Skip to main content
NCBI home page
The Journals of Gerontology Series B: Psychological Sciences and Social Sciences logoLink to The Journals of Gerontology Series B: Psychological Sciences and Social Sciences
. 2019 Aug 14;75(1):85–95. doi: 10.1093/geronb/gbz097

Age Differences in Intertemporal Choice: The Role of Task Type, Outcome Characteristics, and Covariates

Corinna E Löckenhoff 1,✉, Gregory R Samanez-Larkin 2
Editor: Nicole Anderson
PMCID: PMC6909431  PMID: 31410482

Abstract

Objectives

Prior research has revealed age differences in the preferred timing of monetary outcomes, but results are inconsistent across studies. The present study examined the role of task type, outcome characteristics, and a range of theoretically implicated covariates that may contribute to variations in age effects.

Method

Two types of intertemporal choice paradigms (temporal discounting and sequence construction) were administered to a diverse life-span sample (n = 287, aged 18–87). The design experimentally manipulated outcome delay (months vs years), amount (hundreds vs thousands), and valence (gain vs loss) while statistically controlling for a range of potential covariates including demographics, affect, personality, time perspective, subjective health, and numeracy.

Results

In the temporal discounting task, no significant age differences were observed and this pattern did not differ by outcome delay, amount, or valence. In the sequence-construction task, age was associated with a preference for sequences of decreasing impact in the gain condition but not in the loss condition, whereas outcome delay and amount did not moderate age effects. Age patterns in discounting and sequences preferences remained unchanged after controlling for covariates.

Discussion

These findings converge with prior studies reporting weak or null effects of age in temporal discounting tasks and suggest that inconsistent results are not due to variations in outcome valence, delay, or amount across studies. Findings also add to the scarce evidence for age differences sequence-preferences. After discussing methodological limitations, we consider implications for future research and practice.

Keywords: Decision making, Intertemporal choice, Temporal discounting

Intertemporal choices, that is, choices involving trade-offs among events or outcomes that occur at different points in time (Loewenstein & Prelec, 1992), are prevalent in people’s lives and can have momentous consequences—from managing one’s cognitive effort over the course of a work session (e.g., Löckenhoff et al., 2018), to scheduling aversive medical procedures or enjoyable experiences over the course of months (e.g., Strough, Bruine de Bruin, & Parker, 2018), to managing ones flow of income and savings over a lifetime (e.g., Bidewell, Griffin, & Hesketh, 2006). Importantly, the timing of an event often affects its magnitude—spending money today means to lose out on compound interest in the future, and failing to fill a small cavity right away may result in major dental work a few years down the road (Frederick, Loewenstein, & O’Donoghue, 2002).

Prior research has reported that intertemporal decisions differ by age (for recent reviews, see Lim & Yu, 2015; Strough et al., 2018), but the direction and size of effects is inconsistent across studies. Such variations may be due to differences in task types and outcomes as well as insufficient consideration of relevant covariates and methodological factors (Lempert & Phelps, 2016; Lim & Yu, 2015). In an effort to address such concerns, the present study experimentally manipulated a key set of task and outcome characteristics that would be expected to moderate age effects while statistically controlling for a range of theoretically implicated covariates.

Theoretical Perspectives

A range of potential mechanisms behind age differences in intertemporal choice have been proposed. Some of them would predict that the tendency to neglect the future increases with age. First, age is associated with shrinking future time horizons and a subjective sense that time is running out (Carstensen, 2006) which might lead older adults to discount the future and favor sequences that implement the most favorable outcomes right away. By the same token, aging is associated with decrements in cognitive and physical functioning and such limitations have been linked with a greater tendency to discount the future (Boyle et al., 2012; Halfmann, Hedgcock, & Denburg, 2013; Huffman, Maurer, & Mitchell, 2017).

Other theoretical perspectives imply that the tendency to neglect the future decreases with age. Emotion-regulatory skills are well-preserved in later life (Cavanaugh & Blanchard-Fields, 2014) and the resulting psychological resources (e.g., better mood and mental health) have been linked with lower temporal discounting (Löckenhoff, O’Donoghue, & Dunning, 2011). Further, experience-based knowledge is thought to accumulate with age and promote more deliberate and less impulsive decision styles which are likely to favor forward-thinking choices (Lim & Yu, 2015).

Finally, some theoretical perspectives suggest that age differences depend on the valence of the outcome. On the one hand, goal orientation theory (Freund, Hennecke, & Mustafić, 2012) argues that age-related developmental losses shift people’s motivational priorities from promoting future growth to maintaining the status quo which leads to an emphasis on loss prevention over gain maximization. In the context of intertemporal choice, this tendency may reduce future neglect, especially when it comes to avoiding long-term losses. On the other hand, there is growing evidence for an age-related “positivity effect” with older adults (as compared with younger adults) favoring positive over negative material in cognitive processing (Cavanaugh & Blanchard-Fields, 2014). Extended to intertemporal choices, the positivity effect implies that age effects are more pronounced for positive outcomes.

In summary, current theoretical perspectives have identified a range of potentially relevant mechanisms leading to a set of partially contradictory predictions. Importantly, the relative role of these factors may differ across different types of intertemporal choice tasks.

Task Type

Temporal Discounting

The most widely researched paradigm in the field of intertemporal choice are temporal discounting tasks (Frederick et al., 2002) which require participants to choose between outcomes that vary in both timing and magnitude/intensity. For instance, participants may be asked whether they prefer to receive $10 in 1 month or $12 in 5 months. Research on younger samples has revealed a robust tendency to advance monetary gains and delay losses (Frederick et al., 2002).

With respect to age differences, prior work has primarily focused on monetary gains. Several studies indicated that age is associated with a decreased tendency to discount future gains (e.g., Green, Fry, & Myerson, 1994; Halfmann et al., 2013; Harrison, Lau, & Williams, 2002; Jimura et al., 2011; Li, Baldassi, Johnson, & Weber, 2013; Löckenhoff et al., 2011; Whelan & McHugh, 2009), but a smaller number of studies found the reverse pattern (Albert & Duffy, 2012; Green, Myerson, Lichtman, Rosen, & Fry, 1996; Huffman et al., 2017), whereas others found curvilinear age effects (Read & Read, 2004; Richter & Mata, 2018), or an absence of age differences (Chao, Szrek, Pereira, & Pauly, 2009; Samanez-Larkin et al., 2011; Seaman et al., 2016).

