Emotion Regulation and the Late Positive Potential (LPP) in Older Adults

Abstract

Emotion regulation (ER) processes in older adults may be important for successful aging. Neural correlates of ER processes have been examined using event-related brain potentials (ERPs), such as the late-positive potential (LPP) during cognitive reappraisal paradigms. The current study sought to extend this research by examining the LPP from an ER task in a sample of 47 community-dwelling older adults between the ages of 60 and 84 years, scoring either high on emotional well-being (as measured by habitual ER use and resiliency; high WB group, n = 20) or low on emotional well-being (as measured by habitual ER use, resiliency, and depression; low WB group, n = 27). Participants viewed unpleasant and neutral images and were instructed to simply react to the images or reappraise their emotional response. Both pre- and post-instruction LPP amplitudes were scored, in addition to self-reported ratings of negative emotion collected during the task. We found greater LPP amplitude to emotionally salient compared to neutral stimuli, reduced LPP amplitude following instructions to reappraise emotional response to stimuli across groups, and a blunted LPP overall for individuals with higher depressive symptoms. Additionally, we demonstrated that older adults with low emotional well-being were less successful at reappraisal according to self-reported ratings of negative emotion, although this was not reflected in the LPP. Collectively, these data suggest that laboratory-based ER tasks might be used to understand abnormal ER use—though the LPP may be more sensitive to depression than individual differences in ER ability.

1. Introduction

Emotion regulation (ER) processes in older adults may be important for successful aging, as adaptively controlling one’s emotions is crucial to one’s ability to plan, pursue, and achieve goals (Gross, 2015; Urry & Gross, 2010). Further, ER difficulties have been identified as an integral component of depressive disorders (Joorman & Quinn, 2014). Lu and Shelley (2021) found approximately 4–7% of older adults meet subthreshold criteria for depression that may substantially impact quality of life. Moreover, identifying patterns of ER processes in depression-prone older adults (e.g., those experiencing subthreshold depressive symptoms) may further our understanding of factors that contribute to depressive symptoms in aging and provide information on potential targets for intervention for depression in older adults.

Over the past decade, neural correlates of ER have been examined using event-related brain potentials (ERPs), which provide direct measures of neural activity that are time-locked to specific events (Jackson & Bolger, 2014; Luck, 2014; Woodman, 2010). ERPs are advantageous for studying distinct neural processes within and between individuals (Hajcak et al., 2019), thus providing a useful tool for examining neural correlates of ER processes during cognitive reappraisal paradigms (e.g., Deng et al., 2019; Hajcak & Nieuwenhuis, 2006; Langeslag & Surti, 2017). The current study sought to extend previous research by examining ERPs from an ER task in a sample of community-dwelling older adults scoring either high or low on emotional well-being as measured by habitual ER use, resiliency, and depressive symptoms.

1.1. Emotion Regulation and Aging

There are a variety of adaptive ER strategies, including those targeted at cognitive change, which refers to modifying the way a person thinks about a situation (Gross, 2015). Much empirical work has focused on cognitive reappraisal, defined generally as the reinterpretation of emotion-eliciting stimuli to attenuate negative affect (Buhle et al., 2014; Cutuli, 2014; Zilverstand et al., 2017). Specifically, this strategy is aimed at changing the way one thinks about a potentially distressing situation to alter its emotional impact. More frequent self-reported use of cognitive reappraisal has been associated with increased positive affect, experiencing and expressing less negative emotion, increased resilience to stressful events, better interpersonal functioning, and greater psychological well-being (Cutuli, 2014; Gross & John, 2003; Mauss et al., 2007; Sachs-Ericsson et al., 2019; Tugade & Fredrickson, 2007). Conversely, lower use of cognitive reappraisal and higher use of maladaptive ER strategies (e.g., rumination) is associated with higher levels of depressive symptoms (Joormann & Quinn, 2014), including in older adults (Kraaij et al. 2002).

Aging is associated with increased risk for many stressful circumstances that threaten well-being (e.g., chronic disease, loss of loved ones, and potential for decline in cognitive functioning; Sachs-Ericsson, Rushing, et al., 2016) and increase depressive symptoms (Hammen, 2005). For older adults, resilience represents the ability to return to equilibrium when stressful events occur (Windle, 2010), and is related to the use of proactive coping strategies that include adaptive ER skills (Tugade & Fredrickson, 2007). Older adults experiencing depressive symptoms may have particular difficulty regulating emotions when experiencing stressors (Joormann & Quinn, 2014; Kraaij et al., 2002; Sachs-Ericsson, Joiner, et al., 2016; Sachs-Ericsson, Rushing, et al., 2016). Although depression in older adults is associated with low levels of social support (Grav et al., 2011; Travis et al., 2004), Sachs-Ericsson et al. (2019) found that this relationship was moderated by cognitive reappraisal such that depressive symptoms were more strongly associated with low levels of perceived social support among those reporting less frequent use of cognitive reappraisal. This finding provides evidence that increasing use of effective ER strategies may increase resilience among older adults.

