Written or Drawn Episodic Future Thinking Cues Improves Delay Discounting in Children

Abstract

Most people prefer immediate to delayed rewards, even when the immediate rewards are smaller, and children have stronger preferences for immediate rewards compared to delayed rewards. Thinking about vivid, positive future events, or episodic future thinking (EFT), increases the extent to which adolescents and adults choose larger delayed rewards over smaller immediate rewards (delay discounting, DD). While EFT has improved DD in adolescents, previous methodology used auditory or written cues, relying on verbal elaboration and reading of cues to improve DD. Research on autobiographical and episodic memory has shown improvements in retention when using self-drawn images. Images and visual cues are frequently used in studies with children and research is needed to study the use of visual EFT cues as a method to improve EFT effectiveness on DD in children. Research in adults showed that both written and drawn EFT cues improved DD compared to a written episodic recent cue control group. In this study 10-13-year-old children (n=58) were randomized to an EFT-drawing, EFT-writing or an episodic recent thinking (ERT) control drawing group and completed a DD task and a delay of gratification task for real monetary rewards. We found that both EFT drawing and EFT writing improved DD and delay of gratification for real monetary rewards. This suggests EFT is effective at improving DD in multiple modalities.

Introduction

Research in both adults and adolescents have shown that episodic future thinking (EFT), or imagining detailed, positive future events, shifts preferences from smaller immediate rewards to larger delayed rewards as measured by delay discounting (DD) tasks (Atance & O’Neill, 2001; Daniel et al., 2013; Peters & Buchel, 2010). In adults, this has shown to improve behaviors in the laboratory and field, including decreasing energy intake (Daniel et al., 2013; O’Neill et al., 2016), increasing healthy choices for grocery shopping (Hollis-Hansen et al., 2018; Hollis-Hansen et al., 2020), and decreasing cigarette smoking (Stein et al., 2016) and alcohol use (Voss et al., 2022). For children and early adolescents, EFT has shown mixed results on delay discounting or delay of gratification tasks, with at least one study showing effectiveness in improving discounting (Daniel et al., 2015), while others show no effects in children under 12 years of age (Bromberg et al., 2017; Burns et al., 2022).

Both prospection and delay discounting develop across childhood through adolescence. Prospection ability begins around ages 4-5, and a series of studies have shown the emergence of prospection in this age group with various observational and experimental methods (Atance, 2008; Atance & Jackson, 2009; Atance & Mahy, 2016; Atance & Meltzoff, 2005, 2006; Atance & O’Neill, 2005; Caza & Atance, 2019; Nigro et al., 2014; Russell et al., 2010). Delay discounting develops later across adolescence and pre-adolescence, with final development happening in the early 20’s (Gollner et al., 2017; Lu et al., 2023; Steinberg et al., 2009). As children are developing the ability to delay gratification, it is possible using EFT during this period may improve delay discounting, promoting preferences for delayed rewards over immediate rewards.

Children and adolescents are not generally future-oriented (Gollner et al., 2017) and a greater portion of children prefer smaller immediate rewards over larger delayed rewards than adults (Green et al., 1999). While episodic memory and future thinking develop as early as preschool ages, using EFT as a way to improve prospection is an ability that comes online around age 7 (Nigro et al., 2014). A study in 4-5 year old children showed EFT did not improve a delayed choice task (Burns et al., 2021a). This study also included past episodic thinking and semantic thinking groups, which had lower scores than a no-cue control group, suggesting that EFT and other cued thinking requires increased cognitive processing that is not yet developed in this age group. Seven to eleven-year-old children did best on a prospective memory task when prompted with both EFT and prospective memory, compared to each individually and a standard control (Cottini et al., 2021). When examining the relationship between EFT and prospective memory, there was a significant correlation for 7-year-olds, but not 4-6-year-old children (Nigro et al., 2014). The younger children were able to form prospective memories and episodic future thoughts, but the ability to use EFT to improve prospection may develop at approximately age seven. For older children, there is a consistent effect of EFT on DD in children and adolescents 12 and older, with most studies showing non-significant effects for 9-11-year-olds (Bromberg et al., 2017; Burns et al., 2022), with the exception of Daniel et. al (Daniel et al., 2015).

Adults can connect EFT cues to real-life goals through narratives (Epstein et al., 2021), increasing the vividness and efficacy of the cues on delay discounting. The current EFT cue creation and administration relies heavily on reading skills, as children and adults are asked to talk about and write sentences describing each event and read these cues during decision tasks. Reading skills develop in individual variable patterns, with children meeting milestones at various ages (Lerkkanen et al., 2004). For younger children, reading skills are tied to executive function skills, while this relationship weakens for children older than 10 (Ferrer et al., 2007). Expanding EFT to younger children may necessitate innovations in EFT cue creation and administration focusing on age-appropriate skills other than reading and writing.