Sequence Construction

A complementary approach to eliciting intertemporal preferences involves sequence-construction tasks. Participants are presented with a set of predetermined outcomes and asked to arrange them over a given time frame in the sequence they prefer (Frederick et al., 2002; Loewenstein & Prelec, 1993). For instance, participants may be told that they will receive payments of $10, $20, and $30 over the course of 3 months and are then asked to indicate in which sequence they prefer to receive the monthly payments.

The research record on age differences in sequencing is very limited. Only two studies examined monetary outcomes. One study asked participants to imagine an unexpected windfall or debt that would be spread out in installments (Strough et al., 2018). Although no specific time frame was given, adjacent tasks discussing nonmonetary outcomes indicated a month-long time frame. Age was associated with a stronger preference for starting with larger amounts, especially for monetary gains. Another study asked participants to select their preferred sequence for monetary gambles (50% chance of gain or loss) that varied in outcome from $1 to $9 over the course of an experimental session (Löckenhoff, Rutt, Samanez-Larkin, O’Donoghue, & Reyna, 2019). For this scenario, which involved a very short time frame (i.e., 30 min), and variations in income probability (rather than actual income), no age differences in sequence-preferences were found. Studies examining nonmonetary outcomes yielded similarly inconsistent age effects (for a review, see Strough et al., 2018).

In summary, the literature on age differences in intertemporal preferences for monetary outcomes shows mixed results. For temporal discounting, there is a tentative trend suggesting that older adults are more “patient” or less likely to discount the future, but not all studies show this effect. For sequencing, research is very scarce with one study indicating an age-related tendency to schedule more impactful events first and another reporting a lack of age effects. In part, discrepant findings across task types may be explained by conceptual differences in the structure of the tasks. Discounting tasks are presented as dichotomous choices which force participants to make trade-offs between the timing and magnitude of a given outcome. This type of task may primarily reflect age differences in psychological resources for impulse control and age-related increases in deliberate decision styles (Lim & Yu, 2015) and therefore show an age-related tendency toward future-optimizing choices. Sequence-construction tasks, in contrast, offer a predetermined range of outcomes and participants merely select the timing of the outcomes without having to make any difficult trade-offs. This is more likely to tap into age differences in the preferred timing of events and may be sensitive to limitations in time horizons or physical health. Conceivably, older adults may prefer to handle the most impactful events right away if they expect that their resources will be waning. However, the effects of task type remain poorly understood because prior research has not compared age differences for the two task types within a single study.

Outcome Characteristics

Apart from broad differences between discounting and sequence-construction tasks, age effects in intertemporal preferences may be moderated by specific task characteristics and this could contribute to inconsistent results in prior work. We now consider three of these characteristics—delay, amount, and valence—which were previously shown to affect intertemporal choice.

Delay

In prior research, temporal delays have ranged from seconds (Jimura et al., 2011) to decades (Green et al., 1994) and evidence indicates that discount rates decrease as the length of the delay increases (for reviews, see Frederick et al., 2002). Since age-related limitations in time horizons should be more salient for longer intervals, one would expect that age effects are more pronounced when longer delays are involved. In the extant literature this pattern is found for temporal discounting tasks (e.g., Richter & Mata, 2018) as well as sequence-construction tasks where age differences are found for monetary sequences spanning months (Strough et al., 2018) but not minutes (Löckenhoff et al., 2019).

Amount

Intertemporal preferences also vary depending on the amount of the outcome, with higher amounts being discounted less steeply—in part, because large amounts are thought to shift the frame of reference from everyday spending to longer term investment which is less susceptible to impulsive choices (Frederick et al., 2002). If older adults are generally less impulsive, they should be less sensitive to outcome amount. Also, age differences should be less pronounced for higher amounts which elicit more deliberate choices regardless of age. With respect to temporal discounting, this pattern is found in some studies (Green et al., 1994; Vanderveldt, 2017) but not in others (Green et al., 1996; Whelan & McHugh, 2009). For sequence-construction tasks, age differences in magnitude effects have yet to be examined.

Valence

Outcome valence may play a role in age effects as well. Discounting tasks show a marked gain-loss asymmetry with gains being discounted more than losses (Frederick et al., 2002). If anything, this effect appears to be more pronounced with age: Several prior studies comparing age differences in temporal discounting of gains and losses found that age-related increases in patient choices were limited to gains (Halfmann et al., 2013; Löckenhoff et al., 2011; Sparrow & Spaniol, 2018; Experiment 1). However, this effect is not consistently found. Sparrow and colleagues (Sparrow, Armstrong, Fiocco, & Spaniol, 2019; Sparrow & Spaniol, 2018) administered the same type of discounting paradigm across three samples, but only found an age by valence interaction in one of them. With respect to age differences in sequence-preferences, research on the role of valence is limited, but Strough and colleagues (2018) found that age differences in sequence-preferences were larger for gains than for losses. Overall, these findings are more consistent with the age-related positivity effect than with motivational shifts favoring loss avoidance.

The Present Study

In summary, the prior literature on age differences in intertemporal choice is equivocal. There is not only a lack of integration across decision paradigms, but findings are inconsistent within a given paradigm and relevant outcome characteristics and covariates are not systematically controlled. To address these issues, the present study simultaneously examined cross-sectional age differences in temporal discounting and sequence construction. We focused on monetary outcomes since this domain has been the primary target of prior research in this area (Lim & Yu, 2015). We experimentally manipulated three outcome characteristics: Length of delay (months vs years), outcome size (hundreds vs thousands of $), and outcome valence (gain vs loss). Further, we screened for a range of the theoretically implicated covariates discussed above (i.e., time perspective, current affect, self-reported mental and physical health, subjective cognition, and numeracy). We also controlled for demographic characteristics since better education, higher income, and being a member of the majority racial group have been linked to reduced temporal discounting rates (e.g., Huffman et al., 2017; Ishii, Eisen, & Hitokoto, 2017). Finally, we controlled for big-five personality traits since both high neuroticism and low conscientiousness were found to be associated with a tendency to discount the future (Manning et al., 2014).

Given the mixed prior evidence on age differences in intertemporal choice, we acknowledge the possibility that age effects may not be observed in the present study. However, if age differences are found, the theoretical considerations outlined above, along with findings from the prior literature, would predict that age effect show the following pattern.

Hypothesis 1: In the temporal discounting task, age is associated with a reduced tendency to discount future outcomes.

Hypothesis 2: In the sequence-construction task, age is associated with a tendency to schedule the most impactful events (i.e., the largest gains and losses) first.

