Skip to main content
PMC 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
. 2018 Aug 10;75(4):762–771. doi: 10.1093/geronb/gby090

Age Effects in Sequence-Construction for a Continuous Cognitive Task: Similar Sequence-Trends but Fewer Switch-Points

Corinna E Löckenhoff 1,✉, Joshua L Rutt 1,2, Gregory R Samanez-Larkin 2, Casey Gallagher 1,3, Ted O’Donoghue 3, Valerie F Reyna 1
Editor: Nicole Anderson
PMCID: PMC7328034  PMID: 30107593

Abstract

Objectives

Many real-life settings require decision makers to sort a predetermined set of outcomes or activities into a preferred sequence and people vary in whether they prefer to tackle the most challenging aspects first, leave them for the last, or intersperse them with less challenging outcomes. Prior research on age differences in sequence-preferences has focused on discrete and hypothetical events. The present study expands this work by examining sequence-preferences for a realistic, continuous, sustained, and cognitively challenging task.

Methods

Participants (N = 121, aged 21–86) were asked to complete 10 min of a difficult cognitive task (2-back), 10 min of an easy cognitive task (1-back), and 10 min of rest over the course of a 30-min interval. They could complete the tasks in any order and switch tasks as often as they wished and they were rewarded for correct performance. Additional measures included affective and physiological responses, task accuracy, time-perspective, and demographics.

Results

The majority of participants constructed sequences with decreasing task difficulty. Preferences for the general trend of the sequence were not significantly related to age, but the number of switches among the tasks decreased with age, and task-switching tended to incur greater accuracy decrements among older as compared to younger adults.

Discussion

We address potential methodological concerns, discuss theoretical implications, and consider potential real-life applications.

Keywords: Decision making, Emotion/emotion regulation, Self-regulation, Time discounting, Working memory

Much of the literature on aging and decision making has focused on choices among two or more alternative outcomes (Hess, Strough, & Löckenhoff, 2015). Although this work has yielded important insights, many real-life decisions are structured differently in that they require people to sort a given set of events or activities into a preferred temporal sequence. Office workers, for example, have to distribute the contents of their to-do list over the course of a busy morning, hospital nurses have to decide whether to dispense the medications before they change the dressings, and amateur athletes have to pace their effort over the course of a workout session. Over the past decade, a small body of research has begun to examine age differences in sequence-preferences, but so far, the research record is equivocal and has focused on sequences consisting of discrete and often hypothetical events (for a review, see Löckenhoff et al., 2017). The present study advances our understanding by examining age differences in the preferred distribution of a continuous and cognitively effortful task over the course of a half-hour time interval while assessing the role of affective and physiological responses, task accuracy, time-perspective, and demographics.

Prior Research

Research on sequence-preferences has its roots in the behavioral economics literature and branched out from an initial interest in the temporal distribution of monetary payouts to a broader range of events and outcomes (for a review, see Frederick & Loewenstein, 2008). Three types of sequence-trends can be distinguished: Worsening sequences that leave the most intense, aversive, or least valuable outcomes for last, mixed sequence that intersperse different outcomes, and improving sequence that save the most pleasant, valuable, or least intense outcomes for last. Studies using young adult samples have found a general preference for improving sequences for a diverse set of outcomes ranging from financial payouts to hypothetical restaurant meals, annoying sounds, and painful medical procedures (Ariely & Carmon, 2000; Frederick & Loewenstein, 2008; Loewenstein & Prelec, 1993), although this tendency appears to be somewhat attenuated for tasks requiring motor effort (Löckenhoff, Rutt, Samanez-Larkin, O’Donoghue, & Reyna, 2017; Yip & Löckenhoff, 2018).

In comparison, research on age differences in sequence-preferences has been limited and results are equivocal. When asked to consider a series of hypothetical restaurant dishes that varied in taste or a sequence of hypothetical jobs that varied in wage, older adults showed a stronger preference for improving sequence-trends than younger adults (Drolet, Lau-Gesk, & Scott, 2011; Loewenstein & Sicherman, 1991); when asked to view a series of actual photos with negative, neutral, and positive valence, older adults showed a weaker preference for improving sequence-trends than younger adults (Löckenhoff, Reed, & Maresca, 2012), and when asked to select sequences for actual electrodermal shocks, bursts of physical effort, and monetary gambles, no age differences were found (Löckenhoff et al., 2017). Finally, when asked to consider hypothetical sequences of monetary gains or losses, enjoyable weekends, and painful medical procedures, older adults did not show a clear preference for improving or worsening sequence-trends (Strough, Bruine de Bruin, & Parker, 2018). Instead, age was associated with a preference to encounter the most impactful event first. These seemingly contradictory findings may be explained by variations in sample characteristics, outcome domain, sequence length, and whether the sequences were realistic or hypothetical in nature. Depending on the contingencies of a specific study, different age-associated mechanisms behind sequence-preferences may have been activated.

Theoretical Mechanisms

A range of theoretical mechanism for understanding age differences in sequence-preferences has been discussed. One perspective draws on theories of affective and physiological aging (Ong & Löckenhoff, 2016), which contend that age-related shifts in homeostatic capacity lead to declining resources for effortful self- and emotion-regulation in later life. As a result, recovery from sustained negative or high-arousal states is more costly for older as compared to younger adults (Charles, 2010; Urry & Gross, 2010). Even within the realm of positive emotions, older adults were found to prefer low-arousal over high-arousal positive affect (Scheibe, English, Tsai, & Carstensen, 2013) and with regard to cognitive performance, older adults were found to actively avoid effortful strategies, especially for tasks that draw on working memory and other fluid abilities that are disproportionately affected by cognitive aging (Mata, Schooler, & Rieskamp, 2011; Touron, 2015). More generally, older adults are thought to show a greater preference for proactive strategies of self-regulation that avoid prolonged negative or high-arousal states before they occur (Charles, 2010). In the context of sequence-construction, improving sequences which “save the best for last” may therefore be less appealing for older adults because they entail a cluster of more negative or arousing outcomes at the beginning. Instead, one would expect older adults to prefer mixed sequences in which stimuli of different affective valence, arousal, and intensity are evenly distributed to offer time for recovery in-between more challenging experiences (Löckenhoff et al., 2012).

