Advances in Health Sciences Education 8: 155–165, 2003. © 2003 Kluwer Academic Publishers. Printed in the Netherlands.
155
Context-Dependent Memory in a Meaningful Environment for Medical Education: In the Classroom and at the Bedside FRANCISKA KOENS∗, OLLE TH.J. TEN CATE and EUGÈNE J.F.M. CUSTERS
University Medical Center Utrecht, School of Medical Sciences, The Netherlands (∗ Author for correspondence: UMC Utrecht, Onderwijsinstituut, Stratenum 0.304, P.O. Box 85060, 3508 AB Utrecht, The Netherlands; Phone: +31 (30) 253 8347; Fax: +31 (30) 253 8200; E-mail:
[email protected]) Abstract. Purpose: Learning-in-context is a much-discussed topic in medical education. Information is said to be better recalled when the learning environment resembles the later retrieval environment. Godden and Baddeley (1975) showed that divers recalled words better when the recall condition matched the original learning environment, i.e. underwater or on land. Though it is unclear whether the findings can be generalized for medical education, medical educators regularly refer to them. We replicated the Godden and Baddeley study in ecologically more valid conditions for medical education and extended it with meaningful subject matter (namely, a patient case description). Method: Sixty-three clerks were randomized over four conditions, contrasting a clinical (bedside) with an educational (classroom) environment as both learning and recall conditions. Students were asked to recall a list of words and a patient case in the same environment or in the opposite environment as where they learned it. Results: We failed to find a significant same-context advantage for free recall of the list of words and the patient case propositions. However, there does appear to be a slight tendency towards better recall of the case description when learning took place in the clinical environment. Discussion: In medical education, the context, if conceived as physical surroundings, does not seem to contribute to a same-context advantage. One should be cautious in generalizing the findings of Godden and Baddeley. However, different forms of ‘context’ other than the physical one used in the Godden and Baddeley study may well enhance learning effects in medical education. Key words: bedside, classroom, context-dependent, free recall, learning in context, meaningful, medical education, memory, patient case
Introduction Medical curriculum developers are increasingly aware of learning-in-context, in which recall is facilitated when study matter is learned in the appropriate context (Norman and Regehr, 1996). For example, in Problem Based Learning (PBL) clinical problems are used to create an appropriate learning context for biomedical knowledge. Medical students in PBL acquire knowledge in a functional context
156
FRANCISKA KOENS ET AL.
that contains problems that closely resemble the problems they will encounter as a physician (Bridges, 1992). According to PBL-proponents, evidence for this assumption can be found in the results of the Godden and Baddeley study (1975), regularly cited in the medical educational literature (e.g., Norman and Schmidt, 1992; Schmidt, 1993; Van der Vleuten et al., 2000). Godden and Baddeley (1975) tested experienced scuba divers of a university diving club for recall of 36 common three-syllable words that were unrelated. Divers had to learn these words either on land or underwater and had to recall the words in the same environment or in the opposite environment. The authors found a significant benefit in word recall when recall took place in the same environment. Several researchers have reported this same context advantage (for a review, see Smith and Vela, 2001). However, other researchers found no significant effect or even a detrimental effect of the same context (Fernandez and Glenberg, 1985; Saufley et al., 1985; Wilhite, 1991). Although attempts have been made to explain the mixed results (e.g. Rutherford, 2000), it still remains unclear what factors have influenced the same context advantage found so convincingly by Godden and Baddeley (1975). In addition, it can be questioned whether their results can be applied to PBL curricula, where paper cases are used. This ‘classroom-type paper-and-pencil context’ may not resemble medical practice adequately (Colliver, 2000). Besides, the stimulus material in the Godden and Baddeley study has no relevance for the practice of medical education. Current medical curricula do not stress learning by heart very much. It is not known what can be expected from similar experiments, when conditions such as underwater and on land are replaced by circumstances more relevant to medical education, and when meaningful materials replace meaningless stimulus materials. We conducted an experiment to investigate both questions. Thus, the aim of the current study is to replicate the experiment of Godden and Baddeley (1975) and extend it with materials meaningful for medical education. Unlike the Godden and Baddeley study (1975), our experiment differed on three aspects. First, instead of underwater or on land, we used environments more closely related to medical education, namely the bedside as an instance of a clinical environment and the classroom as an instance of an educational environment. Second, next to the words, we used a patient case as to-be-remembered material. In addition, half of the words were common words and half of the words were biomedical words. We hypothesized that biomedical words and even more strongly, a patient case description could decrease the influence of a physical context by providing a meaningful context of itself. A third difference between the Godden and Baddeley study (1975) and our study is the use of a between-subjects design instead of a within-subjects design. In a within subjects design, as employed by Godden and Baddeley, participants might have understood the goal of the experiment. We did not want this possibility to interfere with our experiment.