One such skill that may be implemented in EFT cues is drawing. Children start to scribble, a precursor to drawing, around 18-months of age, with representational drawings occurring around 36 months (Morra & Panesi, 2017). Milestones in drawing are connected to improvements in executive function (Panesi & Morra, 2016) and children are able to create drawings that are more easily differentiated and represent a range of concrete concepts about age 6 (Panesi & Morra, 2016). In a study examining the relationship between executive systems and drawing output, there was a relationship between verbal skills and drawing in 2-7 year old children, but not 8-11 year old children (Toomela, 2016). This suggests that verbal ability and drawing skills are unrelated in children around middle childhood, which may offer a way to include additional details and vividness in an EFT cue separate from the child’s verbal description of the event.

An innovation in EFT cue administration is using self-drawn cues to improve larger-later reward decision making in a delay discounting task. Two published studies have examined self-drawn cues improving DD in adults, with one study focusing on an experimental laboratory manipulation (Carr et al., 2021), while the second examined within subject changes in adult women enrolled in a substance use rehabilitation program (Hudson et al., 2024). Both studies showed EFT effectiveness with self-drawn cues on delay discounting, compared to written cues (Carr et al., 2021), or compared changes in discounting pre- and post-EFT (Hudson et al., 2023).

Episodic future thinking is related to episodic memory (Atance & O’Neill, 2001), as imagining future events involves considering previous life experiences as well as imagining new ones to establish a credible future event. Thinking about the future involves episodic memory, as one needs to remember past events to simulate realistic future events (Martin-Ordas et al., 2014). The constructive episodic simulation hypothesis suggests that a single system can recall and recombine memories of past events to imagine future events (Schacter & Addis, 2007). A recent study in children suggests that drawing is a more effective technique than verbal cues for recalling an event, as drawing allows the interviewer more opportunities to ask questions and prompt details (Søndergaard & Reventlow, 2019), in addition to allowing children to create ‘self-generated’ cues (Barlow et al., 2011). The authors suggest that drawing is best for events that are distinctive, emotional and interesting to the child, with additional benefits for narratively coherent events (Barlow et al., 2011), characteristics that may make drawing uniquely suitable for episodic future thinking.

Drawing has also been used for mood inductions or mood alteration (Drake & Winner, 2013; Monnier et al., 2018), suggesting that drawing improves emotional self-regulation in young children better than using verbal cues to report emotions (Brechet et al., 2022; Drake & Winner, 2013; Monnier et al., 2018). Drawing has been shown to improve mood after a negative mood induction in children and improved feelings of calmness (Brechet et al., 2022; Monnier et al., 2018). This study showed that drawing improved children’s ability to regulate their emotions. While emotional self-regulation is distinct from DD, drawing allows for better emotional regulation, which may contribute to decision making in a ‘cool’, or lowered emotional arousal, rather than ‘hot’ or high emotional arousal state (Zelazo & Carlson, 2012). Based on this previous research, it could be hypothesized that the act of drawing itself could improve ability to choose delayed versus immediate rewards.

While delay discounting is the main outcome when examining EFT effectiveness in adolescents and adults (Ye et al., 2022), no studies have examined EFT on real monetary rewards in children and preadolescents. The current study made use of a delayed monetary choice task (Wulfert et al., 2002b), previously used in 14–18-year-old adolescents, in which children are provided with options for real monetary rewards immediately or delivered 1-week later. Few studies have examined EFT on real world behavior in children, with just one study examining EFT on laboratory energy intake in 9–14-year-old children (Daniel et al., 2015).

The goal of this study was to compare drawn versus written episodic cues on DD and a delayed monetary choice task in 10-13-year-old children. This age group was chosen to include children who have the ability to complete DD tasks (Staubitz et al., 2018), and previously demonstrated effectiveness of EFT on DD using the methodology described here (Daniel et al., 2015). Children were randomized to drawing EFT cues, writing EFT cues or a control episodic recent thinking (ERT) drawing condition, in which children used events from the previous 5 days as cues. ERT drawing was chosen as the control, as opposed to ERT writing, to control for the potential for drawing to provide an effect on emotional regulation. Children completed a cued version of a DD task, in addition to a real-reward delayed monetary choice task. Individual differences in prospection and future orientation were measured. We hypothesized that drawing and writing EFT cues would result in lower rates of discounting in children compared to drawing ERT cues. Exploratory analyses included examining potential moderators of the effect of EFT on DD.

Methods

Transparency and Openness

We report how we determined our sample size, all data exclusions, all manipulations and all measures in the study and the study follows JARS (Journal Article Reporting Standards). Data, analysis code and research materials are available by reasonable request to corresponding author. Data was analyzed using SYSTAT 11.0 (Systat Software, 2004) and JASP version 16.3 for Bayesian analysis (JASP Team, 2022). The study design and analysis were not pre-registered.