Importantly, age may not only affect the general trend of the sequence but also the number of times participants switch among different stimulus types. Prior research on cognitive aging indicates that switching among different types of tasks is more costly for older adults’ performance because the activation and deactivation of different tasks sets is affected by age decrements in executive functioning (for a meta-analysis see Wasylyshyn, Verhaeghen, & Sliwinski, 2011). In voluntary task-switching paradigms, that allow participants to freely choose which task to work on, older adults were found to be less likely to switch tasks than younger adults and this benefitted their performance (e.g., Ardiale & Lemaire, 2012; Butler & Weyward, 2013; Terry & Sliwinski, 2012). However, prior research on age differences in task-switching did not examine the overall trend of the resulting sequences and prior research on age differences in sequence-preferences did not examine task-related resources. Thus, the two lines of literature have yet to be reconciled.

Prior research on aging and sequence preferences has also considered the role of age differences in time-perspective (Löckenhoff & Rutt, 2015). As people move towards a more advanced position in the life-span, they perceive their subjective future as increasingly limited, and, as a result, they become more likely to prioritize emotional well-being in the present moment over the pursuit of temporally distant goals (Carstensen, 2006). In the context of sequence choices, this could make older adults less likely to save the most pleasant or least aversive experiences for last. Consistent with this idea, Löckenhoff et al. (2012) found that—for outcomes spaced over weeks and months—more advanced position in the life span was indeed associated with a lower preference for improving sequences. However, such effects could be offset by age differences in self-continuity, the perceived connection between one’s current sense of self and one’s past and future selves, which was shown to be positively associated with age (for a review see Löckenhoff & Rutt, 2017). Although potential links between self-continuity and sequence-preferences have not been directly examined, research among younger adults suggests that those with higher self-continuity are more willing to save for the future (Ersner-Hershfield, Garton, Ballard, Samanez-Larkin, & Knutson, 2009) and wait for delayed payouts (Ersner-Hershfield, Wimmer, & Knutson, 2009). In the context of sequence choices, older adults’ higher levels of self-continuity may translate into stronger preferences for improving sequences because they feel more connected to their future selves who will eventually reap the benefits of saving the best for last.

In summary, sequence-construction may be sensitive to age-related shifts in affective and physiological resources, task-related resources, and aspects of time-perspective, but different experimental paradigms may activate some of these mechanisms more than others. Hypothetical outcomes spread over weeks or months (e.g., Loewenstein & Sicherman, 1991; Strough et al., 2018) are more likely to activate broad variations in time-perspective, whereas realistic outcomes encountered during a single session are more likely to activate differences in homeostatic and task-related preferences. However, among the two prior studies that implemented realistic and proximal stimuli (Löckenhoff et al., 2012, 2017) only one reported significant age effects (Löckenhoff et al., 2012). This inconsistency is likely due to methodological limitations. The studies used discrete stimuli instead of a continuous task and this constrained possible sequence patterns. In addition, total sequence length was about 5 min in each domain, which does not qualify as sustained arousal. Further, only one of the studies (Löckenhoff et al., 2017) assessed physiological responses, and the range of age-related variations in homeostatic capacity may have been limited because the study utilized electrodermal shocks that required stringent health-based exclusions resulting in a sample of disproportionately healthy older adults. Thus, the interplay between sequence-trend, task-switching, and age-related variations in regulatory resources has not been conclusively examined and the present study was designed to address such limitations.

The Present Study

The present study examined age differences in sequence-preferences for the distribution of a realistic, continuous, and physiologically taxing task over the course of a single experimental session. To address the aforementioned limitations in prior work, we focused on a working memory task. This type of task has been consistently shown to elicit physiological arousal (Backs & Seljos, 1994; Veltman & Gaillard, 1998) but does not require any health-based exclusions. Specifically, we focused on the letter-based n-back paradigm (Ragland et al., 2002) in which participants are shown a series of letters and asked to indicate whether the currently visible letter is identical to the letter shown a specified number of steps beforehand. The task has a short trial length, can be performed continuously over extended periods of time, and can be varied in difficulty by changing the number of steps that have to be recalled.

Over the course of a 30-min interval, participants were asked to complete 10 min of a difficult version (2-back), 10 min of an easy version (1-back), and 10 min of rest. To promote sustained engagement, participants were given a monetary incentive for accurate responses. To allow participants to respond to moment-to-moment variations in regulatory capacity, they could switch among the tasks at any point in time without penalty. This task set-up allowed us to derive two indicators of sequence-preferences: (a) The overall sequence-trend capturing participants’ relative preference for increasing, decreasing, or flat cognitive load, and (b) the number of times that participants switched among the tasks.

To capture self-regulatory resources (i.e., homeostatic capacity and emotion-regulatory processes), we tracked physiological responses (i.e., heart rate and skin conductance) and self-reported affective valence and arousal. To capture task-related cognitive resources, we assessed participants’ relative accuracy for the 1-back and 2-back tasks. To assess age differences in time perception, we assessed subjective life position and self-continuity.