CONTEXT-DEPENDENT MEMORY IN MEDICAL EDUCATION
157
Our hypothesis was that students who recall words in the same environment as where they learned it, would show better recall than students who reproduce it in a different environment. As for the patient case, we hypothesize that the meaningful text would dominate the differential effects of the physical environment and we would thus not expect a same context advantage. Method PARTICIPANTS Sixty-three medical students volunteered (i.e. 84% of the population that was approached). The experiment took place a few days after their first ten-week clerkship in internal medicine. Every two weeks, groups of six to ten students participated in our experiment. They received a small monetary reward for their participation. S TIMULI / MATERIALS The test-material consisted of a list of 20 words and a patient case about biliary pathology (gallstones). All test-materials were presented on tape and recorded by a clinical teacher, whose voice was familiar to the participants. The list of words that was constructed consisted of 10 common, unrelated and mostly three-syllable words and 10 biomedical words. The biomedical words came from different disciplines and could not be retrieved as elements of a superordinate category. The 20 words were randomly ordered, under the restriction that maximum three words from each source (common or biomedical) appeared consecutively on the list. The list of words is shown in Appendix A. The words appeared at three-second intervals. In addition, the list of words was split in two, with one half of the list being presented first for half of the participants, whereas the other half of the list was presented first for the other half of the participants. In this way, possible order effects were controlled for. All students listened to the list of words twice. The patient case (shown in Appendix B) was presented once to the participants. The case included age and gender of the patient, chief complaint, history, physical examination, lab test results and diagnosis. Before listening to the case, the diagnosis was revealed to the students to avoid making it a diagnostic task. Included in the case was collateral information not specifically relevant to the diagnosis. The duration of the 471-word patient case was 2:53 min. A pilot test was done to check feasibility and to control for possible floor and ceiling effects. E XPERIMENTAL CONDITIONS The experiment was carried out in two distinct contexts: the bedside (B) and the classroom (C). The bedside environment (B), which was used to represent a clinical context, was a small room in the Internal Medicine ward of the University
158
FRANCISKA KOENS ET AL.
Table I. The experimental conditions
Learning environment
Recall environment Bedside Classroom
Bedside
BB (N = 16)
BC (N = 15)
Classroom
CB (N = 16)
CC (N = 16)
Medical Center Utrecht where a standardized patient was present. The standardized patient matched the patient in the patient case in terms of age, sex and appearance. Relevant clinical materials such as a moveable cupboard with bandages and needles were visible. The classroom environment (C), representative for an educational context, was a small group teaching room with red floor covering, a clean blackboard and two computers in two corners of the room. In the middle, there was a large oval table with chairs around it. A neutral environment, a meeting-room in the ward with chairs and tables, was used for briefing. D ESIGN The experiment consisted of a learning phase, an interim phase and a recall phase. During the learning phase, half of the participants listened to the tape in the classroom; the other half listened to it at the bedside. Subsequently, during the interim phase, the experimenter took the students to the neutral environment in the ward. After the interim phase, the participants were assigned to either the same or the opposite environment, i.e. the classroom or the bedside. Participants were asked to recall as much as they had remembered of the list of words and the patient case. Thus, four conditions were included, which consisted of all four combinations of learning environment (classroom or bedside) and recall environment (classroom or bedside) (see Table I). P ROCEDURE Each session took place in the afternoon and lasted approximately one hour. Participants gathered in the meeting room where they were informed that they would participate in a learning experiment. Next, participants were assigned and guided to one of the two learning environments (bedside or classroom). As the classroom was slightly farther away than the bedside, the bedside group stayed in the meeting room a few minutes longer. Upon arrival in their respective learning environments participants listened to the tape. This learning phase took approximately 6 minutes. After listening to the tape, all participants returned to the neutral environment,