Hypothesis 3: Age differences are more pronounced for outcomes involving longer delays (3a), lower amounts (3b), and positive versus negative valence (3c).

Hypothesis 4: Age differences are accounted for by variations in time perspective, current affect, mental and physical health, and cognitive variables but remain significant when accounting for variations in demographic characteristics and personality traits.

Although prior studies on age differences in intertemporal choice predominantly report linear effects, exploratory analyses screened for the presence of curvilinear effects.

Method

Participants

Participants were drawn from two sources. We initially recruited an online survey panel (n = 311). In response to reviewer requests, we later added a group of local participants who completed the study in a laboratory setting (n = 61). The Supplementary Material (p. 2) provides details on these data sources and the specific recruitment criteria.

To ensure high data quality, we implemented a series of screening procedures (Supplementary Material, p. 4) including attention screeners, a comprehension screener, and a consistency check comparing self-reported chronological age and birth year. Finally, we removed participants with any missing values on the study variables.

Although the samples differed in some of the demographic and background variables, they did not differ significantly in discounting and sequencing scores or patterns of age effects in these variables (for the full analyses, see Supplementary Material, p. 5). Subsequent analyses were therefore conducted on the pooled sample (final n = 287) which is described in Table 1 (for power analyses, see Supplementary Material, p. 2).

Table 1.

Sample Characteristics and Their Associations With Age and Intertemporal Choice Preferences

Variable M (SD) or % r age τ dis-gain τ dis-loss τ seq-gain τ seq-loss
Age 48.7 (17.4) — .00 −.03 −.15** −.01
% Female 52% −.36** −.07 −.08 −.02 −.07
% Non-Hispanic white 76% .07 −.01 −.02 −.08 .02
Education level 4.06 (1.39) .00 .16** .13** −.07 .02
Income level 3.88 (1.75) −.01 .05 −.02 −.02 −.07
Current affect (valence) 5.09 (1.26) .14* −.08 .01 −.01 .02
Current affect (arousal) 3.45 (1.57) −.05 −.05 −.03 −.03 .02
BFI—neuroticism 2.8 (0.99) −.31** −.04 −.07 .00 −.05
BFI—extraversion 2.89 (0.9) .01 .06 .01 −.05 .02
BFI—openness 3.4 (0.84) −.02 .05 .02 −.08 .05
BFI—agreeableness 3.7 (0.76) .18** .01 −.03 −.07 .07
BFI—conscientiousness 4.02 (0.78) .06 −.04 −.07 −.13* −.06
Limited future time 3.93 (1.83) .13* −.05 −.02 .05 −.01
SRH—physical 3.28 (0.95) −.20** .03 .02 .00 .00
SRH—emotional 3.39 (1.05) .20** .01 −.04 −.04 −.03
Self-rated learning 3.89 (0.89) −.14* .01 .04 .03 −.07
Self-rated memory 3.49 (1.01) −.10 .01 .04 .01 −.04
Numeracy 1.55 (0.99) .06 .22** .13* −.04 .00
Open in a new tab

Notes: Education ranges from 1 = less than high school to 7 = doctorate. Income ranges from 1 = less than $10,000 to 8 = more than $110,000 in increments of $20,000, rage: Pearson correlations with age, τdis-gain/τdis-loss: Kendall’s τ rank order correlations with average annual discount rate for gains/losses, τseq-gain/τseq-loss: Kendall’s τ rank order correlations with dichotomized sequence trend scores for gains/losses. SRH = self-rated health.

*p < .05, **p < .01.

Intertemporal Choice Tasks

The temporal discounting and sequencing task were adapted from existing paradigms to be as comparable as possible in terms of wording and visual appearance across task types (Supplementary Material, p. 8). Delay (months vs years), outcome size (hundreds vs thousands), and outcome valence (gain vs loss) were varied consistently across both tasks. The maximum delay was 10 months in the short delay and 10 years in the long delay condition. The monetary gain or loss was $150 in the low outcome condition and $1,500 in the high outcome condition. Delay and outcome amount were manipulated between subjects, whereas outcome valence was manipulated within subjects.

Discounting task

In the discounting task, adapted from Benzion, Rapoport, and Yagil (1989), participants were told to imagine that they had lost or won money. They were then asked to indicate how much they would have to be paid (gain condition) or would be willing to pay (loss condition) to delay the due date of the payment (for details, see Supplementary Material, pp. 8–10).

Following prior work using this paradigm (Benzion et al., 1989), we computed the average annual discount rate (R) across delays based on the formula R = (F/P)1/t − 1, where F denotes future value, P denotes present value, and t denotes the delay. We dropped outliers at two standard deviations above the mean (10 in the gain condition and 5 in the loss condition). The results are comparable when outliers are instead winsorized. The resulting scores were right-skewed (Gain: S = 5.52, SE = .15 Loss: S = 4.14, SE = .15) and significantly deviated from normality (Dgain = .35 and Dloss = .34, ps < .001), even after common transformations (e.g., log, square-root) were applied.

Sequence-construction task

In the sequence-construction task (adapted from Loewenstein & Sicherman, 1991), participants were told to imagine that they had lost or won money and that the payment schedule involved five installments of different size. They were asked to sort these installments in the order they preferred (for details, see Supplementary Material, pp. 9–12).

Following Löckenhoff and colleagues (2012, 2019), we obtained sequence trend scores by computing Spearman’s rank order correlations between order in the sequence (1–5) and amount of the outcome ($10/100–$50/500). The resulting scores ranged from 1 (steadily increasing) to −1 (steadily decreasing) with neutral scores indicating a preference for mixed sequences. Thus, a score of 1 would indicate a preferences for steadily increasing monetary earnings in the gain condition, and a preference for steadily increasing monetary losses in the loss condition. Descriptive analyses indicated that the distributions were bimodal with the majority of respondents (67% in the gain condition and 66% in the loss condition) preferring monotonously increasing or decreasing scores. We therefore dichotomized the scale into increasing (sequence trend > 0, effect-coded as 1) and decreasing sequences (sequence trend ≤ 0, effect-coded as −1).

Measures

Demographic variables included age, gender, education (from 1 = less than high school to 7 = doctoral degree), race/ethnicity, and yearly household income range (from l = less than 10,000 to 7 = more than 110,000).