In general, older adults are thought to report higher levels of well-being (Charles et al., 2001; Scheibe & Carstensen, 2010) in part due to what has been described as a positivity effect whereby older adults attend to and remember positive information more than negative information compared to younger adults (Mather & Carstensen, 2005). This shift toward positive emotion may be a result of increased motivation to focus on emotional goals (e.g., maintaining social relationships; socioemotional selectivity theory, Carstensen et al., 1999), leading to an overall increase in the use of some ER skills with aging (Charles et al., 2009; Urry & Gross, 2010; Nashiro et al., 2012). Additionally, this positivity effect seems to be most prominent among older adults who are able to engage greater cognitive control (see review, Mather & Cartensen, 2005).

Researchers have documented changes in brain activation, as measured by functional magnetic resonance imaging (fMRI), that may underlie older adults’ ability to better utilize ER strategies while processing emotional stimuli. Findings demonstrate a decrease in amygdala activation to negative stimuli with age, a phenomenon that has been linked to ER skills (see review, Nashiro et al., 2012). Further, compared to younger adults, older adults exhibit greater prefrontal cortex (PFC, a region implicated in ER processes; Nashiro et al., 2012) activation while viewing emotionally salient stimuli, suggesting greater spontaneous engagement in ER strategies. These findings support the conceptualization that adaptive changes in older adults are not due to age-related cognitive deficits, but rather a greater focus on ER goals (Nashiro et al., 2012; Kryla-Lighthall & Mather, 2009).

Based on the above literature, emotional well-being in older adulthood seems to encompass greater use of adaptive ER strategies, higher resilience to stressful experiences, and lower levels of depression. Although older adults may demonstrate increased use of some ER skills overall (Charles et al., 2009; Urry & Gross, 2010; Nashiro et al., 2012), those with depressive symptoms may have difficulty implementing ER skills. In the current study we were interested in older adults with low and high levels of emotional well-being as measured by self-reported adaptive ER use, resiliency, and depressive symptoms – so that we could investigate differences in the ability to use proactive coping strategies (e.g., cognitive reappraisal) when instructed to do so. This work could further our understanding of ER processes and inform targets for intervention with the goal of increasing well-being in older adulthood. This goal is particularly important when considering evidence that suggests ER ability may buffer the negative effects of depression in older adults (Sachs-Ericsson et al., 2019).

1.2. Studying Emotion Regulation with ERPs

The late positive potential (LPP) is a protracted positive-going ERP component that becomes evident within 300 milliseconds following emotional stimulus onset and is typically sustained for the duration of stimulus presentation (Cuthbert et al., 2000; Hajcak et al., 2010). Research shows that the LPP is modulated by bottom-up image content: it is enhanced by arousing pleasant and unpleasant emotional content compared to neutral content (see Hajcak et al., 2010). For example, even within pleasant images, the amplitude of the LPP is larger for more motivationally salient image content (Weinberg & Hajcak, 2010). It has also demonstrated good-to-excellent reliability across trials (i.e., internal consistency; Moran et al., 2013). Thus, the LPP is a neural response that reflects engagement toward motivationally significant stimuli (Hajcak & Foti, 2020) that has good psychometric properties (Moran et al., 2013).

Further, the LPP is modulated relatively quickly by top-down processes (Moser et al., 2006). Indeed, several studies have measured the LPP to investigate ER processes, such as cognitive reappraisal (Baur et al., 2015; Deng et al., 2019; Hajcak et al., 2010; Hajcak & Nieuwenhuis, 2006; Harrison & Chassy, 2017; Langeslag & Surti, 2017; Myruski et al., 2019, Pedersen & Larson, 2016). Hajcak and Nieuwenhuis (2006) first demonstrated that among college-aged participants, the LPP elicited by unpleasant pictures was reduced after cognitive reappraisal instructions. Indeed, the modulation of the LPP was positively correlated with self-reported reductions in emotional intensity. Other studies have replicated these findings using variations of similar paradigms (Dunning & Hajcak, 2009; Foti & Hajcak, 2008; Hajcak et al., 2006). However, not all studies have consistently found a reduction in the LPP following instructions to down-regulate emotion (Baur et al., 2015; Langeslag & Surti, 2017; Langeslag & van Strien, 2010).

