Using Testing to Enhance Learning: A Comparison of Two Hypotheses

Using Testing to Enhance Learning: A Comparison of Two Hypotheses Michael C. Mozer Department of Computer Science & Institute of Cognitive Science University of Colorado Boulder, CO 80309 USA Michael Howe Department of Computer Science & Institute of Cognitive Science University of Colorado Boulder, CO 80309 USA Harold Pashler Department of Psychology University of California at San Diego La Jolla, CA 92093 USA this paradigm, the outcome is ambiguous (Carrier & Pashler, 1992): self testing outperforms study in some experiments (e.g. Hogan & Kintsch, 1971), but not oth- ers (e.g., McDaniel & Masson, 1985). One explanation for the inconsistency is that the rate of retrieval success on self test trials varies among experiments, and the mechanisms of learning are likely to be dependent on retrieval success. The experiments have other problems, including different amounts of time for study-only and self-test conditions, and failure to control the time spent on individual items (Carrier & Pashler, 1992). To overcome these metholological difficulties, Car- rier and Pashler (1992) compared a study-only or SO condition in which each cue-target pair was presented for ten seconds to a test/study or TS condition in which the cue was presented alone for five seconds and then the target appeared for the final five seconds. In TS tri- als, participants were supposed to use the cue to retrieve the target, but even if retrieval failed, the trial still had value due to the presentation of cue and target together for five seconds. Consequently, the dependence on retrieval success rate is minimized. Also, the paradigm matches the total time per item in SO and TS conditions. If anything, self testing is at a disadvantage because the total viewing time for cue plus target was lower. In Experiments 1 and 2 of Carrier and Pashler, 40 cue-target pairs were used, half each assigned to the SO and TS conditions. The experiment began with a study only phase in which participants viewed each of the 40 pairs once for ten seconds. Then two more passes were made through the pairs, presented in the manner desig- nated for that pair—SO or TS. For both conditions, par- ticipants were instructed to say aloud the target. In the TS condition, this instruction required that participants recall the target, or if they failed to recall, to wait until the target appeared. Following the three presentations of each pair, a final test phase evaluated performance in the two conditions via cued recall. In Experiment 1, the cues were consonant-vowel- consonant trigrams and the targets were two digit num- bers. For the sake of ecological validity, Experiment 2 used a language learning task with English language word cues and the corresponding Siberian Eskimo Yupik language translation targets. Table 1 shows the percentage of error responses. In both Experiments 1 and 2, performance was better in the TS condition than Abstract Students learning facts such as foreign language vocabu- lary often rely on a self-testing procedure in which they cue themselves with the English word and try to recall the foreign language target, instead of simply memorizing cue-target pairs. The value of this strategy has been empirically verified by a long history of research, yet existing computational models of human learning do not address the enhancing-learning-through-testing phenom- enon. Using a simple, well studied model—a feedforward neural net with no hidden units—we propose two differ- ent hypotheses for characterizing the phenomenon. Hypothesis 1 is that self-testing generates a target which is used for additional training. Hypothesis 2 is that self- testing produces a more reliable error signal for training than rote memorization. Through simulation studies, we find that hypothesis 2 readily explains the phenomenon whereas hypothesis 1 does not. Further, hypothesis 2 makes predictions worthy of further empirical study, and can be viewed as a natural consequence of temporal dif- ference learning. When learning foreign language vocabulary and other facts, students often study using index cards that have an English vocabulary word (or cue) on one side and a for- eign language vocabulary word (or target) on the other. The intuition is that by testing oneself, the associations are better learned and retained. This intuition has been supported by a long history of empirical demonstrations (e.g., Izawa, 1966; Young, 1971). For example, Bartlett and Tulving (1974) asked participants to learn a list of paired associates (the study phase), and later tested retention of the pairs using free recall or recognition (the final test). Before the final test, subjects were given a cued-recall test (a self test) of some of the paired associates. Retention was better on the final test for those items that received the self test. In this paradigm, it is unclear whether the benefit of the self test is attributable to attempting retrieval per se, or to the fact that successful retrieval of an associate also results in a re-presentation of the pair—an addi- tional training trial. An obvious strategy for examining the effect of retrieval is to conduct and experiment with, in addition to the initial study phase and the final test, an interven- ing phase in which participants are given either a self test or an experiment-provided re-presentation of the paired associate (which we’ll refer to as study only). In

[1]  W. Kintsch,et al.  Differential effects of study and test trials on long-term recognition and recall , 1971 .

[2]  James L. McClelland On the time relations of mental processes: An examination of systems of processes in cascade. , 1979 .

[3]  Alan S. Brown,et al.  Information Processing and Cognition: The Loyola Symposium , 1976 .

[4]  E. Tulving,et al.  Effects of Temporal and Semantic Encoding in Immediate Recall upon Subsequent Retrieval. , 1974 .

[5]  A G Barto,et al.  Toward a modern theory of adaptive networks: expectation and prediction. , 1981, Psychological review.

[6]  H. Pashler,et al.  The influence of retrieval on retention , 1992, Memory & cognition.

[7]  Peter Dayan,et al.  A Neural Substrate of Prediction and Reward , 1997, Science.

[8]  Lars-Göran Nilsson,et al.  Perspectives on memory research : essays in honor of Uppsala University's 500th anniversary , 1979 .

[9]  E. Custers,et al.  Psychology of Learning. , 1921 .

[10]  Richard C. Anderson,et al.  Effects of practice time within prompting and confirmation presentation procedures on paired associate learning , 1968 .

[11]  R. Duncan Luce,et al.  Individual Choice Behavior: A Theoretical Analysis , 1979 .

[12]  Mark A. McDaniel,et al.  Altering memory representations through retrieval. , 1985 .

[13]  R. Luce,et al.  Individual Choice Behavior: A Theoretical Analysis. , 1960 .

[14]  C. Izawa Reinforcement-Test Sequences in Paired-Associate Learning , 1966 .

[15]  E. R. Guthrie The psychology of learning, Rev. ed. , 1952 .

[16]  Robert A. Bjork,et al.  Optimum rehearsal patterns and name learning , 1978 .

[17]  Joseph L. Young,et al.  Reinforcement-test intervals in paired-associate learning , 1971 .