Abstract
Feelings of happiness have been associated with better performance in creative and flexible thinking and processing. Less is known about whether happier individuals have better performance on basic cognitive functions and slower rate of cognitive decline. In a large sample from the UK Biobank (N=17,885; Age 40–70 years), we examine the association between baseline happiness and cognitive function (speed of processing, visuospatial memory, reasoning) over four assessment waves spanning up to 10 years of follow-up. Greater happiness was associated with better speed and visuospatial memory performance across assessments independent of vascular or depression risk factors. Happiness was associated with worse reasoning. No association was found between happiness and the rate of change over time on any of the cognitive tasks. The cognitive benefits of happiness may extend to cognitive functions such as speed and memory but not more complex processes such as reasoning, and happiness may not be predictive of the rate of cognitive decline over time. More evidence on the association between psychological well-being and different cognitive functions is needed to shed light on potential interventional efforts.
Keywords: cognitive function, happiness, positive affect, well-being, longitudinal study
Introduction
Happiness is a critical component of subjective well-being (Diener et al., 2018). While the definition of happiness varies in different contexts (Diener et al., 2018), in general, being happy indicates the possession of positive and pleasurable feelings and not merely the absence of negative feelings or psychological distress (Lyubomirsky et al 2005). In particular, generally happier individuals are characterized by frequent experience of positive affect (Lyubomirsky et al., 2005). When people are happy, they are “doing better than they need to” (Carver, 2003, p. 246), not coping with immediate danger, and are more relaxed (Carver, 2003). Living these happier moments, and a happier life over time, people are situated to approach and strive for new goals as opposed to focusing on avoiding threat (Carver, 2003; Lyubomirsky et al 2005). According to the broaden-and-build theory (Fredrickson, 2001), happiness is also an enduring psychological resource that helps people meet physical and psychosocial demands more successfully. Given these benefits, happiness, and well-being more generally, has been suggested to be a health asset that motivates health behaviors, buffers against the impact of stress, and promotes general and cardiovascular health as well as longevity (Kubzansky et al., 2018; Pressman & Cohen, 2005).
Happiness has also been linked to brain and cognitive functioning (Subramaniam & Vinogradov, 2013). The dopaminergic (Ashby et al., 1999) and broaden-and-build (Fredrickson, 2001) theories suggest that positive affect is associated with greater dopamine release in brain areas, such as the anterior cingulate and prefrontal cortex, which is related to expanded attention, flexible information processing, and creative problem solving. Besides these cognitive functions related to innovation and creativity, however, less is known about whether the benefits of happiness extend to other cognitive functions and less cognitive decline over time.
Existing studies have examined happiness or positive affect, related or composite measures of well-being in association with cognitive health. One study (N=171,197, age 39–72) examined a single-item measure of happiness (“In general, how happy are you?”) in the UK Biobank, and found that happiness was associated with lower risk of all-cause dementia and vascular dementia (Zhu et al., 2022a). In a cross-sectional study of Dutch individuals (N=241, age 40–82 years), positive affect measured as 10 affect words was associated with better performance in free recall of words in older individuals (65–82 years), although no association was found for free recall in the younger group (40–64 years) or recognition in either age group (Hill et al., 2005). As part of the same Dutch project (not the exact same sample), a longitudinal study (N=258, age 40–82.2 years) found that baseline positive affect was not related to the rate of change in episodic memory, speed of processing, or executive function during up to 12 years of follow-up (Berk et al., 2017). Another longitudinal study (Rawtaer et al., 2017) examined a composite measure of well-being that included happiness and related constructs. In a cohort from Singapore (N=1,601, Mean age=64.9, SD=6.8), baseline well-being (a composite of happiness, interest in life, loneliness, and ease of living) was associated with slower decline in cognitive functioning assessed with Mini Mental State Examination (MMSE) and lower risk of incident dementia during up to eight years (Rawtaer et al., 2017). However, because of the multidimensional nature of the well-being measure, the role of happiness is not clear. Related to happiness, satisfaction with life has been associated with cognitive outcomes, but the pattern is more complex. In a Korean sample (N=8,021; age 45–93 years), for example, baseline life satisfaction was prospectively associated with better MMSE scores and lower risk of possible dementia (MMSE<18) during up to 12 years, but it was unrelated to the rate of decline in MMSE (Zhu et al., 2022b). Life satisfaction was also associated with lower risk of incident dementia during a five-year follow-up in a Canadian sample (N=1,024, age 65+; Peitsch et al., 2016). In a study of older adults in Berlin (N=516, Mean age=85), well-being (a composite measure including life satisfaction, aging satisfaction, and nonagitation) predicted less subsequent decline in perceptual speed, whereas the reverse association was not found (Gerstorf et al., 2007). On the other hand, there is also evidence suggesting a negative association between life satisfaction and cognitive function. For example, in a study in Singapore, college students with higher life satisfaction had worse performance in inhibitory control, one type of executive functions (Toh & Yang, 2022). Taken together, limited available evidence suggested positive or negative associations between happiness or related well-being measures with the level of cognitive functioning, whereas the few studies on cognitive decline over time indicated possible protective value of well-being.