Other research has examined individual differences in the LPP to investigate the natural occurrence of ER to emotionally salient stimuli. For example, habitual use of reappraisal as measured by the Emotion Regulation Questionnaire (ERQ; Gross & John, 2003) was associated with a reduced LPP (i.e., decreased reactivity) in response to threatening pictures – even without explicit instruction to regulate emotions (Harrison & Chassy, 2017). These findings raise the possibility that individuals who habitually engage in ER may do so spontaneously when presented with emotionally salient information, thus resulting in reduced LPP amplitude.

The use of an ERP paradigm to measure emotional reactivity may be a useful methodology for investigating emotional processing in older adults. ERPs have excellent temporal resolution, and the LPP may reflect a more direct measure of ability to perform ER compared to self-reported use of various ER strategies. That is, it is possible that ER-related modulation of the LPP may reflect ER ability more than self-reported ER use.

The current study utilizes a paradigm in which participants first view an image and are subsequently instructed to either respond naturally to the image or to reframe the image content to reduce their emotional response (similar to Dunning and Hajcak 2009; Hajcak et al., 2006; Hajcak and Nieuwenhuis, 2006; and Parvaz and MacNamara, 2012). This paradigm allows for the separation of initial reactivity to the image and modulation of the LPP by subsequent regulation instructions. This distinction is important as ER ability may affect general reactivity to stimuli, regulation to those stimuli, or both.

Importantly, studies show that depression dampens neural reactivity to emotional stimuli (Bylsma et al., 2008; Canli et al., 2005; Proudfit et al., 2015). Proudfit et al. (2015) concluded that depressed individuals may disengage from emotionally salient content. In this regard, there is a growing body of literature indicating that both clinical diagnoses of depression and depressive symptoms are associated with reduced LPP to both positive and negative emotional stimuli (Foti et al., 2010; Hajcak & Foti, 2020; Hill et al., 2019; Klawohn et al., 2020; Proudfit et al., 2015; Weinberg et al., 2016). Thus, the presence of depressive symptoms in the current sample may lead to disengagement with emotional content, reflected in an attenuated LPP to emotional stimuli. Prior to the current study, there have been no investigations of how depressive symptoms in older adults may affect ability to engage with emotional stimuli during an ER task.

1.3. The Current Study

The current study leveraged the LPP to index emotional reactivity and regulation among an older adult sample who reported high or low emotional well-being (high WB group and low WB group, respectively). All participants completed an ER task while a continuous electroencephalogram (EEG) was recorded. Participants viewed unpleasant or neutral images, and were subsequently instructed to simply react to, or reappraise their emotional response to those images. Self-reported levels of negative emotion in response to task images were collected in addition to LPP amplitude. Assessing both reactivity to emotionally salient stimuli and the regulation thereof in older adults with variable emotional wellbeing may provide insight into individual differences in ER ability for older adults that could inform potential therapeutic targets for this population.

We predicted increased self-reported negative emotion ratings and greater LPP amplitude for unpleasant image blocks compared to neutral image blocks. Further, we expected retrospective ratings of self-reported negative emotion and LPP amplitude to be lower to reappraise trials compared to react trials, particularly for unpleasant image blocks. These effects were hypothesized to be stronger for the high WB group than the low WB group, as the former were selected based on greater ER use and higher resiliency, and resiliency is related to the use of proactive coping strategies such as adaptive ER (Tugade & Fredrickson, 2007). Further, the low WB group was selected in part, based on depressive symptoms, which have been associated with less use of adaptive ER strategies such as cognitive reappraisal (Kraaij et al., 2002l Joorman & Quinn, 2014). Additionally, we predicted that the high WB group would report lower levels of negative emotion overall compared to the low WB group. This was based on prior literature demonstrating an association between ER use and lower levels of negative emotion (Cutuli et al., 2014; Gross & John, 2003; Mauss et al., 2007) as well as the presence of higher depressive symptoms in the low WB group. Further, the high WB group were selected in part, based on higher resiliency, suggesting that this group might more easily recover from viewing unpleasant images. Finally, given prior work demonstrating a blunted LPP response associated with depression (see Foti et al., 2010; Hill, South et al., 2019; Klawohn et al., 2020; MacNamara et al., 2016; Proudfit et al., 2015; Weinberg et al., 2016), exploratory analyses were conducted to assess potential relationships between depressive symptoms and LPP responses in an older adult sample.

