Does Positivity Operate When the Stakes are High?: Health Status and Decision-Making Among Older Adults

Abstract

Research and theory suggest that emotional goals are increasingly prioritized with age. Related empirical work has shown that, compared to younger adults, older adults attend to and remember positive information more than negative information. This age-related positivity effect has been eliminated in experiments that have explicitly demanded processing of both positive and negative information. In the present study, we explored whether a reduction of the preference for positive information over negative information appears when the material being reviewed holds personal relevance for the individual. Older participants whose health varied from poor to very good reviewed written material prior to making decisions about health related and non-health related issues. As predicted, older adults in relatively poor health (compared with those in relatively good health) showed less positivity in review of information while making health-related decisions. In contrast, positivity emerged regardless of health status for decisions that were unrelated to health. Across decision contexts, those individuals who focused more on positive information than negative information reported better post-decisional mood and greater decision satisfaction. Results are consistent with the theoretical argument that the age-related positivity effect reflects goal-directed cognitive processing and, furthermore, suggests that personal relevance and contextual factors determine whether or not positivity emerges.

Decision making surrounding physical health is often cognitively and emotionally taxing (Luce, Payne, & Bettman, 2000). The process requires gathering information and making trade offs, such as choosing between two imperfect health care plans or doctors when neither option satisfies all preferences. Choices that demand consideration of unpleasant information are likely to be especially challenging because negative emotions are elicited while acquiring potentially useful information (Luce, 1998; Trope, Ferguson, & Ragunathan, 2001).

There is a large body of work suggesting that decision making may be particularly unpleasant for older people, who approach health-related decisions differently than do younger adults (for a review see Mather, 2006). Older adults, for example, are more likely to avoid making health-related decisions, often delegating the decision to others, such as their doctor or a family member. Prior research suggests that when older people do engage in decision-making, they tend to gather less information before making a decision and process information in a relatively biased fashion, focusing more on positive than negative information (Löckenhoff & Carstensen, 2007).

Motivational factors are especially likely to play a role in age differences in contexts where gathering information generates negative affect. According to socioemotional selectivity theory, emotional goals are increasingly prioritized later in life (Carstensen, 2006). Consistent with this motivational shift, an age-related positivity effect has been documented showing that older adults process a greater proportion of positive relative to negative material compared to younger adults (Mather & Carstensen, 2005). Theoretically, the positivity effect reflects goal-directed cognitive processing. A recent meta-analysis based on 100 studies observed the age-related positivity effect to be robust, reliable, and to operate in theoretically specified ways (Reed, Chan & Mikels, 2014). That is, the positivity effect is eliminated under experimental conditions in which informational goals are explicitly primed (e.g., Löckenhoff & Carstensen, 2007) or when cognitive processing is constrained (Mather & Knight, 2005). Findings such as these support the contention that age-related positivity results from a top-down, controlled process and suggests that older adults can process negative and positive material similarly to younger adults.

Although focusing relatively more on positive information may promote emotional satisfaction, information processing that selectively focuses on positive information may have negative consequences in some contexts. Notably, the age-related positivity effect has been observed in the context of decision-making, such as when choosing consumer products (Kim, Healey, Goldstein, Hasher, Wiprzycka, 2008) and when making hypothetical decisions about health care (Löckenhoff & Carstensen, 2007, 2008; Mather et al., 2005), raising concern that older adults may, at times, make poor decisions because they avoid relevant, but negative, information. Evidence that decision quality is compromised by positivity has not been systematically examined, however (e.g., Finucane, Slovic, Hibbard, Peters, Mertz, & MacGregor, 2002; Yates, & Patalano, 1999). The question remains, however, whether older adults naturally reduce their focus on positive material under conditions where attention to negative material is adaptive.

There is reason to expect that this may be the case. Hess and colleagues (Hess, Queen, & Ennis, 2012) found that age differences in information search were reduced for decisions that are personally relevant. In particular, older adults were more likely to engage in systematic search strategies when faced with a decision about prescription drug plans (higher in relevance to older adults) compared to a decision about wireless phone plans (lower in relevance for older adults). These findings suggest that older adults may be more likely to engage in systematic, effortful processing of information for decisions that are personally relevant.

