & Begg, I. M. (1991). Problem-oriented training promotes spontaneous analogical transfer: Memory-oriented training promotes memory for training. Memory and Cognition, 19(6), 543–557. Norman, G. R. (2009). Teaching basic science to optimize transfer. Medical Teacher, 31(5), 807–811. Norman, G. R., Dore, K., Krebs, J., & Neville, A. J. (2007). The power of the plural: Effect of conceptual analogies on successful transfer. Academic Medicine, 82(10 Suppl), S16–S18. Paas, F. G. W. C., & van Merrie¨nboer, J. J. G. (1994). Variability of worked examples and transfer of geometrical problem solving skills: A cognitive-load approach. Journal of Educational Psychology, 85(1), 122–133. Reeves, L. M., & Weisberg, R. W. (1994). The role of content and abstract information in analogical transfer. Psychological Bulletin, 115(3), 381–400. Roher, D., & Taylor, K. (2007). The shuffling of mathematics problems improves learning. Instructional Science, 35(6), 481–498. Rohrer, D., & Pashler, H. (2010). Recent research on human learning challenges conventional instructional strategies. Educational Researcher, 39, 406–422. Ross, B. H. (1987). This is like that: The use of earlier problems and the separation of similarity effects. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13(4), 629–639. Ross, B. H., & Kennedy, P. T. (1990). Generalization from the use of earlier examples in problem solving. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16(1), 42–55. Salomon, G., & Perkins, D. N. (1989). The rocky road to transfer: Rethinking mechanisms of a neglected phenomenon. Educational Psychologist, 24(2), 113–142. Schmidt, R. A., & Bjork, R. A. (1992). New conceptualizations of practice: Common principles in three paradigms suggest new concepts for training. Psychological Science, 3(4), 207–217. Simon, D. A. (2008). Scheduling and learning. Advances in Psychology, 139(1), 61–72. Sweller, J. (1994). Cognitive load theory, learning difficulty, and instructional design. Learning and Instruction, 4(4), 295–312. Sweller, J., van Merrie¨nboer, J. J. G., & Paas, F. G. W. C. (1998). Cognitive architecture and instructional design. Educational Psychology Review, 10(3), 251–296. van Merrie¨nboer, J. J. G., Clark, R. E., & de Croock, M. B. M. (2002). Blueprints for complex learning: The 4C/ID-model. Educational Technology Research and Development, 50(2), 39–64. van Merrie¨nboer, J. J. G., Kester, L., & Paas, F. (2006). Teaching complex rather than simple tasks: Balancing intrinsic and germane load to enhance transfer of learning. Applied Cognitive Psychology, 20(3), 343–352. Woods, N. N. (2007). Science is fundamental: The role of biomedical knowledge in clinical reasoning. Medical Education, 41(12), 1173–1177. Woods, N. N., Brooks, L. R., & Norman, G. R. (2005). The value of basic science in clinical diagnosis: Creating coherence among signs and symptoms. Medical Education, 39(1), 107–112.
123
K. Kulasegaram et al. Woods, N. N., Neville, A. J., Levinson, A. J., Howey, E. H., Ockowski, W. J., & Norman, G. R. (2006). The value of basic science in clinical diagnosis. Academic Medicine, 81(10 Suppl), S124–S127. Woods, N. N., Brooks, L. R., & Norman, G. R. (2007a). The role of biomedical knowledge in diagnosis of difficult clinical cases. Advances in Health Sciences Education: Theory and Practice, 12(4), 417–426. Woods, N. N., Brooks, L. R., & Norman, G. R. (2007b). It all make sense: Biomedical knowledge, causal connections and memory in the novice diagnostician. Advances in Health Sciences Education: Theory and Practice, 12(4), 405–415.
123
The mediating effect of context variation in mixed practice for transfer of basic science Kulamakan Kulasegaram • Cynthia Min • Elizabeth Howey • Alan Neville • Nicole Woods • Kelly Dore • Geoffrey Norman
Received: 25 April 2014 / Accepted: 22 November 2014 Springer Science+Business Media Dordrecht 2014
Abstract Applying a previously learned concept to a novel problem is an important but difficult process called transfer. Practicing multiple concepts together (mixed practice mode) has been shown superior to practicing concepts separately (blocked practice mode) for transfer. This study examined the effect of single and multiple practice contexts for both mixed and blocked practice modalities on transfer performance. We looked at performance on near transfer (familiar contexts) cases and far transfer (unfamiliar contexts) cases. First year psychology students (n = 42) learned three physiological concepts in a 2 9 2 factorial study (one or two practice contexts and blocked or mixed practice). Each concept was practiced with two clinical cases; practice context was defined as the number of organ systems used (one system per concept vs. two systems). In blocked practice, two practice cases followed each concept; in mixed practice, students learned all concepts before seeing six practice cases. Transfer testing consisted of correctly classifying and explaining 15 clinical cases involving near and far transfer. The outcome was ratings of
K. Kulasegaram (&) Department of Family & Community Medicine & The Wilson Centre, University of Toronto, 200 Elizabeth St 1ES-565, Toronto, ON M5G 2C4, Canada e-mail: [email protected] C. Min Centre for Health Education Scholarship, University of British Columbia, Vancouver, BC, Canada E. Howey K. Dore G. Norman Programme for Education Research and Development, Faculty of Health Sciences, McMaster University, Hamilton, ON, Canada A. Neville Department of Oncology, Faculty of Health Sciences, McMaster University, Hamilton, ON, Canada N. Woods Department of Surgery & The Wilson Centre, Faculty of Medicine, University of Toronto, Toronto, ON, Canada