As noted previously, sequence-preferences in this type of setting (i.e., realistic outcomes encountered during a single experimental session) are likely to reflect age-related decrements in self-regulatory and task-related resources (which would imply an age-related preference for flat sequences and fewer switch-points) rather than age differences in broader time horizons. We therefore proposed the following hypotheses:

H1: On average, participants prefer improving sequence-trends, but this tendency is attenuated with age.

H2: The number of switch-points is negatively associated with age

H3: Compared to younger adults, older adults have lower self-regulatory resources as manifested in stronger task-related affective and physiological responses (3a), and these factors statistically account for age differences in sequence-trends (3b).

H4: Compared to younger adults, older adults have lower task-related resources as manifested in lower accuracy on the n-back tasks (4a), more frequent switch-points are associated with lower accuracy among older adults (4b), and these factors statistically account for age differences in switch-points.

H5: Compared to younger adults, older adults report a more advanced position in the life span and greater self-continuity (5a), but because of the short sequence length involved in the present study, broad age-related variations in long-term time perspective do not account for age differences in sequence-trends or switch-points (5b).

Based on prior findings (e.g., Loeckenhoff et al., 2012), we expected that age effects would show a linear pattern, but supplemental analyses were conducted to rule out curvilinear effects. Finally, to statistically control for the role of demographic characteristics and commonly reported age differences in general health (Cavanaugh & Blanchard-Fields, 2015), we assessed gender, race/ethnicity, education levels, and subjective mental and physical health.

Method

Participants

Participants were community-dwelling adults recruited via existing participant databases and public advertisements in Tompkins County, NY. To make age groups as comparable as possible, undergraduate students were not included and selective recruitment was used to hold the gender ratio consistent across the age range. One participant was excluded because she did not achieve sufficient proficiency in the experimental task, another was excluded because of equipment failure. Participants received $25 as base pay for their participation.

The final data set consisted of 121 participants, ranging in age from 21 to 86 years, that were evenly distributed across the age range (aged 21–39: n = 40, M = 29.38, SD = 5.72; aged 40–59: n = 42, M = 50.76, SD = 5.18; aged 60 years and over: n = 39; M = 70.05; SD = 6.61). Further analyses therefore used age as a continuous variable. Table 1, left columns, reports demographic characteristics and their associations with age. Consistent with the local demographic distribution, older adults were less likely to be Hispanic or Non-White. Subsequent analyses therefore controlled for Race/Ethnicity.

Table 1.

Sample Characteristics and Their Associations With Age, Sequence-trend, and Switch-points

Variable M (SD) or % r age r sq_trend r sq_switch
Age 49.91 (17.50) −.02 −.28**
Female 48.8% .03 −.07 .02
Non-Hispanic White 83.5% .20* −.19* −.07
Education level 4.93 (1.85) .04 .15 .21*
SF12 – Mental Health 48.45 (10.66) .25** −.21* −.08
SF-12 – Physical Health 50.84 (9.04) −.26** .09 .17
Subjective life position 57.19 (21.11) .66** .04 −.18*
Self-continuity 4.52 (1.48) .26** −.03 −.10
1-back (proportion correct) .91 (.07) −.21* −.06 −.01
2-back (proportion correct) .88 (.07) −.36** −.06 .00
Open in a new tab

Note: Associations with categorical measures show point-biserial correlations, all others are Pearson correlations. Education ranges from 1 = less than High School to 8 = graduate degree. rage = correlations with age; rsq_trend= correlations with sequence-trend; rsq_switch = correlations with switch-points.

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

Sequence Selection Task

Overview

Participants were told that, over the next 30 min, they would complete 10 min of a difficult cognitive task (2-back), 10 min of an easy cognitive task (1-back), and 10 min of rest. Specifically, we implemented a letter-based n-back task (Ragland et al., 2002). Participants were told that they could switch among the activities at any time and as often as they preferred without penalty, but that they would have to eventually complete all of the activities. To promote continuous effort on the tasks, participants were told that they would be paid up to $10 in addition to the base participant fee, depending on the percentage of correct responses. At the end of the study, 10 cents were paid out for each percent of correct responses.

Interface

Figure 1 shows the interface implemented in E-Prime Software (Psychology Software Tools). The top part of the screen showed the three activities (Rest, 1-back, 2-back) and the time remaining in each of them. The current activity was indicated with a bold frame. Participants switched among the activities using the ring, middle and index finger of their left hand to press keys labeled “R” for Rest, “1” for 1-back, and “2” for 2-back respectively (keys Z, X, and C on a standard keyboard).

Figure 1.

Figure 1.

Open in a new tab

Screen shot of the sequence selection task.

The lower part of the screen showed the interface for the current activity (during the rest phase, the word “REST” was displayed). Participants responded to the 1- and 2-back tasks using the pointer and middle finger of their right hand to press keys labeled “y” for yes and “n” for no (keys 1 and 2 on the number keypad of a standard keyboard). A countdown timer below each activity kept track of the time remaining. When participants switched among activities, the interface immediately acknowledged the input with a beeping sound of 500 ms. It then completed the current trial and switched to the new activity with the next trial.

The sequence task always started with REST. Once the time for a given task was used up, the interface automatically switched to an activity that still had time remaining. During the rest phases, participants were instructed to sit quietly, look at the screen, and refrain from talking or other activities. To ensure thorough understanding of the task environment, training proceeded in two phases which are detailed in the Supplementary Material (Section 1).