To our knowledge, prior research has been based on healthy people making decisions, leaving open answers to questions about findings generalizing to older adults in relatively poor health. Advancing the understanding of how individuals in poor health make health-related choices is important because their decisions may have more immediate and serious consequences. Presumably, prioritizing emotionally positive information over negative information reflects selective cognitive processing in the service of emotional goals. If so, people in poor health may adaptively adjust information review for decisions that have health related consequences if they are focused on other, health-related goals in this context. To the extent that focusing disproportionately on positive information may have detrimental effects on decision quality, relatively unhealthy adults may review information in a more balanced manner in order to improve their health outcomes. Older adults in relatively good health might not make this adjustment to information review if their default emotion-focused goals continue to take priority.

The Present Study

We postulate that although emotional goals are chronically activated in older adults, goals likely shift when the stakes are high. In such situations, an even-handed search of information may better serve goals related to decision-making. In the present study we tested the hypothesis that positivity would be reduced among older adults in relatively poorer health when reviewing material related to health. We did not, however, expect health status to moderate positivity when reviewing material about non-health related decisions: Relatively healthy older adults and those in relatively poor health were expected to display similar levels of positivity in that context. Notably, the age-related positivity effect is conceptualized as an age by valence interaction, such that older adults focus more on positive material than negative material compared to relatively younger adults. In order to test our hypotheses about the positivity effect, we computed an index of attention to positive relative to negative material then tested whether physical health moderates the effect of age when making decisions that are health related and non-health related. We expected an interaction between age and physical health for health related decisions, but only a main effect of age for non-health related decisions. Because there are equivocal findings about the potential affective consequences of the age-related positivity effect (Issacowitz & Blanchard-Fields, 2012), we also examined mood and decision satisfaction. We hypothesized that individuals who focused relatively more on positive information than negative information during review of the choice options would report enhanced mood after making their decisions and would report more satisfaction with their decisions.

Method

Participants

The sample was comprised of 134 patients recruited from a low-income clinic (67% women) ranging in age from 60 to 92 years (M = 72.46, SD = 7.15). The sample was racially and economically diverse. Sixty-five percent of patients self-identified as European-American, 27% identified as African-American, 4% identified as Asian-American, 4% identified as American-Indian, and 3% identified as Hispanic. In terms of self-reported socioeconomic status, 58% self-identified as lower income, 26% self-identified as lower middle income, 14% self-identified as middle-income, and 2% self-identified as upper middle or upper income. Education ranged from 8 to 26 years (M = 15.61, SD = 2.98).

Procedure

After providing informed consent, participants completed health and baseline mood measures. They were then presented with the decision task that contained four hypothetical decisions, two health-related (health plan and physician) and two non-health-related (car and neighbor). Decision types were presented in blocks and counterbalanced across participants.

For each decision, participants were presented with a grid on a computer screen that contained information about 4 options (labeled with the letters A, B, C, or D) that were each described in terms of the same 5 characteristics (i.e., a grid with 20 cells of information). For example, the health plan decision grid contained a row for each of the plan options (“Plan A,” “Plan B,” “Plan C,” and “Plan D”) and a column for each of the following characteristics: preventative care, after-hours care, prescription drugs, appointment availability, and consumer satisfaction (see Figure 1 for a depiction of the health plan decision grid and the car decision grid). They were told to search the grid until they were ready to choose one of the options. In order to make the decision more emotionally challenging, all options were described as average on a global index of quality (e.g., consumer satisfaction) and had two positive characteristics (one “good” and one “very good”) and two negative characteristics (one “poor” and one “very poor”). In addition, the importance of each characteristic was highlighted before the information was presented (e.g., “Preventive care coverage is important because it may help to catch health problems early or avoid them altogether”). The cell content (e.g., very good) was concealed until participants clicked on the cell. The cells were shaded to indicate their emotional valence, such that white cells contained positive information (i.e., good or very good), gray cells contained neutral information (i.e., average), and dark cells contained negative information (i.e., poor or very poor).