Current affect was assessed by asking participants to rate how they felt “right now” using 7-point Likert scales for valence (1 = extremely negative to 7 = extremely positive) and for arousal (1 = very quiet/still to 7 = very activated/aroused; adapted from Nielsen, Knutson, & Carstensen, 2008).

Five-factor personality traits were screened with the BFI-10 which includes two Likert-type items for each of the big-five personality traits (Rammstedt & John, 2007). Cronbach’s α for the two-item scales were .69 for neuroticism, .58 for extraversion, .25 for openness, .26 for agreeableness, and .50 for conscientiousness which is comparable to the values reported in validation studies (for a review, see Balgiu, 2018).

Limited future time perspective was assessed with a single item (included in the scales by both Carstensen & Lang, 1996 and Brothers, Chui, & Diehl, 2014) asking participants to rate on a 7-point scale whether they had the sense that time was running out.

Subjective health was screened with two items (adapted from Bowling, 2005) asking participants to rate their “physical health” and “emotional health” on a 5-point scale from “poor” to “excellent.”

Subjective cognition was screened with two items (adapted from Sorokowski, Sorokowska, Frackowiak, & Löckenhoff, 2017) asking participants to rate their “ability to learn new things,” and “ability to remember things” on a 5-point scale from “poor” to “excellent.”

Numeracy was assessed with a three-item scale assessing basic understanding of probabilities and percentages (Lipkus, Samsa, & Rimer, 2001). Cronbach’s α in the present sample was .50.

Procedure

The study was approved by the Cornell University Institutional Review Board and participants provided informed consent before their participation. All participants received a flat fee. For the online survey, the amount was set by the survey provider. The laboratory sample received $20. In addition, the laboratory participants were told that one participant would be picked at random to receive an online gift certificate of up to $50 and that the exact amount of money and the timing of the payout would be based on their choices in the task (Supplementary Material, p. 7).

After providing informed consent, participants in the online study reported their demographics and their current affect. Next, they were randomly assigned to one of the following between-subject conditions: low outcome/short delay, high outcome/short delay, low outcome/long delay, or high outcome/long delay. Within each of these conditions, participants were presented with both the sequencing and the discounting task in randomized order. Within each task type, participants were randomly assigned to see either the gain or the loss condition first. Next, participants completed measures assessing personality, future orientation, as well as the single-item screeners for mental, physical, and cognitive functioning. Finally, they completed the numeracy scale. Participants in the laboratory study completed the same tasks, but in contrast to the online sample, the randomization procedure and intertemporal choice tasks occurred at the end of the study.

Analyses

To account for the non-normal distribution of the outcome variables, descriptive analyses used nonparametric approaches and hypotheses were tested via generalized linear mixed models (GLMMs, SPSS Version 25) which can accommodate non-normal repeated-measures data. For discount rates, which showed a strongly right-skewed distribution, we specified a gamma distribution and a log link (i.e., Gamma GLMM, see Myers, Montgomery, Vining, & Robinson, 2012). The gamma distribution assumes that all values are positive, but 7% of participants reported discount rates of 0. We therefore added 1 to all values before inclusion in the Gamma GLMM. For the dichotomized sequence trend scores, we specified a binomial distribution and a logit link (i.e., Logistic GLMM).

To test the effect of age on intertemporal choices (Hypotheses 1 and 2) we added age as a fixed factor. Exploratory analyses also tested for curvilinear effects of age. We then added amount and delay as between-subjects fixed factors and valence as a within-subject fixed factor. Interactions among the task characteristics were explored but only retained in the model if they reached statistical significance. To test whether age effects were moderated by task characteristics (Hypothesis 3) we added interaction terms between age and delay, amount, and valence. To test whether age differences were accounted for by variations in demographics or other background variables (Hypothesis 4), we added these as covariates the model.

To complement frequentist techniques, we applied a Bayesian approach to test the primary hypotheses about age effects in intertemporal choice (Hypotheses 1 and 2). This allowed us to obtain an estimate of the relative strength of support for the null versus alternative hypothesis. Analyses were conducted in JASP 0.8.5.1 following the procedures outlined in van Doorn et al. (2019). Given the mixed research record on age differences in intertemporal choice we adopted the default prior distribution suggested in the extant literature (stretched β prior width = 1; Ly, Verhagen, & Wagenmakers, 2016).

Results

Preliminary Analyses

Table 1 shows descriptive information for the background variables and their associations with age. A programming error (Supplementary Material, p. 2) led to an unequal distribution of gender across the age range in the online sample such that older participants were less likely to be female. Otherwise, age differences in the covariates were largely consistent with the aging literature (Cavanaugh & Blanchard-Fields, 2014). Specifically, age was associated with lower neuroticism, higher agreeableness, lower self-rated physical health and new learning, but higher self-rated emotional health as well as more positive current affect. Also consistent with prior research (Carstensen, 2006), age was associated with more limited perceptions of the future.

In the discounting task, the median yearly discount rate was 0.08 for gains and 0.02 for losses. As given in Table 1, nonparametric correlations with age (Kendall’s τ) were not significant for either condition. Thus, the raw correlations did not support Hypothesis 1. The Jeffreys-Zellner-Siow (JZS) Bayes factors in support of the null hypothesis were BF01 = 12.68 for the association between age and the discounting of gains and BF01 = 9.46 for the association between age and the discounting of losses. According to convention (Jarosz & Wiley, 2014), this would be considered as “positive” or “strong” evidence for an absence of age effects.

In the sequencing task, 81% of participants preferred decreasing (over increasing) sequences in the gain condition and 49% preferred decreasing sequences in the loss condition. As given in Table 1, the gain condition showed a significant association between age and a preference for decreasing sequences (p < .01). In other words, older adults were less likely to postpone gains than younger adults. In the loss condition, there was no evidence of an age effect. Thus, the raw correlations offered partial support for Hypothesis 2. Further, the JZS Bayes factor in support of the alternative hypothesis in the gain condition was BF10 = 88.83, which would be considered as “very strong” evidence for the presence of an age effect (Jarosz & Wiley, 2014). For losses, the JZS Bayes factor in support of the null hypothesis was BF01 = 12.18 which would be considered as “strong” evidence for the absence of an age effect.

GLMMs for the Discounting Task

To confirm the absence of age effects in discount rates and examine the potential role of task characteristics (Hypothesis 3) and covariates (Hypothesis 4), we computed a series of Gamma GLMMs (Table 2). Consistent with the correlations given in Table 1, the basic model (Table 2, Model A) did not show a significant main effect of age. Including the quadratic and cubic effects of age did not change this pattern (ps > .05).

