Effect of delay interval and cue focality on prospective memory

Abstract

Background

Prospective memory refers to the ability to perform planned things in the appropriate future situations. Due to pointing towards the future, prospective memory usually has a certain amount of time delay. This study investigated the impact of a delay interval on prospective memory under different cue focality conditions and used the multinomial processing tree model to determine which component of prospective memory was affected.

Methods

Participants were assigned to 2-back delay group, no task delay group, or no-delay group, and then assigned to perform either focal prospective memory tasks or non-focal prospective memory tasks.

Results

The results showed that, compared to the non-delay group and no task delay group, participants in the 2-back delay group exhibited worse prospective memory performance, but better ongoing task performance after the delay interval task. The multinomial processing tree model analysis showed that the delay interval only affected the prospective component of prospective memory and had no impact on the retrospective component.

Conclusions

The results indicate that delay intervals affect the accuracy of prospective memory under different cue focality conditions. Meanwhile, delay intervals only impair the prospective component of prospective memory.

Introduction

Prospective memory is the ability to carry out action plans at a specified time in the future [1], such as taking medication after finishing dinner. It is categorized into focal prospective memory and non-focal prospective memory, based on the nature of the cues [2]. The processing of focal prospective memory task has a high degree of overlap with the processing of ongoing task, while the processing of non-focal prospective memory task has a low degree of overlap with ongoing task. For example, if the ongoing task is to recognize colors, then the focal prospective memory task must involve the recognition of colors, while the non-focal prospective memory task can be a processing unrelated to color recognition. Therefore, the cue recognition of focal prospective memory, due to its consistency with ongoing task processing, primarily relies on bottom-up spontaneous retrieval, while the cue recognition of non-focal prospective memory, which differs from ongoing task processing, mainly depends on top-down attentional monitoring [3].

In daily life, prospective memory tasks typically require a delay between the initial formation of an intention and its execution. Longer delays may have a negative impact on prospective memory. Delays can lead to a reduction in the activation level of prospective memory [4], and this can reduce the degree of strategic processing, thereby leading to less monitoring of prospective memory cues after the delay [5]. At the same time, the delay interval can lead to memory decay [6], by which the memory of prospective memory intentions may gradually be forgotten. Therefore, delays can interfere with performance of the prospective memory either by impairing the monitoring of prospective memory cues or the retrieval of intention content.

To date, a limited number of studies have focused on the impact of delay intervals on prospective memory, but they have yielded inconsistent results. Most studies have found that delay intervals lead to a decline in prospective memory performance [7,8,9,10], but a few have suggested that delay intervals actually improve prospective memory performance [11, 12]. This inconsistency in the results may be related to the compactness of the delay tasks. Compact delay tasks typically have strong continuity, high density, and short intervals between trials, leaving no time for the rehearsal of prospective memory intentions. Martin et al. systematically studied the impact of different types of delay tasks on prospective memory, including a lexical evaluation task and a color judgment task. They found that when a lexical evaluation task was used as the filler task, delays improved prospective memory performance, but color judgment tasks impaired prospective memory tasks. They speculated that this might be because the rhythm of lexical evaluation task was not compact, participants had longer reaction time, giving the participants an opportunity to rehearsal during the delay [11]. By contrast, the color judgment task was more compact, requiring participants to complete the task within a short reaction time, and there was no long delay between trials, offering participants no opportunity to reconsider their intentions, thus reducing the activation level of prospective memory and leading to a decrease in active monitoring. This was confirmed by Mahy et al. who used an item categorization task and Raven’s Progressive Matrices as delay interval tasks. They found that compared to the less compact rhythm of the Raven’s Progressive Matrices, participants who performed the simple item categorization task showed worse prospective memory performance after a delay. In addition, post-experiment questionnaires indicated a positive correlation between prospective memory performance and the frequency of intention contemplation during the filler task, with participants contemplating their intentions more often during the Raven’s Progressive Matrices, which had less compact rhythm [12]. In summary, setting compact tasks as fillers during the delay period may be a prerequisite for the detrimental effects of delay.