The present study
Against the background reviewed above, the goal of the present study is to examine the association between happiness and level and rate of change in cognitive function. We aim to contribute to the literature in several ways. First, regarding the predictor, we focus on trait-like general happiness as one distinct dimension of well-being to examine whether it is predictive of cognitive function as it is for general health outcomes (Pressman & Cohen, 2005). Notably, single-item measures of happiness are not only economical and face valid to use in large cohort studies, but they also demonstrate acceptable reliability and construct validity (Abdel-Khalek, 2006; Moldovan, 2018; see Measures below). Second, regarding the outcome, we are interested in whether the known benefits of happiness for creative cognitive processes extend to other aspects of cognition. In particular, while previous research examined global cognitive status (indexed with a composite score like MMSE) or dementia status, evidence regarding individual cognitive functions may inform more targeted interventions. Only one project reviewed above (Berk et al., 2017; Hill et al., 2005) examined individual cognitive functions (i.e., memory, speed, etc.) separately, so we aim to add to this body of literature. Third, we seek to understand the association of happiness with both level of cognitive function and rate of cognitive decline over time. Fourth, we examine data from a large cohort with multiple longitudinal assessments over about a decade. Individual differences in cognitive trajectories are likely to be influenced by multiple factors and any single predictor is unlikely to have a large effect. Therefore, the multiple assessments in a large cohort provide a relatively robust test for a predictor of differences in cognitive trajectories. Fifth, because depression and cardiovascular risk factors are associated with cognitive functioning (Sheline et al., 2006), we examine whether the association of happiness is independent of these factors. Sixth, because age and sex are important demographic factors associated with cognitive function and aging (Nichols et al., 2021), we explore age and sex differences to provide further insights into the associations between happiness and cognitive functioning and evaluate the generalizability of the associations.
Based on the above rationale, we examine the association between happiness and cognitive functions across four waves of assessment in a large sample from the UK Biobank. We include three fundamental and relatively distinct functions that are commonly assessed in cognitive aging research – speed of process, memory, and reasoning. We predict that greater happiness will be associated with better performance on the tasks as well as slower decline over time. We expect that the association will be independent of vascular risk factors and depressive symptoms. Finally, exploratory analyses test whether the association of happiness is moderated by age and sex.
Method
Participants and Study Design
The UK Biobank (http://www.ukbiobank.ac.uk) is an ongoing longitudinal study conducted in the UK population with an interest in health and diseases. The recruitment at baseline (first assessment center visit) involved over 500,000 individuals who were registered with the UK National Health Service (NHS). The baseline assessment was conducted in 22 assessment centers across the UK between 2006 and 2010. After the baseline, three additional in-person assessments (referred to as follow-ups) were conducted: The second (2012–2013) was a repeat assessment center visit, the third (2014+) was a visit for a neuroimaging substudy, and the fourth (2019+) was a repeat imaging visit (See Supplementary Table S1 for detailed information on sample size). In the current analysis, happiness and covariates were from baseline and the cognitive measures were from the four assessments. The UK Biobank obtained ethical approval from the North West Multicenter Research Ethics Committee and informed consent from all participants. This research was conducted using the UK Biobank Resource (Application Reference Number 57672).
The analytic sample included 17,885 participants who reported baseline happiness, were free from prevalent dementia at baseline, and had complete cognitive data (e.g., completed all three cognitive tasks) at both baseline and at least one follow-up (Fig. 1; Supplementary Table S1). Prevalent dementia was ascertained through linked NHS health records at baseline. The sample was followed for up to 10.86 years. Compared to the analytic sample (N=17,885), participants (N=145,721) who reported baseline happiness, were free from prevalent dementia at baseline, and had complete cognitive data at baseline but none of the follow-ups were older (d=0.11, p<.05), reported lower happiness (d=−0.07, p<.05), had slower speed of processing (d=−0.18, p<.05), worse visuospatial memory (d=−0.14, p<.05) and worse reasoning (d=−0.40, p<.05) at baseline; They were also more likely to be women (χ2=112.32, p<.001), have no college degree (χ2=1514.37, p<.0001), have vascular conditions (χ2=387.84, p<.001), and report more depressive symptoms (χ2=222.17, p<.001) at baseline.
Fig. 1.
Flow chart of study sample
Measures
Happiness.
Happiness was measured with the item “In general how happy are you?” The item was answered on a six-point Likert scale of 1 (Extremely happy) to 6 (Extremely unhappy) and reverse coded such that higher scores indicated greater happiness. Responses of “Do not know” and “Prefer not to answer” were coded as missing. This measure is similar to commonly used single-item measures of happiness, which demonstrate high test-retest reliability (r=.86, Abdel-Khalek, 2006). They are significantly correlated with multi-item measures of happiness (e.g., Oxford Happiness Inventory, OHI; Argyle et al., 1995), suggesting high construct validity (Abdel-Khalek, 2006). They are also positively corelated with measures of positive affect (Positive and Negative Affect Schedule, PANAS; Watson et al 1988; r=.57), and negatively correlated with measures of negative affect (PANAS, r=−.38) and depressive symptoms (Patient Health Questionnaire, Kroenke et al., 2003; r=−.42), suggesting high convergent and divergent validity (Moldovan, 2018; see Results section for correlation with depressive symptoms in the current sample).
Cognitive function.
Participants completed tasks that measured speed of processing, visuospatial memory, and reasoning on a touch screen.
Speed was assessed with a simple reaction time (RT) task. Participants were asked to press a button to identify matching pairs of cards as quickly as possible. The outcome variable was the mean RT (ms) to correctly identify matches (Detailed information is in https://biobank.ndph.ox.ac.uk/showcase/ukb/docs/Snap.pdf). RTs less than 100 ms were excluded (Lyall et al., 2019). Longer RTs indicated worse performance. Intraclass correlations (ICC) computed for the task indicated good test-retest reliability (ICC=0.57, 95% CI=0.56–0.58; Lyall et al., 2016).
Visuospatial memory was measured with a pairs-matching task. Participants were presented with three (first trial) or six (second trial) pairs of matching cards arranged randomly. The cards were then turned face down, and participants needed to recall and correctly touch the position of matching pairs in the fewest tries. The outcome of interest was the total number of incorrect matches made (errors; https://biobank.ndph.ox.ac.uk/showcase/ukb/docs/Pairs.pdf). Over 70 percent of participants made no errors (scores=0) on the first trial. Following previous research (Folley et al., 2019), we used scores from the second trial, which were more normally distributed. More errors indicated worse performance. This task demonstrated relatively modest test-retest reliability (ICC = 0.16; 95% CI = 0.15 to 0.17; Lyall et al., 2016) with large within-person variability (see Results).
Reasoning was measured with 13 logical reasoning questions (e.g., “If Truda’s mother’s brother is Tim’s sister’s father, what relation is Truda to Tim?”). Participants were given two minutes to answer as many questions as possible. This task is referred to as a measure of reasoning or fluid intelligence in the UK Biobank; we follow previous research and use the term “reasoning” (Lyall et al., 2019). The outcome variable was the total number of questions correctly answered (range=0–13; for more information, see https://biobank.ndph.ox.ac.uk/showcase/ukb/docs/Fluidintelligence.pdf). Higher scores indicated better performance. This task had good test-retest reliability (ICC=0.65, 95% CI=0.63–0.67; Lyall et al., 2016).
Sociodemographic covariates.
Baseline age (years), sex (0=female, 1=male), and education levels (1=college/university degree or equivalent, 0=no) were included in all models.
Vascular conditions.
Participants reported whether they had doctor-diagnosed heart attack, angina, stroke, or high blood pressure (1=at least one condition, 0=none).
Depressive symptoms.
Participants responded to two questions in the Patient Health Questionnaire-2 (Kroenke et al., 2003): “Over the past 2 weeks, how often have you had little interest or pleasure in doing things?” and “Over the past 2 weeks, how often have you felt down, depressed or hopeless?” (0=not at all to 3=nearly every day; sum scores range=0–6).
Statistical analyses
Due to the nested data structure (assessments within individuals), two-level multilevel models (MLMs; Raudenbush & Bryk, 2002) were conducted. Level 1 was the assessment level, and Level 2 was the person level. In line with our research questions, we used the intercept-and-slope-as-outcomes model (Raudenbush & Bryk, 2002): The intercept (level) and slope (within-person change) of cognitive function were modeled as the outcomes of baseline happiness for each person. Random intercept and slope were used, allowing both parameters to vary freely between people. Specifically, cognitive scores were the Level 1 outcomes, and linear time was the Level 1 predictor to model the slope of cognitive change. For each individual, linear time was coded as the duration (years) from the date of attending the baseline assessment (0) to the date of attending each follow-up (see Table 1). Happiness, as well as all covariates, were Level 2 predictors of both the intercept and slope of each cognitive function. Continuous Level 2 variables were grand-mean centered. We also compared individuals in the top and bottom of the distribution of happiness. Detailed information including equations is in supplementary material (Analytic Approach). Fully unconditional models (with no predictors; Raudenbush & Bryk, 2002) were conducted before proceeding to the primary analyses.
Table 1.
Descriptive statistics for study variables at baseline
| Mean (SD)/N (%) | |
|---|---|
|
| |
| Age (40–70) | 55.93 (7.57) |
| Female | 9,077 (50.8%) |
| College (yes) | 8,460 (47.3%) |
| Vascular conditions (yes) | 4,129 (23.1%) |
| Depressive symptoms (0–6) | 0.45 (0.94) |
| Happiness (1–6) | 4.48 (0.69) |
| Speed of processing (RT; ms) | 547.31 (106.87) |
| Visuospatial memory (#errors) | 3.70 (2.94) |
| Reasoning (0–13) | 6.75 (2.02) |
| Years from baseline to assessment 2 | 3.02 (0.32) |
| Years from baseline to assessment 3 | 7.94 (1.56) |
| Years from baseline to assessment 4 | 9.91 (0.39) |
Note. N=17,885, except for vascular conditions (N=17,864) and depressive symptoms (N=17,296).
Results
Descriptive statistics are in Table 1. Correlations among all variables at baseline are in Supplementary Table S2. Baseline happiness was negatively correlated with depressive symptoms (r=−.43, p<.0001) and vascular conditions (r=−.01, p=.048). Happiness was weakly and negatively correlated with reasoning performance at baseline (r=−.03, p=.0006), but was not correlated with speed (r=.004, p=.63) or visuospatial memory (r=−.005, p=.49).
Fully unconditional models indicated significant variance in cognitive function at both the within-person and between-person levels. For speed, 52% of the variance was within-person (σ2=6162.15, z=101.73, p<.0001), and 48% of the variance was between people (τ00= 5758.37, z=59.96, p<.0001). For visuospatial memory, 80% of the variance was within-person (σ2=6.94, z=101.83, p<.0001), and 20% of the variance was between people (τ00= 1.69, z=27.57, p<.0001). For reasoning, 35% of the variance was within-person (σ2=1.48, z=101.79, p<.0001), and 65% of the variance was between people (τ00= 2.69, z=73.82, p<.0001).
Results of primary analyses for speed, visuospatial memory, and reasoning are in Tables 2–4. Happiness was associated with better performance in processing speed (shorter RT; Intercept) across assessments (Table 2, Model 1). Compared to individuals with lower happiness (scores 1 to 3, 4% of the sample), those who reported greater happiness (score 6, 5.8% of the sample) had faster speed (β=−20.16, SE=5.40, t=−3.74, p=.0002). The association remained significant when vascular conditions (Model 2) or depressive symptoms (Model 3) were included. There were significant decreases in speed over time for the sample and between-people variability around the slope (τ11= 14.77, z=5.43, p<.0001), but happiness did not predict the slope. Neither age nor sex interacted with happiness to predict the intercept or slope. Depressive symptoms were associated with slower speed across assessments, but neither vascular conditions nor depressive symptoms were associated with the slope of speed. The models explained 13% of the within-person variance in speed and 12% of the between-person variance in speed.
Table 2.
Multilevel models predicting speed of processing (reaction times)
| Model 1 |
Model 2 |
Model 3 |
|
|---|---|---|---|
| Fixed effects | Coefficient (SE) | Coefficient (SE) | Coefficient (SE) |
|
| |||
| Level of reaction time | |||
| Intercept | 558.78**** (1.26) | 558.14**** (1.32) | 558.44**** (1.28) |
| Age | 3.97**** (0.10) | 3.93**** (0.10) | 4.02**** (0.10) |
| Sex | −19.81**** (1.51) | −20.12**** (1.52) | −20.05**** (1.51) |
| College | −8.28**** (1.50) | −8.13**** (1.51) | −7.81**** (1.51) |
| Happiness (1–6) | −4.50**** (1.09) | −4.42**** (1.10) | −3.11* (1.21) |
| Vascular conditions | 2.92 (1.83) | ||
| Depressive symptoms (0–6) | 2.33** (0.90) | ||
| Slope of reaction time change | |||
| Time | 6.25**** (0.17) | 6.21**** (0.17) | 6.24**** (0.17) |
| Age × time | 0.04*** (0.01) | 0.04** (0.01) | 0.05**** (0.01) |
| Sex × time | −0.01 (0.20) | −0.04 (0.20) | 0.02 (0.20) |
| College × time | −0.73*** (0.20) | −0.73*** (0.20) | −0.72*** (0.20) |
| Happiness × time | −0.001 (0.15) | 0.02 (0.15) | 0.12 (0.16) |
| Vascular conditions × time | 0.25 (0.24) | ||
| Depressive symptoms × time | 0.23 (0.12) | ||
| Random effects (variance components) | |||
| Residual (σ2) | 4893.66**** (83.53) | 4875.37**** (83.33) | 4881.47**** (84.65) |
| Variance in cognitive level (τ00) | 5583.72**** (129.40) | 5591.41**** (129.30) | 5577.41**** (131.27) |
| Variance in cognitive slope (τ11) | 14.77**** (2.72) | 15.19**** (2.72) | 14.76**** (2.76) |
| R2 Level 1 | 13% | 13% | 13% |
| R2 Level 2 | 12% | 12% | 12% |
Note. Higher values of the outcome indicated worse performance. Model 1: N=17,885, Model 2: N=17,864, Model 3: N=17,296. Sex: 0=female, 1=male; College and vascular conditions: 0=no, 1=yes.
p<.01.
p<.001.
p<.001.
p<.0001
Table 4.
Multilevel models predicting reasoning
| Model 1 |
Model 2 |
Model 3 |
|
|---|---|---|---|
| Fixed effects | Coefficient (SE) | Coefficient (SE) | Coefficient (SE) |
|
| |||
| Level of reasoning | |||
| Intercept | 6.13**** (0.02) | 6.15**** (0.03) | 6.16**** (0.02) |
| Age | −0.0004 (0.002) | 0.001 (0.002) | −0.002 (0.002) |
| Sex | 0.21**** (0.03) | 0.21**** (0.03) | 0.20**** (0.03) |
| College | 1.15**** (0.03) | 1.15**** (0.03) | 1.13**** (0.03) |
| Happiness (1–6) | −0.05* (0.02) | −0.05* (0.02) | −0.13**** (0.02) |
| Vascular conditions | −0.08* (0.03) | ||
| Depressive symptoms (0–6) | −0.14**** (0.02) | ||
| Slope of change in reasoning | |||
| Time | −0.02**** (0.003) | −0.02**** (0.003) | −0.02**** (0.003) |
| Age × time | −0.001**** (0.0002) | −0.001**** (0.0002) | −0.002**** (0.0002) |
| Sex × time | 0.005 (0.003) | 0.005 (0.003) | 0.005 (0.003) |
| College × time | 0.005 (0.003) | 0.005 (0.003) | 0.005 (0.003) |
| Happiness × time | 0.002 (0.002) | 0.002 (0.002) | 0.001 (0.003) |
| Vascular conditions × time | 0.0001 (0.004) | ||
| Depressive symptoms × time | −0.002 (0.002) | ||
| Random effects (variance components) | |||
| Residual (σ2) | 1.47**** (0.01) | 1.47**** (0.01) | 1.47**** (0.01) |
| Variance in cognitive level (τ00) | 2.33**** (0.03) | 2.33**** (0.03) | 2.31**** (0.03) |
| R2 Level 1 | 9% | 9% | 9% |
| R2 Level 2 | 12% | 12% | 12% |
Note. Higher values of the outcome indicated better performance. Model 1: N=17,885, Model 2: N=17,864, Model 3: N=17,296. Sex: 0=female, 1=male; College and vascular conditions: 0=no, 1=yes. The slope is constrained to be equal across individuals.
p<01.
p<001.
p<.001.
p<.0001.
A similar pattern was found for visuospatial memory: Happiness was associated with better performance across assessments independent of vascular conditions or depressive symptoms (Table 3). Compared to less happy individuals, happier individuals had fewer memory errors (β=−0.38, SE=0.15, t=−2.63, p=.009). There were no significant changes in memory performance over time for the sample, but people varied in their rates of change (τ11= 0.02, z=5.00, p<.0001). Happiness was not a significant predictor of the slope, and age and sex did not interact with happiness to predict the intercept or slope. Neither vascular conditions nor depressive symptoms were associated with the intercept or slope of visuospatial memory. The models explained 2% of the within-person variance in visuospatial memory and 5% of the between-person variance in visuospatial memory.
Table 3.
Multilevel models predicting visuospatial memory (errors)
| Model 1 |
Model 2 |
Model 3 |
|
|---|---|---|---|
| Fixed effects | Coefficient (SE) | Coefficient (SE) | Coefficient (SE) |
|
| |||
| Level of visuospatial memory | |||
| Intercept | 3.71**** (0.03) | 3.70**** (0.04) | 3.70**** (0.04) |
| Age | 0.05**** (0.003) | 0.05**** (0.003) | 0.05**** (0.003) |
| Sex | −0.01 (0.04) | −0.02 (0.04) | −0.01 (0.04) |
| College | −0.09* (0.04) | −0.09* (0.04) | −0.08 (0.04) |
| Happiness (1–6) | −0.09** (0.03) | −0.09** (0.03) | −0.07* (0.03) |
| Vascular conditions | 0.06 (0.05) | ||
| Depressive symptoms (0–6) | 0.04 (0.03) | ||
| Slope of visuospatial memory | |||
| Time | −0.001 (0.006) | −0.003 (0.006) | −0.001 (0.01) |
| Age × time | 0.0004 (0.0005) | 0.0003 (0.0005) | 0.0004 (0.0005) |
| Sex × time | 0.01 (0.01) | 0.01 (0.01) | 0.01 (0.01) |
| College × time | −0.01 (0.01) | −0.01 (0.01) | −0.01 (0.01) |
| Happiness × time | −0.001 (0.005) | −0.0005 (0.005) | −0.002 (0.01) |
| Vascular conditions × time | 0.01 (0.01) | ||
| Depressive symptoms × time | −0.004 (0.004) | ||
| Random effects (variance components) | |||
| Residual (σ2) | 6.49**** (0.11) | 6.49**** (0.11) | 6.46**** (0.11) |
| Variance in cognitive level (τ00) | 1.88**** (0.12) | 1.88**** (0.12) | 1.91**** (0.13) |
| Variance in cognitive slope (τ11) | 0.02**** (0.003) | 0.02**** (0.003) | 0.02**** (0.004) |
| R2 Level 1 | 2% | 2% | 2% |
| R2 Level 2 | 5% | 5% | 5% |
Note. Higher values of the outcome indicated worse performance. Model 1: N=17,885, Model 2: N=17,864, Model 3: N=17,296. Sex: 0=female, 1=male; College and vascular conditions: 0=no, 1=yes.
p<.01.
p<.001.
p<.001.
p<.0001.
Happiness was negatively associated with performance on the reasoning task (Table 4). However, there was no significant difference between individuals in the bottom and top of the distribution of happiness (β=−0.16, SE=0.10, t=−1.60, p=.11). There were significant decreases in reasoning performance over time for the sample, but people did not vary significantly in their rates of decline (τ11= 0.0008, z=0.98, p=.16). Therefore, the slope of reasoning was constrained to be equal between people. Happiness was not a significant predictor of the slope. Exploratory analyses indicated an interaction between sex and happiness in predicting the slope (sex × happiness × time, β= −0.01, SE=0.005, t=−2.17, p=.03; demonstrated in Supplementary Fig. S1). However, this interaction was only found for reasoning and we therefore refrain from interpreting it. Models with random slopes yielded similar results (data not shown). Both vascular conditions and depressive symptoms were associated with worse reasoning across assessments, although neither was related to the slope of reasoning. The models explained 9% of the within-person variance in reasoning and 12% of the between-person variance in reasoning.
Discussion
In a large sample from the UK Biobank, we examined the association between happiness and the level and change in three cognitive functions over up to 10+ years of follow-up. Greater happiness was associated with better processing speed and visuospatial memory but worse reasoning across assessments; these associations were independent of vascular conditions or depressive symptoms. Happiness was not related to change in any of the cognitive outcomes over time.
For both speed and memory, happier individuals tended to maintain better levels of performance across years compared to less happy individuals. This pattern is broadly consistent with previous findings on positive affect (Hill et al., 2005) and satisfaction with life (Zhu et al., 2022b). Together with Hill and colleagues’ (2005) findings on positive affect and recall, our findings indicate that the cognitive benefits of happiness may extend from creative endeavors, as suggested by the broaden-and-build theory (Fredrickson, 2001), to more basic functions such as speed and memory. Several interconnected neuropsychological, behavioral, and cardiovascular factors may underlie the association between happiness and superior processing speed and memory. Positive affect is suggested to enhance momentary attention and cognitive flexibility through increased dopamine levels in the brain (Ashby et al., 1999) and build long-term cognitive and psychological resources (Fredrickson, 2001). Happiness has also been shown to be a driving force for living a healthier and cognitively beneficial lifestyle, including eating healthy and being physically active (Van Cappellen et al., 2018). Further, the presence of positive well-being is consistently associated with better cardiovascular health, a key risk/protective factor for cognitive wellness (Kubzansky et al., 2018). All these may be possible pathways through which generally happier individuals maintain faster speed of processing and better visuospatial memory.
The null association between happiness and the rates of decline in speed and memory over time was also broadly consistent with previous findings on positive affect (Berk et al., 2017) and life satisfaction (e.g., Zhu et al., 2022b), pointing to the difficulty of predicting the slope of change in cognitive function from well-being. In a previous study, composite measures of well-being that include both happiness and related constructs were related to slower rates of cognitive decline in a sample from Singapore (Rawtaer et al., 2017). Given our current findings, it seems that the protective effect against cognitive decline may be driven by other measures of well-being than by happiness. Nevertheless, although happier individuals decline at similar rates compared to their less happy counterparts, they may still maintain consistently better functioning over time. This also suggests that they may be less likely to fall below the threshold of impairment. Indeed, well-being measures such as life satisfaction (Zhu et al., 2022b) and purpose in life (Sutin et al., 2021) have been associated with lower risk of dementia.
The association between happiness and worse performance in reasoning was surprising. When we compared individuals in the top and bottom of the distribution of happiness, however, there was no significant difference. Several factors might contribute to the pattern of negative association. First, a similar pattern is found for dispositional traits related to well-being, such that higher extraversion tends to be associated with better processing speed (Sutin et al., 2019) but worse reasoning (Sutin et al., 2022). Scoring higher on the trait is associated with talking (Armon & Shirom, 2011) and thinking more quickly (Mairesse et al., 2007), but it may become a disadvantage for more complex tasks like reasoning in which multiple cognitive processes need to be employed to manipulate information (Sutin et al., 2022). Relatedly, higher life satisfaction has also been associated with worse performance in inhibitory control, the ability to deliberately suppress irrelevant distractions (Toh & Yang, 2022). Taken together, it may be speculated that well-being might provide a boost for tasks that require greater speed to perform well but may worsen performance on tasks that require greater deliberation. A related perspective is the speed-accuracy trade-off (Salthouse, 1979), which originally describes the tendency that individuals trade accuracy for speed and vice versa in cognitive tasks that gauge speed. Applying the speed-accuracy trade-off explanation to the current finding, it is possible that the faster speed in happier individuals may have a toll on accuracy in their reasoning performance. Lastly, the reasoning task in the UK Biobank requires that the participants can read and comprehend the text-based questions within a time limit, and therefore performance may be influenced by factors such as vision and reading ability. The robustness of findings needs to be tested with other forms of reasoning tasks, such as questions based on numbers and symbols as well as those presented via audio. Together, more evidence is needed to better understand the association between well-being-related measures with reasoning performance.
Strengths of the study include the large sample, three measures of cognitive function, and the four longitudinal assessments. The analyses considered both the overall cognitive level as well as cognitive decline, tested whether the associations were accounted for by competing risk factors, and whether associations were moderated by age and sex. Extending the association between positive affect and creative cognitive performance, our findings indicate that general happiness may also contribute to better processing speed and visuospatial memory.
We also note several limitations and opportunities for future research. First, in line with our goal of examining whether happiness has predictive value for cognitive function as it does for other health outcomes (Pressman & Cohen, 2005), we only analyzed the association in this direction. Although we excluded individuals with prevalent dementia, it is possible that less cognitively healthy people are more likely to report lower happiness at baseline. Nevertheless, our findings suggest that baseline happiness predicts cognitive functioning across up to 10+ years, but not cognitive decline as would be expected from reverse causality. Future research should examine possible (or null) reverse (Toh et al., 2020) or bidirectional effects (Gerstorf et al., 2007), particularly by treating well-being as a time-varying covariate to model well-being-cognition coupling over time (Allerhand et al., 2014; Zhu et al., 2022b). Second, only three cognitive tasks that were available across four assessments were examined. In particular, given the noted limitations with the reasoning task in the UK Biobank, future studies using other tasks may provide more evidence about the association with happiness or related well-being measures. Third, because the amount of variance in cognitive performance explained by our models was small to moderate, results should be interpreted with caution and replicated in future work. Finally, the sample included only individuals in the UK. Given the evidence from other populations on cognitive benefits associated with positive affect and related well-being measures (e.g., Dutch, Hill et al., 2005; Koreans, Zhu et al., 2022b), examining cultural similarities and differences is an important future direction.
Implications
At the theoretical level, the present study points to the complex links between well-being and cognitive function. On the one hand, while the broaden-and-build theory (Fredrickson, 2001) suggests the benefits of positive affect for creative cognitive endeavors, the present findings indicate that the benefits may extend to more basic cognitive processes like speed and memory. In this respect, the present study also adds cognitive evidence to the broad literature on happiness and well-being in general serving as health assets (e.g., Kubzansky et al., 2018; Pressman & Cohen, 2005). On the other hand, the negative association found between happiness and reasoning performance, if replicated in future research, would suggest that conceptual frameworks would need to account for not only benefits but also possible costs (e.g., lower performance in tasks that require greater deliberation) associated with happiness. At the practical level, more empirical evidence is needed to inform interventional strategies. While the study indicates a possibility that general happiness may contribute to better speed of processing and memory over time, we refrain from speculating on happiness-based interventions given the negative association found for reasoning and the null associations with rate of change in all three functions. It is critical for professionals to pay attention to what specific cognitive processes are targeted, since the same intervention strategy may not have uniform effects on all cognitive functions.
Conclusion
In a large sample of middle-aged and older adults followed for up to 10+ years, we found that greater happiness at baseline was associated with higher levels of processing speed and visuospatial memory across four assessment occasions. However, happier individuals did not have slower cognitive decline over time. Happiness was associated with worse performance in reasoning. The different patterns of association highlight the complex links between psychological well-being and cognitive function, as well as the importance of examining different cognitive functions individually.
Supplementary Material
Acknowledgment
This work was supported by the National Institute on Aging at the National Institutes of Health (grant numbers R01AG068093, R01AG053297, R01AG074573). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This research has been conducted using the UK Biobank Resource (Application Reference Number 57672).
Footnotes
Statements and Declarations
Conflicts of Interests
The authors have no relevant financial or non-financial interests to disclose.
Ethical approval
This study used publicly available, de-identified data from the UK Biobank (http://www.ukbiobank.ac.uk), and was thus exempted from additional review by the Institutional Review Board at the Florida State University. The UK Biobank obtained ethical approval from the North West Multicenter Research Ethics Committee.
Consent
All individual participants of the UK Biobank signed informed consent on participation.
Data availability
Data used in this study are publicly available from the UK Biobank.
References
- Abdel-Khalek AM (2006). Measuring happiness with a single-item scale. Social Behavior and Personality. 34(2), 139–150. 10.2224/sbp.2006.34.2.139 [DOI] [Google Scholar]
- Allerhand Gale, C. R., & Deary IJ (2014). The dynamic relationship between cognitive function and positive well-being in older people: A prospective study using the English Longitudinal Study of Aging. Psychology and Aging, 29(2), 306–318. 10.1037/a0036551 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Argyle M, Martin M, & Lu L (1995). Testing for stress and happiness: The role of social and cognitive factors. In Spielberger CD, Sarason IG, Brebner JMT, Greenglass E, Laungani P, & O’Roark AM (Eds.), Stress and emotion: Anxiety, anger, and curiosity (pp. 173–187). Taylor & Francis. [Google Scholar]
- Armon G, & Shirom A (2011). The across-time associations of the five-factor model of personality with vigor and its facets using the bifactor model. Journal of Personality Assessment, 93(6), 618–627. 10.1080/00223891.2011.608753 [DOI] [PubMed] [Google Scholar]
- Ashby FG, Isen AM, Turken, and U. (1999). A neuropsychological theory of positive affect and its influence on cognition. Psychological Review, 106(3), 529–550. 10.1037/0033-295X.106.3.529 [DOI] [PubMed] [Google Scholar]
- Berk L, van Boxtel M, Köhler S, & van Os J (2017). Positive affect and cognitive decline: a 12-year follow-up of the Maastricht Aging Study: Positive affect and cognitive decline. International Journal of Geriatric Psychiatry, 32(12), 1305–1311. 10.1002/gps.4611 [DOI] [PubMed] [Google Scholar]
- Carver C (2003). Pleasure as a sign you can attend to something else: Placing positive feelings within a general model of affect. Cognition and Emotion, 17(2), 241–261. 10.1080/02699930302294 [DOI] [PubMed] [Google Scholar]
- Diener E, Lucas RE, & Oishi S (2018). Advances and open questions in the science of subjective well-being. Collabra: Psychology, 4(1), 15. 10.1525/collabra.115 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Folley S, Zhou A, Llewellyn DJ, & Hyppoenen E (2019). Physical activity, APOE genotype, and cognitive decline: exploring gene-environment interactions in the UK Biobank. Journal of Alzheimer’s Disease, 71(3), 741–750. 10.3233/JAD-181132 [DOI] [PubMed] [Google Scholar]
- Fredrickson BL (2001). The role of positive emotions in positive psychology: The broaden-and-build theory of positive emotions. The American Psychologist, 56(3), 218–226. 10.1037/0003-066X.56.3.218 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gerstorf D, Lövdén M, Röcke C, Smith J, & Lindenberger U (2007). Well-being affects changes in perceptual speed in advanced old age: Longitudinal evidence for a dynamic link. Developmental Psychology, 43(3), 705–718. 10.1037/0012-1649.43.3.705 [DOI] [PubMed] [Google Scholar]
- Hill RD, van Boxtel MP, Ponds R, Houx PJ, and Jolles J (2005). Positive affect and its relationship to free recall memory performance in a sample of older Dutch adults from the Maastricht Aging Study. International Journal of Geriatric Psychiatry, 20(5), 429–35. 10.1002/gps.1300 [DOI] [PubMed] [Google Scholar]
- Kroenke K, Spitzer RL, & Williams JBW (2003). The Patient Health Questionnaire-2: Validity of a two-item depression screener. Medical Care, 41(11), 1284–1292. 10.1097/01.MLR.0000093487.78664.3C [DOI] [PubMed] [Google Scholar]