Table 2.

Gamma GLMMs Predicting Discount Rates Based on Age, Task Characteristics, and Covariates

Model A Model B Model C Model D
β (SE) p β (SE) p β (SE) p β (SE) p
Intercept 0.15(0.04) <.001 0.21(0.04) <.001 0.22 (0.06) <.001 0.21 (0.14) .15
Age −0.001 (0.001) .28 −0.001 (0.001) .32 −0.001 (0.001) .42 −0.000 (0.001) .67
Delay −0.14 (0.02) <.001 −0.19 (0.07) <.01 −0.13 (0.02) <.001
Amount −0.03 (0.02) .26 0.001 (0.07) .99 −0.03 (0.02) .18
Valence 0.22 (0.04) <.001 0.18 (0.12) .15 0.22 (0.04) <.001
Age × Delay 0.001 (0.001) .42
Age × Amount −0.001 (0.001) .69
Age × Valence 0.001 (0.002) .67
Female 0.02 (0.03) .57
White −0.01 (0.03) .77
Education 0.02 (0.01) .06
Income −0.01 (0.01) <.05
Valence −0.001 (0.01) .9
Arousal −0.001 (0.01) .89
Neuroticism −0.002 (0.02) .92
Extraversion −0.02 (0.01) .25
Openness −0.001 (0.01) .97
Agreeableness −0.01 (0.02) .48
Conscientiousness 0.01 (0.02) .71
Time limited 0.003 (0.01) .64
SRH—physical 0.000 (0.01) .98
SRH—emotional −0.01 (0.02) .48
Self-rated learning 0.02 (0.02) .38
Self-rated memory −0.01 (0.01) .52
Numeracy 0.02 (0.01) .07
Information criteria
 Aikaike (corr.) 689.23 463.49 493.06 543.88
 Bayesian 697.85 472.11 501.66 552.43
Open in a new tab

Note: β = fixed coefficients, SE = standard error; corr. = corrected.

Adding the task characteristics (Table 2, Model B) revealed main effects of delay and valence. Consistent with the prior literature (Frederick et al., 2002) discount rates were more pronounced for shorter versus longer delays and gains versus losses. None of the interactions among task characteristics reached significance (all ps > .05, Supplementary Material, p. 14) and they were therefore omitted from further analyses.

To test Hypothesis 3 which proposed that age effects depend on outcome delay, amount, and valence, we added interactions between age and each of these task characteristics (Table 2, Model C). None of the interaction effects reached statistical significance and the effect of age remained nonsignificant even when the interactions were included (all ps > .05). Thus there was no support for Hypothesis 3.

To test Hypothesis 4, we added the demographic variables and the various covariates under consideration (Table 2, Model D). There was a significant effect indicating that those with lower income were more likely to discount the future, but the effect of age remained nonsignificant even when covariates were accounted for. Thus, there was no support for Hypothesis 4.

In summary, these analyses corroborated the absence of age effects that was observed in the correlational analyses (Table 1) and Bayesian analyses. Also, even though we replicated previously observed main effects of delay and valence (Frederick et al., 2002), these factors were unrelated to age effects.

GLMMs for the Sequence-Construction Task

To examine curvilinear effects of age as well as the potential role of task characteristics (Hypothesis 3) and covariates (Hypothesis 4) for age differences in sequence-preferences, we computed a series of Logistic GLMMs (Table 3). Consistent with the correlations given in Table 1, the basic model (Model A) showed a significant main effect of age. The quadratic effect of age (Table 3, Model B) reached statistical significance as well, but there was no significant cubic effect (p > .05). As shown in Figure 1 (gray bars), age was associated with a greater preferences for decreasing sequences, but this effect leveled off in the oldest age groups.

Table 3.

Logistic GLMMs Predicting Sequence-Preferences Based on Age, Task Characteristics, and Covariates

Model A Model B Model C Model D Model E
β (SE) p β (SE) p β (SE) p β (SE) p β (SE) p
Intercept −0.22 (.27) .12 1.12 (.72) .12 2.14 (0.78) <.01 1.86 (0.87) <.05 5.44 (1.62) <.001
Age −0.01 (.01) <.05 −0.08 (.03) <.05 −0.08 (0.03) <.05 −0.08 (0.04) <.05 −0.10 (0.04) <.01
Age2 0.001 (0.00) <.05 0.001 (0.00) <.05 0.001 (0.00) <.05 0.001 (0.00) <.05
Delay 0.23 (0.19) .24 −0.42 (0.55) .44 0.28 (0.20) .17
Amount −0.23 (0.19) .22 −0.01 (0.55) .98 −0.22 (0.20) .29
Valence −1.49 (0.19) <.001 −0.43 (0.56) .44 −1.55 (0.20) <.001
Age × Delay 0.01 (0.01) .20
Age × Amount 0.00 (0.01) .69
Age × Valence −0.02 (0.01) <.05
Female −0.42 (0.24) .08
White 0.05 (0.25) .85
Education −0.06 (0.08) .46
Income −0.03 (0.06) .60
Valence 0.09 (0.09) .32
Arousal −0.02 (0.07) .75
Neuroticism −0.30 (0.14) <.05
Extraversion −0.04 (0.12) .73
Openness 0.02 (0.13) .88
Agreeableness 0.11 (0.15) .47
Conscientiousness −0.28 (0.14) .05
Time limited 0.02 (0.06) .75
SRH—physical 0.07 (0.13) .61
SRH—emotional −0.20 (0.14) .15
Self-rated learning −0.02 (0.16) .92
Self-rated memory −0.11 (0.13) .41
Numeracy −0.10 (0.11) .37
Information criteria
 Aikaike (corr.) 2573.07 2592.59 2608.88 2650.75 2677.75
 Bayesian 2581.75 2601.26 2617.55 2659.40 2686.35
Open in a new tab

Note: β = fixed coefficients; SE = standard error; corr. = corrected.

Figure 1.

Figure 1.

Open in a new tab

Percentage of participants preferring decreasing sequences in the sequence-construction task by age group and valence.

Adding the task characteristics (Table 3, Model C) revealed a significant effect of valence. Consistent with the prior literature (Strough et al., 2018), the preference for decreasing sequences was stronger in the gain as compared to the loss condition. The effects of delay and amount were not significant. Also, none of the interactions among task characteristics reached statistical significance (all ps > .05, Supplementary Material, p. 15) and they were therefore omitted from further analyses.