Figure 1.

Decision grids for the health plan choice (panel A) and car choice (panel B), with the content of the cells revealed.

After choosing one of the four options, participants indicated their current emotional experience and decision satisfaction. After the decision task, participants completed a cognitive battery and demographic questionnaire. The study session took place in a quiet room at the health clinic where participants received their primary care.

Measures

Physical health

Participants completed a health questionnaire (Hultsch et al., 1993) that assessed multiple facets of physical health, including self-rated overall health, illness episodes, instrumental health, chronic illnesses, and number of medications. An index of physical health was computed by averaging across these five dimensions (alpha = .81). There was substantial variability across individuals in their health status (see Table 1 for descriptive statistics).

Table 1.

Descriptive Statistics for Predictor Variables

	Mean	SD	Range
Age	72.46	7.15	60 – 92
Physical health
Subjective health	2.02	0.84	1 – 4
Illness episode	3.93	2.42	0 – 12
Instrumental health	5.27	5.65	0 – 22
Chronic illness	4.99	4.12	0 – 25
Medications	3.19	2.15	0 – 11
Cognitive ability
Verbal fluency	18.57	5.60	6 – 36
Digit symbol	40.08	12.24	14 – 76
Digit span	15.43	4.78	7 – 28
Vocabulary	50.08	13.68	10 – 66
Baseline emotion (positive – negative)	2.75	1.64	−2.95 – 6.00

Note. Physical health components indicate scores on subscales of a comprehensive self-report health questionnaire (Hultsch et al., 1993). Fluency indicates scores on Category Naming Task (Lindenberger, Mayr, & Kliegl, 1993). Digit symbol, digit span, and vocabulary indicate scores on the respective Wechsler tests (Wechsler, 1997). Baseline emotion reflects composite score of 6 positive states minus 6 negative states, each rated on 7-point scales.

Decision satisfaction

After making each decision, participants completed a 6-item decision satisfaction scale (Sainfort & Booske, 2000). An example item is “I am satisfied with my decision”. Across decisions, average alpha = .79 (range = .74 – .81).

Emotion experience

Before the decision task and after each decision, participants indicated how much they were feeling each of 6 positive emotions (happy, excited, proud, calm, content, peaceful) and 6 negative emotions (anxious, irritated, frustrated, concerned, sad, bored), on a 7-point scale ranging from 1 (not at all) to 7 (very much). Positive and negative emotion composites were created by averaging across emotion ratings at baseline and after each of the four decisions (i.e., post-decisional emotion); average alpha for positive emotion = .87 (range = .82 – .89) and negative emotion = .81 (range = .74 – .84).

Cognitive performance

Cognitive functioning was assessed with four instruments (all of which have been normed for older adults): the Wechsler Vocabulary Subtest (Wechsler, 1997), the Wechsler Digit Symbol Test (Wechsler, 1997), the Wechsler Digit Span Test (Wechsler, 1997), and the Category Naming Task (Lindenberger, Mayr, & Kliegl, 1993). The Wechsler Vocabulary Subtest assesses verbal intelligence by having participants provide definitions for words presented in both written and spoken form. The Wechsler Digit Symbol Test assesses visual-motor speed by having participants match as many symbols with letters as they can in 90-seconds. The Digit Span Test assesses working memory by having participants repeat numerical strings frontward and backwards. The Category Naming Task assesses verbal fluency by having participants name as many animals as they can in 90-seconds.