123
K. Kulasegaram et al.
quality of explanations on a 0–3 scale. The repeated measures analysis showed a significant near versus far by organ system interaction [F(1,38) = 3.4, p \ 0.002] with practice with a single context showing lower far transfer scores than near transfer [0.58 (0.37)–0.83 (0.37)] compared to the two contexts which had similar far and near transfer scores [1.19 (0.50)–1.01 (0.38)]. Practicing with two organ contexts had a significant benefit for far transfer regardless of mixed or blocked practice; the single context mixed practice group had the lowest far transfer performance; this was a large effect size (Cohen’s d = 0.81). Using only one practice context during practice significantly lowers performance even with the usually superior mixed practice mode. Novices should be exposed to multiple contexts and mixed practice to facilitate transfer. Keywords design
Basic science Transfer Cognition Teaching strategies Instructional
Introduction A central role of preclinical medical education is to provide students with the baseline knowledge they will use to manage clinical issues during clinical training. However, there is an abundance of evidence from other domains that even if students possess the knowledge needed to solve a problem, they may have difficulty accessing that knowledge to solve a new, unfamiliar version of the problem (Ross 1987). This phenomenon, accessing old knowledge to resolve new problems, is called transfer. In undergraduate medical training, the issue of transfer is particularly serious as many different domains of knowledge such as basic sciences are taught with the expectation of future utility for problem solving. Several studies have established (Goldszmidt et al. 2012; Baghdady et al. 2009, 2013; Woods 2007; Woods et al. 2005, 2006, 2007a) that use of basic sciences knowledge enables students to more accurately and efficiently grasp clinical skills such as diagnosis and interpreting examination findings. However, the usefulness of basic science is not simply dependent on retention (Baghdady et al. 2009). Learners must recognize that their previous knowledge from basic science can be used to understand the clinical skill or task they are learning, and must be able to retrieve the particular knowledge they require to understand the clinical situation. In other words, the transfer of basic science knowledge to the clinical task is essential if learners are to derive benefit from basic science teaching. Over a century of research supports the view that transfer is a difficult cognitive process for learners (Eva et al. 1998). This is not a novel problem in medical education, but rather a near universal finding in many domains (Salomon and Perkins 1989). Despite this, curriculum planners and instructors are rarely able to identify transfer as an explicit curriculum goal and devise strategies to facilitate it (Laksov et al. 2008). Developing strategies to teach learners to transfer concepts requires an understanding of why students find spontaneous transfer difficult. One of the most significant reasons for failure to transfer is the crucial impact of context specificity (Eva et al. 1998). For example, illustrating the application of fluid flow dynamics (the concept) with an example from the respiratory system (e.g. asthma or bronchitis) may enable learners to solve similar problems arising in the respiratory system. Fluid flow dynamics is relevant to many body systems, but when learners are presented with novel problems situated in a new organ system context (e.g. cardiovascular), evidence suggests that learners fail to recognize the usefulness of the previous concept in the new context (Kulasegaram et al. 2012). Problems for which the context of learning and that of transfer have similar, familiar, surface details
123
Effect of context variation on mixed practice
are near transfer (e.g. respiratory to respiratory) and are invariably easier than far transfer, which occurs when the surface details are different between the context of learning and the context of transfer (e.g. transfer from respiratory to cardiovascular). Transfer theory suggests that learners tend to classify and retrieve concepts based on the superficial or surface level similarity based on their evaluation of the context of the problem (e.g. organ system or case setting) instead of conceptual similarity (Gick and Holyoak 1980; Holyoak and Koh 1987; Ross and Kennedy 1990). Our previous work (Kulasegaram et al. 2012) has shown that use of contextual familiarity to cue recognition of concepts is predictive of transfer success for near transfer but is inaccurate for far transfer. Moving learners from over-reliance on superficial contextual similarity and appropriate integration of contextual information is a focus of many transfer interventions. Context variation The recognition that the context of learning is meaningful to learners has led to practice strategies that have focused on manipulating the relationship between conceptual knowledge and the context in which concepts are practiced (Kaminski et al. 2013, 2009). One approach is structuring practice for context variation, i.e. multiple examples with dissimilar surface details (Ross and Kennedy 1990). To illustrate, simultaneously practicing the application of a basic science concept such as fluid dynamics in the cardiovascular and respiratory organ system is practice with contextual variation. Viewing conceptual knowledge in multiple contexts forces learners to discount superficial features and focus on conserved elements of the concept present in both examples. Practice in multiple contexts has shown dramatic improvement in transfer performance for novices and is a common strategy in medical education (Norman 2009). Mixed practice Another innovation has been to redesign practice to promote mixed or interleaved practice. Traditional practice involves learning a concept followed by practice with multiple examples of that concept. This strategy is known as blocked practice. Mixed practices involves practicing examples of multiple concepts together. Originally developed and studied in motor skills learning (Lee et al. 1985; Magill and Hall 1990) mixed practice has been extensively investigated for concept and procedural skills learning (Helsdingen et al. 2010; Roher and Taylor 2007; Hatala 2011). Head to head studies of transfer performance for mixed and blocked practice have consistently demonstrated an advantage for mixed practice (Roher and Taylor 2007). The reason for the advantage of mixed practice is that in traditional blocked practice, learners do not have to identify the type of concept to complete the practice problems; it is already known that the concept is the one from the preceding lesson (Rohrer and Pashler 2010). In contrast, mixed practice requires that learners must actively learn to distinguish which concept correctly applies to each problem. As Roher and Taylor (2007) notes: whereas blocked practice requires students to know how to perform a procedure, it does not require them to know which procedure is appropriate. Evidence for the benefit of mixed practice has been demonstrated for clinical learning. Hatala et al. (2003) studied diagnosis of ECGs by novices who were asked to learn to interpret ECGS of three different types of disorders and then given practice in either blocked or mixed fashion. Transfer testing showed participants in the mixed condition successfully diagnosed 46 % cases while in the blocked condition accuracy was 30 %.
