The spacing effect in young children’s free recall: Support for automatic-process explanations

Memory & Cognition, Mar 1991

The effect of spacing repetitions on children’s free recall was investigated in two experiments. In Experiment 1, both 4-year-old children and 7-year-old children exhibited a spacing effect in free recall, and the magnitude of the effect did not change with age. In Experiment 2, free recall was examined as a function of spacing, age (3 years old vs. 4 years old) and presentation rate (1 vs. 2 vs. 5 sec per stimulus). A spacing effect was obtained that did not differ as a function of age or presentation rate. Of particular interest was the fact that 3-year-olds exhibited a strong spacing effect even when stimuli were presented at a very rapid 1-sec rate. The results support the hypothesis that fundamental memory mechanisms that operate relatively automatically are sufficient to produce a spacing effect in free recall.

A PDF file should load here. If you do not see its contents the file may be temporarily unavailable at the journal website or you do not have a PDF plug-in installed and enabled in your browser.

Alternatively, you can download the file locally and open with any standalone PDF reader:

http://link.springer.com/content/pdf/10.3758%2FBF03197112.pdf

The spacing effect in young children’s free recall: Support for automatic-process explanations

THOMAS C. TOPPINO 0 1 0 This research was supported by Grant HD21209 from the National Institute of Child Health and Human Development. I thank Pamela Blewitt , Ronald Fisher, Robert Greene, and Barry Stein for their help ful comments on an earlier draft of this paper. I also thank Kristine Krajnak, Anna PhilIips, and Maria Schwendinger for assisting with the preparation of materials, datacoUection andIor dataanalyses. Correspon dence should be addressed to Thomas C. Toppino, Department of Psy chology, Villanova University , Villanova, PA 19085 1 Villanova University , Villanova, Pennsylvania The effect of spacing repetitions on children's free recall was investigated in two experiments. In Experiment 1, both 4-year-old children and 7-year-old children exhibited a spacing effect in free recall, and the magnitude ofthe effect did not change with age. In Experiment 2, free recall was examined as a function of spacing, age (3 years old vs. 4 years old) and presentation rate (l vs. 2 VS. 5 sec per stimulus). A spacing effect was obtained that did not differ as a function of age or presentation rate. Ofparticular interest was the fact that 3-year-olds exhibited a strong spacing effect even when stimuli were presented at a very rapid I-sec rate. The results support the hypothesis that fundamental memory mechanisms that operate relatively automatically are sufficient to produce a spacing effect in free recall. - A widely accepted generalization is that repetition im proves memory, but the effeetiveness of repeated exposure to infonnation depends on the spacing between repetitions. When repetitions are massed (presented in immediate suc cession), their beneficial effect is often minimal. How ever, memory for repeated infonnation typically improves with increases in the number of other events or items of information separating repetitions. This spacing effect is of interest in its own right and assumes additional importance because understanding it should provide insight into the more far-reaching ques tion of why repetition facilitates memory in the first place. Considerable research over the last two decades has indicated that the spacing effect is unusually gen eral and robust. It has been obtained in virtually every standard memory task from free recall to recognition (Hintzman, 1974). It has been demonstrated with a wide variety of materials, both verbal (Kraft & Jenkins, 1981) and nonverbal (Hintzman & Rogers, 1973). It occurs with intentional memory procedures (Underwood, 1969) and with many incidental leaming procedures (e.g., Glenberg & Srnith, 1981; McFarland, Rhodes, & Frey, 1979). It seems to be relatively unaffected by variation in task parameters such as rate and modality of stimulus presentation (Melton, 1970), and attempts to elirninate or attenuate the spacing effect often have been unsuccessful (Bird, Nicholson, & Ringer, 1978; Hintzman, Summers, Eki, & Moore, 1975; Jensen & Freund, 1981; Maskarinec & Thompson, 1976; Shaugh nessy, 1976). Most hypotheses about the mechanism(s) responsible for the spacing effect fall into one oftwo broad categories: encoding-variability theories and deficient-processing the ories. According to encoding-variability theories (e.g., Bower, 1972; Glenberg, 1979; McFarlandet al., 1979), repetition improves memory to the extent that a repeated item is encoded differently on each presentation. The spac ing between repetitions is assumed to influence the degree to which repeated items are encoded differently on each occurrence, with longer spacings leading to more exten sive differential encoding and better memory. According to deficient-processing theories (e.g., Jacoby, 1978; Shaughnessy, Zimmerman, & Underwood, 1972), sub jects do not fully process both occurrences of a repeated item when the spacing between repetitions is very short. Consequently, memory for the item suffers. As the spac ing between repetitions increases, however, subjects be come more likely to process both occurrences fully, and memory performance improves eorrespondingly. One issue that euts aeross theories of the spaeing effeet concems whether the effeet is produced by voluntary (stra tegic) or by involuntary (automatically activated) processes. Some encoding-variability theorists have pro posed that the phenomenon is the result of sophisticated mnemonic strategies. For example, organizational strate gies have been hypothesized to underlie the spacing ef feet in free recall (e.g., Glenberg, 1977). That is, the spac ing effeet is assumed to occur