Semantic and episodic predictions of memory for plans†‡§
Abstract: This study was conducted to investigate the planning process within memory for plans. Experiment 1 examined the effect of self-involvement on memory for a 1-day plan, in which approximately half of the 104 undergraduate participants were instructed to assume that the plans were their own while the other half were told only to memorize them. A U-shaped curve, indicating better recall for morning and evening plans than for afternoon ones, was observed only with the first group. Experiment 2 was designed to further clarify the planning strategies of memory for plans, in which approximately half of the 77 undergraduate participants were asked to imagine executing the plans while the other half were given a more limited strategy. U-shaped curves were observed in both conditions, with memory performance being higher for the first group. These results are interpreted as indicating that the semantic prediction of future plans conforms to a U-shaped curve, while the episodic prediction of future plans enhances memory performance.
The remembering of future plans is called “prospective memory” (see Brandimonte, Einstein, & McDaniel, 1996, for a review). Although most studies of prospective remembering have focused upon a prospective component (i.e., remembering to do something), “very little research has been devoted to investigating the nature of the retrospective representation, or memory for the intention itself” (Marsh, Hicks, & Bink, 1998, p. 350; see also Goschke & Kuhl, 1993; Marsh, Hicks, & Bryan, 1999; Watanabe, 2003; Watanabe & Kawaguchi, in press, for further discussion). One of the reasons for this has been that the procedures that participants have been given only involve memorizing very simple (one or just a few) plan(s) (e.g., press the F10 key, or make a telephone call to the experimenter). With these procedures, the participants have to remember the timing(s) of when they carried out the plan(s), but they do not actually need to remember the content of the plans, because they are very simple.
Recently, to investigate the nature of retrospective representations of plans, Watanabe and Kawaguchi (2000) examined memory for a fictitious day of plans, and found that memory performance for plans varies according to the time of day. In their experiment, participants were asked to memorize plans that consisted of an “hour” and an “action phrase” (e.g., “10:00,”“make a telephone call”). When participants were asked to memorize plans (from 08:00 to 22:00 hours) as their own “future” (tomorrow) plans, both the plans for the morning and for the evening were recalled better than those for the afternoon.5 This pattern can be called a U-shaped curve or an effect of the time of day. Moreover, this phenomenon was only observed in the “future” plan condition, in which the participants were asked to memorize the plans as future (tomorrow) plans, and not in the “past” event condition, in which the participants were asked to memorize the plans as past (yesterday) events. This suggests that the U-shape effect is due to one's intention for the future rather than memory for the paired associate stimulus (i.e., hour and action phrase). It also implies that the processes involved in memory for plans include more than just a “memorizing” of the stimulus. Dobbs and Reeves (1996) argue that “prospective memory” is “more than memory,” and that it includes some other aspects (e.g., metaknowledge, planning, monitoring, and so on). In the present study, the goal was to investigate what aspects of memory for plans contribute to the U-shaped effect.
In the studies of Watanabe and Kawaguchi (2000, 2002), participants were instructed to regard the plans for tomorrow as their own plans. Did this instruction to treat the plans as their own lead participants to construct future images of the plans involving themselves? To examine this question, in the following experiments, the effect of self-involvement in the plans on memory was examined.
Experiment 1 was designed to examine whether the time of day effect appears exclusively when participants assume the plans are their own plans or not. In this experiment, approximately half of the participants were told to regard the plans as their own plans for tomorrow (the self-plan condition), while the other half were told only to memorize the plans (the control condition). If self-involvement in plans affects the memory for plans, then memory performance in the self-plan condition might differ from that in the control condition.
One hundred and four undergraduate students participated in the experiment. They were recruited from undergraduate psychology courses and received a course research credit for their participation. Fifty-four participants were assigned to the self-plan condition and 50 to the control condition.