123
K. Kulasegaram et al.
Contextual variation in optimizing mixed practice While mixed practice exposes learners to different conceptual structures, context variation exposes learners to the variability of how a single concept may present during future problem solving. Both strategies emphasize variability but at different levels in order for learners to develop the appropriate recognition of conceptual knowledge necessary for future transfer and problem solving. As such, it is possible that these interventions may have similar mechanisms of action and adopting one or the other maybe sufficient to support concept transfer by learners. Given the increasing evidence for the benefits of mixed practice, simply reorganizing existing basic science practice material may be sufficient to promote transfer. On the other hand, novice learners may still intuitively make close associations with contextual surface features if practice does present variations of context and surface features for basic science concepts. To our knowledge, few studies have examined context variation and its role in impacting of mixed practice of conceptual knowledge. The studies to do so, (Simon 2008) have not examined transfer but focused on outcomes more closely related to recall or memory for taught concepts. Furthermore, these studies did not examine the impact on near and far transfer. The central question of this study was whether the efficacy of mixed practice of multiple concepts when it occurs with and without context variation on near and far transfer of basic science concepts. We hypothesized that context variation is essential for the advantages of mixed practice and that without it, mixed practice may lose its efficacy. Specifically, if context variation plays a role in mediating the effects of practice strategy, the effect of multiple contexts should be to encourage participants to discount context as a cue to categorization, and focus on the underlying concept. Consequently, we expected to see a significant reduction in the classic difference between near and far transfer performance for participants who practiced with context variation regardless of practice strategy. On the other hand, if context variation did not play a role, we expected to see a reduction in the difference between near and far transfer for participants with mixed practice compared to blocked practice regardless of context variation.
Methods Participants This study was conducted with forty-four (N = 44) first year undergraduate students in the Department of Psychology at McMaster University during 2010–2011. Participants received either $10.00 in compensation for participation or course credit. The study received institutional ethics approval from the McMaster University Research Ethics Board. Procedures, materials, and other aspects of the design are similar to those used in previous studies (Kulasegaram et al. 2012; Norman et al. 2007). Materials We used materials previously used for previous transfer studies (see Norman et al. 2007; Kulasegaram et al. 2012). Participants in this study were asked to read explanations written by an expert clinician (AN) for the three physiology related principles.: Fluid dynamics, Laplace’s law, and Starling’s law. Fluid dynamics illustrates the principles of laminar and turbulent flow, which has application in multiple organ systems including the respiratory, cardiovascular, gastrointestinal, and genitourinary tract. Laplace’s law describes
123
Effect of context variation on mixed practice Table 1 Example of explanation: Laplace’s law and teaching analogy Analogy supplemented clinical explanation Laplace’s law applies in many situations, particularly in the gastrointestinal tract. The intestines are tubes with a radius, intestinal wall tension, and a pressure exerted by the tube’s contents: digesting food. The tension in the intestinal wall is proportional to the radius of the intestine and the pressure of the contents. This relationship can be explained by: T = PR where T is the wall tension, P is the pressure, and R is the radius. Thus, having a small radius decreases the tension in the wall for the same internal pressure. Laplace’s law explains why in normal circumstances intestines don’t rupture their contents and why in some conditions, serious complications can follow dilatation or an increase in the radius of the bowel, as occurs in inflammatory conditions of the bowel A diagrammatic way of explaining this relationship of wall vessel tension and the pressures across the vessel wall is shown in the left diagram. In order to maintain the same downward force (the pressure), the wall tension T must increase as the wall becomes less curved. This is analogous to a weight hanging on a string as in the diagram at the right. When a weight is suspended from a string, the tension in the two ends of the string increases as the string becomes closer and closer to horizontal, because the vertical component of the tension must equal the weight
relationship between tension in the wall of a cylindrical vessel and the radius of the vessel and pressure across the wall; it applies to the same organ systems as the principles of laminar and turbulent flow. Starling’s law describes the elastic behavior of the heart in response to filling of the ventricles and volume of the ejected blood. Because it only applies to the heart, Starling’s law was used to increase the number of concepts participants had to learn and was tested only in near transfer cases. All explanations included a detailed outline of the concept, its relevance to physiology and a brief example of its application, and emphasized that the concept applied in multiple organ systems (except for Starling’s law). Each explanation was also augmented by the provision of a teaching analogy, which illustrated the basic principles of the concept (Kulasegaram et al. 2012). For example, the relationship of tension, pressure, and vessel radius described by Laplace’s law was illustrated using the analogy of a weight suspended in the middle of a string. The increase in vessel wall tension due to increasing the radius of a cylindrical vessel is akin to the increase in the tension of the string as the ends of the string are pulled apart. We have previously used teaching analogies to study transfer performance (Kulasegaram et al. 2012; Norman et al. 2007) and all of the participants in the study were given teaching analogies for each concept (see Table 1). Practice for each concept consisted of two clinical vignettes written by an expert clinician (AN). Participants were asked to provide an explanation of how the concept related
123
K. Kulasegaram et al. Table 2 Example of Laplace’s law’s practice vignette and feedback The patient complains of increasing abdominal pain. On examination, his abdomen is distended and an abdominal X-ray shows a dilated colon. The patient complains of being unable to move his bowels for the past four days and of previously severe pain the area. The doctor is concerned that the patient might suffer a bowel perforation and prescribes anti-inflammatory drugs. He recommends surgery to constrict the size of the stomach. Explain why these measures may help and what the risk is to the patient? The dilation of the colon is an increase in the radius of the gastric tube. While the pressure of the contents may stay the same, the increase in the radius will increase the tension of the colon (T = PR). If the tension continues to increase the colon’s walls might actually rupture from the excess stress. While this is a rare possibility, it is severe. Similarly, the increased size of the bowel reduces the ability of the muscles to generate enough force to push through the contents. Reducing the radius or size should eliminate this difficulty
to the signs and symptoms—in other words, how the concept applied to the problem. They were then provided with the correct answer. See Table 2 for an example. Design In a 2 9 2 design, students were randomized to learn and practice three physiology concepts with either mixed or blocked practice as well as practice with either one or two organ system contexts for two of the concepts (see Fig. 1). Manipulation for mixed and blocked practice In the blocked practice conditions, practice cases followed each explanation so that a participant starting with Laplace’s law would read the explanation for the concept and would then complete the two practice cases before moving onto the next concept. Practice cases in blocked conditions required students to explain how the concept explained the features of the case; participants were then given feedback in the form of the correct answer as written by the expert clinician. In the mixed practice condition, participants read all three explanations prior to any practice; in the practice phase, participants were presented with six practice cases in random order and told to identify which law applied and how it explained the features. As in the blocked group, participants were then given the correct answer for each case.