The detrimental effect of delay intervals on prospective memory may be influenced by the focality of the cues as well. According to the multiple processing theory, the processing of cues for focal prospective memory overlaps to a great extent with the ongoing task, and thus it relies mainly on bottom-up spontaneous retrieval. By contrast, the processing of cues for non-focal prospective memory overlaps less with the ongoing task, and thus relies mainly on top-down active monitoring [13]. During a long delay interval, the activation level of an individual’s prospective memory intentions gradually declines, leading to a decrease in the attentional monitoring of prospective memory cues during the intention maintenance period [14]. Note that the processing of non-focal cues relies more heavily on attentional monitoring, while the retrieval of focal cues is less dependent on attention. Therefore, delay intervals would be expected to more significantly impair the performance of non-focal prospective memory. McBride et al. used ongoing tasks as delay filler tasks and examined the impact of delay under different focality conditions on prospective memory. They found that non-focal prospective memory performance gradually declined with increasing delay interval time, indicating that delay intervals impair non-focal prospective memory [5]. However, that study had only 14 participants for each experimental condition, and the participants in the long delay condition group practiced more ongoing tasks than those in the short delay condition group. Therefore, this study might have been biased by exercise effects and greater random errors. Moreover, McBride et al. did not focus on the specific reasons for the delay’s impairment of prospective memory. To more effectively and specifically study the impact of delay on prospective memory, the current study treats delay as a separate stage without practicing the tasks in the experiment, in order to avoid any possible exercise effects during the delay [15]. In addition, we use a higher number of participants and a higher frequency of cue appearances, to reduce random errors. By exploring the influencing factor of cue focality and improving the experimental design, this study aims to more effectively verify the extent to which delay interferes with the performance of prospective memory by affecting top-down processing.

Determining which component of prospective memory is affected by delay intervals would clarify the processing mechanism by which delay intervals impair prospective memory. Prospective memory consists of two components: the prospective component and the retrospective component [1]. The former mainly involves cue monitoring, while the latter involves the retrieval of intention content [16]. Some studies have suggested that delay intervals disrupt participants’ attentional monitoring, such that prospective memory cues cannot be successfully recognized when they appear [5]. However, the content of retrospective memory can also be forgotten over time, and so delays can also damage the retrieval of intention content in prospective memory. In summary, the evidence to date indicates that delays could have a detrimental effect on either the prospective or retrospective components of prospective memory, or both. The multinomial processing tree (MPT) model is a modeling method that combines cognitive psychology and statistics. This model is based on the real reactions for analysis, and takes into account the possibility of guessing, making it more advantageous in separating different components of prospective memory [17, 18].

The current study explored the effect of delay intervals under different cue focality conditions on prospective memory. The 2-back task was used as the delay task, as this type of task has been verified to be compact, individuals need to react to each trial quickly, which prevents them from having the opportunity to repeat the prospective memory task during the delay period [19]. In order to separate the different effects of delay time and filler tasks during the delay period on prospective memory, we set up two delay groups: the 2-back delay group with filler tasks and the no task delay group without filler tasks during the delay period. On the one hand, we compared the prospective memory performance between the no task delay group and the no-delay group without delay time to determine the effect of delay time on prospective memory. On the other hand, we compared the prospective memory performance between the no task delay group and the 2-back delay group to determine the effect of filler tasks during the delay period on prospective memory. At the same time, in order to reveal the processing mechanism of the delay interval affecting prospective memory, the MPT model was used to determine specifically which processing stage of prospective memory was affected. Based on existing evidence, we speculated that 2-back delay group had lower accuracy and prospective components than the other two groups under non-focal conditions.

Methods

Participants and design

One hundred and eighty-one students participated in the experiment. Their ages ranged from 19 to 25 years old (M_age=22.78, SD = 1.96). They were randomly assigned to one of six conditions: 2-back delay focal group (N = 30), 2-back delay non-focal group (N = 30), no task delay focal group (N = 27), no task delay non-focal group (N = 28), non-delay focal group (N = 33), non-delay non-focal group (N = 33). This study was reviewed and approved by the Institutional Review Board of the Henan Provincial Key Laboratory of Psychology and Behavior, before the experiment, all participants signed an informed consent form and received course credits or other compensation after the experiment concluded.