Results

Data Reduction and Analysis Plan

For each decision, we computed an index of positivity in review (Löckenhoff & Carstensen, 2007). The number of negative cells viewed (i.e., “poor” or “very poor”) was subtracted from the number of positive cells viewed (i.e., “good” or “very good”) and divided by the total number of emotional cells viewed (positive and negative cells). Positive values indicate a greater focus on positive features, negative values indicate a greater focus on negative features, and zero indicates comparable focus on positive and negative information. The total number of cells visited and the number of repeat visits to cells (i.e., number of cells viewed multiple times) were also calculated for each decision. On average, people visited about 31 cells before making a decision (health decisions: M = 31.28, SD = 22.75; non-health decisions: M = 30.45, SD = 22.30), with about half of the cells being repeat visits (health: M = 16.74, SD = 19.11; non-health: M = 15.89, SD = 19.10). Overall, a similar number of positive and negative cells were viewed: about 13 positive cell views (health decisions: M = 13.75, SD = 10.53; non-health decisions: M = 12.93, SD = 10.27) and 13 negative cell views (health decisions: M = 12.81, SD = 10.80; non-health decisions: M = 13.06, SD = 10.47). The positivity index scores ranged from 1 to −1 for health decisions (M = .07, SD = .39) and non-health decisions (M = .00, SD = .38).

Analyses were conducted using multilevel modeling in the linear MIXED MODELS function in SPSS. We ran a two-level model, in which decisions were nested within persons. Positivity in review, as well as the amount of information viewed (total number of cells and number of repeat visits), was predicted from decision type (0 = health decisions, 1 = non-health decisions), age, health, and their interactions. Results are reported as unstandardized MLM coefficients in Table 2. In addition, post-decisional mood and decision satisfaction were predicted from positivity, age, decision type and their interactions. Results are reported as unstandardized MLM coefficients in Table 3. The proportion of within- and between-person variance explained by each model was estimated by subtracting the variance in the conditional model (with level 1 and level 2 predictors included) from the variance in the baseline model (viz., intercept only) and then dividing by the variance in the baseline model (Raudenbush & Bryk, 2002). Semi-partial R² values were computed as estimates of effect size (Edwards, Muller, Wolfinger, Qaqish, & Schabenberger, 2008).

Table 2.

Results of Multilevel Modeling Predicting Aspects of Information Seeking as a Function of Age, Physical Health, and Decision Type

	Positivity in Review	Amount of Information	Repeat Visits
Intercept	0.084 (.028)**	31.199 (1.724)**	16.662 (1.450)**
Age	0.071 (.030)*	−0.543 (1.819)	−0.671 (1.531)
Physical Health	−0.001 (.039)	−0.714 (1.370)	0.105 (1.989)
Decision Type	−0.083 (.025)**	−0.556 (1.299)	−0.852 (1.156)
Age X Health	0.091 (.039)*	−0.852 (2.399)	−0.878 (2.016)
Age X Decision Type	−0.015 (.026)	1.743 (1.370)	1.642 (1.231)
Health X Decision Type	−0.029 (.034)	−0.147 (1.784)	−0.830 (2.016)
Decision Type X Age X Health	−0.096 (.035)**	3.026 (1.807)	2.554 (1.627)

Note. Unstandardized estimates are presented with standard errors in parentheses. Age and physical health are z-scored. Health index keyed such that higher scores indicate better physical health. Decision type coded 0 for health-related decision and 1 for non-health decisions. Positivity indicates proportion of positive relative to negative information reviewed. Amount of information indicates the total number of cells reviewed (including repeat visits to the same cell). Repeat visits indicates the number of cells viewed that had already been viewed previously.

^*p < .05.

^**p < .01.

Table 3.

Results of Multilevel Modeling Predicting Post-Decisional Mood and Decision Satisfaction as a Function of Positivity in Attention, Age, and Decision Type

	Positive Emotion	Negative Emotion	Satisfaction
Intercept	4.159 (.121)**	2.174 (.100)**	3.579 (.059)
Positivity	0.268 (.147) +	−0.462 (.125)**	0.254 (.109)*
Age	0.109 (.122)	0.118 (.101)	0.002 (.059)
Decision Type	0.064 (.063)	−0.001 (.054)	−0.0002 (.050)
Positivity X Age	−0.002 (.133)	−0.404 (.112)**	0.046 (.010)
Positivity X Decision Type	−0.0002 (.181)	0.0004 (.154)	−0.152 (.142)
Age X Decision Type	−0.067 (.065)	0.067 (.055)	−0.108 (.052)*
Positivity X Age X Decision Type	−0.037 (.181)	−0.054 (.153)	0.015 (.141)

Note. Unstandardized estimates are presented with standard errors in parentheses. Positivity indicates proportion of positive relative to negative information reviewed in the decision task (0 indicates equal review of positive and negative information). Decision type coded 0 for health-related decision and 1 for non-health decisions. Age is z-scored.