Data processing

Over the course of the sequence task, E-Prime automatically logged the current activity as well as the accuracy of responses for each 3-s trial interval. For further analyses, we computed the following scores:

A sequence-trend score captured the temporal distribution of task types over the 30-minute interval. For each participant, we computed the Spearman’s Rank order correlation between trials in the testing period (1–600) and task difficulty (coded as rest = 1, 1-back = 2, 2-back = 3). The resulting scores ranged from −1 (monotonously decreasing cognitive load) to 1 (monotonously increasing cognitive load) with neutral scores indicating a preference for mixed sequences.

A switch-point score indicated the number of times that participants switched among the different activities. Specifically, we computed the number of trials in which activity type differed from the activity type in the trial before. For each participant, this included two automatic switch-points that were generated when the time ran out on a task component. Two outliers scoring more than three standard deviations above the mean were deleted from analyses involving switch points.

Accuracy scores indicated the total number of accurate responses for the 1-back and the 2-back task, respectively.

Measures

Current affect was assessed with virtual sliders capturing the two dimensions of the affective circumplex (Posner, Russell, & Peterson, 2005): Valence (from 0 = “very negative” to 100 = “very positive”) and arousal (from 0 = “not aroused at all” to 100 = “very aroused”; Nielsen, Knutson, & Carstensen, 2008).

Subjective life position, was assessed by asking participants to mark their current position on a line ranging from “birth” on the left to “death” on the right (Hancock, 2010).

Self-continuity was assessed with a visual scale (Ersner-Hershfield, Garton, et al., 2009; Rutt & Löckenhoff, 2016) that showed seven pairs of intersecting circles which ranged from complete separation (scored as 1) to almost complete overlap (scored as 7). Using this scale, participants were asked to rate the perceived overlap of their present selves with their past and future selves. For each temporal direction (past vs future), we assessed three temporal distances (1 month, 1 year, and 10 years). The six items showed high internal consistency (Cronbach’s α = .86) and an average score was computed.

Demographic characteristics included age, gender, education level, and race/ethnicity.

Subjective mental and physical health were assessed with the SF-12 (Ware, Kosinski, & Keller, 1998), a widely used screening measure capturing aspects of physical health (including general health, pain, and functional limitations) and mental health (including mood and energy levels).

Physiological Assessments

Physiological assessments were obtained before the training phase, before the main sequence task, and at the end of the study. Participants were asked to sit quietly for two minutes and look at a fixation cross on the screen. Following common practice (Boucsein et al., 2012) and consistent with our prior work (Löckenhoff et al., 2017) we computed the average heart rate, average tonic skin conductance level (SCL), and the number of nonspecific skin conductance responses (SCRs) per minute to capture physiological activation within each of these intervals. Physiological data were lost for nine participants due to equipment failure, experimenter error, or excessive motion artifacts. Details about the acquisition and processing of physiological data are reported in the Supplementary Material (Section 2).

Procedure

After providing informed consent, participants were connected to the electrodes for physiological recording. Then they completed assessments of demographic characteristics and subjective health. Next, they completed the training phase of the sequence task, followed by the 30-min main phase of the task. Finally, they completed the time-perspective measures. Current affect was assessed four times over the course of the study session: At the very beginning of the session and immediately before each of the physiological baselines (i.e., before the training phase of the sequence task, before the main phase of the sequence task, and immediately after the sequence task). At the end of the session, participants were debriefed and paid. At the end of the same experimental session, participants also completed a battery of measures that were unrelated to the present study including a decision task involving hypothetical electrical shocks and a series of questionnaires assessing aspects of self-regulation (Löckenhoff et al., 2016, Study 2). None of these variables showed significant associations with age or sequence-trends (all rs<|.17|, ps>.05) and they were omitted from further analyses.

Results

Sequence Preferences

The average sequence-trend score was −.46 (SD = .29) indicating that, in general, participants preferred sequences of decreasing cognitive load (i.e., improving sequences). To construct their preferred sequences, participants switched an average of 11.08 times (SD = 6.52) among the tasks. There was a significant correlation between sequence-trend score and the number of switch-points (r = .32, p < .001) indicating that participants who distributed the cognitive load more evenly over the 30-min interval were more likely to switch among tasks.

Contrary to H1, the association between age and the sequence-trend score was not significant (r = −.02, p > .8). Regression analyses predicting sequence-trend scores based on the centered age variable and a quadratic age term did not find any evidence of curvilinear age effects (p > .6, see scatterplot in the Supplementary Material, Section 3). Further analyses revealed a JZS Bayes factor of 8.65 in support of the null hypothesis for the correlation between age and sequence-preferences (default stretched β prior width = 1; Wagenmakers, Verhagen, & Ly, 2016).

Consistent with H2, the number of switch-points was negatively associated with age (r = −.28, p < .01). Again, there was no evidence of quadratic age effects (p > .1, see scatterplot in the Supplementary Material, Section 3). Further analyses revealed a JZS Bayes factor of 10.45 in support of the alternative hypothesis for the correlation between age and switch-points (default stretched β prior width = 1; Wagenmakers et al., 2016).

In summary, although the overall trend of the constructed sequence did not vary by age, older adults were less likely to switch among tasks in order to achieve their desired sequence.

This pattern of associations remained the same when analyses of sequence-trend statistically controlled for switch-points and when analyses of switch-points statistically controlled for sequence-trend.

Affective and Physiological Responses

Table 2 shows affect ratings and physiological markers obtained during the 2-min resting phases that were administered before the n-back training phase (pretraining), between the training and the 30-min sequencing task (presequence), and immediately after the sequencing task (postsequence).