- Kubzansky LD, Huffman JC, Boehm JK, Hernandez R, Kim ES, Koga HK, Feig EH, Lloyd-Jones DM, Seligman MEP, & Labarthe DR (2018). Positive psychological well-being and cardiovascular disease: JACC Health Promotion Series. Journal of the American College of Cardiology, 72(12), 1382–1396. 10.1016/j.jacc.2018.07.042 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lyall DM, Cullen B, Allerhand M, Smith DJ, Mackay D, Evans J, Anderson J, Fawns-Ritchie C, McIntosh AM, Deary IJ, & Pell JP (2016). Cognitive Test Scores in UK Biobank: Data Reduction in 480,416 Participants and Longitudinal Stability in 20,346 Participants. PloS One, 11(4). 10.1371/journal.pone.0154222 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lyall LM, Cullen B, Lyall DM, Leighton SP, Siebert S, Smith DJ, & Cavanagh J (2019). The associations between self-reported depression, self-reported chronic inflammatory conditions and cognitive abilities in UK Biobank. European Psychiatry, 60, 63–70. 10.1016/j.eurpsy.2019.05.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lyubomirsky S, King L, & Diener E (2005). The benefits of frequent positive affect: Does happiness lead to success? Psychological Bulletin, 131(6), 803. 10.1037/0033-2909.131.6.803 [DOI] [PubMed] [Google Scholar]
- Mairesse F, Walker MA, Mehl MR, & Moore RK (2007). Using linguistic cues for the automatic recognition of personality in conversation and text. Journal of Artificial Intelligence Research, 30, 457–500. 10.1613/jair.2349 [DOI] [Google Scholar]
- Moldovan CP (2018). AM Happy Scale: Reliability and validity of a single-item measure of happiness. (Ph.D dissertation). Loma Linda University, Loma Linda, California. [Google Scholar]
- Nichols ES, Wild CJ, Owen AM, & Soddu A (2021). Cognition across the Lifespan: Investigating Age, Sex, and Other Sociodemographic Influences. Behavioral sciences (Basel, Switzerland), 11(4), 51. 10.3390/bs11040051 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Peitsch L, Tyas SL, Menec VH, and St John PD (2016). General life satisfaction predicts dementia in community living older adults: A prospective cohort study. International Psychogeriatrics, 28(7), 1101–1109. 10.1017/S1041610215002422 [DOI] [PubMed] [Google Scholar]
- Pressman SD & Cohen S (2005). Does positive affect influence health? Psychological Bulletin, 131(6), 925–971. 10.1037/0033-2909.131.6.925 [DOI] [PubMed] [Google Scholar]
- Raudenbush SW, & Bryk AS (2002). Hierarchical linear models: Applications and data analysis methods. 2nd ed (Vol. 1). Sage. [Google Scholar]
- Rawtaer I, Gao Q, Nyunt MSZ, Feng L, Chong MS, Lim WS, Lee TS, Yap P, Yap KB, and Ng TP (2017). Psychosocial risk and protective factors and incident mild cognitive impairment and dementia in community dwelling elderly: Findings from the Singapore Longitudinal Ageing Study. Journal of Alzheimer’s Disease, 57(2), 603–611. 10.3233/JAD-160862 [DOI] [PubMed] [Google Scholar]
- Salthouse TA (1979). Adult age and the speed-accuracy trade-off. Ergonomics, 22 (7), 811–821. 10.1080/00140137908924659 [DOI] [PubMed] [Google Scholar]
- Sheline YI, Barch DM, Garcia K, Gersing K, Pieper C, Welsh-Bohmer K, ... & Doraiswamy PM (2006). Cognitive function in late life depression: Relationships to depression severity, cerebrovascular risk factors and processing speed. Biological Psychiatry, 60(1), 58–65. 10.1016/j.biopsych.2005.09.019 [DOI] [PubMed] [Google Scholar]
- Subramaniam K, & Vinogradov S (2013). Improving the neural mechanisms of cognition through the pursuit of happiness. Frontiers in Human Neuroscience, 7. 10.3389/fnhum.2013.00452 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sutin AR, Aschwanden D, Luchetti M, Stephan Y, & Terracciano A (2021). Sense of purpose in life is associated with lower risk of incident dementia: A meta-analysis. Journal of Alzheimer’s Disease, 83(1), 249–258. 10.3233/JAD-210364 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sutin AR, Stephan Y, Luchetti M, Strickhouser JE, Aschwanden D & Terracciano A (2022). The association between five factor model personality traits and verbal and numeric reasoning. Aging, Neuropsychology, and Cognition, 29 (2), 297–317. 10.1080/13825585.2021.1872481 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sutin AR, Stephan Y, Luchetti M, & Terracciano A (2019). Five-factor model personality traits and cognitive function in five domains in older adulthood. BMC Geriatrics, 19. 10.1186/s12877-019-1362-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Toh WX, & Yang H (2022). Executive function moderates the effect of reappraisal on life satisfaction: A latent variable analysis. Emotion, 22(3), 554–571. 10.1037/emo0000907 [DOI] [PubMed] [Google Scholar]
- Toh WX, Yang H, & Hartanto A (2020). Executive function and subjective well-being in middle and late adulthood. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 75(6), 69–77. 10.1093/geronb/gbz006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Van Cappellen P, Rice EL, Catalino LI, & Fredrickson BL (2018). Positive affective processes underlie positive health behaviour change. Psychology & Health, 33(1), 77–97. 10.1080/08870446.2017.1320798 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Watson D, Clark LA, & Tellegen A (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54(6), 1063–1070. [DOI] [PubMed] [Google Scholar]
- Zhu X, Luchetti M, Aschwanden D, Sesker AA, Stephan Y, Sutin AR, & Terracciano A (2022a). Multidimensional assessment of subjective well-being and risk of dementia: Findings from the UK Biobank Study. Journal of Happiness Studies. Advance online publication. 10.1007/s10902-022-00613-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhu X, Luchetti M, Aschwanden D, Sesker AA, Stephan Y, Sutin AR, & Terracciano A (2022b). Satisfaction with life and risk of dementia: Findings from the Korean Longitudinal Study of Aging. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences. Advance online publication. 10.1093/geronb/gbac064 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data used in this study are publicly available from the UK Biobank.