Tesng
Learning One organ system
Mixed Pracce Two organ systems
Blocked Pracce
One organ system
Two organ systems
Fig. 1 Summary of manipulation and general design
123
10 T/F
6 MCQ, 4 short answer
15 Transfer cases (near and far)
Effect of context variation on mixed practice
Manipulation of context variation Practice contexts were manipulated using organ systems involved in the vignettes depicted in the practice cases, which could involve the same or different organ systems. In the single practice context condition, fluid dynamics cases involved only respiratory disorders and Laplace’s law cases involved only gastrointestinal disorders. In the multiple practice context condition, fluid dynamics had one respiratory and one cardiovascular case while Laplace’s law was practiced with a case from the gastrointestinal and one from the cardiovascular system. Starling’s law only applied to the cardiovascular system and was included to increase the difficulty of the learning and transfer tasks. Procedure This study was conducted in two phases, learning–practice and testing, which occurred in the same session. Phase 1: Learning and practice In phase 1, participants were exposed to the principles and their assigned practice condition. Order of presentation of principles and thus practice of principles was randomized. After viewing the explanations and completing practice with feedback, all participants were asked to complete a ten question True/False quiz on the concepts. Participants were required to obtain a minimum score of 8/10 to move on to a knowledge test or were asked to repeat the quiz before moving onto phase 2: Testing. To avoid test-enhanced learning, participants were not given feedback. Phase 2: Testing Recall and knowledge test Phase 2 began with a knowledge test of the concepts in the form of six multiple choice and four short answer questions for a total score out of 10. Questions directly elicited information presented about the concepts in the explanations. Participants only completed the test once and were not given feedback. Transfer testing All participants were presented with 15 transfer cases in the form of clinical vignettes with similar structure to those in the practice session. Participants were asked to identify the concept that best applied to the vignette, then provide an explanation for how the concept accounted for the clinical signs and symptoms. Responses were scored on a 0–3 scale with one point awarded for correctly identifying the concept involved and additional points for accuracy and depth of explanation. This scoring system was used in previous transfer studies (Norman et al. 2007; Kulasegaram et al. 2012). A sub-sample of cases was scored by two raters to determine inter-rater reliability. See Table 3 for an example of the test cases and scoring. To determine if differences between groups were due increases in correctly recognizing concepts or increasing depth of explanations, we created a secondary measure on the proportion of cases correctly classified by concept. As in our previous study (Kulasegaram et al. 2012) this allowed us to examine if gains in transfer were purely due to participants developing more complex representations of the concepts as expressed through better
123
K. Kulasegaram et al. Table 3 Example of Laplace’s law testing case, response and appropriate scoring A patient complained of vomiting copious amounts at the end of each day. Investigations revealed he had a lax, poorly contracting stomach. He was given a drug that constricted the size of the stomach, resulting in more efficient emptying. Explain why constricting the size impacts stomach emptying Assigned score
Response
Response
Rationale
0
Goethe/ starling
1
Laplace
Not provided or incorrect
Correct law but no explanation or wrong explanation
2
Laplace
This can be due to a decreased radius of the stomach. The drug must have dilated the pathway of the stomach, allowing the food to stay in the stomach for further digestion
Correct law but superficial explanation
3
Laplace
The drug might have decreased tension within the stomach walls. From the knowledge of Laplace’s law, we know that as the tension decreases, so does the pressure. The drop in the pressure from the outside walls, stomach’s radius increases and thus, resulting in more efficient space
Correct response, response captures the deep structure of Laplace’s law
Wrong response
scoring on the explanations or if the mechanism of the interventions worked at the purely better recognition of concepts. To create near and far transfer cases, vignettes involved disorders of organ systems that were familiar (near transfer) and unfamiliar (far transfer) for participants. Near transfer for all participants in the study were Flow dynamics cases involving the respiratory organ system, Laplace’s law cases involving the gastrointestinal system, and Starling’s law cases involving the cardiovascular organ system. Flow dynamics and Laplace’s law cases involving the cardiovascular organ system were considered near transfer for half of the participants who saw this organ system and concept pairing in practice as part of the multiple contexts of practice manipulation. These cases were far transfer for the half that did not. Flow dynamics cases involving the urinary tract and Laplace’s law involving the reproductive system were consider common far transfer for all participants. Outcomes and analysis To control for time on task and as a proxy for effort, we measured time spent during the learning phase and analyzed average time spent per case using 2 9 2 9 2 repeated measures ANOVA with average Near and Far transfer case time as the repeated measures factor, practice strategy (mixed vs. blocked) and practice contexts (one vs. two) as the between subject factors. The outcome for the knowledge test was a total score out of 10; a 2 9 2 ANOVA was conducted on the knowledge test with practice strategy and number of practice contexts as the between subjects factors. The main analysis of the hypothesis involved outcomes on the transfer test analyzed using average score on case (0–3) and using 2 9 2 9 2 repeated measures ANOVA with average Near and Far transfer performance as the repeated measures factor, practice strategy (mixed vs. blocked) and practice contexts (one vs. two) as the between subject
123
Effect of context variation on mixed practice
factors. The secondary analysis was of the recognition or classification score (percent of cases correctly classified) using a 2 9 2 9 2 repeated measures ANOVA with the same design as the main analysis. We determined a priori to use the knowledge test and time spent per case as covariates in the analysis if there were significantly associated with the outcome.