⁺p < .10

^*p < .05.

^**p < .01.

Cognitive ability and baseline mood were included as person level covariates in follow-up analyses because age and health have been shown to be associated with these variables. In order to simplify these models, we created single indices of baseline mood and cognitive ability. A baseline mood index was created by subtracting individuals’ negative emotion composite score from their positive emotion composite score (Carstensen et al., 2011). An index of cognitive performance was created by z-scoring participants scores for each of the four cognitive measures then averaging across them (alpha = .78).

In the current sample, age was associated with lower cognitive ability (r = −.32, p < .001) but was not significantly correlated with baseline mood (r = .04, p = .619), and health was associated with greater cognitive ability (r = .29. p = .001) and being in a better mood at baseline (r = .21; p = .014). There was not a significant association between age and physical health (r = −.12, p = .173). Preliminary analyses revealed that order of the decision blocks (i.e., health first versus non-health first) did not influence the core findings so it is not discussed further.

Does Health Status Moderate Age-Related Positivity?

First, we examined whether health status moderated age-related positivity in attention during the decision task. There was a main effect of age such that relatively older adults showed a greater focus on positive information than did younger old adults (γ = 0.071, SE = .030, p = .018, semi-partial R² = 0.028). The interaction between age and health was also significant reflecting the fact that positivity in attention was reduced for older adults in poorer health (γ = 0.091, SE = .039, p = .022, semi-partial R² = 0.027). However, there was also a significant three-way interaction between age, health, and decision type (γ = −0.096, SE = .035, p = .006, semi-partial R² = 0.019), suggesting that the effect of health on age-related positivity varied across the two types of decisions. For the overall model, the within-person pseudo R² was .041 and the between-person pseudo R² was .059. To decompose the 3-way interaction, we ran a follow-up analysis examining the effects of age, health, and their interaction separately for health-related decisions and non-health-related decisions. The findings are depicted in Figure 2. As expected, relatively older adults in poorer health showed less positivity for health-related decisions (Age x Health: γ = 0.092, SE = .040, p = .025, semi-partial R² = 0.038), but not non-health decisions (Age x Health: γ = −0.007, SE = .038, p = .858, semi-partial R² = 0.0002). For non-health-related decisions there was only a marginally significant main effect of age (γ = 0.056, SE = .029, p = .054, semi-partial R² = 0.027). When controlling for cognition and baseline emotion, the interaction between age, health, and decision type was robust (γ = −0.096, SE = .035, p = .006, semi-partial R² = 0.019), as was the Age x Health interaction for health-related decisions (γ = 0.092, SE = .040, p = .023, semi-partial R² = 0.039).¹

Figure 2.

Positivity during review of information for health-related decisions (panel A) and non-health-related decisions (panel B). Age and physical health plotted 1 SD above and below the mean.

Next, we examined whether a similar pattern emerged for other aspects of information, including the amount of information viewed and the number of repeat visits to cells (i.e., how often a cell was revisited after it had been opened the first time). There were no significant effects of age, health, or decision type, nor was there a significant interaction between age, health, and decision type for number of cells viewed (γ = 3.026, SE = 1.81, p = .095, semi-partial R² = 0.007) or repeat visits (γ = 2.554, SE = 1.627, p = .117, semi-partial R² = 0.006).

Does Positivity Predict Post-Decisional Emotion and Decision Satisfaction?