2. Method

2.1. Participants

Participants were recruited from the Institute for Successful Longevity (ISL) participant registry, which consists of community-dwelling older adults in North Florida. Participants were sent an email requesting completion of an online survey via Qualtrics which was used to determine their eligibility for the current study. A total of 910 participants completed the entire online survey. For more details and information about recruitment of the initial survey sample see Sachs-Ericsson et al. (2019).

Based on the survey sample, two groups (high WB and low WB) were identified for recruitment. The high WB group (n = 20) was comprised of individuals falling in the upper 25^th percentiles (within the sample) on both ER (M = 6.55, SD = 0.62) and resiliency (M = 91.55, SD = 4.9) as measured by the cognitive reappraisal facet of the ERQ (Gross & John, 2003) and the Connor-Davidson Resilience Scale (CD-RISC; Connor & Davidson, 2003), respectively. It should be noted that 22 individuals were originally recruited for the high WB group; however, two individuals dropped out of the study – one after their first office visit, and the other prior to their first office visit. The low WB group (n = 27) was comprised of individuals in the lower 25^th percentiles (within the sample) on both ER (M = 3.89, SD = 0.79) and resiliency (M = 54.59, SD = 9.35), as well as the upper 25^th percentile (within the sample) on depressive symptoms¹ (M = 9.89, SD = 4.82) as measured by the 8-item Patient Health Questionnaire (PHQ-8, Kroenke et al., 2009). It should be noted that 28 individuals were originally recruited for the low WB group; however, one dropped out prior to the first visit. Given the exploratory nature of the current study, there were no exclusionary criteria.

2.2. Measures

2.2.1. Emotion Regulation.

ER was assessed using the cognitive reappraisal subscale of the ERQ (Gross & John, 2003), which is comprised of six items (e.g., “I control my emotions by changing the way I think about the situation I’m in”) on a Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree). The ERQ was scored by calculating the average response to the items in the facet of interest, and thus the possible score range was 1.0 – 7.0. The reliability for the current sample was α = .94.

2.2.2. Resilience.

Resiliency was assessed using the CD-RISC (Connor & Davidson, 2003), which is comprised of 25 items designed to measure how effectively individuals recover from adverse situations. Participants indicated how accurate a statement (e.g., “I can deal with whatever comes my way”) was for them based on the past month by responding on a 5-point Likert-type scale from 0 (not at all true) to 4 (true nearly all of the time). Possible score on the CD-RISC ranges from 0 – 100. The reliability for the current sample was α = .97.

2.2.3. Depressive Symptoms.

Depressive symptoms were assessed using the PHQ-8 (Kroenke et al., 2009), which has been shown to be a valid and reliable measure that is effective at screening for a diagnosis of depression consistent with the longer PHQ-9 (Kroenke et al., 2001; Wu et al., 2019). Participants responded on a 4-point Likert-type scale ranging from 0 (not at all) to 3 (nearly every day) to indicate how often they’ve been bothered by depressive symptoms (e.g., “Little interest or pleasure doing things”) over the past 2 weeks, with a possible score range of 0 – 24 (10+ indicating current depression; Kroenke et al., 2009). The reliability for the current sample was α = .87.

2.3. Procedure

The current study was approved by Florida State University’s Institutional Review Board. All participants came to the researchers’ offices for an initial 1-hour session for completion of consents and assessments. All participants then attended a subsequent EEG visit (procedures described below) and were compensated $40 for their time.