Results Forty-two students completed the study. One participant did not complete the test phase and another randomized to the Blocked practice with two organ systems scored 0 in all transfer cases; these were excluded from analyses for a final sample size of 11 per group in the mixed with two and blocked with one group and ten per group for the mixed with one and blocked with two. Inter-rater reliability for rating of explanations was high (ICC = 0.85). Effect of time on task and time per case Participants on average took 2,378 s on the learning phase (SD = 202 s); participants in the mixed condition took longer [2,734 s (190 s)] than in the blocked condition [2,023 s (134 s)] and this difference was significant [F(1,38) = 5.7, p \ 0.02, Cohen‘s d = 4.3]. Time spent during learning and practice did not correlate with scores on knowledge test or on transfer cases (r = 023, p \ 0.35). Average time per far transfer case was 152.3 s (61 s) and for near transfer case was 92.2 s (54). Analysis of time per case showed no significant differences due to practice condition, organ system but a main effect of far transfer time being greater than near transfer time [F(1,38) = 15.8, p \ 0.0001, d = 1.03]. There was no significant correlation between average case time and performance for near transfer (r = 0.210, p \ 0.18) or for far transfer (r = -0.19, p \ 0.11). Performance on knowledge test Scores on the multiple choice and short answer knowledge tests are shown in Table 4. Analysis showed no effect of practice strategy [F(1,38) = 0.993, p \ 0.35] or number of organ systems during practice [F(1,38) = 2.3, p = 0.21] as well as no significant interaction [F(1,38) = 2.1, p \ 0.19]. Scores on knowledge testing did not correlate significantly with average performance with near transfer cases (r = 0.19, p \ 0.43) and far transfer cases (r = 0.33, p \ 0.18) separately and thus were not included as covariates in the analysis of transfer performance. Performance at transfer testing The repeated measures analysis showed a significant near versus far by organ system interaction [F(1,38) = 3.4, p \ 0.002, g2p = 0.16] with individuals practicing with a single organ system showing lower average far transfer scores than near transfer; mean (SD) for far 0.58 (0.37) versus 0.83 (0.37) for near (d = 0.68). The two organ systems of practice group had similar far and near transfer scores [1.19 (0.50) for far vs. 1.01 (0.38) for near; d = 0.4]. Post-hoc testing using the Least-Significant-Differences method showed no significant differences between the groups on near transfer performance (see Fig. 2) but a significant main effect for two organ systems over one on far transfer (see Fig. 3)
123
K. Kulasegaram et al. Table 4 Average total score on multiple choice and short answer knowledge test by condition
Practice type Blocked Mixed
No of organ systems
Mean (SD)
1
7.50 (0.32)
2
8.10 (0.45)
1
7.30 (0.44)
2
7.66 (0.56)
[F(1,38) = 20.3, p \ 0.0001, d = 0.81]. There was no significant main effect of practice strategy [F(1,38) = 2.4, p \ 0.128, d = 0.1] or for the practice by organ system by near versus far interaction [F(1,38) = 0.07, p \ 0.78, g2p = 0.03]. Overall, the participants who completed mixed practice with two organ systems had the highest performance for near [1.16 (0.30)] and far transfer [1.24 (0.41)]. Blocked practice participants with two organ systems had similar far transfer scores at mean (SD) of 1.14 (0.32) and outperformed participants in mixed practice and blocked practice with a single organ system as 0.44 (0.51) and 0.73 (0.41) respectively. Classification of transfer cases On average participants classified more near transfer cases correctly at [mean (SD)] 65.35 % (19 %) compared to far transfer cases [55.1 % (22 %); F(1,38) = 7.68, p \ 0.01, d = 0.5]. There was a significant interaction with organ system [F(1,38) = 19.14, p \ 0.0001, g2p = 0.3] but no interaction with practice condition [F(1,38) = 1.5, p \ 0.22, g2p = 0.02]. On average, participants in the two organ systems condition classified 60 % (15 %) of near and 65 % (18 %) of far transfer cases correctly. Participants in the one organ system group classified 70 % (20 %) of near transfer cases but only 43 % (19 %) of far transfer cases. The practice by organ system interaction was marginally non-significant (F1,38) = 3.87, p \ 0.06, g2p = 0.11. The mixed practice with two organ systems group had similar rates of success at classifying near and far transfer cases at 66 % (13 %) versus
Fig. 2 Average score on near transfer cases by condition
123
Effect of context variation on mixed practice
Fig. 3 Average score on far transfer cases by condition
70 % (19 %). The mixed practice with one organ system group classified 72 % (23 %) of near transfer cases correctly but only 37 % (22 %) of far transfer cases.