Table 2.

Variations in Affective and Physiological Measures Across Times of Assessment

Variable r age Pretraining M (SD) Presequence M (SD) Postsequence M (SD) F
Affective valence .14 73.17 (17.7) 63.23 (20.19) 74.17 (16.7) 30.07**
Affective arousal −.06 38.52 (25.58) 48.98 (26) 44.66 (25.88) 17.28**
Heart rate .02 74.43 (12.19) 74.6 (12.22) 72.61 (11.24) 8.39**
SCL −.28** 6.95 (4.95) 8.53 (5.52) 8.58 (5.76) 34.54**
SCRs/minute −.19* 2.96 (2.33) 3.29 (2.2) 2.81 (2.7) 3.85*
Open in a new tab

Note: rage = correlations between age and pretraining scores for each of the affective and physiological measures; SCL = tonic skin conductance level; SCRs/minute = number of nonspecific skin conductance responses per minute. F-score is based on repeated-measures analyses of variance comparing affective and physiological activation across times of assessments.

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

As would be expected, repeated-measures analyses of variance (Table 2, column 6) with Greenhouse Geisser corrections indicated that each of the affect ratings and physiological markers showed significant variations over the course of the training phase and sequencing task. However, an examination of within-subject contrasts indicated that the trajectories varied across individual variables. Affective valence and SCRs showed quadratic effects (ps < .001) such that affective valence dropped and the number of SCRs per minute increased from pretraining to presequence, but returned to pretraining levels after the task was completed. Affective arousal, heart rate, and SCL, in contrast showed both linear and quadratic effects (ps < .001), such that affective arousal and SCL increased from pretraining to presequence and continued to remain elevated even after the sequence task was complete. Heart rate, finally, remained stable from pretraining to pretask but dropped from presequence to postsequence.

Table 2 (second column) shows correlations between age and affective and physiological markers at pretraining. Age was not significantly related to pretraining affective valence, affective arousal, or heart rate, but—consistent with the prior literature—age was negatively associated with both SCL and SCRs per minute (Barontini, Lazzari, Levin, Armando, & Basso, 1997; Löckenhoff et al., 2017). To examine if there were age differences in temporal trajectories of affective and physiological responses (H3a), we added age as a covariate to the repeated measures analyses reported in Table 2. The time by age interaction reached significance for affective valence, F(2, 236) = 9.94, η2p= .08, p < .00, and for heart rate F(2, 216) = 10.20, η2p = .09, p < .001, indicating that older adults experienced a steeper drop in affective valence and heart rate in response to the experimental task than younger adults. For the remaining measures, the time by age interactions did not reach significance (ps > .05). This provides partial support for H3a.

To examine if age variations in affective and physiological measures accounted for age patterns in sequence-construction (H3b), we computed a series of regression analyses. The predictor variables were age and each of the affective and physiological markers showing associations with age (i.e., pretraining scores for SCL and SCRs/minute, and pretraining minus presequence difference scores for affective valence and heart rate). Sequence-trend scores and switch-point scores were the dependent variables. To address concerns about collinearity, the covariates were included one at a time. Contrary to H3b, even when age variations in affective and physiological responses were accounted for, the pattern of age effects did not change: The age effect in sequence-trend scores remained nonsignificant (all ps > .6), and age decrements in the number of switch-points remained significant (all ps < .05).

Accuracy

Consistent with H4a, task accuracy was negatively associated with age for the 1-back task (r = −.21) and for the 2-back task (r = −.36, ps < .05). To test H4b, we examined whether the association between the number of switch-points and task accuracy was constant across the age range. We conducted regression analyses with accuracy for the 1-back and 2-back task as the dependent variables and age, the number of switch-points, and a centered age by switch-point interaction term as predictors. For accuracy in the 2-back task, we found a significant main effect of age (β = −.40, p < .001), but the main effects of switch-point (β = −.13, p = .17), and the interaction term (β = −.06, p = .49) were not significant. For accuracy in the 1-back task, we found a main effect of age (β = −.25, p < .01) and a marginally significant age by switch-point interaction (β = −.18, p = .06). The main effect of switch-point was not significant (β = −.12, p = .2). To further explore this interaction trend, we performed a median-split of the file into a younger and older group and examined the correlations between accuracy in the 1-back task and switch-points within each age group. We found a positive association in the younger group (r = .14) but a negative association in the older group (r = −.22), and the difference in correlation coefficients between the age groups was marginally significant (z = 1.96, p = .05). Thus, in partial support of H4b, switching more frequently tended to be associated with worse performance, but only among older adults and only for the 1-back task. However, contrary to H4c, the effect of age on switch-points remained significant even when 1-back and 2-back scores were added as predictors in the same regression analyses (ps < .01).

Time-Perspective and Other Covariates

Table 1 (left columns) shows descriptive information for time-perspective, demographic, and other background variables as well as their correlations with age, sequence-trends, and switch-points (for a full table of intercorrelations among study variables, see the Supplementary Material, Section 4).

As seen in Table 1, sequence-preferences showed few associations with the covariates, and the effects that reached statistical significance were small in size (all rs < .3). With regard to sequence-trends, non-Hispanic White participants and those with better mental health created sequences that were more negative and thus showed steeper declines in cognitive load. With regard to switch-points, participants with higher education switched more frequently, whereas participants with a more advanced life position switched less frequently.

Age differences in covariates were generally consistent with the prior literature (Cavanaugh & Blanchard-Fields, 2015). Consistent with H5a, age was associated with a more advanced position in the life span and greater self-continuity. Age was also associated with better subjective mental health but worse subjective physical health.