because, in comparison with massed repetitions, subjects are more likely to organize each occurrence of a spaced repetition into a different sub jective unit, thereby increasing the likelihood of success ful retrieval during free recall. Sophisticated strategie processes also have been assigned a central role in some deficient-processing theories. For example, it has been proposed that one presentation of a repeated item receives less rehearsal when the repetitions are massed than when they are spaced. Thus, re1ative1y poor memory for massed repetitions is attributed to 1ess total rehearsal (e.g., Rundus, 1971). According to another deficient-processing hypothesis, subjects allocate 1essattention (study time and resources) to the second occurrence of massed repetitions than they do to the second occurrence of spaced repeti tions (e.g., Shaughnessy et al., 1972). In one version of this theory (Zechmeister & Shaughnessy, 1980), subjects are assumed to adjust the al1ocation of study resources in accordance with metarnemory judgments about how well items are 1eamed. In any event, however, the spac ing effect is assumed to occur because subjects choose to study (attend to) massed repetitions 1ess than they do spaced repetitions. In contrast with the strategy-based hypotheses, some researchers have noted the uncommon generality and per sistence of the spacing effect and have proposed that the phenomenon reflects the operation of automatic memory processes (e.g., Jensen & Freund, 1981) that may be so fundamental that they are hard-wired into the memory system. With respect to encoding-variability theories, for example, Glenberg (1979) proposed that automatie, encoding-variability processes are sufficient to produce the spacing effect in free recall. During study, contex tual information is assumed to be encoded automatically, with more varied contextual information being encoded with spaced, as opposed to massed, repetitions. Thus, spaced repetitions are assumed to have an advantage over massed repetitions in free recall, because retrieval in free recall depends to a great degree on contextual cues. A similar hypothesis with respect to free recall was offered by Greene (1989) as part of a two-process account of the spacing effect. Some deficient-processing theories also offer autornatic process explanations. For example, some investigators have proposed that short spacings between repetitions in terfere with ongoing consolidation processes (e.g., Lan dauer, 1969; Peterson, 1966). Others have proposed that, when an item is encoded, astate of habituation automati cally ensues that prevents proper encoding of the same item until recovery is complete (e.g., Hintzman, 1974). The issue of voluntary versus involuntary processes has implications for the developmental course of the spacing effect. Research on memory development indicates that very young (preschool) children do not voluntarily em ploy the kinds of sophisticated strategies that have been proposed to underlie the spacing effect (e.g., Appel et al., 1972; Myers & Perlmutter, 1978; Paris, 1978). In fact, for the kinds of tasks involved in spaced-repetition research, even when the precursors of strategies have been found in preschool children, they have proved to be mne monically ineffective (Baker-Ward, Omstein, & Holden, 1984; Omstein, Baker-Ward, & Naus, 1988). Thus, if voluntary strategie processing is necessary to produce the spacing effect, very young children should not exhibit the effect. Instead, the spacing effect should emerge with the development of children's abilities to use such strategies. However, if relatively primitive memory mechanisms that operate automatically are sufficient to produce the spac ing effect, even very young children should manifest the effect. The research reported here addresses this issue in the context of free recall. Two previous studies have exarnined the effect of spac ing repetitions on the free-recall performance of young children (Rea & Modigliani, 1987; Toppino & DiGeorge, 1984). In both studies, a spacing effect was obtained with elementary-school children. However, the studies produced conflicting findings with preschool children. Rea and Modigliani obtained a spacing effect in 4-year-old preschoo1 children, whereas Toppino and DiGeorge did not. The source of the discrepant findings is not obvious. Comparing Toppino and DiGeorge's procedures with the procedures of Rea and Modigliani's first experiment, which was methodologieally most similar, revea1s numer ous procedural differences. Among the most salient are the following: First, in Toppino and DiGeorge's studies, 1ists were composed of either 22 positions (Experiment 1), including positions for three massed- and three spaced repetition items, or 27 positions (Experiment 2), includ ing positions for four massed- and four spaced-repetition items. In contrast, Rea and Modigliani's 1ists were com posed of only 16 positions, inc1uding positions for two repeated items in each of three spacing conditions (tags of 0, 1, and 3 intervening items). Second, Toppino and DiGeorge presented stimuli manual1y, whereas Rea and Modig1iani controlled stimulus presentation and timing electronical1y. Third, Toppino and DiGeorge presented children with color pictures taken from picture books, and, as each pieture was presented, each child supplied his or her own label, pronouncing it aloud. In contrast, Rea and Modigliani presented b1ack-and-white line drawings in two of their four lists and typed words in the other two lists. In both picture lists and word lists, each visual stimulus was accompanied by its auditorily presented label, and children simp1y repeated this experimenter-supplied label aloud. Of course, the discrepant findings might be explainable in terrns of population differences or sampling error rather than in terms of methodological differences. Therefore, it was decided that, instead of