Verbal, visual, and motoric encoding of prospective memory instructions
To date, only two studies have evaluated the effects of different encoding modalities on children’s PM performance (Li & Wang,
2015; Passolunghi et al.,
1995). Both have compared groups of children who were asked to encode PM instructions verbally, visually, or by enacting the intention. For instance, Passolunghi et al. (
1995) conducted a series of experiments in which they manipulated the modality of PM instruction encoding (i.e., verbal, visual, or enacted). Primary school children were presented with a series of word lists on a computer screen and asked to press a specific key on the keyboard whenever the word “
boat” appeared. The authors showed that 7- and 8-year-old children who were shown the visual representation of the PM target (i.e., the picture of a boat) obtained higher PM scores than their peers who were shown the written word “
boat” during the instructions. By contrast, 10- and 11-year-old children’s PM performance benefited most when they could enact the PM intention (i.e., press the key after the PM instructions) rather than being verbally or visually presented with the PM target.
Similarly, Li and Wang (
2015) showed that only 7-year-old children profited from the presence of the objects representing the PM targets (e.g., showing a mirror for the PM target “
looking in the mirror”), while 8- and 9-year-olds benefited most from enacting the PM intention (e.g., pretending to look in a mirror or looking in a real mirror for the PM target “
looking in the mirror”). The authors argued that 7-year-old children might still develop the cognitive abilities required to imagine the specific object while enacting the intention (see also Mecklenbräuker et al.,
2011). Alternatively, they suggested that imagining the specific object during the enactment might overload young children’s still developing cognitive resources. Consequently, younger children would benefit from the object's presence to better encode an intention, as has been shown by Passolunghi et al. (
1995). On the other hand, older children profit more from enacting the intention. Because no ongoing task performance costs depended on encoding modality, the authors suggested that this strategy is likely to enhance the automaticity of intention retrieval by increasing cue distinctiveness or the link between the cue and the related action (Li & Wang,
2015).
Implementation intentions
During the encoding phase, when we form an intention and think of what we must do and when we must do it, one strategy is to rehearse and visualize the plan that must be enacted. This task can be accomplished by implementing intentions in statements that specify the plan, such as “If situation x appears, I will do y” (Gollwitzer,
1999). This encoding strategy has been shown to improve goal attainment in everyday-life tasks (Duckworth et al.,
2013; Gollwitzer,
1999; Wieber et al.,
2015) and PM performance (Chasteen et al.,
2001; Lee et al.,
2016; Zimmermann & Meier,
2010). Participants are usually required to verbally reformulate PM instructions in the typical if–then format and repeat the sentence aloud thrice (e.g., Zimmermann & Meier,
2010). Alternatively, participants were asked to visualize the PM instructions for 30–45 s (e.g., Brewer et al.,
2011) or to perform the verbal and the imagery form together (e.g., McDaniel et al.,
2008). A meta-analysis on the effects of implementation intention on PM performance of healthy young adults revealed that all three versions are effective; however, combining the verbal and the imagery form produced larger effects than the two formats applied alone (Chen et al.,
2015). The authors also found that implementation intention enhanced PM performance in both focal and non-focal PM tasks.
Concerning the underlying mechanisms, it remains unclear whether implementation intentions elicit automatic retrieval or strategic monitoring processes. To date, three main positions have been elaborated (see Chen et al.,
2015). The first postulates that implementation intention establishes a robust cue-intention association by eliciting spontaneous retrieval of the intention without bearing on cognitive resources (Brewer et al.,
2011; Gollwitzer,
1999; McDaniel et al.,
2008). The second assumes that implementation intention influences the PM task's importance perception, inducing participants to engage more cognitive resources to perform the task. Accordingly, a PM performance boost induced by implementation intentions would be accompanied by an ongoing task performance cost (Brewer & Marsh,
2010; Meeks & Marsh,
2010; Smith et al.,
2014). The third hypothesis argues that implementation intentions would strengthen the cue-intention link and simultaneously elicit the engagement of cognitive resources (McDaniel & Scullin,
2010).
In developing populations, implementation intentions have been investigated in three studies. The first study examined the effect of the verbal version of implementation intentions on PM performance across the lifespan, comparing 10- to 14-year-olds with young and older adults (Zimmermann & Meier,
2010). Participants were required to perform a lexical decision task as ongoing task. For the PM task, they were required to press a key whenever the word of an animal was presented (i.e., focal and categorical PM cue). Implementation intentions improved PM performance across the lifespan with a significantly higher advantage for older than younger adults and a numerical but non-significant benefit for children and adolescents. Monitoring costs were similarly displayed across all three age groups and in both conditions. Consequently, implementation intentions did not further affect monitoring costs, suggesting that this strategy does not involve additional cognitive resources.