CONTEXT-DEPENDENT MEMORY IN MEDICAL EDUCATION
159
where they formally received their final mark for their clerkship. This was not a stressful event, since students already knew what mark they would receive. The interim phase took about 15 to 20 minutes of the experiment. Next, about half of the participants went to their initial learning environment; the other half went to the opposite environment. Participants were given blank sheets of papers and were instructed to write down all they remembered of the words and the patient case as literally as possible. There was no time limit. All students finished their work within 20 minutes. S CORING One author (FK), uninformed about the condition to which each participant was assigned, counted the number of words correctly recalled. A score form for the patient case description was developed and consisted of all elements of the case, broken down into 141 propositions (single words or short single phrases). For each correct case proposition, one point was scored. Some participants used propositions that were paraphrases. In case of doubt, an independent clinician was consulted to decide whether the paraphrase was correct. A NALYSIS The number of biomedical and common words was analyzed by a 2∗2∗2 ANOVA with learning context (bedside or classroom) and recall context (bedside or classroom) as between subject factors and type of words (common and biomedical) as a within subjects factor. The number of case propositions correctly recalled was analyzed by a 2∗2 ANOVA with learning context (bedside or classroom) and recall context (bedside or classroom) as between subjects factors. To investigate an overall effect of learning and recall context on the words and patient case, we conducted a MANOVA. The number of words and the number of case propositions were analyzed with a MANOVA with the words and the patient case as dependent variables and with learning environment and recall environment as fixed factors. Results Analysis of variance with words as the dependent variable revealed no main effect of learning environment [F(1,59) = 0.342, p = 0.561)], recall environment [F(1,59) = 0.496, p = 0.484)] and no significant interaction [F(1,59) = 0.315, p = 0.577]. Students recalled on average 10.9 words (s = 3.9) when tested in the same context, against 10.4 words (s = 2.9) when tested in the opposite context. Even though some students volunteered that they would definitely remember more biomedical than common words, they recalled on average 5.2 (s = 2.1) common words and
160
FRANCISKA KOENS ET AL.
Table II. Mean number of words recalled (standard deviation in parentheses)
Learning context
Recall context Bedside Common Biomedical
Mean
Classroom Common Biomedical
Mean
Bedside Classroom Mean
5.7 (2.3) 5.1 (1.9) 5.4 (2.1)
11.4 (3.9) 10.4 (3.1) 10.9 (3.5)
5.0 (1.8) 5.0 (2.3) 5.0 (2.1)
10.3 (2.7) 10.3 (3.9) 10.3 (3.3)
5.8 (1.9) 5.3 (2.0) 5.5 (1.9)
5.3 (2.1) 5.3 (2.2) 5.3 (2.1)
Table III. Mean number of propositions of the patient case recalled (total 141)
Learning context
Recall context Bedside [M (SD)]
Classroom [M (SD)]
Average [M (SD)]
Bedside Classroom Average
54.9 (17.0) 48.2 (12.3) 51.5 (15.0)
55.0 (11.2) 49.2 (19.4) 52.0 (16.0)
54.9 (14.3) 48.6 (16.0) 51.8 (15.4)