attempting to systemati cally sort out the critica1 differences between the two methods, the first step shou1dbe to reexamine the effects of spacing repetitions by using the essentials of Toppino and DiGeorge's methodology. This was accomplished as part of Experiment 1, which was a factorial exarnination of grade level (preschoo1 vs. second grade), spacing of repetitions (massed vs. spaced), and frequency ofrepeti tions (two vs. three presentations). When repeated items occurred twice, lists were very similar to those used by Toppino and DiGeorge. In addi tion, the procedures of both experiments reported here were very similar to those ofToppino and DiGeorge, with the exception of one potentially important improvement. Toppino and DiGeorge had presented pictures to children manual1y. This technique is often used in developmental research with young children, because it is thought that the social interaction involved helps to keep children's at tention focused on the task. However, the procedure rnay also introduce other sources of variability. Therefore, in the present experiments, a procedure was used that was very successful in maintaining children's attention but that, at the same time, allowed stimulus presentation and tim ing to be controlled electronically. Finally, frequency of repetition was examined in Ex periment I, because it was thought that the magnitude of the spacing effect rnight increase as a function of this vari able (Underwood, 1969). Thus, if the spacing effect is relatively weak in young children when repeated items occur twice (as suggested by Toppino and DiGeorge's failure to find an effect of spacing), the effect rnight be stronger and more easily detected with a greater number of presentations. EXPERIMENT 1 Method Subjects and Design. Subjeets were 24 preschool children (mean age = 53.3 months, age range 49-58 months) and 24 secend-grade children (mean age = 92.3 months, age range 86-105 months) who attended nursery school and elementary school, respectively, in suburban Philadelphia. Data were collected from an additional 6 preschoolers, but these subjects were eliminated from the analyses in order to equate the number of subjects at each grade level. Within each grade level, 12 children were randomly assigned in blocks of 2 to each of two groups that differed with respect to the order in which they studied and recalled a list containing twice presented items and a list containing thrice-presented items (2P-3P vs. 3P-2P). The design of the experiment was a 2 (grade) x 2 (order) x 3 (repetitionJspacing condition) x 2 (type oflist: 2P vs. 3P) mixed factorial, with the last two factors varied within-subjects. Levels of the repetitionJspacing factor were: once-presented items, repeated items with massed presentations (MP), and repeated items with spaced or distributed presentations (D P). Materials. Four list structures were generated-two 23-position structures for lists of twice-presented repeated items and two 29 position structures for lists of thrice-presented repeated items. Each structure began and ended with three slots for once-presented items serving as primacy and recency buffers, respectively. The middle portion of the list structure contained slots for two once-presented fillers and for three items representing each of the levels of the repe titionJspacing variable (once-presented, MP, and DP items). The presentations of DP items were separated by lags of 3, 4, or 5 in tervening items, with each lag used at least once in each list struc ture. Within each list structure, the mean serial positions of the fi nal occurrences of items in the MP and DP conditions were equated with one another and with the mean serial position of the critical once-presented items. Each subject received one list containing twice-presented items (2P list) and another list containing thrice-presented items (3P list). With respect to list structures, two combinations of 2P and 3P lists were employed, such that a 2P list containing the first 2P list struc ture was always paired with a 3P list containing the first 3P list structure, and a 2P list containing the second 2P list structure was always paired with a 3P list containing the second 3P list struc ture. Half of the subjects in each group received one of these pair ings; the remaining half of the subjects in each group received the other pairing . Stimuli presented in the lists were colored pictures of common objects taken from children's picture collections. Two sets of 17 pietures were selected so that each pieture would be familiar to young children but would bear no strong or obvious semantic relation ships to other pictures in the same set. Items within each set of 17 pictures were randomly assigned to serve specific functions (e.g., critical items, buffers, fillers). Then, pietures in one set of items were assigned to positions in one of the list structures of twice presented repetitions and to positions in one of the list structures of thrice-presented repetitions. Pictures in the other set of 17 items were assigned to positions in the remaining two list structures. For each of the four originallists constructed in this fashion, two new Iists were generated in which the critical items were reassigned to repetitionJspacing conditions within the same list structure accord ing to a Latin-square principle. This allowed the assignment of lists to children to be done in such a way that, both between and within order groups, the same items served equally often in each repeti tionJspacing condition. To prepare lists for presentation, the picture stimuli were pre pared as slides, and electronically timed slide sequences were video taped with a Panasonie WV3260 color video camera. Proceclure. The children participated individually in the experi ment, which was presented as a "memory game." They sat before a 9-in. color-television monitor as Iists were presented by