Participants

Participants were 61 (25 males, 36 females) children ages 10-13 years old (Table 1). Children were recruited from a database of families interested in studies at the [Research laboratory], online advertisements (e.g. Facebook) and ads on the [University] Research website from January 2022–July 2022 in [City, State]. Parents completed a screening survey and children were excluded if the child (1) was diagnosed with depression (James et al., 2021), (2) <10 years old or >13 years old, (3) participated in a study using future thinking methods in the previous 6 months, (4) did not have any personal mobile electronic devices (based on an hypothesis not presented in this paper), (5) was being medicated for ADHD or had developmental delays or (6) a sibling was enrolled. Three participants were excluded from analysis; one participant was excluded due to being unable to follow directions (drew all events in writing group), and two for completing the adjusting DD incorrectly, for a total n = 58.

Table 1.

Participant Characteristics by Group (Mean ± SD).

	Total	EFT-write	EFT-draw	ERT-draw	p-value
N	58	19	19	20
Males/Females (n)	25/33	9/10	8/11	8/12	0.89
Age (years)	11.9 ± 1.2	11.8 ± 1.2	11.7 ± 1.2	12.1 ± 1.1	0.71
Age Range (years)	10.1 – 13.9	10.1 – 13.9	10.2 – 13.7	10.1 – 13.8	--
Race (n)					0.50
Black	7	2	1	4
Asian	5	3	2	0
White	42	13	15	14
More than one	4	1	1	2
Household Education (years)	16.8 ± 2.2*	17.1 ± 2.7	16.6 ± 2.1*	16.8 ± 1.9	0.76
Household income (US $)	137004 ± 108506*	121194 ± 66817*	159210 ± 156150	130137 ± 82629	0.84
Violated JB Rules
JB Rule 1	18	6	6	6	0.99
JB Rule 2	3	2	1	0	0.33
Either	19	6	7	6	0.89
Consideration of future consequences
Future	24.9 ± 4.3	24.6 ± 3.8	25.1 ± 4.0	25.0 ± 5.3	0.93
Immediate	21.9 ± 5.1	20.6 ± 4.8	21.4 ± 5.8	23.7 ± 4.4	0.15
Average Paper Cue Generation Time (seconds)	214 ± 36	199 ± 42	229 ± 22	213 ± 35	0.028

^*One datapoint missing

Power and Sample size

Previous research on effect of EFT on DD in children (Daniel et al., 2015), showed an effect size (Cohen’s d) of 1.069. With 80% power and alpha =0.05, this effect can be detected with 15 per group. As there is no previous data on modality type effects on discounting in children, 20 per group were estimated, for a total of 60 children. The main goal of this study was to examine if drawing EFT cues is effective in children on delay discounting and it was decided not to include the full 2x2 design, excluding the possible ERT writing group.

Measures

Delay Discounting.

Adjusting amount DD (Frye et al., 2016) was used with five time points, 1-week, 1-month, 3-months, 6-months and 1-year, presented in an ascending fixed order, and a delayed reward of $50. Each time point included five questions for which the immediate monetary amount adjusted based on the child’s previous choice. Adjustments were +/− 50%, 25%, 12.5%, 6.75%, 3.38%, and 1.69% of the delayed reward for each question, with the percentage added when choosing the delayed reward and the percentage subtracted when choosing the immediate reward. For example, the first question for all children was “would you prefer $25 now or $50 in 1 week.” If the child chose $ 25 now, 25% of the delayed reward was subtracted, for a new question asking $12.50 now or $50 in 1 week. If the child chose $ 50 in 1 week, 25% of the delayed reward was added to the immediate reward for a new question of $ 37.5 now or $ 50 in 1 week. This repeated for 5 trials for each time delay, allowing indifference points to be calculated by adjusting the last immediate reward by +/− 50%. Prior to the task, children completed a practice trial for a $10 delayed reward and completed control questions including “would you rather have $0 now or $10 tomorrow” and “would you rather have $10 now or $0 tomorrow.” All children answered the control questions correctly. During the task, written and drawn paper cues were displayed on a stand next to the computer screen before each time period and during each choice. We chose not to display the written cues on the screen, to control for cue presentation between the drawing and writing groups. Before each new time period, children were asked to think about their cue for 30 seconds prior to completing the delay questions for that time period. After each delay time period, children were asked to indicate how much and how vividly they imagined their cues during the previous choices on a Likert-type scale from 1 – 5 (1-not at all – 5-very much). An average vividness rating was calculated by averaging the 10 vividness questions.

Indifference points were calculated for each time delay. An indifference point is the amount of money that is approximately equivalent to the maximum delayed reward and is found for each delay. Area under the curve (AUC) was calculated (Myerson et al., 2001), in addition to ordinal AUC (Borges et al., 2016) as the dependent measures. Area under the curve is calculated by calculated the trapezoidal area between the y-axis and the indifference point on the y-axis at each time delay on the x-axis. Ordinal area under the curve prevents long delays from unduly influencing the number, with the x-axis being converted to consecutive numbers. Johnson-Bickel rules are used to identify non-systematic responding (Johnson & Bickel, 2008) and uses two rules, (1) no consecutive indifference point is less than 10% less than the previous indifference point and (2) the last indifference point is at least 20% less than the first indifference point. Recent research has shown that removing non-systematic data may bias results (Gelino et al., 2025), so sensitivity analyses were conducted and the full dataset was analyzed.