The time schedule stimuli consisted of 19 plans, and each plan consisted of an “hour” and an “action phrase” (e.g., “at 10:00, make a telephone call”). Fifteen of the 19 plans were target plans, with four plans as buffer plans. Two of the buffer plans were presented at the beginning and two at the end of the list to eliminate primary and recency effects. The “hour” element of each target plan was set at hourly intervals ranging between 08:00 and 22:00 hours, while the “hour” components of the buffer plans were at 06:00 and 07:00 hours and at 23:00 and 24:00 hours, respectively. The “action phrases” were general activities that occur in daily life (e.g., “go shopping”).6 There were two sets of time schedules, consisting of different combinations of the “hour” and “action phrase” elements.
Participants were tested in a group and the experiment was carried out over two different classes. In the first phase (i.e., the study phase), participants were asked to read instructions that were printed on a sheet of paper (in this way, both the self-plan and the control conditions could be included within each class), and they were also told to memorize the plans for 1 day, which consisted of “hours” and “action phrases.” The participants in the self-plan condition were told to regard the plans as their own plans for tomorrow. In contrast, the participants in the control condition were told only to memorize the plans. Following these instructions, two buffer plans, 15 target plans, and a further two buffer plans were presented successively on a screen using a projector. Each plan appeared for 10 s in the middle of the screen and the interstimulus interval (ISI) was 3 s. One of the two time schedules was assigned to each class. The 15 target plans in each set were presented in a randomized order.
In the second phase (i.e., the distraction task phase), the participants carried out a distraction task for 1 min. The participants were asked to remember as many of the names of areas in the prefecture in which their university is located as possible.
In the third phase (i.e., the free recall phase), the participants recalled the plans as “hour” and “action phrase” pairs, writing them down on a blank sheet in the order they remembered them. There was no time limit for response, but the participants on average took approximately 10 min to complete the free-recall task.
No participants in the control condition reported using the strategy of explicitly regarding the plans as their own.
The 15 target plans were divided into three time zones, each of 5-h duration: time zone 1 (morning) was from 08:00 to 12:00 hours; time zone 2 (afternoon) was from 13:00 to 17:00 hours; and time zone 3 (evening) was from 18:00 to 22:00 hours.7Table 1 (upper half) shows the results of the free-recall task. Correct recall was defined as correctly remembering both the time and the action phrases. A 2 × 3-way ANOVA with one between-subjects factor of instruction (self/control) and one within-subjects factor of time zone (morning/afternoon/evening) was carried out on the proportions of correct recall. There were significant main effects of instruction (F(1,102) = 7.433, p < 0.01) and of time zone (F(2,204) = 35.935, p < 0.01). In addition, the interaction between instruction and time zone was significant (F(2,204) = 16.293, p < 0.01). Regarding the interaction, there were significant simple main effects of instruction for the morning time zone (F(1,306) = 14.134, p < 0.01) and the evening time zone (F(1,306) = 16.941, p < 0.01), in which plans in the self-plan condition were recalled better than plans in the control condition. Further, there was a significant simple main effect of time zone in the self-plan condition (F(2,204) = 50.152, p < 0.01). Ryan's multiple comparisons revealed that the recall rates in time zone 1 (morning) and in time zone 3 (evening) were higher than in time zone 2 (afternoon) for the self-plan condition (MSE = 0.028, p < 0.01). These results indicate that the U-shaped curve was only observed in the self-plan condition.
Table 1. Mean proportion of correct recall of plans as a function of instruction (Experiment 1)/strategy (Experiment 2) and the time of day
| Self (SD)||0.604 (0.172)||0.340 (0.184)||0.644 (0.213)|
| Control (SD)||0.456 (0.212)||0.415 (0.188)||0.481 (0.219)|
| Self-image (SD)||0.635 (0.173)||0.390 (0.195)||0.625 (0.185)|
| Clock-image (SD)||0.416 (0.182)||0.351 (0.176)||0.422 (0.179)|
In the results of Experiment 1, the U-shaped curve was observed only in the self-plan condition, which suggests that the memorizing process for future plans is influenced by self-involvement.