To test Hypothesis 3, we added interactions between age and each of the task characteristics (Table 3, Model D). Consistent with Hypothesis 3c, there was a significant age by valence interaction indicating that age effects were more pronounced in the gain condition than in the loss condition (see Figure 1, black and textured bars). There were no significant interactions between age and delay or age and amount, thus offering no support for Hypotheses 3a and 3b, respectively.

To test Hypothesis 4, we added the demographic variables and the various covariates under consideration. There was a significant effect of neuroticism (p < .05) and a marginal effect of conscientiousness (p = .05) such that those with higher neuroticism and conscientiousness showed a greater preference for decreasing sequences. However, the effects of age remained significant even when the covariates were accounted for, thus offering no support for Hypothesis 4.

In summary, these analyses corroborated the age effects that were observed in the correlational analyses (Table 1) and Bayesian analyses. They also provided qualified support that age effects in sequence-preferences are curvilinear and more pronounced for positive versus negative outcomes (Hypothesis 3c). However, these latter findings need to be interpreted with caution since, according to the information criteria, none of the advanced models provided a better fit than Model A which simply accounted for age.

Discussion

This study advanced the literature on aging and decision making by comparing age differences in two intertemporal choice tasks (i.e., temporal discounting and sequence construction) within the same study and by manipulating three key task characteristic (i.e., outcome delay, amount, and valence) while controlling for a range of theoretically implicated covariates and background variables.

For the discounting task, we found no evidence of age differences, and Bayesian analyses indicated “positive” or “substantial” evidence in favor of the null hypothesis. Also, there was no evidence that age differences were moderated by task characteristics, and controlling for a host of potential covariates did not change the pattern of results. These findings are in line with several prior studies which failed to find significant age differences in monetary discounting tasks (Chao et al., 2009; Samanez-Larkin et al., 2011; Seaman et al., 2016) but conflict with other studies which reported that temporal discounting rates were higher (e.g., Huffman et al., 2017) or lower (e.g., Löckenhoff et al., 2011) in older as compared with younger adults.

For the sequencing task, in contrast, we found a significant age effect indicating a preference for decreasing sequences that begin with the most impactful events. Consistent with Strough and colleagues (2018), we also found an age by valence interaction with gains showing more pronounced age effects than losses. In fact, Bayesian analyses revealed “very strong” evidence for the presence of an age effect in the gain condition but “strong” evidence for the absence of an age effect in the loss condition. Logistic GLMM analyses also hinted at the presence of a curvilinear effect with age differences in sequences preferences being most pronounced between young and middle adulthood and leveling off in later life. However, this findings should be interpreted with caution since, even though the quadratic age term reached statistical significance, it did not improve model fit.

Of course, potential methodological limitations need to be considered. One potential concern is the nature of our sample with the majority of participants being drawn from an online survey panel. Although we included multiple attention and consistency screeners, online participants are not likely to be representative of the general population and may have been distracted during study completion which could have increased noise. Conversely, the participants in the smaller, laboratory-based sample scored higher than the online sample on health, education, and financial resources (Supplementary Material, p. 5) and may therefore have shown different aging trajectories than the general population. Thus, it is reassuring that the patterns of age effects in the outcome variables were consistent across the two samples.

The present findings are also limited because, in order to include a wide range of covariates, we had to rely on brief screening measures. Some of these measures showed suboptimal reliability and may not have been sensitive enough to pick up on subtle variations relevant for understanding age effects. It is reassuring to note that the patterns of age differences in covariates were generally consistent with the prior literature (Cavanaugh & Blanchard-Fields, 2014), but future studies should include more comprehensive and reliable measures.

Another limitation were the adaptations to the intertemporal tasks that were required to limit demand characteristics, maximize participant choice, and make the two paradigms similar in outward appearance. Note, however, that the tasks showed the same sensitivity to outcome delay and valence as reported in the prior literature (Frederick et al., 2002; Strough et al., 2018), indicating that they functioned as intended. Nonetheless, it is conceivable that other versions of the tasks—especially those entailing higher cognitive load (e.g., inferring discounting rates from a series of binary choices, Löckenhoff et al., 2011), may show more pronounced age effects. Moreover, the presentation of a discounting and a sequencing task in the same study may have led to spillover effects. For instance, if the sequencing task was presented first, participants may have subjectively reframed the discounting task as a sequence as well, and this could have suppressed discounting effects (Magen, Dweck, & Gross, 2008). It is therefore reassuring that we found divergent patterns of age effects across the tasks. Finally, the study focused on monetary outcomes and patterns of age effects are likely to differ for other outcome domains (Seaman et al., 2016).

In spite of these limitations, the present findings have implications for our understanding of age differences in intertemporal choice. For temporal discounting, a broad body of prior research is available, but patterns of age effects are strikingly inconsistent. On the one hand, this may point to a lack of insight into relevant moderators and covariates, and the present study contributes to the literature by testing several promising candidates (Lempert & Phelps, 2016) and finding no evidence that any of them modulate age effects. On the other hand, inconsistent findings could mean that there are, in fact, no age differences in temporal discounting, with the present study adding further evidence in favor of a null effect. Conceivably, the well-known tendency to relegate studies with nonsignificant effects into file drawers (Franco, Malhotra, & Simonovits, 2014), may have led to a situation where spurious effects showing significant age differences in discounting (regardless of their direction) have a higher likelihood of getting published, leading to an underrepresentation of the evidence for null effects in the public record. Resolving this situation will require a combination of strategies including not only editorial practices that explicitly encourage the publication of null results of age (Isaacowitz, 2018), but also open science techniques that encourage pre-registration of hypotheses, a priori power analyses to ensure sufficient sample size, and data sharing to facilitate meta-analytic integration.

With respect to sequence construction, in contrast, the present findings replicate those of the only other comparable study (i.e., Strough et al. 2018) and contribute further evidence that sequencing is susceptible to the age-related positivity effect. As noted in the introduction, sequence-construction tasks may offer a clearer view of age-related differences in time preferences because—other than discounting tasks—they do not entail a trade-off between outcome amount and timing. Future studies are needed to strengthen this scarce body of research.

In conclusion, additional data collection along with a reconsideration of current research and editorial practices will be required to offer conclusive evidence about the presence or absence of age effects in different types of intertemporal choices. However, given the ubiquity of financial planning tasks in people’s lives (from structuring mortgage schedules to distributing retirement funds) and the precipitous aging of the U.S. population, we simply cannot afford to leave this important issue unresolved.