To examine if any of these covariates played a role in the observed age patterns in sequence-construction, we computed a series of regression analyses which included age and each of the variables showing significant associations with age (see Table 1, third column) as predictors and sequence-trend scores and switch-point scores as the dependent variables. To address concerns about collinearity, the covariates were included one at a time. As predicted in H5b, statistically controlling for life position and self-continuity did not change the pattern of age effects in sequence-trends or switch-points. The same was true when controlling for age differences in mental and physical health. For all analyses, the null effect of age in sequence-trend scores remained nonsignificant (all ps > .6), and age decrements in the number of switch-points remained significant (all ps < .05).

Discussion

Age Differences in Sequence Construction

The present study adds to the literature by examining age differences in sequence-construction for a realistic, continuous, and cognitively challenging task in an adult life-span sample. We differentiated between two indicators of sequence-construction: General sequence-trend and the number of switch-points. We also explored a wide range of potential mechanisms behind age effects including the role of affective and physiological responses, task-related resources, time-perspective, as well as demographics and health.

We did not support our prediction (H1) that older adults would show a greater preference for flat sequences. Instead, most participants showed a preference for improving sequences that started with the 2-back task and left most of the rest period for last, and this tendency did not vary by age. However, consistent with H2, we found that the number of task-switches was negatively associated with age.

In terms of affective and physiological responses we found that—as intended—the n-back task posed a challenge for emotional well-being and homeostasis. Consistent with our expectations, these effects varied by age such that older adults showed more pronounced task-related responses for two of the measures: Affective valence and heart rate (H3a). Contrary to our predictions, however, age variations in affective and physiological responses did not statistically account for age patterns in the sequencing task (H3b).

With regard to task-related resources, we found the expected age decrements in accuracy (H4a), and also confirmed that task-switching tended to incur greater costs (i.e., reduced accuracy) among older as compared to younger adults (H4b). Thus, although age groups did not differ in the overall trend of the sequence, older adults created their sequences in a pattern that was conducive to better task performance. Note, however, that older adults’ performance only benefited from fewer task-switches in the 1-back task and that this effect was only marginally significant. It is possible that in the more difficult 2-back condition the benefit of fewer shifts in response sets was offset by fatigue after being exposed to a prolonged series of trials of this task. This pattern would be consistent with Guastello et al. (2015) who found that fatigue effects in the n-back task vary by cognitive load. Further studies are needed to corroborate the selective association between task-switching and accuracy in the 1-back task and examine if effects are maintained when task-switching is mandatory rather than voluntary.

With regard to the remaining covariates, we found the expected age patterns in time-perspective (H5a) with older adults showing greater self-continuity and a more advanced position in the life span than their younger counterparts. However, as expected, age differences in time-perspective could not account for age patterns in sequence-construction (H5b) nor could health or demographic factors.

Taken together, we found significant age differences in sequence-construction, but these were limited to switch-points and did not manifest themselves in the general sequence-trend. Further, even though we included a wide range of theoretically implicated covariates that showed the expected associations with age, none of them were found to mediate the age patterns in switch-points. This raises concerns about potential limitations of our approach.

Limitations

Even though null effects play a key role in advancing the scientific process (Franco, Malhotra, & Simonovits, 2014), potential methodological limitations need to be considered. As noted above, the experimental paradigm functioned as intended and elicited the expected physiological and affective responses. Further, although most participants preferred decreasing sequences, there was significant variation in the steepness of sequence-trends across individuals, and there was no evidence of floor or ceiling effects. Thus, the lack of age differences cannot be explained by a restriction of range. With regard to sample size, our study was sufficiently powered (.96) to detect effects of moderate size (r > .3), and the JZS Bayes factor for a null effect of age in sequence-preferences, was 8.65. According to Andraszewicz et al. (2015) this is considered as “moderate” evidence for the null hypothesis.

Of course, other methodological limitations remain. For instance, the n-back task tapped into select aspects of working-memory, and results would need to be replicated across a wider range of cognitive tasks. Further, even though the 30-min interval involved in the present task was substantially longer than in previous studies, it may still have been too short to tap into age differences in sustained homeostasis. Future studies should consider employing even longer intervals—ideally over the course of a full working day.

Broader Implications and Future Directions

Although some questions remain, the observed null effects of age differences in preferred sequence-trends converge with several other recent studies by ourselves and others which failed to find age differences in sequence-preferences and other forms of intertemporal choice, such as temporal discounting, when outcomes are real, nonmonetary, and occur in the immediate future (Jimura et al., 2011; Löckenhoff et al., 2016; Löckenhoff et al., 2017). Conceivably, temporal preferences for concrete and proximal outcomes may emerge early in life (perhaps with the establishment of independent work and study habits) and remain comparatively stable thereafter. Conversely, age differences in sequence-trends may be more pronounced in scenarios involving hypothetical events and longer time frames because they are more sensitive to age-related shifts in goals and time horizons. To examine this possibility, future research should systematically manipulate both outcome type (i.e., outcome domain, hypothetical vs real) and temporal delay (ranging from minutes to years).