Watanabe and Kawaguchi (2002) investigated knowledge for the time of day using a “time-action association test,” in which participants were asked to recall usual activities and/or events associated with times from 01:00 to 24:00 hours at hourly intervals. The result showed that the number of associations with personal activities is greater for the morning and the evening than for the afternoon, which is similar to the effect of the time of day on remembering, represented by the U-shaped curve, that is, both plans for the morning and the evening are recalled better than those for the afternoon. This suggests that the U-shape effect on memory for plans might be based on knowledge of one's own daily activities. It is possible that a relatively fixed or routine pattern to one's own activities (i.e., knowledge of daily activities) might be represented in the mental process as “a mental to-do-list,” into which new and dynamic plans can be inserted. Therefore, a plan that has to be executed at a highly specific time might be encoded by linking to it to one's knowledge of daily activities, and such a process might enhance memory performance. In contrast, a plan that can be executed at a less specific time cannot be encoded by associating it with knowledge of daily activities and consequently memory performance for such a plan would be inferior.
The results of Experiment 1 suggest that when participants are asked to assume that plans for tomorrow are their own plans they do not simply “memorize” the plans, but actively “predict” the future events. It would appear that they are “preexperiencing” a future event. Indeed, after the experimental session, many participants reported that they “imagined actually carrying out their plan” during the study phase. Consequently, in the next experiment, the effect of concrete predictions or imagination on memory for future events is examined.
Experiment 2 addressed planning strategies in memory for plans. In this experiment, approximately half of the participants were asked to imagine executing the plans at the planned time (the self-image condition), while the second group were, instead, asked to imagine assigning the plans to the hours on a clock face (the clock-image condition). A previous study (Watanabe & Kawaguchi, 2002) has suggested that when people memorize future actions as a plan, they use knowledge of their own daily activities (e.g., what one routinely does at that time, where she usually is). Therefore, it was expected that the strategy of imagining or predicting the actual execution of the plans might enhance memory performance and produce the U-shape effect.
Seventy-seven undergraduates participated in the experiment. They were recruited from undergraduate psychology courses and received a course research credit for their participation. Forty participants were assigned to the self-image condition and 37 to the clock-image condition.
The plans to memorize were identical to those of Experiment 1 with the following exception. A round clock, 2 cm in radius, was printed on the instruction sheet and used in the clock-image condition.
The procedure was generally similar to that of Experiment 1. In the study phase, participants in both the self-image and clock-image conditions were told to assume that the plans for tomorrow were their own. In the self-image condition, the participants were asked to imagine actually executing the plans at the planned time, while in the clock-image condition, the participants were asked to imagine assigning the plans to the times of the clock face printed on the instruction sheet.
As in Experiment 1, the 15 target plans were divided into three time zones. Table 1 (lower half) shows the result of the recall task. A 2 × 3-way anova with one between-subjects factor of strategy condition (self-image/clock-image) and one within-subjects factor of time zone (morning/afternoon/evening) was carried out on the proportions of correct recall. There were significant main effects of strategy (F(1,75) = 21.294, p < 0.01) and time zone (F(2,150) = 31.843, p < 0.01). In addition, the interaction between strategy and time zone was significant (F(2,150) = 10.067, p < 0.01). Regarding the interaction, there were significant simple main effects of strategy for the morning zone (F(1,225) = 27.069, p < 0.01) and the evening time zone (F(1,225) = 23.391, p < 0.01), in which plans in the self-image condition were recalled better than plans in the clock-image condition. Further, there were significant simple main effects of time zone in the self-image condition (F(2,150) = 38.822, p < 0.01) and in the clock-image condition (F(2,150) = 3.088, p < 0.05). Ryan's multiple comparisons revealed that the recall rates in time zone 1 (morning) and in time zone 3 (evening) were higher than those in time zone 2 (afternoon) for both the self-image and the clock-image conditions (MSE = 0.019, p < 0.01 for the self-image condition and p < 0.05 for the clock-image condition). These results showed, first, that the U-shaped curve was evident in both the self-image and the clock-image conditions and, second, that recall was better for the participants in the self-image condition than those in the clock-image condition.