This experiment used a 3 (delay condition: non-delay, 2-back delay, no task delay) × 2 (prospective memory cues: focal, non-focal) between-subjects design.

Materials and tasks

The task stimuli consisted of 24 uppercase English letters (excluding F and J), displayed in black font on a white background, presented in random order.

The practice task, ongoing task, and prospective memory task stimuli comprised 140 words, all presented in nine colors (black, white, red, turquoise, blue, yellow, gray, purple, and green) against a silver background, with the words “cake” and “wreath” serving as non-focal cues, and words in yellow font serving as focal cues.

E-prime 2.0 was used to write the program and present the experimental task instructions, stimuli, and data collection on a computer. All participants individually carried out the tasks in a soundproof room pressing keys on the keyboard.

Procedure

The experiment was divided into a practice phase, a delay phase, and a prospective memory phase. In the practice phase, the ongoing task instructions were presented and the participants were then asked to complete 20 ongoing tasks, in which they were required to judge whether the color of the word’s font matched the color of any of the four-color blocks that had appeared on the previous screen. A match was indicated by pressing the “J” key; otherwise, they were to press the “F” key. At the beginning of each ongoing task, a fixation point (+) was presented for 250ms, followed by a color block for 800 ms, and then a word was presented for 5000ms. Participants responded by pressing keys on the keyboard when a word appeared, after which the word disappeared. Finally, a blank buffer interface appeared for 250ms (see Fig. 1). There was a total of 20 trials in the practice phase. Participants advanced to the next phase if their accuracy in the practice phase was above 0.6; otherwise, they practiced again. After completing the practice phase, participants were presented with instructions for the prospective memory task: the focal group was instructed, in addition to performing the ongoing task, to press the “SPACE” key when they encountered words in yellow font; while non-focal group was instructed, in addition to performing the ongoing task, to press the “SPACE” key when they encountered the words “cake” and “wreath.” To ensure their understanding, the participants were required to correctly repeat the instructions for the prospective memory task to proceed to the next phase.

Fig. 1

The flowchart of the ongoing task

During the delay phase, the 2-back delay group was informed that they had to perform a 2-back task before the formal experiment. In the 2-back task, the participants were required to compare whether the current letter on the screen was the same as the letter presented two steps back. If they were the same, participants pressed the “J” key; if they were different, they pressed the “F” key. In the 2-back task, a fixation point (+) was first presented for 500ms, followed by an uppercase English letter for 2500ms, and then a blank buffer interface for 500ms. The total duration of the 2-back task was 5 min. In total, participants needed to perform more than 120 2-back tasks. After competition, the 2-back delay group proceeded to the prospective memory phase. The specific procedure of the task is shown in Fig. 2. The no task delay group was informed that they needed to sit quietly for 5 min without doing anything during the delay phase. The non-delay group did not have a delay phase.

Fig. 2

The flowchart of the 2-back task

Fig. 3

The entire experimental flowchart for different groups

After completing the delay phase, the two delay groups entered the formal experimental phase, while the no-delay group entered the formal experimental phase after completing the practice phase. The formal experimental phase included 56 ongoing task trials and 4 prospective memory task trials, for a total of 60 trials. In the focal cue group, the yellow font word stimulus was presented 4 times; in the non-focal cue group, the words “cake” and “wreath” were presented 2 times each, for a total of 4 times (see Fig. 3). After the experiment, the participants were asked to confirm whether they remembered the existence of the prospective memory task, to ensure that they were aware of it.

Results

The data in this study were analyzed using SPSS 20.0, and the analysis method was analysis of variance (ANOVA).

2-back task performance

2-back task accuracy

Table 1 Performance of the 2-back task in 2-back delay group (M ± SD)

The ANOVA on accuracy of the 2-back task revealed no significant difference in accuracy between the delayed focal and delayed non-focal groups, p = 0.14 (see Table 1).

2-back task reaction time

The ANOVA on reaction time of the 2-back task showed no significant difference between the accuracy rates of the delayed focal and delayed non-focal groups, p = 0.99.