Delayed Monetary Choice Task.

A real rewards task was adapted from (Wulfert et al., 2002a) in which children were offered a bonus payment for their participation of $5 cash now or $10 digital gift card emailed in one week. For this age group, we selected dollar amounts in which the delayed reward was double the immediate reward for ease of child understanding. The task was completed immediately after the cued DD task and the cues remained on display but were not re-cued during the decision, as children did not create a 1-week cue to match the delay time period used in the choice task.

Consideration of future consequences scale.

The 14-item consideration of future consequences scale was administered, with items rated on a scale from 1-not at all like me to 5-extremely like me (Joireman et al., 2012). The consideration of future consequences has been administered to 11-17 year old children (Simmons et al., 2022) and has been validated and shown to have good internal reliability (Joireman et al., 2012). Both the immediate and future subscales were scored, by adding up the question scale ratings.

Demographics.

Parents self-reported race and ethnicity for themselves and their child, employment status, household income and education level, in addition to subjective social status using a standard questionnaire (Stewart, 2004).

Procedure

Eligible children were scheduled for a laboratory appointment and randomized to one of three groups (EFT-write, EFT-draw, Episodic recent thinking (ERT)-draw) in pre-randomized blocks of six (1:1:1 allocation ratio), stratified by gender and race/ethnicity (White/non-Hispanic, non-White or Hispanic).

Parents and children completed consent and assent documents upon arrival to the laboratory. Parents were escorted to another room to complete demographic and child screen use questionnaires. Children first completed the Consideration of Future Consequences scale (Joireman et al., 2012), and were then asked to create five episodic cues, guided by the experimenter as described (see Figure 1 for timeline). First, the experimenter had the child view a calendar outlining the range of dates for the cue, and describe in one sentence the event, starting with the time frame (e.g. In about 1 week or About 1 day ago). Children then rated the event on enjoyment, personal importance, excitement, imagery and vividness on a Likert-type scale from 1-not at all to 5-very much. If children rated cue vividness less than 3, they were asked to choose a different event. After rating their initial event cue, children were asked to provide details about the event in sentences, answering questions about who they are/were with, what they are/were doing, where they are/were and how they are/were feeling. Experimenters recorded the children’s verbal cues in the computerized survey but preserved each child’s use of words and grammar and did not correct or paraphrase with the exception of tenses or to ensure all details about the event were positive to the child. This was to ensure that written cues were personalized to the child. The experimenter filled out a checklist for each detailed cue, indicating use of the correct tense, inclusion of specific details and a focus on positives. After creating each detailed cue, children were provided with 4-minutes to complete a paper cue in which they were either asked to draw or write their typed cue, depending on their randomized group. Cues displayed to children were the full cue and children were asked to create their paper cues using the full cue details.

Figure 1. Cue Generation Procedures.

This is an example of the procedure for an episodic future thinking (EFT) cue of 1-week. Procedures are repeated for each subsequent EFT cue. For ERT cues the procedure is the same except cues are written in past tense (e.g. About 1 day ago I was…)

After cue generation, children completed an adjusting amount DD task. Prior to beginning the task, they were asked questions to ensure they understood differences in time and amount of money, then completed a practice trial with the experimenter. Written or drawn cues were displayed next to the screen during the DD task. Immediately after completing the DD task, children were offered a bonus payment of $5 cash now or $10 digital gift card in 1 week (Wulfert et al., 2002a) as delayed monetary choice measure of real rewards. Additionally, parents completed a screen-time measurement questionnaire, a delayed screen task and children had height and weight measured (results are not reported). Children were paid $25 and parents $10 at the end of the appointment and, if chosen, bonus gift cards were delivered electronically to parents after 1 week. All procedures were approved by the [University] Internal Review Board.

Episodic Cue Groups.

Both the EFT and ERT groups followed procedures outlined by O’Donnell et al. (O’Donnell et al., 2017). The EFT cue generation task used time periods of 1 week, 1 month, 3 months, 6 months and 1 year. The ERT cue generation task used time periods of 1 day, 2 days 3 days, 4 days and 5 days ago. Cue checks were completed by the experimenter at various points in cue creation (Figure 1), with EFT cue checks emphasizing, use of present tense (e.g., “I am” not “I will be”) while ERT cue checks used past tense (e.g., “I was”). Both cue types were checked for positive details, inclusion of specific types of details (who, what, where, how feeling) and that the event occurred/would occur over <24 hours. ERT was chosen to control for both episodic thinking and the personal nature of events. Recent thinking has been shown not to orient adults to the future or the past in previous work (Daniel et al., 2016), allowing for a control that does not use retrospection to activate prospection (Prabhakar & Ghetti, 2020; Schacter & Addis, 2007).