5.4 (s = 2.0) biomedical words (see Table II). In general, the hypothesis that there would be a same context advantage for the words could not be confirmed. Analysis of variance with the patient case as the dependent variable revealed no main effect of learning environment [F(1,59) = 2.591, p = 0.113], recall environment [F(1,59) = 0.021, p = 0.885] and no significant interaction [F(1,59) = 0.013, p = 0.911]. The mean number of correct propositions of the patient case for the four groups is shown in Table III. Participants recalled on average 51.8 (s = 15.4) propositions of the patient case description. Participants in the same context recalled on average 52.0 propositions (s = 18.2) whereas participants in the opposite contexts recalled 51.5 propositions (s = 12.1). A MANOVA with words and patient case as dependent variables and with learning environment and recall environment as fixed factors revealed no significant main effects (F = 1,288, p = 0.284 for learning environment; F = 0.301, p = 0.741 for recall environment), neither a significant interaction (F = 0.155, p = 0.856). Discussion We replicated the Godden and Baddeley (1975) study in medical educational contexts (at the bedside or in the classroom) and extended it with meaningful material, i.e., a patient case description. One of our hypothesizes was that recall of the words in the same environment as where they were learned would facil-
CONTEXT-DEPENDENT MEMORY IN MEDICAL EDUCATION
161
itate recall. Surprisingly, we were unable to replicate the findings of the Godden and Baddeley study (1975). Basically, we found no differences in word or patient case recall between participants who learned and recalled in the same environment (classroom as well as bedside) compared to participants who learned and recalled in opposite environments. Why were we unable to replicate Godden and Baddeley’s results in a more (albeit limited) ecologically valid study for medical education? First, our study was designed like the Godden and Baddeley study, except that we did not use a within subjects design. Since Godden and Baddeley’s results were so convincingly significant, we thought that it would be able to replicate the same context advantage even when a between subjects design was used. However, participants in the Godden and Baddeley study (1975) experienced all four combinations of the learning and recall environment, so they may have noticed the aim of the experiment and therefore, this may have influenced their scores. In the Godden and Baddeley (1975) study itself, a second experiment in which one of the conditions was replicated with or without disruption, revealed a substantially less impressive same context advantage than did the first experiment. A second, more practically oriented explanation could be that our participants did not receive instructions how to memorize the words and the patient case. This may have lead to the use of participants’ own encoding strategies, and may have ‘overshadowed’ effects due to context (Davies and Thomson, 1998, p. 337). One of the strategies our participants used was closing their eyes to concentrate on the tape recording. This could have resulted in a performance improvement (Glenberg et al., 1998) and a decrease of environmental influence. Another possible reason for the lack of evidence for same context advantage in our study is that in the Godden and Baddeley study (1975), physiological conditions may have added to the effect. Divers experience physiological changes underwater, like being weightless and being restricted of sight. Even when participants in the Godden and Baddeley study closed their eyes as an encoding strategy, these physiological differences in the context may have contributed to environmental differences. Miles and Hardman (1998) found that students, who recalled the words in the same physiological state (i.e., with regulated heartbeats) as where they learned them, performed significantly better than students who experienced opposite states. So, perhaps our contexts did not differentially influence physiological states to the same extent as the contexts in the Godden and Baddeley study. A fourth explanation is that much same-context advantage research has been conducted with participants who were tested individually or in pairs. Perhaps this is a factor that could contribute to a contextual effect. The group size (three to five participants) and composition may have influenced the context and diminished differences between the classroom and the bedside. Group testing may, however, reflect the natural environment of the bedside and the classroom better than individual testing.
162
FRANCISKA KOENS ET AL.