videotape. After the children studied and recalled a short practice list, the first experimental list was presented. One picture was presented every 5 sec with a .S-sec interstimulus interval, and the children labeled each picture as it appeared. Following list presentation, the children were allowed 2 min for oral free recall. They were credited with correctly recalling an item if they provided, during the recall test, either the item 's conventional label or an idiosyncratic label that they had given previously during list presentation. When the recall period was over, there was a 2-min rest period during which the children worked on a simple puzzle. Then, the procedure was repeated with the second experimental list. Results and Discussion Table 1 presents the mean percentage of correct free recall as a function of grade level, type of list, and repe tition/spacing condition. The repetition/spacing factor is actually a combination of two logically separable vari ables: repetition (once-presented vs. repeated) and the spacing between repetitions (MP vs. DP). Therefore, the effects of these two variables were examined in separate analyses. Note-IP = once-presented items, MP = rnassed presentations, DP = distributed presentations, 2P lists = Iists containing twice-presented items, 3P lists = lists containing thrice-presented items. The first analysis assessed the effect of repetition per se on the percentage of items correctly recalled, with the use of a 2 X 2 X 2 X 2 (grade X order X repetition X type of list) ANOV A with repeated measures on the last two factors. Neither the main effect of order nor any interactions involving this variable were significant [all Fs(1,44) S 1.06]. The repetition variable compared the mean percentage of critical once-presented items reealled with the mean percentage of repeated items recalled, in cluding both MP and DP items. The effect of this factor was significant [F(1,44) = 39.69, MS. = 495.72, p < .001], indicating that recall was better for repeated items than it was for once-presented items. Recall was better in lists involving thrice-presented repetitions than it was in lists involving twice-presented repetitions [F(1,44) = 5.68, MS. = 265.77, P < .025]. Although this difference was primarily due to recall of repeated as opposed to once-presented items in these lists, the repe tition X type-of-list interaction did not reach conventional levels of significance [F(1,44) = 2.77, MS. = 528.47, p = .10]. Finally, recall was better for second graders than it was for preschoolers [F(1,44) = 7.46, MS. = 704.77, p < .0 1], and there was a grade X repetition interaction [F(1,44) = 8.00, MS. = 495.72, p < .01], indicating that second graders benefited more from repe tition than did preschoolers. No other interactions were significant [all Fs(1,44) s 1.10]. The effeet of spacing repetitions was examined in a see ond 2 X 2 X 2 X 2 (grade X order X spacing X type of list) ANOVA with repeated measures on the last two factors. Again, neither the main effect of order nor any of the interactions involving order approached significance [all Fs(1,44) s 1.09]. A main effect of list type (which is equivalent to frequency of repetition in this case) indi cated that recall was better for 3P items than it was for 2P items [F(I,44) = 7.90, MS. = 753.06, P < .01]. However, list type was not systematically related to any other variable, as refleeted in the fact that no interactions involving this factor even approached significance (Fs .s 1.12). Second graders recalled better than preschoolers [F(1,44) = 18.48, MS. = 993.21, P < .001], but the most important finding was a significant spacing effect [F(I,44) = 5.01, MS. = 788.94, P < .05], with DP items being recalled better than MP items. Neither grade level nor spacing interacted significantly with any other variable (all Fs s 1.12). Thus, the results ofthis experi ment indicate that young children exhibit a spacing ef feet that undergoes no significant change between the ages of 4 and 7 years old. EXPERIMENT 2 Experiment 1 included conditions that were method ologically very similar to those of Toppino and DiGeorge (1984), whose results were consistent with the hypothe sis that the spacing effeet is produced by strategies that emerge with development. However, the results of Ex periment I, like those of Rea and Modigliani (1987), did not support this hypothesis. Preschoolers exhibited a spac ing effeet, and the effect did not change with develop ment between the preschool and early elementary-school years. Although the preschoolers in all of the above experi ments were approximately the same chronological age (i.e., 4 years old), one cannot rule out the possibility that the subjeets in Experiment I and in Rea and Modigliani' s study were cognitively more advanced in their develop ment than were the children sampled by Toppino and DiGeorge. That is, the former studies may have failed to deteet the developmental change predicted by the strategy hypothesis because their 4-year-old subjeets may have been so advanced that they had already undergone the predicted developmental changes in cognitive func tioning. This possibility would be greatly reduced, however, if even younger children could be shown to ex hibit a spacing effect in free recall. Therefore, in Experi ment 2, the age of preschoolers was varied to include both 3- and 4-year-old children. We further examined the issue of strategic versus au tomatic processes by varying the presentation rate (I, 2, or 5 sec per picture). Voluntary processes require time to exeeute. Some investigators (e.g., Stoff & Eagle, 1971; Wilkinson & Koestler, 1983) have suggested that al-sec presentation rate may be too fast for adults to implement a voluntary strategy. This seems especially likely in the case of children as young as 3 years old. Thus, ifthe spac ing effect in preschool children is, by any chance, the result of some undetected