Cue Creation Modalities.

Immediately after creating each cue, children were instructed to either draw their event or hand-write their event based on their randomized group. All groups were provided a blank sheet of white paper inscribed with a large rectangle, along with pens, pencils and a set of colored pencils and markers. Children were read instructions aloud by the experimenter and given four minutes to create a written or drawn cue. Instructions provided to children included asking children to include event details, focusing on positive parts of events and a requirement that the child should be able to read or recognize the cue later in the session, as outlined in previous research (Carr et al., 2021). The full detailed, typed cue was displayed on the computer screen along with a 4-minute timer. Children were asked to include all the details in the cue and, if in the written group, to write the entire cue. Children were allowed to move onto the next time period when they indicated they were done with their paper cue. Timing of cue creation is shown in Table 1 by group. After creating the paper version of each cue, a manipulation check asked if (1) participants could read or identify their cue, (2) they included all the details from the typed cue, (3) they added extra details and (4) they focused on positive parts of the cue. Cues were rated by experimenters at the end of the study for the use of representational imagery (e.g. stick figures, lack of relative size or perspective or use of symbols) versus true-to-life imagery (use of perspective, relative size, images of a scene from a point of view) with each cue receiving a 0 (did not use abstract image) or 1 (did use abstract images) as a proxy for drawing sophistication. Cues were coded for extra details by indicating a 1 if the child self-rated that they had added extra details during their paper cue creation and adding these scores across the five cues for a range of 0-5 for each child. Examples of paper cues from all groups are included in supplemental files.

Analytic Plan

Child demographics were examined by group using analysis of variance for continuous variables and chi-square for categorical variables. Group differences in participant characteristics and DD cue vividness were considered as potential covariates in subsequent analyses. For scoring the DD task, area under the curve (AUC) was calculated and the Borges AUC adaption (ordinal AUC), was used (Borges et al., 2016). Pearson correlations were examined between DD, consideration of future consequences immediate and future scales with child anthropometrics, and demographics. Correlations were examined to identify any potential covariates to include in the main analyses.

ANOVA was used to examine main effects of group on DD AUC, with post-hoc linear contrasts to compare differences between groups and Cohen’s d effect sizes were calculated for group contrasts. Sensitivity analyses for the ANOVA were conducted by examining the sample with and without children whose DD task violated one or both Johnson-Bickel rules (Johnson & Bickel, 2008). The real delayed monetary task was a binary choice task and effects of group on were examined using chi-square analysis.

As an exploratory analysis, potential moderators of the effect of group on AUC was examined using regression models with group dummy-coded (ERT-drawing as the reference group) (West et al., 1996). First, a base model with the dummy coded groups was created. Then potential moderators were entered starting with the ones with the highest correlation with delay discounting. Then the interaction was added. Incremental R-squared tests were used to examine if the moderator or the interaction significantly increased the variance accounted for (Recchia, 2010). Immediate and future time perspective, child age, average vividness during DD task, and household education were examined as potential moderators of the effect of group on ordinal AUC one at a time.

Results

Participant characteristics are presented in Table 1. There were no significant differences by group in demographics or consideration of future consequences. There were also no differences in cue vividness ratings during the DD task. Analysis of variance for ordinal AUC showed a significant effect of group (p = 0.001, see Table 2, Figure 2). Post hoc comparisons showed significant differences between the ERT-drawing group and both EFT-drawing (p = 0.003) and EFT-writing (p = 0.006). There was not a significant difference between the two EFT groups (p = 0.97). Examining the ANOVA with exclusions based on J-B rules (Johnson & Bickel, 2008), significant effects of group were found with excluding J-B-rule 1 (n=40, F(2,37) = 6.99, p = 0.003; 0.734 vs. 0.759 vs. 0.490), JB rule 2 (n = 55, F(2,52) = 6.76 p = 0.002; 0.723 vs. 0.714 vs. 0.518) and both (n=39, F(2,36) = 6.578, p = 0.004; 0.718 vs. 0.759 vs. 0.490) (Supplemental Table 2).

Table 2.

Results of Analysis of Variance and Hierarchical Regression Analysis for Ordinal AUC