In a more recent study, Kretschmer-Trendowicz et al. (
2021) investigated the impact of implementation intentions in the combined verbal and imagery format on 9-, 12-, and 15-year-old participants’ PM performance in two experiments with varying resource demands. In Experiment 1, children and adolescents received standard PM instructions or implementation intention instructions. The importance of the PM task was manipulated within participants when giving instructions to induce them to allocate their attentional resources either to the PM or the ongoing task. As ongoing activity, participants had to perform a letter comparison task in which they had to compare pairs of five-letters strings and decide whether they were equal. The PM task consisted of pressing a key when one of four specific letter strings appeared; this task was focal to the ongoing task with specific PM cues. Importance was varied across participants by saying either that it was essential to perform the PM task successfully or that high ongoing task performance was central. For implementation intentions, half of the participants were first asked to form an if–then statement for each PM cue in the first person and repeat it aloud three times. Afterward, they were requested to visualize themselves while executing the PM and the ongoing task. Neither implementation intentions nor importance of the PM task affected PM accuracy, while the importance of the PM task affected PM RTs by slowing them down. The authors argued that the missing effects of implementation intentions and importance manipulation might be due to the nature of the PM task. In fact, participants’ PM performance was quite high (i.e., near the ceiling).
In the second experiment, the authors again assigned participants either to an implementation intention or to a standard instruction condition while manipulating the task-switching demands of the ongoing task within participants. As the ongoing task, participants were required to decide whether sequentially presented black and white drawings were either living or non-living and small or big. The classification rule was specified by a cue presented jointly with the line-drawing. In the non-switching condition, participants had to classify pictures according to either one or the other criterion, whereas the classification criterion was mixed in the switching condition. The PM task consisted in pressing a key whenever two specific pictures were presented (appearing twice). Consequently, the PM task was again focal with specific PM cues. While task-switching demands negatively affected PM performance, implementation intentions had no impact on PM accuracy or PM RTs.
These outcomes contrasted with the authors’ expectations and with results obtained from previous studies including healthy young and older adult participants in which implementation intention was effective even when the PM cues were focal (Chen et al.,
2015). In addition to the ease of the PM task as a possible cause for the null effect, Kretschmer-Trendowicz et al. (
2021) proposed another possible explanation; they hypothesized that the lack of experience with this kind of strategy would not have permitted its beneficial effect (Einstein & McDaniel,
2007). Accordingly, children need more occasions than adults to become familiar with the strategy (i.e., metacognitive utilization deficit; Miller,
1994; Miller & Seier,
1994). This claim is supported by studies evaluating the effect of implementation intentions in children on different cognitive or non-cognitive tasks revealing contrasting results with some producing benefits (e.g., Duckworth et al.,
2013) and others producing no effects (e.g., Peach & Martin,
2017). The authors also argued that, in the standard condition, children were asked to repeat the instructions several times to maintain a similar length of the PM encoding phase across conditions; this request might have similarly benefited PM performance as implementation intentions (Kretschmer-Trendowicz et al.,
2021).
In response to these contrasting and limited results, a very recent study has investigated the effect of this strategy in children between 7 and 11 years of age on different PM tasks (i.e., event-, time-, and activity-based), as well as underlying processes and possible moderators of the effect (Yang et al.,
2023). Participants were required to play the Fishing Game, a computerized PM task developed for children (Yang et al.,
2011). While the children were asked to catch as many fish as possible (i.e., ongoing task), they were asked to click on the cat every minute to feed it (i.e., time-based task), to click on the cat whenever a striped fish appeared (i.e., focal and specific event-based task), or to press a button at the end of the game to pull the fishing boat ashore (i.e., activity-based). In line with the previously described studies including older children and adolescents, the children in the combined verbal and visualization group did not outperform those in the standard group in either PM task. However, a significant interaction between WM and PM task type, suggested that children with high WM capacity benefited from implementation intentions in the time-based PM task. The authors have argued that implementation intentions strengthen the intention and that children with higher WM are better able to keep the strengthened intention in mind and to engage monitoring resources more efficiently. Moreover, they have argued that their results are in line with those from a study by Geurten et al. (
2016) showing that children’s time-based PM is predicted by metamemory only when their cognitive resources are high. Finally, the findings were also in accordance with the metacognitive utilization deficiency hypothesis, according to which both metamemory and cognitive resources are needed to efficiently use a strategy (DeMarie et al.,
2004).