voluntary strategy, the spacing effect should be evident when presentation rates are rela tively slow (e.g., 5 sec) but should be eliminated with a I-sec presentation rate. In contrast, if fundamental, au tomatic, memory processes are sufficient to produce the spacing effect in free recall, a spacing effect should be evident even when to-be-remembered stimuli are pre sented at a I-see rate. Method Materials and Procedures. Three 22-position list structures were generated. These began and ended with slots for three primacy buffers and three recency buffers, respectively. The middle region of the structures contained slots for I once-presented filler item and for 3 items representing each ofthe levels ofthe repetition/spacing variable (once-presented, MP, and DP items). All repeated items were presented twice, and presentations of DP items were sepa rated by a lag of 3 intervening items. The serial position of critical items within the list structure was controlled, as in Experiment I. Stimuli consisted of either 48 color pictures of common objects taken from children's picture collections or 48 black-ink drawings on a white background taken from a set of pictures published by Snodgrass and Vanderwart (1980). Each black-and-white picture corresponded to one of the color pictures in thatit depieted the same type of object having the same label (e.g., another hammer, chair, dog, etc.). A given list contained only one type of stimulus, and two sets of lists were constructed that were identical, except for the type of stimuli that were employed in each. The sets of 48 pictures comprised three subsets of 16 such that the depicted objects had no strong or obvious semantic relation ships to other members ofthe same subset. Each subset of 16 items was assigned to a different one ofthe three list structures. Following the procedures outlined in Experiment I, stimuli within each subset of items were assigned functions (e.g., MP items) and then were assigned to slots in the designated list structure. By using a Latin-square principle, we generated two more lists with each itern-subset/list-structure combination such that, across the three lists involving a given list structure, the same stimulus items served equally often in each ofthe three repetition/spacing conditions. By this procedure, we generated nine lists of color pictures and a cor responding nine lists of black-and-white pictures. Materials were prepared for presentation by videotaping aseries of electronically timed slides, as in Experiment I. The presenta tion rates of I, 2, and 5 sec refer to the interval from the onset of one stimulus to the onset of the next. However, there was a .8 sec blank interval between the offset of one stimulus and the onset of its successor. The subjects received three successive lists with one list presented at each of the three presentation rates. Therefore, for each type of stimulus, we used a complex counterbalancing scheme to generate and record 27 sequences of three lists so that each sequence inc1uded one list of each of the three different subsets of 16 pictures and one list presented at each of the three different presentation rates (1,2, or 5 sec per item). Across all 27 sequences, each of the nine basic lists (see above) occurred equally often at each of the three presentation rates and was presented at each of these presentation rates equally often as the first, second, and third list in a sequence. Procedures were the same as in Experiment I, except that three, rather than two, successive lists were presented and recalled. Subjects and Design. Subjects were 54 3-year~ld children (rnean age = 42.7 months, age range = 36-47 months) and 54 4-year old children (rnean age = 53.9 rnonths, age range = 48-59 months) who attended nursery school in suburban Philadelphia. Within each age group, each of the 27 list sequences of color stimuli and each of the 27 list sequences of black-and-white stimuli were assigned randomly to one child. The overall design can be conceptualized as a 2 (age) x 2 (type of stimulus) x 3 (presentation-rate order) x 3 (presentation rate) x 3 (repetition/spacing) mixed factorial, with the last two factors varied within subjects. The presentation rare-erder variable refers to the presentation rate of successive lists in the sequence of three lists presented to a subject (1, 2, 5 vs. 5, I, 2 vs. 2, 5, I). Neither this variable nor the type-of-stirnulus vari able was of theoretical interest. The former was inc1uded for pur poses of control; the latter was inc1uded to increase generality. Results and Discussion Table 2 presents the mean percentage of correct free recall as a function of age, presentation rate, and repeti tion/spacing condition. As in the previous experiment, the results were analyzed by two ANOVAs. The first ex arnined the effect of repetition per se in a 2 x 2 x 3 x 3 x 2 (age x stimulus type X presentation-rate order x presentation rate x repetition) mixed ANOVA with repeated measures on the last two factors. The second ANOVA, which exarnined the effect of the spacing be tween repeated items, was identical to the first except that the repetition factor (once-presented vs. twice-presented items) was replaced by spacing (MP vs. DP items). The first ANOVA indicated that recall was better for twice-presented iterns than it was for once-presented iterns [F(l,96) = 23.73, MSe = 492.61, p < .001], and this effect of repetition did not interact significantly with any other variable. Other effects paralleled those obtained with the second ANOVA, to which we now turn. As expected, the percentage of repeated items recalled was greater for 4-year-olds than it was for 3-year-olds [F(l,96) = 6.87, MSe = 99O.21,p = .01]. Also, the chil dren's performance did not improve systematically with presentation rate. The main effect of presentation rate was not reliable [F(2, 192) = 2.08, MSe = 562.63, p > .10], with the mean