Analysis of Variance
Effect	F (DF)	p	ηP2	Contrast Tests	t (DF)	p-value*	Cohen’s d
Group (n = 58)	7.62 (2,55)	0.001	0.217	ERT-draw vs. EFT-draw	3.49 (55)	0.003	1.95
				ERT-draw vs. EFT-write	3.23 (55)	0.006	1.86
				EFT-draw vs. EFT-write	0.25 (55)	0.97	0.081
Hierarchical Regression Analysis
Model	R2	ΔR2	p-value	Variables	B**	p-value
Group	0.217	--		Dummy-coded EFT-drawing	0.221	<0.001
				Dummy-coded EFT-writing	0.200	0.002
Immediate time perspective***	0.307	0.09	0.011		−0.014	0.012
	0.311	0.004	0.847	x EFT-drawing	0.002	0.884
				x EFT-writing	0.007	0.590
Household Education	0.351	0.042	0.071		0.021	0.071
	0.367	0.016	0.536	x EFT-drawing	−0.034	0.270
				x EFT-writing	−0.015	0.579
Future time perspective	0.317	0.010	0.383		−0.005	0.383
	0.318	0.001	0.966	x EFT-drawing	0.004	0.798
				x EFT-writing	0.000	0.973
Child age	0.270	0.002	0.737		0.007	0.737
				x EFT-drawing	0.013	0.822
				x EFT-writing	0.043	0.438
Cue Vividness	0.307	0.000	0.933		−0.002	0.933
	0.332	0.025	0.387	x EFT-drawing	0.080	0.212
				x EFT-writing	0.092	0.304
Use of Additional Details	0.312	0.005	0.395		0.009	0.532
	0.321	0.009	0.328	x EFT-drawing	−0.024	0.530
				x EFT-writing	−0.025	0.454
Use of representational imagery&	0.375	0.020	0.289		−0.016	0.289
	0.375	0.000	0.947	x EFT-drawing	−0.002	0.947

^*Tukey p-value correction,

^**Unstandardized B,

^***Immediate time perspective main effects were included in subsequent models,

^&Model for use of representational imagery did not include the EFT-writing group, as this group did not produce images that could be coded; Note, dummy coded variables includes the ERT-drawing group as the comparison, so significant effects for dummy coded variables are interpreted as a significant difference between the named group and the comparison group.

Figure 2.

A. Indifference points by group. Mean + SEM. B. Ordinal Area Under the Curve by group. *p<0.05

For the delayed money choice task, there was a significant effect of group (X²(2) = 6.69, p = 0.035). Children in the EFT-writing group all chose the delayed reward, 16/19 children in the EFT-drawing group chose the delayed reward and 14/20 children chose the delayed reward in the ERT-drawing group. When comparing each group, there was a significant difference between the EFT-writing and ERT-drawing group (X²(2) = 6.73, p = 0.009), and there was not a significant difference between EFT-drawing and ERT-drawing groups (X²(2) = 1.11, p = 0.29). Examining the EFT-writing and EFT-drawing groups, there was also not a significant difference (X²(2) = 3.26, p = 0.07).

Correlations were examined as an initial step in determining covariates to include in the regression analysis. Significant correlations (Supplemental Table 2) were found between ordinal AUC and CFC immediate scale (r = −0.40), delayed choice task and CFC immediate scale (r = −0.40), and between delayed choice task and cue vividness during the DD task (r = −0.26). When examining the relationship between delayed choice task and cue vividness, the correlation is driven by the relationship within the ERT group (ERT group r = −0.57, p = 0.0083, n = 20; EFT groups r = −0.10, p = 0.55, n = 38).

Hierarchical regression analyses were conducted to examine potential moderators of group effects on ordinal AUC. All main effects and interactions between dummy-coded group and the moderators were included. Potential moderators included participant characteristics that may impact delay discounting, including immediate time perspective, future time perspective, child age, and household education, as well as task characteristics that may impact the effectiveness of EFT, including average cue vividness during the DD task, use of representational or abstract drawing cues and inclusion of extra details not in the typed cue. Potential moderators with the highest correlations with the discounting task were entered first into the hierarchical regression and R² analyses were examined to see if they significantly increased the variance accounted for. Results of the hierarchical regression are shown in Table 2. There were no significant interactions with group. Immediate time perspective was the only moderator that significantly improved the variance accounted for and was significantly related to discounting while controlling for group. In addition, to examine potential non-linear effects of age, children were split into two age groups, 10-11 and 12-13 and examined the dichotomous variable as a potential moderator. There were also no moderation effects of age group on ordinal AUC.

Discussion

This study shows both writing and drawing EFT cues decreased discounting compared to an ERT-drawing control condition. We had originally hypothesized that drawing EFT cues would be more effective at improving DD than written EFT cues in children, and found that they showed no significant differences. EFT has previously been shown to lower discounting in children over 12 (Bromberg et al., 2017; Daniel et al., 2015), with one study showing effects in children ages 9-12 (Daniel et al., 2015) Methodological concerns in previous studies may explain differences in EFT effectiveness on delay discounting in children. First, some studies in children use episodic tags (Burns et al., 2022), as previously used in adults (Peters & Buchel, 2010). This study and the Daniel et al (Daniel et al., 2015), both included full cues during delay discounting, and cue generation consisted of a variety of details. In addition, most other studies in EFT haven not used present tenses to describe the future events, allowing children to state “I am going to be” rather than “I am” as done in this study. If the goal of EFT is to make the events feel as if they are happening now, use of different tenses may cue children differently. This is also the first study to examine how drawing may improve episodic cues in children, while previous research on children and drawing have focused on semantic (Sykula, 2019) and retrospective memory (Barlow et al., 2011).