Prospective memory task performance

Prospective memory task accuracy

Table 2 Prospective memory performance and ongoing task performance(M ± SD)

The ANOVA on the accuracy of the prospective memory showed significant main effect of delay, F(1, 175) = 3.69, p = 0.03, η_p² = 0.04, with 2-back delay group having lower accuracy of the prospective memory than both the non-delay group (p = 0.01) and no task delay group (p = 0.04). The main effect of cue focality was also significant, F(1, 175) = 5.67, p = 0.02, η_p² = 0.04, with higher accuracy being observed for focal prospective memory tasks than non-focal tasks. The interaction between delay and cue focality was not significant, p = 0.91 (see Fig. 4; Table 2).

Fig. 4

The accuracy on the prospective memory task under different delay and cue focality conditions. The “*” represents p < 0.05, and the error bar represents the mean of the standard error

Prospective memory task reaction time

The ANOVA on the reaction time of the prospective memory showed that the main effect of delay was not significant, p = 0.89. Similarly, the main effect of cue focality was not significant, p = 0.43; and the interaction between delay and cue focality was also not significant, p = 0.73.

Ongoing task performance

Ongoing task accuracy

The ANOVA on the accuracy of the ongoing task showed that the main effect of delay was significant, F(1, 175) = 5.02, p = 0.008, η_p² = 0.05, with the 2-back delay group having lower accuracy on the ongoing task than both the non-delay group (p = 0.003) and no task delay group (p = 0.02). However, the main effect of cue focality was not significant, p = 0.24, and the interaction between delay and cue focality was also not significant, p = 0.86 (see Fig. 5).

Fig. 5

The accuracy of the ongoing task under different delay and cue focality conditions. The “*” represents p < 0.05, the “**” represents p < 0.01, and the error bar represents the mean of the standard error

Ongoing task reaction time

The ANOVA on the reaction time of the ongoing task showed that the main effect of delay was not significant, p = 0.87; similarly, the main effect of cue focality was not significant, p = 0.22; and the interaction between delay and cue focality was also not significant, p = 0.69.

MPT model analysis

This study used the multinomial processing tree (MPT) Model to analyze the frequency of different types of components, thereby directly verifying which component of prospective memory is impaired by delays. The MPT model includes independent parameters P and M for preparatory attention processing and retrospective memory processing in the prospective memory task, as well as ongoing task parameters C₁ and C₂. Because participants may make successful guesses while performing the task, the MPT model also includes a guessing parameter c for the ongoing task and a guessing parameter g for the prospective memory task. Thus, the model has a total of 7 free parameters: C₁, C₂, P, M₁, M₂, g, c. Smith and Bayen were the first to use the MPT model to measure event-based prospective memory [18]. They assumed M₁ = M₂, adjusted the values of g and c to observe the changes in P and M, and ultimately determined the assignments for g and c to be g = 0.1, c = 0.5. Thus, through adjustment of the model parameters, a new 4-parameter model was obtained: P, M, C₁ and C₂. The MPT model directly takes the actual response frequency of participants as the object of analysis and, because it also takes into account the possibility of guessing, is more reliable in distinguishing the two components of prospective memory. Use of the MPT model is premised on the fact that the prospective memory task must occupy attentional resources [17]. The monitoring of focal cues overlaps highly with the processing of the ongoing task and is not highly dependent on attention, while, by contrast, the monitoring of non-focal cues does not overlap as much with the ongoing task and definitely requires self-initiated attentional resources [20]. Therefore, we only conducted MPT model analysis on the response frequency of participants under non-focal cue conditions.

First, a model fit analysis was conducted on the frequency of participants’ responses, to determine whether there was a significant difference between the actual response data and the estimated data, and also to determine whether the actual response data was suitable for analysis by the MPT model. The model fit test was conducted on the response frequency data of the 2-back delay group, and the results showed a good fit, G²(4) = 7.26, p = 0.12 (P = 0.32, M = 0.89). The model fit test was then conducted on the response frequency data of the non-delay group, and those results also showed a good model fit, G²(4) = 6.90, p = 0.14 (P = 0.53, M = 0.94). The model fit test was then conducted on the response frequency data of the no task delay group, and those results also showed a good model fit, G²(4) = 4.22, p = 0.38 (P = 0.47, M = 0.93).