2.4. EEG Procedures

2.4.1. Emotion Regulation Task.

Stimuli consisted of 50 unpleasant images (e.g., threatening scenes; upsetting images such as a mutilated body) and 50 neutral images (e.g., household objects) selected from the International Affective Picture System (IAPS; Lang et al., 1999)². Images were presented against a black background on a computer using Presentation software (Neurobehavioral Systems, Albany, California) to control presentation and timing of images. Neutral and unpleasant images differed in their normative ratings of valence (t(98) = −23.19, p < .001; neutral: M = 5.02, SD = .44, unpleasant: M = 2.39, SD = .67) and arousal (t(98) = 22.22, p < .001; neutral M = 3.44, SD = .41, unpleasant: M = 5.88, SD = .65). These values are comparable to those reported in previous studies of the LPP (Hajcak & Nieuwenhuis, 2006). Participants sat approximately 31 cm from the computer monitor and images were presented at a visual angle of approximately 15 by 20 degrees. Participants were told they would be viewing a series of images, and during each trial they would be instructed by an auditory stimulus to either “react” or “reframe” in response to the image. An experimenter explained to participants that to “react,” they should simply view and think about the image as they normally would, and to “reframe,” they should either reinterpret the image in a less negative way or view the image from a more detached, objective point of view. To familiarize participants with the task, a total of nine practice trials were administered. The first three practice trials included one neutral image (i.e., staircase) and two unpleasant images (i.e., violent law enforcement situation, crying child) during which an experimenter coached the participant on how to react and reframe appropriately. On the third image, the experimenter asked the participant to reframe the image out loud to ensure the participant understood the task. Six more practice trials were then presented with the accompanying auditory instruction provided through headphones to either “react” (i.e., react trial) or “reframe” (i.e., reappraise trial) presented 1000 ms after image presentation. Sound files consisted of a human female voice providing the one-word instruction and were approximately 600ms long (i.e., 610 ms & 630 ms, respectively). The task lasted approximately 15 minutes and consisted of two block types based on image type (unpleasant, neutral) each presented twice for a total of four blocks containing 25 trials each. Images were pulled at random exactly once from the pool of 50 images for each image type (unpleasant, neutral). Block presentation order was completely randomized. Each image was presented for a total of 7000 ms with a fixed inter-stimulus-interval of 2000 ms during which a fixation cross was presented at the center of the screen.

2.4.2. Self-Reported Negative Emotion.

Self-reported negative emotion was assessed throughout the ER task. After each block of images (i.e., four time points), participants were asked three questions. First, participants were asked to rate their current emotional state on a 5-point scale ranging from “not at all negative” to “extremely negative,” with “moderately negative” at the midpoint. For analysis, scores were averaged based on image block type to obtain post-block ratings of negative emotion for unpleasant and neutral images separately. Using the same scale, participants were also asked to retrospectively rate how negatively they felt after they reacted to the images and after they reframed the images in the preceding block. The same averaging procedure was used to obtain retrospective ratings of negative emotion based on instruction type for unpleasant and neutral images separately.

2.4.3. EEG Recording and Data Processing.

Continuous EEG was recorded while participants completed the ER task. An active electrode system (ActiCHamp, Brain Products, GmbH, Gilching, Germany) was used and included 32 scalp electrodes, arranged according to the 10/20 system (Homan, 1988). Two of these electrodes were placed directly behind the left and right ears on the mastoid bone. Vertical electrooculogram (VEOG) activity was recorded from electrodes placed 1 cm above and below the left eye, and horizontal electrooculogram (HEOG) activity was recorded from electrodes placed adjacent to the outer canthi of the left and right eyes. EEG signal was pre-amplified at the electrode to improve signal-to-noise ratio, and data were digitized at a 24-bit resolution with a sampling rate of 1000 Hz using a low pass filter of 100 Hz. Impedance levels were reduced to below 25 kOhms prior to recording. EEG was referenced online to scalp site Cz.

All EEG data were processed offline using Brain Vision Analyzer version 2.1 (Brain Products GmbH, Gilching, Germany). Raw EEG data were re-referenced offline to the average activity of the left and right mastoids, and band-pass filtered from 0.01 to 30 Hz. Data were segmented for each trial creating a window from 200 ms pre-stimulus onset to 7000 ms post-stimulus onset to capture the full image presentation. Eye-blink and ocular movement artifacts were corrected using an established regression-based algorithm developed by Gratton et al. (1983). Artifact rejection was performed for each individual channel using a semi-automated procedure which identified if a voltage step greater than 50 μV was present between sample points, if a deflection greater than 300 μV occurred within an epoch, or if a voltage difference of less than 0.50 μV was detected within 100 milliseconds intervals. Visual inspection of remaining trials was conducted to detect any remaining artifacts. The 200 ms pre-stimulus interval was used for baseline correction.

Based on a collapsed localizers approach³ (Luck & Gaspelin, 2017), the LPP was scored as the average activity where it was maximal on the scalp at a pooling of parietal-occipital (P3, Pz, P4, O1, Oz, O2) electrode sites in two separate time windows: 500 ms – 1000 ms (pre-instruction) and 1500 ms – 7000 ms (post-instruction). The LPP was scored separately for react and reappraise trials in the unpleasant and neutral blocks, resulting in the LPP to four different trial types (i.e., react unpleasant, reappraise unpleasant, react neutral, reappraise neutral), at the two time windows (pre-instruction, post-instruction).