percentage of recall equal to 24.5, 29.2, and 27.3 for the 1-, 2-, and 5-sec presentation rates, respectively. This finding is consistent with the assump tion that, at this age, children do not employ effective mnemonic strategies. However, two interactions must be mentioned, even though they do not change the interpretation of the presentation-rate data. First, there was a significant presentation rate X presentation-rate order interaction [F(4,192) = 5.17, MSe = 562.63, p < .001], which sirn ply reflected the fact that recall performance declined across the three successive lists given to each child. Thus, the relative performance among the three presentation rates depended on the order in which lists involving the different presentation rates were presented. Second, there was a marginally reliable interaction between presenta tion rate and type of stimulus [F(2,192) = 2.96, MSe = 562.63, p = .054]. The mean percentages of recall for the 1-, 2-, and 5-sec presentation rates were, respectively, 27.5,26.5, and 27.5 for color stimuli, and 21.6,31.8, and 27.2 for black-and-white stimuli. Thus, the marginal interaction seems to stern from relatively poor perfor mance when children encountered black-and-white stimuli at al-sec presentation rate. However, the effect of presen tation rate with black-and-white stimuli was not reliable in a probe of simple main effects [F(2, 192) = 2.50, MSe = 562.63, p > .05]. There are other reasons, as weIl, for believing that the interaction was spurious. Although no higher order in teraction approached significance, poor recall with black and-white stimuli presented at a I-sec rate was traceable to the fact that, in this condition, 4-year-olds performed unusually poorly on spaced repetitions. For no discern ible reason, 4-year-olds performed more poorly than did 3-year-olds in this one condition, whereas the opposite pattern was obtained in every other one of the 12 conditions generated from two types of stimuli, three presen tation rates, and two levels of spacing. Finally, if the in crease in performance from 1 to 2 sec per pieture indicated that children used the extra time to engage in additional mnemonie processing, one should expect a fur ther increase in performance at even slower rates. One should also expect the strategy to be employed with color, as well as black-and-white, stimuli. However, neither of these expectations was supported by the data.' Tuming to the most important findings revealed by the second ANOV A, there was a significant main effect of spacing [F(1,96) = 12.67, MSe = 802.50, p < .001] such that children recalled DP items better than they did MP items. Moreover, the spacing between repetitions did not interact significantly with age [F(1,96) < 1.00], presentation rate [F(2,192) < 1.00], or any other vari able or combination of variables (all Fs s 1.37). Of particular interest was whether or not the youngest children (3-year-olds) would exhibit a spacing effect with al-sec presentation rate. It was thought that this rate was fast enough to preclude virtually any undetected strategie processing in which these children might engage. A separate analysis of 3-year-olds' performance with the 1 sec presentation rate showed that the spacing effect in these children was significant [F(1,53) = 7.76, MSe = 530.34, P < .01]. GENERAL DISCUSSION Preschool children, who are unsophisticated with respect to mnemonic strategies, exhibited a spacing ef feet in both of the experiments reported here. In Experi ment 1, 4-year-olds manifested a spacing effect that did not differ from the spacing effect produced by second graders. In Experiment 2, 3-year-olds, as well as 4-year olds, were found to exhibit a spacing effect. These findings are consistent with previous findings reported by Rea and Modigliani (1987), but they are in conflict with the findings of Toppino and DiGeorge (1984), who obtained no spacing effect in preschool chil dren. The cause of the discrepant results is far from ob vious. The primary methodo1ogical difference between the current experiments and those of Toppino and DiGeorge is that stimulus presentation was more precisely controlled in the present experiments. The fact that 3-year olds produced a spacing effect in Experiment 2 makes it extremely unlikely that Toppino and DiGeorge's results are attributable to a sampie of cognitively less advanced 4-year-olds. Although one cannot rule out the possibility that Toppino and DiGeorge's results are systematically rep1icable under some highly specific and subtle set of circumstances, the present findings, in conjunction with those reported by Rea and Modigliani, clearly indieate that preschool children generally exhibit a spacing effect in free recall under a wide variety of experimental con ditions. The fact that preschool children manifest a spacing ef feet in free recall is difficult to reconcile with encodingvariability theories based on the use of organizational strategies (e.g., Glenberg, 1977) and with deficient processing theories emphasizing the roIe of rehearsal (e.g., Rundus, 1971). Preschoolers do not spontaneously engage in organizational strategies (e.g., Bjorklund & Jacobs, 1985; Moely, 1977), nor do they evidence ver bal rehearsal or elaboration (Wellman, 1988) in tasks such as those used to study spaced-repetition effects. In addition, the performance of preschoolers creates at least three problems for deficient-processing theories in whieh the spacing effect is attributed to voluntary alloca tion of attention (study time and resources) to the second occurrence of repeated items (e.g., Hintzman, 1974; Shaughnessy et al., 1972; Zechmeister & Shaughnessy, 1980). First, the theories are vague. When subjects allo cate attention to an item, some form of postperceptual mnemonic processing seems to be assumed, but its na ture is uncIear. To account for findings obtained from preschoo1ers, one must hypothesize an