Finally, we included the response frequencies of the participants under the 2-back delay and non-delay conditions into the MPT model for comparison, and the results showed that the P value of the delay group was significantly smaller than that of the non-delay group, △G²(1) = 10.72, p = 0.001. However, there was no difference in M value between the delay group and the non-delay group, △G²(1) = 1.17, p = 0.28. We included the response frequencies of the participants under the 2-back delay and no task delay conditions into the MPT model for comparison, and the results showed that the P value of the delay group was significantly smaller than that of the non-delay group, △G²(1) = 5.09, p = 0.024. However, there was no difference in M value between the no task delay group and the 2-back delay group, △G²(1) = 0.63, p = 0.43 (see Table 3).

Table 3 Response category frequencies for all participants for each trial type in the non-focal cue condition

Discussion

This study manipulated the delay interval to explore the impact of the delay interval on prospective memory. In prior studies, non-focal cues have typically been assumed to be processed in a top-down manner, relying more on active monitoring and requiring individuals to invest more attentional resources, while focal cues were assumed to be processed in a bottom-up manner, relying more on spontaneous retrieval [3]. Therefore, we hypothesized that the delay interval would only impair non-focal tasks. This study used a 2-back task as the filler task during the delay period, as such a task is compact [19, 21], thus preventing participants from having the opportunity to rehearsal during the delay period. This study found that under non-focal cue conditions, the accuracy of prospective memory in the 2-back delay group was lower than that in the no-delay group, and thus the delay interval did indeed impair non-focal prospective memory performance, consistent with both our hypothesis and previous research findings [5, 7]. However, under the focal cue conditions, the delay interval also impaired prospective memory performance, which is not consistent with our hypothesis, and indicates that the effect of the delay interval on prospective memory was not affected by cue focality. At the same time, we found no difference in the accuracy of prospective memory between the no task delay group and the no delay group, indicating that delay time did not impair prospective memory performance. This may be because no task delay group did not perform any filler tasks during the delay period, and they could use this time to repeat prospective memory tasks multiple times, which may have resulted in a high level of activation of prospective memory intentions, thereby hindering the decline in prospective memory performance. Finally, we found that the accuracy of prospective memory in the 2-back delay group was also higher than that in the no task delay group, indicating that the filler task during the delay interval interfered with the performance of prospective memory. The results of this study showed that the accuracy of prospective memory in the 2-back task delay group decreased under both focal and non-focal conditions, but the response speed of prospective memory did not change. This may be because when prospective memory cues appear, individuals undergo multiple processing processes, such as inhibition, attention shift, etc [2, 15]., which are subsequent processes after successful cue monitoring and are not easily affected by delays. Therefore, the 2-back delay group did not show a slowing down of prospective memory response speed. At the same time, there was no difference in response speed between focal and non-focal prospective memory in the 2-back delay group. This should be because focal prospective memory and non-focal prospective memory mainly differ in cue properties [3], and their dependence on attention is mainly reflected in the cue monitoring stage. When prospective memory cues appear, individuals mainly extract prospective memory intentions. When performing focal prospective memory tasks and non-focal prospective memory tasks, once prospective memory cues are successfully identified, participants need to suppress ongoing tasks and shift their attention to extract the execution content of prospective memory. The processes involved in the retrieval stage of focal prospective memory tasks and non-focal prospective memory tasks are very similar, so there is no difference in the response speed between focal prospective memory tasks and non-focal prospective memory tasks.