In addition, in this study EFT had effects on a real-reward delayed monetary choice task, with more children in the EFT writing condition waiting for the delayed bonus payment than in the ERT control group. It is notable that there were not significant differences between EFT-drawing and ERT-drawing, but the study was originally powered to examine differences in EFT effects on discounting. Previous research in younger children have examined effects of EFT on a delayed choice task using non-monetary rewards, including toys, stickers and candy. In a study in 4-5-year-olds, using EFT increased choices for immediate reward (Burns et al., 2021b). The authors pointed out that all the non-control conditions, including EFT, past thinking and imagination, actually did worse on the task compared to a no-cue control condition in which children completed a coloring activity to control for cue generation time. This suggested that children in this age range did not have the working memory capacity to improve delayed choices using future thinking. In a sample of 6-7-year-olds, EFT cueing using breakfast tomorrow as the episodic event for all children did not improve a delayed choice task for watching videos (Tanguay et al., 2023). All children were prompted with the same event, breakfast, which, while personal, is not distinctive as it occurs daily. Additionally, this event may not be far enough in the future to prompt future thinking in this age group. Finally, in a sample of 7-10-year-olds, the delayed choice task was examined within two EFT interventions, the first intervention was reward related, in which children were asked to think about spending toy coins on the delay choice rewards, and the second intervention was reward-unrelated, in which children were asked to think of three events that would make them happy versus a no-cue control (Canning et al., 2024). The authors did not find any effects of cues on the delay choice task. The cues used in this study were recorded by the experimenter on an index card and did not indicate tense, if it was written in the child’s own words, or if it was the full details of the cue. Additionally, the directions to imagine receiving the reward may have biased children towards immediate rewards, as seen in previous work by Mischel and colleagues (Mischel & Ebbesen, 1970; Mischel & Moore, 1973; Peake et al., 2002). This expands the use of EFT in children to real delayed monetary rewards with EFT-writing improving choices compared to EFT-drawing or ERT-drawing.

Research examining EFT effectiveness on improving DD in children have had mixed results. Several studies have shown the effectiveness of using EFT to improve discounting in adolescents (Bromberg et al., 2017; Bromberg et al., 2015). For children, three studies examined EFT’s effect on DD in age groups <14 years old. The first study showed a significant effect of EFT on discounting for children 9-14 years old, in addition to decreasing energy intake in a snack food ad libitum task (Daniel et al., 2015). In the second study, EFT discounting effects were found with pre-adolescents 12-14 years old, but not for children 8-11 years old, using no-cue control and episodic past memory group comparisons (Burns et al., 2022). For the younger children, the measures were conducted across two sessions, with cue generation occurring at the first session and the cued DD task occurring the next day. The separation of cue generation from use may have decreased cue effectiveness. In our sample of 10–13-year-old children, age did not moderate the effects of EFT on DD as either a continuous variable or when groups into 10-11-and 12–13-year-old age groups. This suggests methodological differences in administering EFT may be important in children but not adolescents, with EFT cue creation occurring shortly before completing cued DD tasks.

These results replicate research on drawing versus writing EFT cues in adults (Carr et al., 2021) and extends the types of cue modalities that can be used for EFT with children. In adults, immediate future time perspective moderated the effect of group, such that drawing improved discounting rates for adults with a high immediate time perspective, but writing about future episodic events did not (Carr et al., 2021). There were no moderation effects for age, sex, household education, child race, child minority status, immediate or future time perspective in children. This suggests that EFT-drawing and EFT-writing have generalizable effects for many types of children. Future research should examine how differences in reading or writing ability or drawing self-efficacy may moderate the effects of EFT modality on discounting and other preventative health behaviors.

Self-reported cue vividness during the DD task was significantly correlated with performance on the real delayed monetary choice task, but not DD. It’s possible that completing the uncued delayed choice task immediately after the cued DD task was influenced by the vividness of the previous cues. While cue vividness was not related to DD, future studies should examining if using drawing to generate an event may improve narrative information in children (Gross et al., 2009), as this been shown to improve EFT effectiveness in adults (Epstein et al., 2021).

EFT has been hypothesized to impact delay discounting through expansion of the temporal window (Bickel et al., 2023), which allows events further in the future to feel like they are happening now. It is likely that there is some amount of working memory capacity, or the ability to simultaneously hold a thought or idea while manipulating other information (Ilkowska & Engle, 2010; Reynolds et al., 2022), that impacts the ability to use EFT to improve choices. This is evident when examining use of EFT in young children, as children ages 4-5 performed worse on discounting choice task when asked to think about future, past or semantic memories, compared to a control in which children colored for 5-minutes to control for cue-creation time and completed delay choice tasks with no cues (Burns et al., 2021b). For studies that have shown no effects of EFT on discounting in children 9-12 years of age, it may be that the cues need to be personal, vivid, use appropriate tenses and include the full cue for each decision to be salient enough to improve delayed choice, as these methods were successful in the current study, as well as in (Daniel et al., 2015), and were not used in the studies discussed above.