Discussion This study examined the impact of varying the number of practice contexts and mixed strategies for transfer of concept knowledge. We found that there was a significant interaction between the number of practice contexts participants were exposed to and the difference between near and far transfer. This suggested that increasing the number of practice contexts in both practice schedules reduced reliance on contextual information. Our results showed that mixed practice performance on far transfer was dependent on the number of practice contexts used during learning thus supporting our initial hypothesis. As the number of practice contexts increased, so did far transfer performance in both practice strategies. While mixed practice with multiple organ systems had the highest overall transfer performance, mixed practice with a single organ system per concept had the lowest far transfer performance. This suggests that without contextual as well as conceptual variability, mixed practice may not always be successful at promoting far transfer. Analysis of accurate recognition or classification indicates the advantage of practice with multiple contexts results from improvement in the ability to recognize concepts in new contexts not with increasing depth of explanation. This is reflected by scores on quality of explanations showed participants in all groups to still have low performance indicating both the difficulty of transfer and lack of strong conceptual understanding. There are several limitations to this study. Studies of mixed and blocked practice typically find an increase in performance for mixed practice with delayed testing (Helsdingen et al. 2010). Thus, the absence of a delayed test limits the inferences from this study to everyday educational practice. However, the purpose of this specific study was to evaluate mechanisms that mediate the impact of practice strategies on transfer
123
K. Kulasegaram et al.
performance. We were not interested in retention of knowledge and thus extrapolation to applied settings maybe limited. Still, comparisons with other studies using these materials and including delayed testing indicated that differences between groups remain robust after delay though there is also significant forgetting (Kulasegaram et al. 2012). A second limitation is that the sample used in this study consisted of psychology students generally naı¨ve to the physiology and clinical conditions that were used in testing. Previous exposure or content knowledge can negate differences and obscure the mechanisms involved in mediating transfer. In our local context, available health sciences trainees are typically knowledgeable in the concepts used to study transfer in this study. Interpretations of these results should once again focus on the mechanisms that impact the efficacy of mixed practice in more complex settings and with more advanced learners. While the sample size is also small, the large effect sizes suggest we were adequately powered to detect differences. A third limitation is that we did not control for prior knowledge using a screening test. Pre-testing could have been a source of bias in this study due to the phenomenon of test-enhanced learning and thus was part of the design. Instead, we chose to control for knowledge gained by students in phase 1. While these limitations as a whole limited the external validity of the results to applied education practice, they do not necessarily impact interpretations of the mechanisms of transfer of conceptual learning in basic sciences. Other studies using these materials and protocol (Norman et al. 2007; Kulasegaram et al. 2012) have investigated transfer mechanisms under similar circumstances. As in those studies, participants had very low performance on our primary outcome. In fact, the highest mean score for any group was still less than half of the possible maximum (1.5/3). While recognition scores were higher, the overall rating of the transfer case explanations suggest some limitations of the design and intervention. In general, transfer is difficult for novice learners and alleviation of this difficulty or building of sophisticated responses to clinical vignettes is unlikely to occur through a brief one-shot learning intervention. Indeed, given the difficulty of the task for the sample, it is remarkable we could detect differences between groups. Furthermore, while raw scores were low, the effects size of differences between comparison groups were moderate to large. Still, a larger study with delayed testing is necessary to address the external validity of the results and the direct applicability to classroom educational practice. There are several explanations for the overall benefits of mixed practice. Generally, these explanations argue that mixed practice poses additional processing requirements for learners in that they must not only process the solution to the practice problems but must also attempt to distinguish the type of problem (Roher and Taylor 2007). For example, this extra processing activity has been characterized as desirable difficulty (Schmidt and Bjork 1992) and is thought to induce task-specific processing for solving transfer problems. A more instrumental explanation is provided by cognitive load theory and its application to instructional design. Cognitive load theory posits that learning tasks draw upon limited mental resources such as working memory and that learning is maximized when this load is optimized to the types of processing necessary for building conceptual schemas for problem solving (Paas and van Merrie¨nboer 1994). In this regard, while extraneous load or distracting processing should be reduced, processing relevant to retention and transfer should be increased according to the learner‘s needs (Sweller 1998). The effects of mixed practice and contextual variation or interference, and more generally, variability during learning can be understood in this framework. Variability increases the mental effort devoted to understanding the differences between concepts (germane load), thus leading to more robust representations of the concept or category (van Merrie¨nboer et al. 2006). Contextual inference in the form of contextual variation (i.e. multiple organ systems for a
123
Effect of context variation on mixed practice