Even in spite of these differences, we expected to find at least a minor same context advantage. However, our results were so far from approaching significance, we can not but conclude that no such effect ws present. As for learning, learning may be not as dependent as we thought. This may be reassuring, because there is a trade off between context dependence and transfer. A strong dependency of context may generate problems of knowledge transfer. If the learning of subject matter can only take place in a specific context, knowledge will be inflexible and transfer limited (Anderson, 1995). PBL-based curricula are supposed to enhance the application of basic science concepts to clinical problems. Students tend to ‘forget’ much of the knowledge they have acquired in preclinical years, even if this knowledge is acquired by means of solving paper cases (Prince et al., 2000), because they are not directly linked to patient care. Some context elements do seem to make a difference. Although we tried to create learning and retrieval conditions that resemble ecologically valid circumstances for medical education, our contexts may not have been entirely representative for medical education. The clinical environment would have been still more realistic if the clerks had been asked to take a history and perform a physical examination, and thus have personal responsibility towards the patient. Our patient was a standardized patient, merely present as part of the context. In real life, clerks may be more motivated; this might be an additional factor in context dependent memory. In addition, clinical reasoning towards solving problems would be more ecologically valid. However, our aim basically was to replicate the Godden and Baddeley study; therefore we deliberately revealed the diagnostic ‘solution’ before the stimulus material was presented. Moreover, participants probably recall information better when instructed to memorize information, instead of solving a problem (Needham and Begg, 1991). The concept of context dependency should, in our view, be further elaborated. From our experiment so far we conclude that context, if restricted to differences in physical surroundings, have limited significance for learning. However, we did find a slight difference in the two learning contexts. Clerks who learned at the bedside tended to recall more patient case propositions than those who learned in the classroom regardless the context in which they were tested. This trend could not be seen with word recall. Perhaps the learning environment can provide some context elements that are relevant for meaningful material but not for unrelated words. If we assume that the learning environment plays a role in context-dependent memory, it is important to notice that it will probably not impair recall for meaningful material, or in other words, that it won’t interfere with transfer of knowledge. We tend to believe, however, that other forms of context may have significant impact on learning. Therefore, further specifications of context, relevant to medical education should be made. Apart from a physical dimension, we should at least distinguish a semantic dimension (e.g., basic science information, presented in conjunction with clinical problems). Also, a dimension related to commit-
CONTEXT-DEPENDENT MEMORY IN MEDICAL EDUCATION
163
ment (learning to solve a problem, because patient care requires a solution to an imminent problem, or a patient presentation is requested), and a definition related to student activities (e.g., ward based activities) could be identified. Since context-characteristics influence learning and memory by interacting with learning processes, with commitment possibly playing an important role, we should explore and understand the nature of these influences. Further research should focus on different characteristics of context and manipulate even more real life situations than we have done in our experiments. Acknowledgements The authors want to thank Jeen Haalboom, Bernard Zonnenberg, Alexander de Graeff and Anneke Tellegen, as well as the 63 students whose data were collected, for their cooperation and participation. References Anderson, J.R. (1995). Cognitive Psychology and Its Implications. New York: W.H. Freeman & Company. Bridges, E.M. and Hallinger, P. (1992). Problem-Based Learning for Administrators. Eugene, OR: ERIC Clearinghouse on Educational Management, University of Oregon. Colliver, J.A. (2000). Effectiveness of problem-based learning curricula: research and theory. Academic Medicine 75(3): 259–266. Davies, G.M. & Thomson, D.M. (1988). Context in context. In G.M. Davies & D.M. Thomson (eds.), Memory in Context: Context in Memory, 335–345. New York, NY: John Wiley & Sons. Fernandez, A. & Glenberg, A.M. (1985). Changing environmental context does not reliably affect memory. Memory and Cognition 13(4): 333–345. Glenberg, A.M., Schroeder, J.L. & Robertson, D.A. (1998). Averting the gaze disengages the environment and facilitates remembering. Memory and Cognition 26(4): 651–658. Godden, D.R. & Baddeley, A.D. (1975). Context-dependent memory in two natural environments: on land and underwater. British Journal of Psychology 66(3): 325–331. Miles, C. & Hardman, E. (1998). State-dependent memory produced by aerobic exercise. Ergonomics 41(1): 20–28. 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. & Regehr, G. (1996). Issues in cognitive psychology: implications for professional education. Academic Medicine 71(9): 988–1001. Norman, G.R. & Schmidt, H.G. (1992). The psychological basis of problem-based learning: a review of the evidence. Academic Medicine 67(9): 557–565. Prince, K.J.A.H., Van de Wiel, M.W.J., Scherpbier, A.J.J.A., Van der Vleuten, C.P.M. & Boshuizen, H.P.A. (2000). A qualitative analysis of the transition from theory to practice in undergraduate training in a PBL-medical school. Advances in Health Sciences Education 5: 105–116. Rutherford, A. (2000). The ability of familiarity, disruption, and the relative strength of nonenvironmental context cues to explain unreliable environmental-context-dependent memory effects in free recall. Memory and Cognition 28(8): 1419–1428. Saufley Jr, W.H., Otaka, S.R. & Bavaresco, J.L. (1985). Context effects: classroom tests and context independence. Memory and Cognition 13(6): 522–528.