It is important to note that this study manipulated the presentation of EFT cues, and did not manipulate the generation of the events for cues. Cue presentation is an important aspect of EFT to study, as various cue types may be easier to present in real-life interventions. For example, using drawn cues to remind one of behavioral goals while at work or school may be easier for many people to use than written or audio. How cues are generated and details are added have not been systematically studied for EFT in children or adults. For children, this is an interesting question, as it has been shown that using drawing to start conversations with young children about events provides more detailed information (Søndergaard & Reventlow, 2019). However, we chose to use previously methodology shown to improve DD in younger children (Daniel et al., 2015), and only manipulate the cue presentation modality.

A common question when considering novel ways to present images as EFT cues is whether to allow self-generated drawings or have experimenters provide stock images. While differences between self-drawn cues and provided stock images have not been examined, drawing may be uniquely suited to EFT. First, research on semantic memory has tested provided images versus self-generated images on to-be-remembered information, and self-generated images improved memory better than written information, read information, or provided images (Wammes et al., 2016). Semantic memory research has also shown drawing improved short- and longer-term recall for items and concepts with self-generated imagery, compared to written notes, tracing or viewing pictures (Fernandes et al., 2018). A more recent study in adults has shown improvements in autobiographical memory, or episodic memory, for adults when events were drawn (Tran et al., 2022). Events recalled after 2-weeks were compared to daily diary entries and drawing improved accuracy and increased detail compared to writing, suggesting that drawing can improve both semantic and episodic memory. For children, studies on art-therapy have shown that self-created images improve therapeutic discussion, increasing children’s provided details to the therapist by discussing their self-created images (Søndergaard & Reventlow, 2019), again suggesting that self-creation may be an important component.

When considering future studies and age groups of children that may benefit from EFT-drawing, the development of prospection is an important consideration. Children show improvements in remembering ‘when’ something needs to be done between the ages of 7 and 10, with ‘remembering-to-do’ a task already developed by age 7 (Guarjardo, 2000; Kavavilashvili, 2001). Prospection continues to improve from childhood to adolescence, with adolescents becoming more future oriented (Steinberg et al., 2009). The ability to use EFT as a way to improve prospection does not seem to develop until about age 7 (Nigro et al., 2014). A study in 4-5 year old children showed EFT did not improve a delayed choice task (Burns et al., 2021a), but also post episodic thinking groups, as well as semantic thinking groups showed lower score than a no-thinking control group. These results suggest that EFT and other cued thinking requires cognitive processing that is not yet developed in this age group. Seven to eleven-year-old children did best on a prospective memory task when prompted with both EFT and prospective memory, compared to each individually and a standard control (Cottini et al., 2021). In another study, there was a significant correlation between EFT and prospective memory for 7-year olds, but not 4-6-year-old children (Nigro et al., 2014). The younger children were able to form prospective memories and episodic future thoughts, but the ability to use EFT to improve prospection may develop at around age seven. Future research should examine how drawing EFT cues impacts EFT on delayed choices in children ages 7-10.

This study has some limitations. This study included a majority of families with middle – high income and education, limiting generalizability. Children from under-resourced households may have more difficulty with DD and prospection and may benefit from a future thinking intervention (Evans & Rosenbaum, 2008). Lower household income or education is associated with greater likelihood of lower reading achievement (So & Chan, 1984). This population may benefit from drawing procedures, as it relies less on reading abilities. The study population included children in middle childhood, in which they have several years of school experience reading and writing. For this age group, there may not be much of a difference between descriptive writing and drawing, as opposed to younger children just learning to read and write. While EFT influenced a delayed money choice task, children may have difficulty choosing delayed rewards when the immediate rewards are their favorite activities, such as video games or socializing with friends. Future research should examine the effects of different delayed rewards, in addition to money, as well as studying EFT across a range of ages. Additionally, future studies on drawn cues for EFT should include written EFT cues as a comparison when looking at other behaviors, as it is possible that cue modality may affect different types of delayed rewards differently.

Drawing is an important developmental milestone for children, representing a complex set of skills involving motor coordination, visual representation, motor planning, imagination and spatial awareness (Frith & Law, 1995; Toomela, 2016). Drawing may offer the opportunity to expand the populations that can benefit from EFT, including young children who may not yet read or write, and children with reading or verbal delays. Using drawing as a technique to both generate and cue future events may prove to be effective for children in improving preferences for delayed rewards. Goals for future research should include both increasing effectiveness of EFT for younger children and applying EFT towards behaviors that improve health and involve delayed rewards, including healthier eating and increased physical activity.

Highlights.

• Episodic future thinking (EFT) is an intervention involving thinking about positive and vivid future events

• EFT improves delay discounting in children, shifting preferences to future delayed rewards

• EFT is usually written and requires reading and writing skills to create cues

• Drawing EFT cues showed similar improvements in delay discounting to written cues in children