elementary form of mnemonie processing to which subjects can voluntarily allocate attentional resources but which differs from what we usually mean by rehearsal and/or elaboration. Second, these theories assurne that, as each item in a list is presented, subjects evaluate the item on some basis and modify their processing ofthe item accordingly. However, this kind of flexible, self-regulated processing is not characteristic of very young children. Third, one specific hypothesis is that the amount of study time and resources allocated to an item depends upon a metamemory judg ment of how well the item is learned (Zechmeister & Shaughnessy, 1980). However, there is reason to believe that the strategy of studying unlearned items more than leamed items may not develop until chi1dren are consider ably older. For example, in the context of simple multi trial tasks, there is evidence that even 6- and 7-year-olds fai1 to differentially allocate study time as a function of whether or not an item was recalled on an immediately preceding test (Bisanz, Vesonder, & Voss, 1978; Masur, McIntyre, & Flavell, 1973). It is possible that some voluntary, but as yet unidenti fied, strategy is necessary to produce the spacing effect. However, this possibility seems remote in light of the results of Experiment 2, whieh varied presentation rate and extended previous work with 4-year-olds to younger, developmentally less advanced 3-year-olds. Given that both initial encoding and subsequent conscious-decision processes take time and that they tend to take longer in younger children (Siegier, 1986), very fast presentation rates should inhibit strategie processes and their effects in children, whereas the effects of voluntary processes should be most apparent when relatively slow presenta tion rates are used. Results indieated that even 3-year old children exhibited a strong spacing effect in free recall with a very rapid I-sec presentation rate. Furthermore, there was no tendency for the spacing effect to become more pronounced with slower rates. These findings are most consistent with the hypothesis that fundamental even primitive-memory mechanisms that operate automatically and are available from a very young age are sufficient to produce a spacing effect in free recall. The results of the present experiments are consistent with several theories that have attributed spacing effects to automatic processes. Among encoding-variability the ories, those of both Glenberg (1979) and Greene (1989) suggest that, as the spacing between repetitions increases, more varied contextual information automatically is en coded with the repeated item. Thus, free recall is facili tated because more contextual cues can serve as effective retrieval cues." Unfortunately, although these theories are compatible with the findings reported here, the fundamen tal assumption underlying encoding-variability theories has not always been supported. According to this assump tion, increasing the number of retrieval routes through varied encoding increases the probability of successful retrieval, thus improving memory performance. Although this assumption has received support under lirnited cir cumstances (e.g., McDaniel & Masson, 1985), other at tempts to provide direct support for this assumption fre quently have failed (e.g., Maki & Hasher, 1975; Postman & Knecht, 1983). The present findings are also consistent with several deficient-processing theories, including consolidation the ory (e.g., Landauer, 1969; Peterson, 1966) and habitua tion theory (Hintzman, 1974). According to the former theory, consolidation of one presentation interferes with consolidation of the other presentation when repetitions occur too closely to one another. The result is a weakened memory trace brought about by less total consolidation (Hintzman, 1974). According to habituation theory, en coding the first presentation habituates some encoding process so that the second presentation cannot be encoded properly until enough time has elapsed for recovery to occur (Hintzman, 1974). Although these theories are con sistent with the findings reported here, which seem to limit the viability of voluntary-strategy explanations of the spac ing effect, they are not without problems. As Rea and Modigliani (1987) noted, some of their hypothesized mechanisms (e.g., the time course of consolidation or recovery from habituation) are not weIl specified, mak ing them difficult to test. Perhaps for this reason, there is little evidence actually supporting these theories (e.g., Hintzman, Summers, & Block, 1975). A final class of deficient-processing theories that will be considered emphasizes the role of retrieval operations (e.g., Cuddy & Jacoby, 1982; Jacoby, 1978). The sec ond presentation of an item is assumed to initiate retrieval of the first presentation (study-phase retrieval) . If repeti tions are massed, retrieval of the first presentation is so trivial that little processing is involved. As the spacing between repetitions increases, successful retrieval requires increased processing, which results in improved memory on a later test. This version of deficient-processing theory does not specify whether the assumed processes are voluntary or involuntary in nature. However, such a theory would be consistent with our findings if one assumed that study phase retrieval is a basic and automatic component of processing repeated information. This assumption seems reasonable when one considers that retrieval is clearly one of the most fundamental processes of the memory sys tem and that it can be triggered automatically by present ing appropriate cues. A deficient-processing mechanism based on retrieval processes is easily reconciled with the complementary view that successful study-phase-retrieval operations con tribute significantly to the beneficial effect of repetition (e.g., Thios & D'Agostino, 1976; Whitten & Bjork, 1977). Also, it