2.5. Analysis

All statistical analyses were conducted using IBM SPSS Statistics 26. Split-half reliability estimates for each ERP were conducted by calculating the Spearman-Brown Coefficient (Farnsworth, 1928) between odd and even trials for each LPP measure. Mixed-model Analyses of Variance (ANOVAs) were conducted to test the hypotheses in the current study. Two participants (1 low WB, 1 high WB) did not have rating data resulting in 45 participants for behavioral analyses. One individual was excluded from analyses due to a lack of task markers rendering the EEG data unable to be processed (high WB group), resulting in 46 participants for all LPP analyses.

3. Results

3.1. Sample Characteristics

The overall sample (N = 47) was between 60 and 84 years of age (M = 67.64, SD = 5.65) and primarily White (87.2%). There were 19 males and 28 females, and all reported completing at least some college (Some college: 36.2%, Bachelor’s: 23.4%, Master’s: 27.7%, Doctorate: 12.8%). There were no significant differences between groups (high WB, low WB) on age, sex, race, or education level (ps > .358). Groups significantly differed on ER, resiliency, and depression scores (ps < .0001). See Table 1 for descriptive means and Table 2 for zero-order correlations between the study variables. See Table S1 of the Supplement for zero-order correlations for un-collapsed study variables.

Table 1:

Descriptive Statistics

	High WB Group (n = 20)	Low WB Group (n = 27)	Total Sample (N = 47)
Continuous Variables
Emotion Regulation*** (Cognitive Reappraisal)	6.55 (.62)	3.89 (.79)	5.02 (1.51)
Depressive Symptoms***	1.80 (2.24)	9.89 (4.82)	6.45 (5.61)
Resiliency***	91.55 (4.90)	54.59 (9.35)	70.32 (20.01)
Age	67.05 (5.88)	68.07 (5.54)	67.64 (5.65)
Categorical Variables
Gender (% female)	65	55.6	59.6
Race (% non-White)	15%	11.1%	12.8

Note. Continuous variables are reported as mean (standard deviation). Categorical variables are reported as percentages.

^***Between group difference of p < .0001

Table 2.

Table 2

Variable	n	1	2	3	4	5	6
1. Emotion regulation (cognitive reappraisal)	47	-
2. Depressive symptoms	47	−0.59**	-
3. Resiliency	47	0.88**	0.75**	-
4. Negative emotion (current)	45	−0.33*	0.50**	−0.48**	-
5. Negative emotion (retrospective)	45	−0.21	0.33*	−0.27	0.63**	-
6. Pre-instruction LPP	46	0.21	−0.09	0.20	0.02	−0.02	-
7. Post-instruction LPP	46	0.34*	−0.31*	0.34*	−0.14	−0.08	0.69**

Note. Ratings of current negative emotion were collapsed across block type. Retrospective ratings of negative emotion in relation to instruction type were collapsed across block type and instruction type. LPP amplitudes were collapsed across image and trial type. Zero-order correlations with non-collapsed variables are reported in the supplementary materials in Table S1.

^*p < .05.

^**p < .01.

3.2. Self-Reported Negative Emotion Ratings

To analyze current negative emotion ratings, we ran an image type (unpleasant, neutral) by WB group (high WB, low WB) mixed model ANOVA, which revealed that participants reported higher levels of negative emotion after unpleasant image blocks (M = 2.13, SD = .71) compared to neutral image blocks (M = 1.36, SD = .50; F(1, 43) = 46.49, p < .001, η_p² = .519), regardless of group membership. Participants in the low WB group reported higher levels of negative emotion (M = 1.99, SD = .48) compared to the high WB group (M = 1.50, SD = .48; F(1, 43) = 11.43, p = .002, η_p² = .210) regardless of image type⁴. There was no significant image type by group interaction (p = .430).