Funding

This research was supported by the National Institute on Aging grant (R21AG043741 to C. Löckenhoff).

Conflict of Interest

None reported.

Supplementary Material

gbz097_suppl_Supplementary_Material
Click here for additional data file. (1.3MB, docx)

Acknowledgments

Data and analysis scripts are available from the first author upon request. Study materials can be found in the Supplementary Material. This research was not preregistered.

References

  1. Albert S. M., & Duffy J (2012). Differences in risk aversion between young and older adults. Neuroscience and Neuroeconomics, 2012(1), 3–9. doi: 10.2147/NAN.S27184 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Balgiu B. A. (2018). The psychometric properties of the Big Five inventory-10 (BFI-10) including correlations with subjective and psychological well-being. Global Journal of Psychology Research: New Trends and Issues, 8(2), 61–69. doi: 10.18844/gjpr.v8i2.3434 [DOI] [Google Scholar]
  3. Benzion U., Rapoport A., & Yagil J (1989). Discount rates inferred from decisions: An experimental study. Management Science, 35(3), 270. doi: 10.1287/mnsc.35.3.270 [DOI] [Google Scholar]
  4. Bidewell J., Griffin B., & Hesketh B (2006). Timing of retirement: Including a delay discounting perspective in retirement models. Journal of Vocational Behavior, 68(2), 368–387. doi: 10.1016/j.jvb.2005.06.002 [DOI] [Google Scholar]
  5. Bowling A. (2005). Just one question: If one question works, why ask several? Journal of Epidemiology and Community Health, 59, 342–345. doi: 10.1136/jech.2004.021204 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Boyle P. A., Yu L., Segawa E., Wilson R. S., Buchman A. S., Laibson D. I., & Bennett D. A (2012). Association of cognition with temporal discounting in community based older persons. BMC Geriatrics, 12, 48. doi: 10.1186/1471-2318-12-48 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Brothers A., Chui H., & Diehl M (2014). Measuring future time perspective across adulthood: development and evaluation of a brief multidimensional questionnaire. The Gerontologist, 54, 1075–1088. doi: 10.1093/geront/gnu076 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Carstensen L. L. (2006). The influence of a sense of time on human development. Science (New York, N.Y.), 312, 1913–1915. doi: 10.1126/science.1127488 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Carstensen L. L., & Lang F. R (1996). Future time perspective scale. Unpublished manuscript, Stanford University. [Google Scholar]
  10. Cavanaugh J. C., & Blanchard-Fields F (2014). Adult development and aging. Stamford, CT: Cengage Learning. [Google Scholar]
  11. Chao L. W., Szrek H., Pereira N. S., & Pauly M. V (2009). Time preference and its relationship with age, health, and survival probability. Judgment and Decision Making, 4, 1–19. [PMC free article] [PubMed] [Google Scholar]
  12. van Doorn J., van den Bergh D., Bohm U., Dablander F., Derks K., Draws T.,…Wagenmakers E. (2019, January 23). The JASP guidelines for conducting and reporting a Bayesian analysis. doi: 10.31234/osf.io/yqxfr [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Franco A., Malhotra N., & Simonovits G (2014). Social science. Publication bias in the social sciences: unlocking the file drawer. Science (New York, N.Y.), 345, 1502–1505. doi: 10.1126/science.1255484 [DOI] [PubMed] [Google Scholar]
  14. Frederick S., Loewenstein G., & O’Donoghue T (2002). Time discounting and time preference: A critical review. Journal of Economic Literature, 40(2), 351–401. doi: 10.1257/002205102320161311 [DOI] [Google Scholar]
  15. Freund A. M., Hennecke M., & Mustafić M (2012). On gains and losses, means and ends: Goal orientation and goal focus across adulthood. In Ryan R. M. (Ed.), The Oxford handbook of human motivation (pp. 280–300). New York, NY: Oxford University Press. [Google Scholar]
  16. Green L., Fry A. F., & Myerson J (1994). Discounting of delayed rewards—A life-span comparison. Psychological Science, 5(1), 33–36. [Google Scholar]
  17. Green L., Myerson J., Lichtman D., Rosen S., & Fry A (1996). Temporal discounting in choice between delayed rewards: The role of age and income. Psychology and Aging, 11, 79–84. [DOI] [PubMed] [Google Scholar]
  18. Halfmann K., Hedgcock W., & Denburg N. L (2013). Age-related differences in discounting future gains and losses. Journal of Neuroscience, Psychology, and Economics, 6, 42–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Harrison G. W., Lau M. I., & Williams M. B (2002). Estimating individual discount rates in Denmark: A field experiment. American Economic Review, 92(5), 1606–1617. [Google Scholar]
  20. Huffman D., Maurer R., & Mitchell O. S (2017). Time discounting and economic decision-making in the older population. The Journal of the Economics of Ageing. Advance online publication.doi: 10.1016/j.jeoa.2017.05.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Isaacowitz D. M. (2018). Planning for the future of psychological research on aging. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 73, 361–362. doi: 10.1093/geronb/gbx142 [DOI] [PubMed] [Google Scholar]
  22. Ishii K., Eisen C., & Hitokoto H (2017). The effects of social status and culture on delay discounting. Japanese Psychological Research, 59(3), 230–237. doi: 10.1111/jpr.12154 [DOI] [Google Scholar]
  23. Jarosz A. F., & Wiley J (2014). What are the odds? A practical guide to computing and reporting Bayes factors. Journal of Problem Solving, 7(1), 2–9. doi:10.7771/1932–6246.1167 [Google Scholar]
  24. Jimura K., Myerson J., Hilgard J., Keighley J., Braver T. S., & Green L (2011). Domain independence and stability in young and older adults’ discounting of delayed rewards. Behavioural Processes, 87, 253–259. doi: 10.1016/j.beproc.2011.04.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lempert K. M., & Phelps E. A (2016). The malleability of intertemporal choice. Trends in Cognitive Sciences, 20, 64–74. doi: 10.1016/j.tics.2015.09.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Li Y., Baldassi M., Johnson E. J., & Weber E. U (2013). Complementary cognitive capabilities, economic decision making, and aging. Psychology and Aging, 28, 595–613. doi: 10.1037/a0034172 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Lim K. T. K., & Yu R (2015). Aging and wisdom: Age-related changes in economic and social decision making. Frontiers in Aging Neuroscience, 7, Article number: 120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Lipkus I. M., Samsa G., & Rimer B. K (2001). General performance on a numeracy scale among highly educated samples. Medical Decision Making, 21, 37–44. doi: 10.1177/0272989X0102100105 [DOI] [PubMed] [Google Scholar]