Finally, we examined whether positivity during review predicted decision satisfaction and post-decisional mood. Individuals who showed greater positivity reported less post-decisional negative emotion (γ = −0.462, SE = .125, p < .001, semi-partial R² = 0.030). There was also a significant interaction between positivity and age (γ = −0.404, SE = .113, p < .001, semi-partial R² = 0.028) reflecting the fact that this link between positivity and negative emotion was stronger for older adults.² For positive emotion, there was a marginal main effect of positivity (γ = 0.268, SE = .147, p = .069, semi-partial R² = 0.007), such that individuals who focused more on positive information during information search reported feeling more positive emotion after the decision than did individuals who showed less positivity during information search. In these models predicting negative and positive emotion, the within-person pseudo R² was .080 and .000, respectively, and the between-person pseudo R² was .025 and .022, respectively. When controlling for cognition and baseline emotion, the effect of positivity on post-decisional negative emotion remained significant (γ = −0.443, SE = .089, p < .001, semi-partial R² = 0.032), as did the interaction between positivity and age (γ = −0.414, SE = .112, p < .001, semi-partial R² = 0.029). In addition, there was still a marginal effect of positivity on post-decisional positive emotion (γ = 0.258, SE = .144, p = .074, semi-partial R² = 0.007). These results suggest that individuals who focused relatively more on positive information than negative information when reviewing choice options were better protected from mood decline after making decisions than were individuals who did not show as much positivity in their review of information.

Positivity also predicted greater decision satisfaction (γ = 0.254, SE = .109, p = .021, semi-partial R² = 0.010; within-person pseudo R² = .022, between-person pseudo R² = .012), even when controlling for baseline mood and cognitive performance (γ = 0.246, SE = .109, p = .024, semi-partial R² = 0.010). However, when controlling for post-decisional mood, this effect of positivity on satisfaction was no longer significant (γ = 0.138, SE = .101, p = .180, semi-partial R² = .003); there was only a significant effect of mood (positive emotion: γ = 0.197, SE = .026, p < .001, semi-partial R² = 0.141; negative emotion: γ = −0.131, SE = .031, p < .001, semi-partial R² = 0.047). In contrast, the effect of positivity on post-decisional negative emotion did remain significant when controlling for decision satisfaction (γ = −0.399, SE = .123, p = .001, semi-partial R² = 0.023). These findings suggest that the effect of positivity on mood is not simply a reflection of being more satisfied with the decision that was made (in contrast to the effect of positivity on decision satisfaction, which could not be separated from the effect of mood on satisfaction). The effect of positivity on mood and decision satisfaction did not vary by decision type or health status.

Discussion

According to socioemotional selectivity theory (Carstensen, 1992), positivity in attention supports chronically activated goals about emotional meaning and satisfaction that older people tend to prioritize. Theoretically, the positivity effect reflects an adaptive accommodation of goal directed cognitive processing. We postulated that in contexts where gathering information is highly personally relevant, rather than focusing disproportionately on the positive, older people will display even-handed review of both positive and negative information. We explored this postulate in the context of health-related decision-making by comparing information processing among individuals who differed in health status. We reasoned that older adults in relatively poor health would be less focused on emotional goals in health decision contexts than relatively younger adults because these types of decisions are particularly self-relevant and therefore may elicit an increased motivation to make an effective decision. We expected health status to predict information search behavior for health-related decisions but not for decisions unrelated to health.

Overall, we found evidence of age-related positivity during information review across the different types of decisions. The fact that an effect of age emerged in this group of older adults is noteworthy given the restricted variance. Although we did not include a younger adult comparison sample, the older adults spanned a range of over 30 years (from age 60 to 92), so we were still able to test for age differences. This finding supports the idea stemming from socioemotional selectivity theory that motivational shifts occur continuously across adulthood (Carstensen, 2006) and highlights the importance of examining development within old age.

One critical question was whether health status moderates positivity for health-related decisions. Consistent with past findings (Löckenhoff & Carstensen, 2007, 2008), when relatively healthy adults were faced with health-related decisions, we found the typical age-related positivity effect in review. As expected, however, positivity was reduced among adults in relatively poor health when making health-related decisions. These participants viewed health-related information in a relatively balanced manner, regardless of age. The health-related decision-making patterns of older-old adults in good health and relatively poor health did not differ in any other way aside from the relative attention they paid to positive versus negative information. Relatively older adults in poorer health did not look at more information or go back to revisit more cells than relatively younger or healthier adults. Therefore, differences in positivity were not driven by differences in these other information search characteristics.