concept) can teach learners about relevant diagnostic information by thinking about how concepts generalize across contexts (van). In this study, the mixed practice with multiple organ systems had two levels of variation that increase learner’s germane load: concept and context. Learners must distinguish between the representations of the concept within each organ system as well as learn that contextual information is an imperfect guide to concept identification. This group thus had the greatest difficulty during practice. The additional processing or germane load—distinguishing structural and surface features during practice—likely forced them to recognize and use structural features during problem solution. This in turn reduces reliance on surface features and possibly decreases the intrinsic load of future problem solving. Accordingly, this group had the highest near and far transfer performance and the difference between near and far transfer performance was lowest for this group. On the other hand, mixed practice with one organ system has the benefit of task variability without contextual interference. With the constraining of one organ system to one concept, using organ system as a cue for identifying concepts is cognitively less demanding activity. While this approach minimizes overall cognitive load, it also decreases the germane processing necessary for successful future transfer and reinforces use of surface level organ system cues as diagnostic of underlying concepts. A complementary, and more fundamental account of the effects seen here can be found in the transfer appropriate processing framework (TAP; Morris et al. 1977; Needham and Begg 1991). Essentially, TAP suggests that transfer occurs when there is congruence in mental processing at learning and at transfer. Many transfer teaching strategies emphasize the importance of fostering increasing abstraction of concepts as the transfer appropriate process when concepts apply in multiple contexts (Holyoak 1989; Lowenstein et al. 2003; Norman 2009). In this study, processing contextual information could have been accurate method for classifying near transfer problems. But, with problems where concepts are not neatly constrained to solely one context, reliance on contextual cues can be misleading. Processing that reinforces reliance on contextual cues for classification does not align with the processing required for transfer in this case. Previous work suggests that novices tend to naively use contextual information in problem-solving unless taught otherwise (Salomon and Perkins 1989). Learners do not create automatically abstract representations or schemas; instead, associations arise with the example used to teach the concept. It is the example that is called to mind when recalling the concept as it is easy and salient to access (Reeves and Weisberg 1994). Use of a single context in practice reinforces the confounding of concepts and contexts; presenting concepts in multiple contexts serves to deemphasize the importance of the context in recognizing concepts. Examining the recognition scores, it is apparent that the mixed with one organ system group used contextual similarity effectively for near transfer problems but could not use this strategy for far transfer problems with accuracy as context was not diagnostic of concept. Transfer appropriate processing to solve these problems requires using the abstraction of the conceptual features and minimizing reliance on contextual cues. The mixed with two organ systems group displays some features of this transfer appropriate processing as evidenced by the reduced gap between near and far transfer. The germane processing induced by variability in both concepts and contexts likely helped foster processing of conceptual structure and thus led to reduced reliance on contextual cues by participants in the multiple organ systems group. There are other benefits for combining conceptual and contextual variability which were not explored in this study. For example, in motor skills training (Magill and Hall 1990), contextual variation is integrated into mixed practice and is the defining feature of creating practice schedules that support transfer. This contextual interference has the impact of
123
K. Kulasegaram et al.
giving learners a range of possible variations of a task or concept and practice with adapting knowledge to multiple contexts. Mixed practice with multiple contexts thus furthers learning by reinforcing adaptation of previous experiences to new contexts. In general, our results reinforce the continuing difficulty of transfer and support the need for extended practice in multiple contexts and focus on teaching deep structure. Most transfer interventions (Norman 2009) focus on enhancing the abstraction of concepts—for example using teaching analogies, active comparison, mixed practice etc. Exposure to multiple contexts during practice forces learners to rely on more abstract, conserved features of concepts and possibly aids their ability to adapt knowledge to new contexts. A judicious mix of deep structure teaching aids that allow abstraction and practice with multiple contexts can push learners to more readily extract the essential features of concepts. This will likely prepare them to recognize these features in future transfer problems and ease the recall and application of that knowledge. Teaching of basic science concepts in undergraduate medical training should find ways to supplement learning along these axes.
Conclusion This study demonstrated that the effects of mixed practice for transfer of conceptual knowledge were dependent on the number of practice contexts. More generally, transfer performance is affected by contextual knowledge and ignoring the effect of context specificity can negate teaching interventions. Mixed practice as a tool to promote transfer should include as many forms of variability as appropriate for learners to appreciate the extent to which concepts generalize across problems. Teaching for transfer of basic science should consider that practice of concepts should expose learners to variability in both conceptual and contextual dimensions in order to maximize the possibility of transfer. Acknowledgments We would like to thank Dr. Wei-Zhen Lee and the Program for Educational Research and Development at McMaster University for assistance with data collection.
References Baghdady, M. T., Pharoah, M. J., Regehr, G., Lam, E. W., & Woods, N. N. (2009). The Role of basic sciences in diagnostic oral radiology. Journal of Dental Education, 73(10), 1187–1193. Baghdady, M. T., Carnahan, H., Lam, E. W., & Woods, N. N. (2013). Integration of basic sciences and clinical sciences in oral radiology for dental students. Journal of Dental Education, 77(6), 757–763. Eva, K. W., Neville, A. J., & Norman, G. R. (1998). Exploring the etiology of content specificity: Factors influencing analogical transfer and problem solving. Academic Medicine, 73(10 Suppl), S1–S5. Gick, M. L., & Holyoak, K. J. (1980). Analogical problem solving. Cognitive Psychology, 12(3), 306–355. Goldszmidt, M., Minda, J. P., Devantier, S., Skye, A. L., & Woods, N. N. (2012). Expanding the basic sciences debate: The role of physics knowledge in interpreting clinical findings. Advances in Health Sciences Education: Theory and Practice, 17(4), 547–555. Hatala, R. (2011). Practice makes perfect… sometimes. Medical Education, 45(2), 114–116. Hatala, R. M., Brooks, L. R., & Norman, G. R. (2003). Practice makes perfect: The critical role of mixed practice in the acquisition of ECG interpretation skills. Advances in Health Sciences Education: Theory and Practice, 8(1), 17–26. Helsdingen, A. S., Gog, T., & van Merrie¨nboer, J. J. G. (2010). The effects of practice schedule on learning a complex judgment task. Journal of Educational Psychology, 103(2), 383–398. Holyoak, K. (1989). Analogical mapping by constraint satisfaction. Cognitive Science, 13(3), 295–355.