To analyze retrospective ratings of negative emotion based on instruction type, we ran an image type (unpleasant, neutral) by instruction type (react, reappraise) by group (high WB, low WB) mixed model ANOVA. Participants reported increased negative emotions after unpleasant compared to neutral image blocks (F(1, 43) = 163.39, p < .001, η_p² = .792; unpleasant: M = 2.79, SD = .80, neutral: M = 1.32, SD = .42) and after react compared to reappraise instructions (F(1, 43) = 13.32, p < .001, η_p² = .237; react: M = 2.19, SD = .62, reappraise: M = 1.92, SD = .52); there was also a significant image type by instruction type two-way interaction (F(1, 43) = 41.01, p < .001, η_p² = .488)⁵, and a significant image type by instruction type by group three-way interaction (F(1, 43) = 6.11, p = .017, η_p² = .124). Breakdown of the three-way interaction by image type revealed a significant instruction type by group two-way interaction after unpleasant image blocks (F(1, 43) = 5.83, p = .020, η_p² = .119). Post-hoc Bonferroni-corrected t-tests (ɑ_corr = .05/2 = .025) indicated higher levels of negative emotion after reappraise instructions for the low WB group (M = 2.79; SD = 0.81) compared to the high WB group (M = 2.21; SD = .73; t(43) = 2.45, p = .018, η_p² = .123)⁶, but no group difference after react instructions (t(43) = .22, p = .825, η_p² = .001; see Table 3). There was no two-way interaction between instruction type and group after neutral image blocks (F(1, 43) = 0.06, p = .815, η_p² = .001); therefore, no further post hoc tests were conducted to decompose this interaction.

Table 3.

Table 3

	Self-reported ratings of negative emotion
Current rating
	Neutral image block		Unpleasant image block
High WB (n = 19)	1.16 (0.33)		1.84 (0.75)
Low WB (n = 26)	1.56 (0.59)		2.42 (0.67)
Retrospective rating
	React	Reappraise	React	Reappraise
High WB (n = 19)	1.24 (0.35)	1.26 (0.48)	3.05 (1.04)	2.21 (0.73)
Low WB (n = 26)	1.37 (0.50)	1.42 (0.50)	3.12 (0.86)	2.79 (0.81)

Note. Mean ratings (current & retrospective) of negative emotion by image block and instruction type. Reported at mean (standard deviation).

3.3. LPP Amplitude

Mean LPP amplitudes for each group, instruction type, image type, and time window included in analyses are presented in Table 4. Seven individuals (4 low WB, 3 high WB) with LPP amplitude values greater than 3 standard deviations above or below the mean were winsorized such that outlying values were adjusted to reflect boundary values. The LPP demonstrated moderate-to-excellent reliability (see Table 5). Grand average ERPs and topographic maps are presented in Figures 1 and 2, respectively.

Table 4:

LPP Amplitudes

	LPP Amplitude (μV)
	Neutral		Unpleasant
	React	Reappraise	React	Reappraise
Pre-Instruction
High WB (n = 19)	2.66 (3.81)	3.41 (3.78)	5.75 (4.84)	5.93 (3.69)
Low WB (n = 27)	1.48 (4.52)	2.10 (4.31)	4.46 (3.95)	4.24 (4.56)
Post-Instruction
High WB (n = 19)	.77 (4.67)	.19 (5.56)	2.61 (4.59)	.69 (3.80)
Low WB (n = 27)	−2.02 (5.60)	−3.02 (6.37)	−.46 (4.25)	−1.54 (5.77)

Note. LPP amplitudes by image type, instruction type, time window (pre-instruction: 500 ms – 1000 ms; post-instruction: 1500 ms – 7000 ms), and group. Reported as mean (standard deviation). Scores are winsorized as described in the main text.

Table 5:

Reliability Coefficients

	Reliability Coefficients
	Neutral		Unpleasant
	React	Reappraise	React	Reappraise
Pre-Instruction LPP	.65	.63	.82	.85
Post-Instruction LPP	.71	.81	.47	.44

Note. Spearman Brown reliability coefficients for odd/even pre- and post-instruction LPP amplitudes across groups (N = 46). Reliability coefficients reflect raw data (not winsorized).

Fig. 1.

Note. ERPs at a parietal-occipital pooling (P3, Pz, P4, O1, Oz, O2) of electrode sites for the high WB group (top) and the low WB group (bottom). The dashed lines (from left to right) indicate image and instruction onset, respectively. Grand averages were created with raw data (not winsorized).

Fig. 2.

Note. Main effects of group (left) and instruction type (right) on the post-instruction LPP (1500 ms – 7000 ms).

3.3.1. Pre-Instruction LPP.

The pre-instruction LPP was analyzed with an image type (unpleasant, neutral) by instruction type (react, reappraise) by group (high WB, low WB) mixed model ANOVA. LPP amplitudes were larger following unpleasant images (M = 5.09, SD = 4.10) compared to neutral images (M = 2.41, SD = 3.80) regardless of group and pending instruction type (F(1, 44) = 50.66, p < .001, η_p² = .535; see Figure S1 of the Supplement). There were no significant main effects of instruction type or WB group and no significant interactions (ps > .221).