Importantly, health status was unrelated to positivity for non-health-related decisions. For these decisions, there was an age-related increase in positivity regardless of health status (although positivity was generally somewhat lower for these decisions compared to health-related decisions). Therefore, in this sample of older adults, the relatively older adults in poorer health did not show less positivity in all contexts, but rather positivity was only reduced for decisions relevant to health. These results support a motivational account of the positivity effect and suggest that older adults’ goals shift in response to situational demands. In the context of health-related decision-making, older adults in good health and those in relatively poor health are likely prioritizing different goals. Older adults in good health may be focused on emotional goals (which are chronically activated) and therefore tend to show positivity in information processing. In contrast, older adults in relatively poor health are likely more focused on making an effective health decision, resulting in a greater openness to considering negative information that might be useful for making a good decision. Future work is needed to see whether similar effects would be found in other domains that are of high self-relevance or personal importance to certain groups of older adults (e.g., poor individuals making financial decisions).

Theoretically, positivity reflects goal-directed cognitive processing (Mather & Carstensen, 2005). It is possible that directing attention to relatively more positive than negative information also has the potential to improve mood. Consistent with this idea, we found that positivity during review was indeed linked to more positive post-decisional mood, as well as greater decision satisfaction. Interestingly, this link between positivity and mood was stronger for relatively older than relatively younger adults suggesting that focusing relatively more on positive materials may hold increasingly greater benefits with age. This finding is consistent with past work showing that older adults demonstrate positivity in attention when they are in a negative mood whereas younger adults show mood-congruent gaze patterns (Isaacowitz, Toner, Goren, & Wilson, 2008) and appear to be protected from mood decline by focusing on positive material (Isaacowitz, Toner, & Neupert, 2009). However, more research is needed to investigate the emotional consequences of positivity in information processing across adulthood.

Limitation and Future Directions

We focused on the role of self-reported physical health status in moderating the age-related positivity effect for information search during hypothetical decisions made in a lab setting. Future work is now needed that examines how individuals actually make health related decisions in the context of their daily lives. Longitudinal designs will be particularly useful in order to test whether changes in health status lead to changes in health-related decision-making across adulthood. Self-report measures of health, like the one used in the present study, may not provide an accurate assessment of health status (e.g., individuals may either exaggerate or downplay the severity of health problems). Integrating more behavioral measures of health and differentiating effects of mental and physical health might provide better insight into how health status may influence the ways in which individuals approach different types of decisions. In addition, we assumed that health related decisions would be of greater relevance to individuals that reported being in relatively worse health, but it might be useful to ask individuals to report on the perceived importance of decision outcomes and test whether these perceptions explain individual differences in information processing.

Finally, the current study was not designed to assess decision quality so we do not know whether relatively older adults in poorer health made better health decisions because they reviewed information in a more even-handed manner. Especially in the context of health care choices, older adults’ tendency to disregard problematic aspects of choice options could lead to potentially serious consequences. However, an increased focus on emotional goals in decision making does not necessarily mean that decision quality will deteriorate (Mikels, Löckenhoff, Maglio, Goldstein, Garber, & Carstensen, 2010). In the present sample of older adults, individuals who showed more positivity in their review of information about choice options were actually more satisfied with their decision, an effect that seemed to be driven, at least in part, by the participant’s improved mood. Therefore, more work is needed to understand how positivity in information processing affects decision outcomes.

Conclusion

Findings from the present study support the postulate that the age-related positivity effect reflects cognitive processing operating in the service of goals. Although older adults often prioritize emotional goals---and therefore focus on information that will make them feel good---there are times when other goals are more highly prioritized. Under such circumstances, positivity is not expected. In a sample of older adults, we found that those who were in relatively poor health did not show the typical age-related increase in positivity when reviewing decisions about health, yet displayed positivity when reviewing information about decisions that were not related to health. Such findings are consistent with the postulate that cognitive resources are directed to high priority goals. Although, as a large body of research suggests, default goals of older people typically concern emotional meaning and satisfaction, goal priorities can and do change when circumstances demand it. When they do, cognitive resources appear to be deployed adaptively.