These three studies similarly show that implementation intentions do not appear to work effectively in children. The first two studies (Kretschmer-Trendowicz et al.,
2021; Zimmermann & Meier,
2010) used focal PM tasks, whereas the third also included activity- and time-based PM tasks (Yang et al.,
2023). Implementation intentions also do not appear to be effective with the latter PM tasks. However, in the time-based PM task, children with high WM appear to obtain some benefit from implementation intentions, thereby supporting the idea that this strategy might be too resource-demanding for children.
Episodic future thinking
Like the visual format of implementation intention, EFT consists of visualizing PM instructions. It has been argued that the two encoding strategies are similar because they both require simulating the future task and involve linking the intention to the context in which it has to be retrieved (Addis et al.,
2008). However, unlike implementation intentions, EFT has produced promising effects on children’s PM performance. For instance, Kretschmer-Trendowicz et al. (
2016) compared a group of 5- and 7-year-old children who obtained EFT instructions with a control group of the same age. Participants were asked to name pictures depicted on cards and say the word “juice” whenever they encountered a picture of a fruit or a vegetable. This PM task was focal to the ongoing task and included a categorical PM cue, thus, it was moderately resource-demanding. Children in the EFT condition were required to close their eyes and were guided to visualize task execution. Children in the EFT group significantly outperformed children in the control group. The authors also found that 7-year-olds profited more from EFT instructions than 5-year-olds. They argued that this finding could be explained by developmental differences in EFT abilities, with older children being better able to imagine the future task and consequently benefitting more. This argument is also supported by a study by Nigro et al. (2014) showing that EFT and PM are related starting from the age of 7 years.
In a further study, investigators asked 10- to 12-year-olds to perform a complex task with real-life task demands (Kretschmer-Trendowicz et al.,
2017). This task consisted of a “sight-seeing tour” in the laboratory that served as ongoing task and which required children to perform activities such as throwing marbles in a bucket or building a tower by piling small wooden sticks on the back of a wooden camel. The PM tasks were either social or neutral and required children, for example, to fill the experimenter’s glass when it was empty (social PM task) or to put on a jacket that was in another room when the experimenter opened the window (neutral PM task). Consequently, the PM tasks were non-focal to the ongoing task and were relatively resource-demanding. Again, the group who obtained EFT instructions significantly outperformed children in the control group, and this effect was independent of the nature of the PM task (neutral vs. social).
Concerning the effect of EFT instructions on ongoing task performance, results of the first study by Kretschmer-Trendowicz et al. (2016) showed that, although children’s ongoing task performance was generally worse in the block containing the PM task compared to the baseline, there were no additional monitoring costs attributable to the encoding strategy. Similarly, there were no differences between the groups regarding ongoing task performance in the second study, suggesting that EFT did not affect monitoring costs (Kretschmer-Trendowicz et al.,
2017). Thus, EFT appears not to affect the amount of attentional resources engaged to perform the PM task and probably to strengthen the association between the cue and the action.
EFT appears to be a useful strategy for improving PM in children, at least starting from the age of 7 years, when the ability to envision future situations appears to be sufficiently developed (Kretschmer-Trendowicz et al.,
2016). These results substantially differ from those related to the implementation intention strategy, which has been argued to be highly similar to EFT, except for the if–then statement. This difference in the effectiveness of the two strategies might be due to methodological differences among studies, such as PM task characteristics. In fact, implementation intentions have not been shown to have any effect on children’s performance on focal and specific PM tasks (Kretschmer-Trendowicz et al.,
2021; Yang et al.,
2023), whereas a numerical but non-significant difference has been observed when the PM task is focal and categorical (Zimmermann & Meier,
2010). Moreover, in Yang et al. ’ (
2023) study, children with higher WM capacity benefitted from implementation intentions in a time-based task. Similarly, EFT efficiently enhanced children’s PM performance in a focal and categorical PM task as well as in a non-focal PM task (Kretschmer-Trendowicz et al.,
2016,
2017). Because responding to focal and specific PM cues appears to require fewer executive resources than responding to categorical, non-focal, and time-based PM tasks, these strategies might be likely to be particularly efficient when the PM task is resource-demanding and when children have sufficiently developed cognitive resources to both perform the task and adopt these strategies. Finally, the if–then statement in the implementation intention strategy might involve metacognitive abilities and additionally overload children’s limited resources. Nevertheless, drawing any conclusion from the few existing studies is difficult, because no study has compared EFT with implementation intensions in children. Moreover, no studies to date have assessed the effect of EFT on children’s performance in less demanding PM tasks, including focal event-based PM targets, or the effect of implementation intentions on non-focal event-based tasks. Thus, more evidence is needed to better understand the reasons for these contrasting results.
The beneficial effect of making performance predictions was first shown in adults (Meier et al.,
2011; Rummel et al.,
2013). These studies found that participants who were asked to predict their PM performance outperformed participants who received standard instructions. Moreover, this positive effect was revealed when the PM task was categorical but not when it was specific (Meier et al.,
2011). Rummel et al. (
2013) showed that making performance judgments prior to performing a PM task also affected ongoing task performance by slowing down RTs, affecting attentional monitoring costs. Cottini et al. (2018) aimed to replicate these outcomes within a developing population. The authors asked 7-year-olds to perform a computerized picture-classification task in which they had to determine whether sequentially presented objects were part of different rooms of a house. Children had to perform the task twice with two different embedded PM tasks, one in which they were required to press a key whenever the picture of a fruit appeared (focal with categorical PM cue) and the other when three specific objects appeared (focal with specific PM cues). Participants were divided into two groups: one was asked to predict PM performance in a yes/no manner (“
Will you remember or not?”) and to express confidence (“
How sure are you to remember/forget?”); the control group received standard instructions. Children obtained higher PM accuracy rates when making PM performance predictions. A prediction advantage was found in the categorical but not in the specific PM task replicating the adult studies. In a subsequent study, in which a specific PM task with high WM demands was used, the authors replicated that outcome and showed that performance predictions improved RTs to the PM cues but not accuracy rates (Cottini et al.,
2019).
Effects on attentional monitoring costs were found only with the categorical PM task, with children in the prediction group displaying slower ongoing task RTs than the control group (Cottini et al.,
2018). To gain deeper insights into this effect, the authors conducted further analyses on the prediction group by exploring the effects of prediction accuracy and confidence judgments. While results on prediction accuracy were inconclusive, confidence judgments were negatively related to ongoing task performance slowing; children expressing lower confidence displayed higher RTs than those who were highly confident to remember the PM task. Similarly, Cottini et al. (
2021b) replicated this pattern by revealing a significant negative correlation between confidence judgments and attentional monitoring costs. Marsh et al. (
2005) and Smith (
2016) hypothesized that when we form an intention, we predict how likely it will be that we will remember or forget. We adjust our attentional resources to execute the task based on these predictions. Whereas adults are inclined to underrate their future PM performance (Kuhlmann,
2019), children frequently overrate it (Cottini et al.,
2018,
2021b; Lavis & Mahy,
2021).
In fact, children’s metacognitive monitoring abilities develop during the school years and become increasingly accurate around 10–13 years (Bertrand et al.,
2017). Nevertheless, asking children to predict their PM performance appears to be effective, although they are inaccurate in their predictions. Encouraging children to reflect on their future PM performance might likely induce them to engage more attentional resources and simultaneously strengthen the encoding of the PM task. From a developmental point of view, this intervention might be more beneficial for older children who have already developed some insight into their PM ability and have more cognitive resources to engage in attentional monitoring. However, developmental studies on the effect of performance predictions, including more extensive age ranges, are lacking.
Episodic future thinking combined with performance predictions
Starting from the evidence demonstrating the beneficial effect on PM of EFT and performance predictions, a recent study investigated the effect of combining the two encoding strategies (Cottini,
2021a). Children between 8 and 11 years were divided into standard, prediction, EFT, and combined EFT + prediction groups. After receiving encoding instructions, all children were required to perform a computerized ongoing picture-classification task. The embedded PM task consisted of pressing a key every time a picture of a fruit appeared during the ongoing task. Children in the combined encoding group obtained significantly higher PM accuracy rates than children in the single-prediction or EFT groups. This PM performance boost was not accompanied by additional attentional monitoring costs, as all groups displayed similar ongoing task performance slowing from the block without and the block with the embedded PM task.
Interestingly, when comparing children in the prediction-only with the EFT + prediction group, children in the combined encoding group who made a lower prediction had a significantly higher PM performance than those who made a higher prediction. Although this was a preliminary finding, the authors suggested that EFT might positively impact PM performance and performance predictions that, in turn, might have had an additional effect on PM performance. Nevertheless, future studies are needed to unravel these two effects and shed light on their nature.