Beyond examining age effects, the present findings add to the general literature on sequence-preferences. First, we show that the tendency to prefer improving sequences, which has been found across a variety of domains (Ariely & Carmon, 2000; Frederick & Loewenstein, 2008; Loewenstein & Prelec, 1993) extends to sequences involving sustained cognitive effort. This has potential implications for managing work-flow in mentally straining tasks such as monitoring production lines or editing technical documents. Second, we add to the growing body of literature examining individual differences in sequence-preferences (Löckenhoff et al., 2012; Löckenhoff et al., 2017; Strough et al., 2018; Yip & Löckenhoff, 2018). In the present study, we found small but significant effects indicating that preferences for improving sequences were stronger among those with better subjective mental health and among non-Hispanic Whites. The latter of these results is consistent with recent findings from our laboratory indicating that preferences for improving sequences are stronger among European-American students than among their Asian or Asian-American counterparts (Yip & Löckenhoff, 2018). If corroborated by future research, potential mechanisms behind ethnic and/or cultural differences in sequence-preferences could be explored.

Funding

This research was supported by National Institute on Aging Grant R21AG043741 to C. E. Löckenhoff. G. R. Samanez-Larkin was supported by National Institute on Aging Pathway to Independence Award R00-AG042596.

Supplementary Material

gby090_suppl_Supplementary_Material
Click here for additional data file. (57.8KB, docx)

Acknowledgments

The authors thank the members of the Cornell Healthy Aging Laboratory and lab managers Katya Swarts and Kyrsten Costlow for help with data collection.

Conflict of Interest

None reported.

References

  1. Andraszewicz S., Scheibehenne B., Rieskamp J., Grasman R., Verhagen J., & Wagenmakers E. J (2015). An introduction to Bayesian hypothesis testing for management research. Journal of Management, 41, 521–543. doi: 10.1177/0149206314560412 [DOI] [Google Scholar]
  2. Ardiale E. & Lemaire P (2012). Within-item strategy switching: An age comparative study in adults. Psychology and Aging, 27, 1138–1151. doi: 10.1037/a0027772 [DOI] [PubMed] [Google Scholar]
  3. Ariely D., & Carmon Z (2000). Gestalt characteristics of experiences. The defining features of summarised events. Journal of Behavioural Decision Making, 13, 191–201. doi: [DOI] [Google Scholar]
  4. Backs R. W. & Seljos K. A (1994). Metabolic and cardiorespiratory measures of mental effort: The effects of level of difficulty in a working memory task. International Journal of Psychophysiology, 16, 57–68. doi: 10.1016/0167-8760(94)90042-6 [DOI] [PubMed] [Google Scholar]
  5. Barontini M. Lázzari J. O. Levin G. Armando I. & Basso S. J (1997). Age-related changes in sympathetic activity: Biochemical measurements and target organ responses. Archives of Gerontology and Geriatrics, 25, 175–186. doi: 10.1016/s0167-4943(97)00008-3 [DOI] [PubMed] [Google Scholar]
  6. Boucsein W. Fowles D. C. Grimnes S. Ben-Shakhar G. roth W. T. Dawson M. E.,…Filion D. L; Society for Psychophysiological Research Ad Hoc Committee on Electrodermal Measures (2012). Publication recommendations for electrodermal measurements. Psychophysiology, 49, 1017–1034. doi: 10.1111/j.1469-8986.2012.01384.x [DOI] [PubMed] [Google Scholar]
  7. Butler K. M. & Weywadt C (2013). Age differences in voluntary task switching. Psychology and Aging, 28, 1024–1031. doi: 10.1037/a0034937 [DOI] [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. Cavanaugh J. C., & Blanchard-Fields F (2015). Adult development and aging (7th ed). Stamford, CT: Cengage Learning. [Google Scholar]
  10. Charles S. T. (2010). Strength and vulnerability integration: A model of emotional well-being across adulthood. Psychological Bulletin, 136, 1068–1091. doi: 10.1037/a0021232 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Drolet A., Lau-Gesk L., & Scott C (2011). The influence of aging on preferences for seqeunces of mixed affective events. Journal of Behavioral Decision Making, 24, 293–314. doi: 10.1002/bdm.695 [DOI] [Google Scholar]
  12. Ersner-Hershfield H. Garton M. T. Ballard K. Samanez-Larkin G. R. & Knutson B (2009). Don’t stop thinking about tomorrow: Individual differences in future self-continuity account for saving. Judgment and Decision Making, 4, 280–286. [PMC free article] [PubMed] [Google Scholar]
  13. Ersner-Hershfield H. Wimmer G. E. & Knutson B (2009). Saving for the future self: Neural measures of future self-continuity predict temporal discounting. Social Cognitive and Affective Neuroscience, 4, 85–92. doi: 10.1093/scan/nsn042 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. 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]
  15. Frederick S., & Loewenstein G (2008). Conflicting motives in evaluations of sequences. Journal of Risk and Uncertainty, 37, 221–235. doi: 10.1007/s11166-008-9051-z [DOI] [Google Scholar]
  16. Guastello S. J. Reiter K. Malon M. Timm P. Shircel A. & Shaline J (2015). Catastrophe models for cognitive workload and fatigue in N-back tasks. Nonlinear Dynamics, Psychology, and Life Sciences, 19, 173–200. [PubMed] [Google Scholar]
  17. Hancock P. A. (2010). The effect of age and sex on the perception of time in life. The American Journal of Psychology, 123, 1–13. doi: 10.5406/amerjpsyc.123.1.0001 [DOI] [PubMed] [Google Scholar]
  18. Hess T. M., Strough J., & Löckenhoff C. E (Eds.). (2015). Aging and decision making: Empirical and applied perspectives. New York: Elsevier Academic Press. doi: 10.1016/B978-0-12-417148-0.00001-7 [DOI] [Google Scholar]
  19. 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]
  20. Löckenhoff C. E. & Rutt J. L (2015). Age differences in time perception and their implications for decision making across the life span. In Hess T. M., Strough J., & Löckenhoff C. E., (Eds.),. Aging and decision making: Empirical and applied perspectives (pp. 213-233). San Diego, CA: Elsevier Academic Press. doi: 10.1016/B978-0-12-417148-0.00011-X [DOI] [Google Scholar]
  21. Löckenhoff C. E., & Rutt J. L (2015). Age differences in time perception and their implications for decision making across the life span. London, UK: Academic Press/Elsevier. [Google Scholar]
  22. Löckenhoff C. E. & Rutt J. L (2017). Age differences in self-continuity: Converging evidence and directions for future research. The Gerontologist, 57, 396–408. doi: 10.1093/geront/gnx010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Löckenhoff C. E., Rutt J. L., Samanez-Larkin G. R., O’Donoghue T., & Reyna V. F (2017). Preferences for temporal sequences of real outcomes differ across domains but do not vary by age. Journals of Gerontology: Psychological Sciences, Advance Access. doi: 10.1093/geronb/gbx094 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Löckenhoff C. E. Rutt J. L. Samanez-Larkin G. R. O’Donoghue T. Reyna V. F. & Ganzel B (2016). Dread sensitivity in decisions about real and imagined electrical shocks does not vary by age. Psychology and Aging, 31, 890–901. doi: 10.1037/pag0000136 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Loewenstein G. F., & Prelec D (1993). Preferences for sequences of outcomes. Psychological Review, 100, 91–108. doi: 10.1037//0033-295x.100.1.91 [DOI] [Google Scholar]
  26. Loewenstein G., & Sicherman N (1991). Do workers prefer increasing wage profiles. Journal of Labor Economics, 9, 67–84. doi: 10.1086/298259 [DOI] [Google Scholar]
  27. Mata R. Schooler L. J. & Rieskamp J (2011). The aging decision maker: Cognitive aging and the adaptive selection of decision strategies. In Gigerenzer G., Hertwig R., Pachur T., Gigerenzer G., Hertwig R., & Pachur T., (Eds.),. Heuristics: The foundations of adaptive behavior. (pp. 455-473).New York, NY, US: Oxford University Press; DOI:10.1093/acprof:oso/9780199744282.003.0022 [Google Scholar]
  28. 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]
  29. Ong A. D., & Löckenhoff C. E (Eds.). (2016). Emotion, aging, and health. Washington, DC: American Psychological Association. doi: 10.1037/14857-000 [DOI] [Google Scholar]
  30. Posner J., Russell J. A., & Peterson B. S (2005). The circumplex model of affect: An integrative approach to affective neuroscience, cognitive development, and psychopathology. Development and Psychopathology, 17, 715–734. doi: 10.1017/S0954579405050340 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Ragland J. D. Turetsky B. I. Gur R. C. Gunning-Dixon F. Turner T. Schroeder L.,…Gur R. E (2002). Working memory for complex figures: An fMRI comparison of letter and fractal n-back tasks. Neuropsychology, 16, 370–379. doi: 10.1037//0894-4105.16.3.370 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Rutt J. L. & Löckenhoff C. E (2016). From past to future: Temporal self-continuity across the life span. Psychology and Aging, 31, 631–639. doi: 10.1037/pag0000090 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Scheibe S., English T., Tsai J. L., & Carstensen L. L (2013). Striving to feel good: Ideal affect, actual affect, and their correspondence across adulthood. Psychology and Aging, 28, 160–171. doi: 10.1037/a0030561 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Strough J., Bruine de Bruin W., & Parker W. D (2018). Taking the biggest first: Age differences in preferences for Monetary and Hedonic Sequences. Journal of Gerontology, Series B: Psychological Sciences and Social Sciences. doi: 10.1093/geronb/gbx160 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Terry C. P. & Sliwinski M. J (2012). Aging and random task switching: The role of endogenous versus exogenous task selection. Experimental Aging Research, 38, 87–109. doi: 10.1080/0361073X.2012.637008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Touron D. R. (2015). Memory avoidance by older adults: When ‘old dogs’ won’t perform their ‘new tricks’. Current Directions in Psychological Science, 24, 170–176. doi: 10.1177/0963721414563730 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Urry H. L., & Gross J. J (2010). Emotion regulation in older age. Current Directions in Psychological Science, 19, 352–357. doi: 10.1177/0963721410388395 [DOI] [Google Scholar]
  38. Veltman J. A. & Gaillard A. W (1998). Physiological workload reactions to increasing levels of task difficulty. Ergonomics, 41, 656–669. doi: 10.1080/001401398186829 [DOI] [PubMed] [Google Scholar]
  39. Wagenmakers E-J., Verhagen J., & Ly A (2016). How to quantify the evidence for the absence of a correlation. Behavior Research Methods, 48, 413–426. doi: 10.3758/s13428-015-0593-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Ware J. E., Kosinski M. & Keller S. D (1998). A 12-item short-form health survey: construction of scales and preliminary tests of reliability and validity. Medical Care, 34, 220–233. doi: 10.1097/00005650-199603000-00003 [DOI] [PubMed] [Google Scholar]
  41. Wasylyshyn C. Verhaeghen P. & Sliwinski M. J (2011). Aging and task switching: A meta-analysis. Psychology and Aging, 26, 15–20. doi: 10.1037/a0020912 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Yip A., & Löckenhoff C. E (2018). Cultural Differences in “Saving the Best for Last”. Journal of Cross-Cultural Psychology, 0022022118793535. doi: 10.1177/0022022118793535 [DOI] [Google Scholar]

Associated Data

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

Supplementary Materials

gby090_suppl_Supplementary_Material
Click here for additional data file. (57.8KB, docx)

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

RESOURCES