123
Effect of context variation on mixed practice Holyoak, K. J., & Koh, K. (1987). Surface and structural similarity in analogical transfer. Memory and Cognition, 15(4), 332–340. Kaminski, J. A., Sloutsky, V. M., & Heckler, A. F. (2009). Concrete instantiations of mathematics: A double-edged sword. Journal for Research in Mathematics Education, 40(2), 90–93. Kaminski, J. A., Sloutsky, V. M., & Heckler, A. F. (2013). The Cost of concreteness: The effect of nonessential information on analogical transfer. Journal of Experimental Psychology: Applied, 19(1), 14–29. Kulasegaram, K., Martimianakis, M. A., Mylopoulos, M., Whitehead, C. R., & Woods, N. N. (2013). Cognition before curriculum: Rethinking the integration of basic science and clinical learning. Academic Medicine, 88(10), 1578–85. Kulasegaram, K., Min, C., Ames, K., Howey, E. H., Neville, A. J., & Norman, G. R. (2012). The effect of conceptual and contextual familiarity on transfer performance. Advances in Health Sciences Education: Theory and Practice, 17(4), 489–499. Laksov, K. B., Lonka, K., & Josephson, A. (2008). How do medical teachers address the problem of transfer? Advances in Health Sciences Education : Theory and Practice, 13(3), 345–360. Lee, T. D., Magill, R. A., & Weeks, D. J. (1985). Influence of practice schedule on testing schema theory predictions in adults. Journal of Motor Behaviour, 17(3), 283–299. Lowenstein, J., Thompson, L., & Gentner, D. (2003). Analogical encoding facilitates transfer in negotiation. Psychological Bulleting Review, 6(4), 586–597. Magill, R. A., & Hall, K. G. (1990). A review of the contextual interference effect in motor skill acquisition. Human Movement Science, 9, 241–289. Morris, C. D., Bransford, J. D., & Franks, J. J. (1977). Levels of processing versus transfer appropriate processing. Journal of Verbal Learning and Verbal Behaviour, 16(5), 519–533. Needham, D. R., & Begg, I. M. (1991). Problem-oriented training promotes spontaneous analogical transfer: Memory-oriented training promotes memory for training. Memory and Cognition, 19(6), 543–557. Norman, G. R. (2009). Teaching basic science to optimize transfer. Medical Teacher, 31(5), 807–811. Norman, G. R., Dore, K., Krebs, J., & Neville, A. J. (2007). The power of the plural: Effect of conceptual analogies on successful transfer. Academic Medicine, 82(10 Suppl), S16–S18. Paas, F. G. W. C., & van Merrie¨nboer, J. J. G. (1994). Variability of worked examples and transfer of geometrical problem solving skills: A cognitive-load approach. Journal of Educational Psychology, 85(1), 122–133. Reeves, L. M., & Weisberg, R. W. (1994). The role of content and abstract information in analogical transfer. Psychological Bulletin, 115(3), 381–400. Roher, D., & Taylor, K. (2007). The shuffling of mathematics problems improves learning. Instructional Science, 35(6), 481–498. Rohrer, D., & Pashler, H. (2010). Recent research on human learning challenges conventional instructional strategies. Educational Researcher, 39, 406–422. Ross, B. H. (1987). This is like that: The use of earlier problems and the separation of similarity effects. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13(4), 629–639. Ross, B. H., & Kennedy, P. T. (1990). Generalization from the use of earlier examples in problem solving. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16(1), 42–55. Salomon, G., & Perkins, D. N. (1989). The rocky road to transfer: Rethinking mechanisms of a neglected phenomenon. Educational Psychologist, 24(2), 113–142. Schmidt, R. A., & Bjork, R. A. (1992). New conceptualizations of practice: Common principles in three paradigms suggest new concepts for training. Psychological Science, 3(4), 207–217. Simon, D. A. (2008). Scheduling and learning. Advances in Psychology, 139(1), 61–72. Sweller, J. (1994). Cognitive load theory, learning difficulty, and instructional design. Learning and Instruction, 4(4), 295–312. Sweller, J., van Merrie¨nboer, J. J. G., & Paas, F. G. W. C. (1998). Cognitive architecture and instructional design. Educational Psychology Review, 10(3), 251–296. van Merrie¨nboer, J. J. G., Clark, R. E., & de Croock, M. B. M. (2002). Blueprints for complex learning: The 4C/ID-model. Educational Technology Research and Development, 50(2), 39–64. van Merrie¨nboer, J. J. G., Kester, L., & Paas, F. (2006). Teaching complex rather than simple tasks: Balancing intrinsic and germane load to enhance transfer of learning. Applied Cognitive Psychology, 20(3), 343–352. Woods, N. N. (2007). Science is fundamental: The role of biomedical knowledge in clinical reasoning. Medical Education, 41(12), 1173–1177. Woods, N. N., Brooks, L. R., & Norman, G. R. (2005). The value of basic science in clinical diagnosis: Creating coherence among signs and symptoms. Medical Education, 39(1), 107–112.
123
K. Kulasegaram et al. Woods, N. N., Neville, A. J., Levinson, A. J., Howey, E. H., Ockowski, W. J., & Norman, G. R. (2006). The value of basic science in clinical diagnosis. Academic Medicine, 81(10 Suppl), S124–S127. Woods, N. N., Brooks, L. R., & Norman, G. R. (2007a). The role of biomedical knowledge in diagnosis of difficult clinical cases. Advances in Health Sciences Education: Theory and Practice, 12(4), 417–426. Woods, N. N., Brooks, L. R., & Norman, G. R. (2007b). It all make sense: Biomedical knowledge, causal connections and memory in the novice diagnostician. Advances in Health Sciences Education: Theory and Practice, 12(4), 405–415.
123