is compatible with findings indicating that spaced-repetition effects may occur only for items that are recognized as repetitions (e.g., Madigan, 1%9) and that, when a single study opportunity is followed by two tests, performance on the final test improves as a func tion of the spacing between studying and the initial test (e.g., Modigliani, 1976; Whitten & Bjork, 1977). In summary, the present findings extend the generality of the spacing effect and place constraints on the form that a satisfactory theory of the spacing effect can take. The results suggest that very fundamental memory mechanisms that operate largely automatically are suffi cient to produce the spacing effect in free recall. However, several caveats are appropriate in closing. First, although voluntary strategic processes do not seem to be necessary to produce the spacing effect in free recall, it remains pos sible that the effect may be augmented or otherwise in tluenced by voluntary strategies after they are acquired. Second, some investigators have suggested that two spaced-repetition effects should be distinguished: a "spac ing effect" and a "lag effect" (e.g., D'Agostino & DeRemer, 1972; Hintzman, 1974). Much ofthe spaced repetition literature has focused on the spacing effect, which includes MP versus DP contrasts, as weIl as the effects of short spacings between repetitions, whereas the lag effect includes the effect of relatively long spacings between repetitions. The results and conclusions of the present research clearly apply to the spacing effect. However, to the extent that these two effects can be func tionally distinguished, the present conclusions may not extend to the lag effect. Third, and finally, much of the research on the spacing effect has involved free recall. However, there is evidence that somewhat different processes may underlie the spacing effect in free recall and in cued-memory tasks such as recognition (Glenberg & Smith, 1981; Greene, 1989). Thus, it remains possi ble that voluntary processes are necessary to produce the spacing effect in cued-memory tasks, as Greene (1989) has recently suggested. ApPEL, L. F., COOPER, R. G., McCARRELL, N., SIMS-KNIGHT, J., Yus SEN, S. R.,. FLAVELL, J. H. (1972). The development ofthe dis tinction between perceiving and rnemorizing. ChildDevelopment, 43, 1365-1381. BAKER-WARD, L., ORNSTEIN, P. A.,. HOLDEN, D. J. (1984). Theex pression of memorization in early childhood. Journal of Experimen tal cuu Psychology, 37, 555-575. BIRD, C. P., NICHOLSON, A. J., RiNGER, S. (1978). Resistance of 1. The lack of any rate effeet, although anticipated in preschoolers, left open the possibility that the conditions of this experiment were in sufficient to produce a rate effect in any subjects-even those known to employ sophisticated strategies in apparently similar circumstances. Therefore, as a manipulation check, the same materials and procedures were employed with 54 college students. These subjeets' recall improved steadily as presentation rate became slower, as would be expected if they were using sophisticated mnemonic strategies [F(2, 106) = 3.30, MS. = 806.79, P < .051. The mean percentage of correct free recall was 66.1 %,71.4%, and 76.0% for the 1-,2-, and 5-sec presentation rates, respectively. It should also be noted that the spacing effect was significant, and that it did not interact significantly with presentation rate. However, these effects of spacing cannot be interpreted unambig uously. For college students, the primacy and recency buffers, which were appropriate for preschoolers, were not adequate to avoid con tarninating the spacing effect with serial position effects. his account of the spacing effect in free recall as a "study-phase retrieval subtheory ." It is classified here as an encoding-variabi1ity theory, be cause the theory accounts for the superior recall of DP iterns over MP items by means of an encoding-variability mechanism. That is, DP items are assumed to be recalled better. because they presumably are encoded with a greater variety of contextual elements. thereby increasing the num ber of possible retrieval cues. CALL FOR PAPERS Memory & Cognition Special Issue "Memory and Cognition Applied" Special Issue Editors: Susan Dumais (Bellcore) Margaret Jean Intons-Peterson (Indiana University) James I. Chumbley (University of Massachusetts) Robert Proctor (Purdue University) Papers relating to applied memory and cognition are invited for this special issue of Memory & Cognition to be published in late 1991 or early 1992. The special issue will focus on fruitful interactions between applied and basic research in cognition. Research involving applications of cognitive psychology to real world problems. as weil as applied solutions to practical problems, which suggest interesting new research areas are of interest. We encourage submission of papers on the following topics: Cognitive-based design of computer interfaces Cognitive problems observed in complex systems Educational applications of memory and cognition Complex decision making Computer or related aids to memory, problem solving, and decision making Medical or legal memory and reasoning Long-term retention of skills and knowledge Prospective memory Please submit papers as soon as possible. Manuscripts should be submitted to Susan Dumais, c/o Memory & Cognition, Department of Psychology, Indiana University, Bloornington, Indiana 47405 before April 15, 1991, to be considered for the special issue. Manuscripts will undergo the normal reviewing process and may be published in a regular issue of Memory & Cognition if appropriate. For more information, contact:


This is a preview of a remote PDF: http://link.springer.com/content/pdf/10.3758%2FBF03197112.pdf

Thomas C. Toppino. The spacing effect in young children’s free recall: Support for automatic-process explanations, Memory & Cognition, 1991, 159-167, DOI: 10.3758/BF03197112