  29. Löckenhoff C. E., O’Donoghue T., & Dunning D (2011). Age differences in temporal discounting: The role of dispositional affect and anticipated emotions. Psychology and Aging, 26(2), 274–284. doi: 10.1037/a0023280 [DOI] [PubMed] [Google Scholar]
  30. Löckenhoff C. E., Reed A. E., & Maresca S. N (2012). Who saves the best for last? Age differences in preferences for affective sequences. Psychology and Aging, 27(4), 840–848. doi: 10.1037/a0028747 [DOI] [PubMed] [Google Scholar]
  31. Löckenhoff C. E., Rutt J. L., Samanez-Larkin G. R., Gallagher B. S., O’Donoghue T., & Reyna V. F (2018). Age effects in sequence-construction for a continuous cognitive task: Similar sequence-trends but fewer switch-points. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences. Advance online publication. doi: 10.1093/geronb/gby090 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Löckenhoff C. E., Rutt J. L., Samanez-Larkin G. R., O’Donoghue T., & Reyna V. F (2019). Preferences for temporal sequences of real outcomes differ across domains but do not vary by age. The Journals of Gerontology: Psychological Sciences and Social Sciences 74( 3), 430–439. doi: 10.1093/geronb/gbx094 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Loewenstein G., & Prelec D (1992). Anomalies in intertemporal choice—Evidence and interpretation. Quarterly Journal of Economics, 107(2), 573–597. [Google Scholar]
  34. Loewenstein G. F., & Prelec D (1993). Preferences for sequences of outcomes. Psychological Review, 100(1), 91–108. [Google Scholar]
  35. Loewenstein G., & Sicherman N. (1991). Do workers prefer increasing wage profiles. Journal of Labor Economics, 9(1), 67–84. [Google Scholar]
  36. Ly A., Verhagen A. J., & Wagenmakers E.-J (2016). Harold Jeffreys’s default Bayes factor hypothesis tests: Explanation, extension, and application in psychology. Journal of Mathematical Psychology, 72, 19–32. doi: 10.1016/j.jmp.2015.06.004 [DOI] [Google Scholar]
  37. Magen E., Dweck C. S., & Gross J. J (2008). The hidden-zero effect: representing a single choice as an extended sequence reduces impulsive choice. Psychological Science, 19, 648–649. doi: 10.1111/j.1467-9280.2008.02137.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Manning J., Hedden T., Wickens N., Whitfield-Gabrieli S., Prelec D., & Gabrieli J. D (2014). Personality influences temporal discounting preferences: Behavioral and brain evidence. Neuroimage, 98, 42–49. doi: 10.1016/j.neuroimage.2014.04.066 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Myers R. H., Montgomery D., Vining G. G., & Robinson T. J (2012). Generalized linear models: With applications in engineering and the sciences (2nd ed.). Hoboken, NJ: John Wiley and Sons. doi: 10.1002/9780470556986 [DOI] [Google Scholar]
  40. Nielsen L., Knutson B., & Carstensen L. L (2008). Affect dynamics, affective forecasting, and aging. Emotion (Washington, D.C.), 8, 318–330. doi: 10.1037/1528-3542.8.3.318 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Rammstedt B., & John O. P (2007). Measuring personality in one minute or less: A 10-item short version of the Big Five Inventory in English and German. Journal of Research in Personality, 41, 203–212. [Google Scholar]
  42. Read D., & Read N. L (2004). Time discounting over the lifespan. Organizational Behavior and Human Decision Processes, 94(1), 22–32. doi: 10.1016/j.obhdp.2004.01.002 [DOI] [Google Scholar]
  43. Richter D., & Mata R (2018). Age differences in intertemporal choice: U-shaped associations in a probability sample of German households. Psychology and Aging, 33, 782–788. doi: 10.1037/pag0000266 [DOI] [PubMed] [Google Scholar]
  44. Samanez-Larkin G. R., Mata R., Radu P. T., Ballard I. C., Carstensen L. L., & McClure S. M (2011). Age differences in striatal delay sensitivity during intertemporal choice in healthy adults. Frontiers in Neuroscience, 5, 126. doi: 10.3389/fnins.2011.00126 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Seaman K. L., Gorlick M. A., Vekaria K. M., Hsu M., Zald D. H., & Samanez-Larkin G. R (2016). Adult age differences in decision making across domains: Increased discounting of social and health-related rewards. Psychology and Aging, 31, 737–746. doi: 10.1037/pag0000131 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Sorokowski P., Sorokowska A., Frackowiak T., & Löckenhoff C. E (2017). Aging perceptions in Tsimane’ Amazonian forager-farmers compared with two industrialized societies. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 72, 561–570. doi: 10.1093/geronb/gbv080 [DOI] [PubMed] [Google Scholar]
  47. Sparrow E. P., Armstrong B. A., Fiocco A. J., & Spaniol J (2019). Acute stress and altruism in younger and older adults. Psychoneuroendocrinology, 100, 10–17. doi: 10.1016/j.psyneuen.2018.09.025 [DOI] [PubMed] [Google Scholar]
  48. Sparrow E. P., & Spaniol J (2018). Aging and altruism in intertemporal choice. Psychology and Aging, 33, 315–324. doi: 10.1037/pag0000223 [DOI] [PubMed] [Google Scholar]
  49. Strough J., Bruine de Bruin W., & Parker W. D (2018). Taking the biggest first: Age differences in preferences for monetary and hedonic sequences. The Journals of Gerontology: Psychological Sciences and Social Sciences, 74, 964–974. doi: 10.1093/geronb/gbx160 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Vanderveldt A. M. (2017). Decision making in older adults: A comparison of delay and probability discounting across ages Arts & Sciences Electronic Theses and Dissertations.805 Retrieved from https://openscholarship.wustl.edu/art_sci_etds/805
  51. Whelan R., & McHugh L. A (2009). Temporal discounting of hypothetical monetary rewards by adolescents, adults, and older adults. Psychological Record, 59(2), 247–258. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

gbz097_suppl_Supplementary_Material
Click here for additional data file. (1.3MB, docx)

Articles from The Journals of Gerontology Series B: Psychological Sciences and Social Sciences are provided here courtesy of Oxford University Press

RESOURCES