Why does a delay interval impair the performance of prospective memory? During a delay interval, individuals must invest attentional resources to maintain the activation and active monitoring of the prospective memory task [5, 14]. When individuals are performing a compact 2-back task, it is very difficult to continuously invest attentional resources to maintain prospective memory, and they do not have the opportunity to think about their prospective memory intentions. Therefore, the activation of prospective memory intentions is likely to decrease over time, and the content of the intention is likely to be forgotten. This study found that the 2-back delay group had higher accuracy of the ongoing task than both the non-delay group and no task delay group, indicating that the attentional cost of the prospective memory task was reduced during the intention maintenance period in the 2-back delay group. During intention maintenance, individuals must continuously monitor cues, which consumes most of their attentional resources [22, 23]. After the delay phase, the activation of prospective memory intention decreases, the active monitoring of prospective memory declines, and the reduction in attentional resources also impairs the recognition of prospective memory cues and the execution of prospective memory tasks [24]. Due to the opportunity to repeat prospective memory tasks during the delay period, the prospective memory intention remains highly activated during the intention retention period. Therefore, we did not find any differences in the performance of ongoing tasks between the non-delay group and the no task delay group. The analysis results for the MPT model showed that the 2-back delay group had lower prospective component than both the non-delay group and no task delay group, while there were no differences in the retrospective components between each two groups. This further confirms that the delay mainly affected the cue monitoring of prospective memory, rather than impairing the performance of intention retrieval. Finally, without rehearsal, delay can lead to the forgetting of memory content [6]. This study did not find that delay affected the retrieval of the content of prospective memory intentions, which is inconsistent with the results from our previous study and everyday experience. This may be because the content of the intention set in this study was relatively simple, and even after the delay, it was not easy for participants to forget. This may be an important reason why this study did not observe the retrospective component to be impaired under delay conditions.

It has been shown that the monitoring of focal prospective memory cues is not highly dependent on attention [13], but the results of this study show that delay can also impair the performance of focal prospective memory. This may be because, although the monitoring of focal prospective memory cues is not highly dependent on attention, it nonetheless requires some attention to confirm the presence of prospective memory cues [25]. Thus, if individuals forget the existence of the prospective memory task during the task execution, focal prospective memory performance will also decrease. The interval reduced the activation of the prospective memory intention, causing individuals to gradually forget the need to confirm the prospective memory cues, which may have led to the observed decrease in prospective memory performance under the delay conditions. The study demonstrated that under focal cue conditions, the 2-back delay group had a higher accuracy on the ongoing task than the other two groups. This indicates that the maintenance of intentions in focal prospective memory requires attention. It also confirmed that during the intention maintenance period, the delay did indeed lead to a reduction in individuals’ attention to prospective memory. We also found no difference in reaction time between the 2-back delay group and the no-delay group, indicating that under delay conditions, individuals’ processing time during intention retrieval was not decreased. Given that this study used the same response setup for both focal and non-focal prospective memory, we speculated that the delay might not have impaired the retrieval of prospective memory intentions. Therefore, it is likely that the delay impaired focal prospective memory through the prospective component rather than the retrospective component.

Contributions and limitations

This study is the first to focus on the effects of delay on different components of prospective memory, which is helpful to reveal the processing mechanism of delay damage prospective memory. This study has some limitations. First, based on the example of other studies, this study used a 5-minute delay. However, in real life, such delays usually last for hours or even days. During a delay, a prospective memory intention may not always remain activated and is instead likely to be in a subthreshold state of activation [26]. Therefore, the mechanism by which delays affect prospective memory in real life may differ from that in the laboratory. Second, the execution of prospective memory tasks is usually accompanied by motivation, and thus individuals will evaluate different prospective memory tasks as having different levels of importance [27]. When a prospective memory task is judged to be more important, individuals may have already fully repeated it during the encoding period; and even during the delay period, no matter how busy they are, they will reserve some attention for repeating the prospective memory task. Therefore, the performance of important prospective memory tasks may not be significantly impaired by the delay interval. Finally, we initially considered that prospective memory tasks usually occur infrequently, and this study only set up four prospective memory tasks in this study. There was evidence to suggest that a small amount of practice could significantly improve prospective memory performance, producing a strong practice effect [28]. To avoid the occurrence of practice effect, we adopted a between-subject design. However, individual differences among participants, such as variations in memory capacity, may affect our results, which is a limitation of this study.

Conclusion

This study explored the effects of delay intervals on prospective memory and its two components. The results indicate that delay intervals affected the accuracy of prospective memory under different cue conditions. Contrary to our hypothesis, the MPT model results show that delay intervals only impaired the prospective component of prospective memory. Delay intervals impaired the activation level of prospective memory, thereby reducing the attentional monitoring of prospective memory. The MPT model results also confirmed that delay intervals affected the cue recognition of prospective memory.