3.3.2. Post-Instruction LPP.

The post-instruction LPP was analyzed with an image type (unpleasant, neutral) by instruction type (react, reappraise) by group (high WB, low WB) mixed model ANOVA. Consistent with the pre-instruction LPP, LPP amplitudes were larger following unpleasant images (M = .33, SD = 3.85) compared to neutral images (M = −1.02, SD = 4.45) regardless of instruction type or WB group (F(1, 44) = 6.18, p = .017, η_p² = .123). As expected, LPP amplitudes were smaller following reappraise trials (M = −.92, SD = 4.17) compared to react trials (M = .23, SD = 4.00) regardless of image type, or WB group (F(1, 44) = 5.45, p = .024, η_p² = .110), see Figure 2. Finally, the low WB group demonstrated lower LPP amplitudes (i.e., less reactivity; M = −1.76, SD = 3.67) compared to the high WB group (M = 1.07, SD = 3.67) across image type, instruction type, and time window (F(1, 44) = 6.61, p = .014, η_p² = .131), see Table 4 and Figures 1, 2, and S1. There were no significant two- or three-way interactions (ps > .674).

Based on the above findings, prior literature suggesting blunted LPP amplitude to be related to depression, and that depression scores were normally distributed across our sample, we ran Pearson correlations between LPP amplitudes and depressive symptoms across groups. Although the pre-instruction LPP (collapsed across image type and instruction type) was not significantly associated with depressive symptoms (r(44) = −.09, p = .540), higher depressive symptoms were significantly associated with lower post-instruction LPP amplitude (collapsed across image type and instruction type; r(44) = −.31, p = .039)⁷, see Figure 3.

Fig. 3.

Note. The relationship between post-instruction LPP amplitude (collapsed across image type and instruction type) and depressive symptoms as measured by the PHQ-8.

4. Discussion

The current study sought to examine emotional reactivity and ER (measured by self-reported negative emotion in response to task images and the LPP) in a laboratory-based ER task among older adults who reported either high emotional well-being (high WB group) or low emotional well-being (low WB group). We found the low WB group reported higher levels of negative emotion overall, greater negative response to unpleasant images after reappraisal instructions, and overall lower LPP amplitudes compared to the high WB group.

Analysis of ratings of negative emotion revealed that the low WB group reported significantly higher levels of negative emotion (across image type) than the high WB group. Follow-up analyses suggested that this effect was driven by shared variance across habitual ER use, resiliency, and depressive symptoms (see footnote 4). This is consistent with existing literature; specifically lower use of adaptive ER strategies is associated with higher levels of depressive symptoms (Joorman & Quinn, 2014; Kraaij et al., 2002), and resiliency is related to proactive coping strategies such as adaptive ER (Tugade & Fredrickson, 2007). Thus, the current study demonstrates that low emotional well-being (i.e., collectively reflected by lower habitual ER use and resiliency, and higher depressive symptoms) is related to increased self-reported negative emotions in our laboratory-based ER task – across all image and instruction types.

Ratings also confirmed that participants experienced lower negative emotion after reappraise instructions compared to react instructions (consistent with: Baur et al., 2015; Deng et al., 2019; Langeslag & Surti, 2017). However, the low WB group reported experiencing significantly more negative emotion compared to the high WB group after using reappraise instructions for unpleasant images. Thus, although both groups reported success in decreasing levels of negative emotion to unpleasant images when utilizing cognitive reappraisal, the high WB group was more successful at utilizing the reappraisal instructions based on self-reported negative emotion. Further, follow-up analyses suggested that this effect was driven by shared variance across habitual ER use, resiliency, and depressive symptoms (see footnote 6). These findings suggest that individual differences in ER use, resiliency, and depressive symptoms predicted both increased negative emotions to stimuli and increased negative emotions following attempts to reappraise unpleasant images. Collectively, these data suggest that individual differences in emotional well-being correspond to self-reported negative emotions in laboratory-based ER tasks.

Analysis of pre- and post-instruction LPP amplitudes revealed greater LPP amplitude to unpleasant images compared to neutral images. These findings are consistent with the existing literature suggesting the LPP to provide an index of emotional reactivity (Hajcak et al., 2010; Hajcak & Foti, 2020; Schupp, Cuthbert, et al., 2004; Schupp, Öhman et al., 2004). Moreover, there was an overall reduction in LPP amplitude following reappraise instructions compared to react instructions. Consistent with several past studies (Dunning & Hajcak, 2009; Foti & