FOREGONE REWARDS AND AGING 1 Information about Foregone Rewards Impedes Dynamic Decision-Making in Older Adults

“Making an informed decision” implies that more information leads to better decisions, yet it may be the case that additional information biases decisions in a systematic and sometimes detrimental manner. In the present study we examined the effect of additional information on older adults’ decision-making using a task for which available rewards were dependent on the participant’s recent pattern of choices. The optimal strategy was to forego the immediately rewarding option in favor of the option that yielded larger delayed reward. We found that providing information about true foregone rewards— the reward that would have been received had the participant chosen the other option—significantly reduced older adults’ decision-making performance. However, false foregone rewards—foregone rewards manufactured to make the long-term option appear more immediately rewarding, led older adults to perform at a level equal to younger adults. We conclude that providing information about foregone rewards biases older adults toward immediate rewards at a greater rate than younger adults, leading to poorer older adult performance when immediate rewards and long term rewards conflict, but intact performance when immediate rewards and long-term rewards appear to align. Key terms: decision-making, aging, foregone rewards, delayed rewards FOREGONE REWARDS AND AGING 3 Information about Foregone Rewards Impedes Dynamic Decision-Making in Older Adults In making daily decisions one is often forced to weigh smaller immediate rewards against potentially greater delayed rewards. For example, an individual desiring to boost their energy level might ponder the choice between a sugary caffeinated energy drink and an alternate healthy snack. Indulging in the caffeinated beverage would provide the individual with higher, more immediate energy boost. On the other hand, choosing to maintain a healthier diet by consistently selecting a healthy item may lead to better-sustained energy over the course of time with a smaller immediate boost of energy. In this situation, foregoing the greater immediate reward can result in a greater long-term benefit. The same type of tradeoff applies to many economic and health choices. Despite the prevalence of these choice scenarios, immediate temptations often prevail over long-term goals (Loewenstein, 1996; Rachlin, 1995). With this in mind, a recurring theme in reward-based decision-making centers on the distinction and preference between immediate and delayed reward (Bechara, Damasio, Damasio, & Anderson, 1994; Hariri, Brown, Williamson, Flory, de Wit & Manuck, 2006; McClure, Laibson, Loewenstein, & Cohen, 2004). The decisions that we make are rarely isolated events—the rewards that are available at any point are often dependent on our previous choices. Consider two individuals who both chose to exercise on the same day: one regularly exercises, and one rarely exercises. Although both individuals make the same single choice, the individual who regularly exercises will observe a greater benefit than the individual who rarely exercises because their reward (good physical health) is contingent on many previous choices. The dependence of available rewards on choice history is reflected in dynamic decision-making tasks that have been widely used to assess preference for future or delayed reward (e.g. Cooper, Worthy, Gorlick & Maddox, 2013; FOREGONE REWARDS AND AGING 4 Gureckis & Love, 2009; Worthy, Otto & Maddox, 2012; Worthy, Cooper, Byrne, Gorlick & Maddox, 2014). In dynamic decision-making tasks in the laboratory, participants repeatedly choose between two reward options, one that is Short-Term optimal, and one that is Long-Term optimal. Selecting the Short-Term option gives a higher immediate reward on each trial, but selecting the Long-Term option causes the rewards available for both options to increase on future trials. The strategy that leads to optimal performance is to repeatedly forego the option that gives the higher immediate reward (Short-Term option) in favor of the larger delayed reward (Long-Term option), directly pitting long-term and short-term rewards against one another. Previous work from our labs using dynamic decision-making reveals an older adult advantage in learning to forego a more immediately rewarding option in favor of an option that leads to larger long-term rewards (Worthy, Gorlick, Pacheco, Schnyer, & Maddox, 2011; see also Cooper et al., 2013). While this advantage is exciting, decision-making environments in the real world are extremely variable, and more work is needed to understand the extent to which older adults are able to perform optimally in dynamic decision-making situations. One critical aspect of real-life decision-making, which has yet to be explored in the older adult population, is that future choices may be informed not only by the rewards that have been received, but also by the rewards that could have been received had an alternative choice been made. This scenario is analogous to an executive who is able to both observe the result of continued investment while simultaneously calculating the hypothetical cost savings of suspending investment (Otto & Love, 2010). Otto and Love (2010) considered a dynamic decision making task in which people learned about the rewards resulting from both chosen and unchosen (foregone) options. Perhaps counter-intuitively, they found that younger adults perform FOREGONE REWARDS AND AGING 5 worse when given information about true foregone rewards. That is, when foregone rewards show that another option would have been more immediately rewarding, long-term performance suffers. This type of task parallels real-life decision making in three key ways. First, similar to dynamic decision making tasks, many real-world decisions are informed by past outcomes and many current situations are determined by past choices (Busemeyer & Pleskac, 2009). Second, and contrary to classic temporal discounting tasks, the participant is neither explicitly told of nor promised a larger delayed reward, but must discover the value of the Long-Term option by exploring the reward environment. Third, participants are able to learn both from the reward that they received, as well as the reward that they could have received had they acted differently. Despite these similarities to real-world decision-making, little is known about the effect of foregone rewards on older adults’ decision-making performance. In the present study we tested older adults’ susceptibility to foregone rewards (called True Foregone Rewards) by providing information about the reward for the option that they chose as well as the reward that would have been received on each trial if the unselected option had been selected. In the True Foregone rewards condition, if a participant selected the choice that represented the “Long-Term” option on a given trial and was awarded 60 points, they would also be told what had they selected the other choice, 70 points (note that participants do not initially know which option is long-term optimal or short-term optimal, but must discover it through trial-by-trial feedback). While information about foregone rewards does not affect the participant’s current decision (since it has already been made), it can still affect their future decisions. In our energy example, this would correspond to someone choosing the healthy snack, foregoing the sugary caffeinated beverage, and observing a much more energized, caffeinated FOREGONE REWARDS AND AGING 6 coworker. This information about true foregone rewards or “what could have been” may cause the individual to make a different decision when presented the same choice in the future. It is possible that older adults, who may have a greater sensitivity to delayed rewards than younger adults (Green, Fry, & Myerson, 1994; Löckenhoff, O'Donoghue, & Dunning, 2011; Samanez-Larkin et al., 2011), are less affected than younger adults by true foregone rewards that highlight the immediate value of the Short-Term option. Another possibility is that older adults are more affected by foregone rewards than younger adults. Providing information about the foregone reward distinctly highlights the higher immediate value, or local superiority, of the Short-Term option, whereas in the absence of true foregone rewards the immediate value of the Short-Term option is not as obvious. Highlighting the local superiority of the Short-Term option leads to poorer younger adult performance in this task (Otto & Love, 2010). Recent work using modified versions of the Iowa Gambling Task has found an age-related increase in hypersensitivity to reward, whereby the decisions of older adults are disproportionately influenced by prospects of receiving rewards as compared to younger adults (Bauer et al., 2013). Taken together, we hypothesize that older adults will be even more affected by the presence of foregone rewards that highlight the greater immediate reward of the Short-Term option, resulting in poorer performance relative to younger adults in this decision-making task when true foregone rewards are present. We included an additional condition in our experiment, False Foregone Rewards, which provides additional information on each trial without highlighting the immediate superiority of the Short-Term option. We hypothesized that the presence of true foregone rewards would systematically shift value from the higher delayed reward (Long-Term option) to the higher immediate reward (Short-Term option) as a result of the highlighted local superiority of the FOREGONE REWARDS AND AGING 7 Short-Term option. However, a second plausible explanation for an age-related performance decline in the true foregone reward condition is that the additional information from foregone rewards simply causes confusion stemming from additional information. While older adults typically perform better than or equal to younger adults on this task under normal conditions (Cooper et al., 2013; Worthy et al., 2011; Worthy & Maddox, 2012), their performance is negatively affected by different types of increased task difficulty, including performance pressure (Cooper et al., 2013) and increasing the number of options (Worthy et al., 2014), thus their performance may decline simply because additional information makes the task more confusing. In the False Foregone Rewards condition, the reward associated with the chosen option was always true to the reward structure, but the foregone reward was fictional and skewed to dull the salience of the immediate reward of the Short-Term option on each trial (Otto & Love, 2010). This False Foregone Rewards condition was also used in Otto and Love to verify that performance in the True Foregone Rewards condition was not simply due to confusion stemming from additional information. In our energy example, the False Foregone Rewards condition is analogous to an individual choosing the healthy snack and (falsely) hearing that the caffeinated beverage was much less effective, encouraging their future selection of the Long-Term option by making the immediate rewards of the Short-Term option less salient. A significant older adult performance deficit relative to younger adults when False Foregone Rewards are presented would suggest that additional information about the reward environment confuses older adults and negatively affects their performance. Alternately, a lack of older adult performance deficit would indicate that the presence of additional information alone does not lead to poorer older adult decision-making. Older adults performing better than or equal to younger adults when False Foregone Rewards are present, but worse when True FOREGONE REWARDS AND AGING 8 Foregone Rewards are present, would provide support for the hypothesis that older adults are more affected by true foregone rewards because they highlight the local superiority of the ShortTerm option, a feature that results in a systematic bias toward the immediately rewarding option. Method Participants Seventy-five younger adults between the ages of 18 and 28 (M=20.28, SD=2.44) and 58 older adults between the ages of 60 and 88 (M=67.17, SD=5.25) were recruited from the Austin, TX community and compensated $10/hour for their participation. Informed consent was obtained from all participants, and the experiment was approved for ethics procedures using human participants. Demographic information was collected for each subject (Table 1). Participants were randomly assigned to one of three conditions: No Foregone Rewards (nyoung = 25, nold = 20), True Foregone Rewards (nyoung = 26, nold = 19), or False Foregone Rewards (nyoung = 24, nold = 19). Materials and Procedure Neuropsychological Testing. Older adults were given a series of standardized neuropsychological tests designed to assess general intellectual ability across attention, executive functioning, and memory. Tests included: Wechsler Adult Intelligence Scale-III Digit Span (WAIS-III; Weschler, 1997); WAIS-III Vocabulary (Weschler, 1997); Trail Making Test A&B (TMT; Army Individual Test Battery, 1944); Stroop Color-Word Test (Stroop, 1935); F-A-S Word Fluency subtest of the Neurosensory Center Comprehensive Examination for Aphasia (Spreen and Benton , 1977); Wechsler Memory Scale Third Edition (WMS-III) logical memory subtests (Weschler, 1997); and California Verbal Learning Test (CVLT; Delis, Kramer, Kaplan, & Ober, 1987). The tests were administered in one two-hour session. FOREGONE REWARDS AND AGING 9 Normative scores for each subject were calculated for each neuropsychological test using the standard age-appropriate published norms. Table 1 shows the means and standard deviations of standardized z scores on each test for older adults in each condition. All WAIS subtest percentiles were calculated according to the testing instructions and then converted to standardized z scores. The CVLT standardized T scores were calculated according to testing directions and then converted to standardized z scores; and the TMT standard z scores were calculated according to the testing instructions. Participants were excluded from participation if they scored more than 2 SD away from the standardized mean on more than one neuropsychological test in the same area (i.e., memory, executive functioning, or attention) or more than 3 standard deviations from the mean in a single area. Three participants were outside of normative ranges and were not included in the analysis. An ANOVA was conducted for each of the neuropsychological tests to determine if there were any differences between the three groups—the p-values of each of these tests are included in Table 1, and significant differences are discussed in the results section. Decision Making Task. The experiment was performed on PC computers using Matlab software with Psychtoolbox 2.54 (Brainard, 1997; Pelli, 1997). Participants were told that they were going to have many trials to select between two decks of cards with a goal of maximizing their cumulative points. For each selection they received points that were added to their cumulative total. Participants in both of the foregone rewards conditions were instructed that in addition to the number of points they received from their current selection they would also be shown the points they would have received had they selected from the other deck, the foregone