164
FRANCISKA KOENS ET AL.
Schmidt, H.G. (1993). Foundations of problem-based learning: some explanatory notes. Medical Education 27(5): 422–432. Smith, S.M. & Vela, E. (2001). Environmental context-dependent memory: a review and metaanalysis. Psychonomic Bulletin and Review 8(2): 203–220. Van der Vleuten, C.P.M., Dolmans, D.H.J.M. & Scherpbier, A.J.J.A. (2000). The need for evidence in education. Medical Teacher 22(3): 246–250. Wilhite, S.C. (1991). Evidence of a negative environmental reinstatement effect. British Journal of Psychology, 82: 325–342.
Appendix A – List of words tablecloth lymphocyte interferon candidate agonist ribosome myeline photographer oncogene adventure dissolvable mitosis stapler synapse annoyance cytoplasm continent porcelain pylorus centre
Appendix B: Patient case gallstones A woman, age 58 years, shows up at the outpatient department complaining of pain in her upper abdomen. She describes the pain, which has been present for approximately six months, as nagging. It is localized in her right upper abdomen. The pain increases upon eating fat food. Sometimes, she experiences attacks of a sharp pain, which she feels to move toward her back before disappearing. Inbetween these attacks, her upper abdomen remains tender. Her feces are occasionally a bit discolored, while her urine appears sometimes a little darker than usual. As a consequence of the pain, she has become somewhat reluctant to eat and has lost 7 kilograms of weight over the last two months. When explicitly asked, she also confirms that the frequency of defecation has differed considerably, that the feces are often hard and that sometimes she fails to defecate for some days. She has no problems with swallowing food, except that she has experienced heartburns a couple of times. The systems review reveals that she has had a contusio cerebri as a child; besides, she has never had problems of a neurological or psychological nature. Her vision has slightly deteriorated during the last couple of years. She has smoked moderately for thirty years, but has quit after the death of her husband in 1995. She is not short of breath and has no problems in performing daily activities, such as shopping. In 1996, she has been admitted to the hospital because it was suspected that she had had a heart attack, though at the time this could not be confirmed. Her blood pressure is average.
CONTEXT-DEPENDENT MEMORY IN MEDICAL EDUCATION
165
However, she has to go to the washroom twice every night and prefers to use an extra cushion at night. Concerning the other systems, she does not have any complaints, nor has she noted any abnormalities. A survey of family problems reveals the presence of gallstones in her relatives: the patient’s mother has undergone surgery for gallstones. Physical examination shows a moderately obese woman, height 1 meter 68, weight 84 kilograms. She has no jaundice and definitely does not appear to be ill. Her pulse is 68 (regular equal) and her blood pressure is 145 over 85. Examination of head, cervix, heart and lungs does not reveal any abnormalities. Her adipose abdomen is pliable; during auscultation, normal, albeit sparse, peristaltic sounds are heard. Percussion shows changing tympany and upon palpation, the patient reports tenderness over the whole upper abdomen. There is a small diastasis of the rectus abdominus; tenderness during pressure is relieved by tightening the abdominal muscles. Liver and spleen are not palpable. Rectal and vaginal examination, as well as examination of the extremities, do not reveal any abnormalities. Preliminary neurological investigation fails to show the Achilles tendon reflex and also reveals a slightly diminished sensibility of the right foot. Additional laboratory investigation shows: Hb 8,6; hematocrit 0,43; leukocytes 4,8; creatinine 79; bilirubin 12; amylase 87; sodium 143; potassium 4,5; ASAT 23; ALAT 45; alcalic phosphatase 78; the protein spectrum shows a small band in the gamma area; the CRP is 12. Ultrasound scan of the abdomen shows no defects of kidneys, pancreas or liver. The wall of the gall bladder has thickened and multiple, relatively large calcified stones are seen. Gastroscopy only shows a minor, grade 1 reflux-esophagitis and a small hiatus hernia.