The results of Experiment 2 suggest the existence of two processes: while one process is immune to the strategy of imagining future situations, the other process is affected by such a strategy. In Experiment 2, the U-shape curve for the time of day was observed in both conditions, independently of whether the participants imagined or predicted their situation in executing the plans. Because Watanabe and Kawaguchi (1999, 2002) demonstrated that the number of time-action association pairs for daily activities is greater for the morning and the evening than for the afternoon (i.e., the U-shaped curve), in both conditions in this experiment, the participants seem to be using their knowledge of personal daily activities. That is, the relatively fixed pattern of one's own activities would be represented in the mental process, in a way that new and dynamic plans can be inserted. The second finding was that the memory performance of the participants who were asked to actually imagine executing the plans was better than that of the participants who did not use such a strategy. This suggests that the strategy of imagining a future situation leads to superior memory for plans. These results suggest that there are two different processes involved in planning, which correspond to the pattern of remembering (i.e., U-shaped remembering) and to the quantity of remembering, respectively.
The results of the two experiments described here demonstrate that knowledge about one's own daily activities is used in memorizing a new future action plan. The results also suggest that there are two planning processes in the “self-involving plan.” One process leads to the U-shape effect and the other process facilitates memory performance. It is proposed that the former process is based on semantic information (i.e., knowledge of one's own daily activities) to execute future action plans, while the latter process is based on concrete episodic information. Atance and O’Neill (2001) propose a concept of “episodic future thinking” that is formulated on the construct of episodic memory (Tulving, 1983). They define episodic future thinking as “a projection of the self into the future to pre-experience an event,” and give as an example “envisaging my forthcoming vacation might require me to consider … how much spending money I will have, how much work I will have completed before I go, and so on” (p. 533). According to Atance and O’Neill's (2001) concept, when the participants in the present experiments memorized plans, they apparently predicted executing the future daily activity plans using both “semantic prediction” and “episodic prediction”.8
For the relation between semantic prediction and episodic prediction in memory for plans, there is a notable result from Experiment 2. In the self-image condition in Experiment 2, memory performance was only enhanced by the use of the imagining or self-involvement strategy for time zones that are knowledge-rich (i.e., the morning and the evening). In contrast, memory performance for plans during the time zone that is knowledge-poor (i.e., the afternoon) did not increase, even though the participants were able to use the self-image strategy.
These results suggest that semantic prediction, that is, linking knowledge and a new plan, contributes to episodic prediction, which in turn activates a concrete image or a “preexperiencing” of a future situation. When the participants were asked to “memorize” plans, they did not “memorize” plans literally, but constructed plans by using the two kinds of predictions. Therefore, this kind of “planning,” using semantic and episodic predictions, is a process in which memory of the past or knowledge is actively associated with future behavior.
Part of this work was supported by a grant from the Research Fellowships of the Japan Society for the Promotion of Science for Young Scientists.
I am grateful to Dr Jun Kawaguchi (Nagoya University) for his instructive comments on previous versions of this article. I am also indebted to Dr Terry Joyce (Tokyo Institute of Technology) for his help in preparing the manuscript. In addition, I would like to thank the anonymous reviewers for their helpful suggestions and their thoughtful and constructive criticism of an earlier version of this article.
This article presents the results of two experiments, which the author conducted at Nagoya University in 2002.
Correspondence concerning this article should be sent to: Hama Watanabe, Department of Physical and Health Education, Graduate School of Education, University of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. (Email: email@example.com)
The same results were observed in similar experiments that the author conducted in which the participants had to memorize plans for two consecutive days (i.e., tomorrow and the day after tomorrow [Watanabe & Kawaguchi, 2000; Experiment 2]) and in which the participants had to memorize plans for a 24-h period (Watanabe & Kawaguchi, 2002).
Action phrases were selected based on preliminary research in which other undergraduates made a list of their actual plans for the coming months.
Even if the division of time zones is shifted by 1 h earlier, the results show the same pattern.
Episodic future thinking is applicable to a wide range of topics, such as prospective memory, judgments and decision-making, goal attainment, future time perspective and future orientation, the ability to think about the future, and positive/negative future thinking, as well as anxiety and depression, neuropsychological cases, and awareness of the future (Atance & O’Neill, 2001). This article proposes the concept of episodic and semantic prediction as a description of the processes involved in the encoding/storage of future plans.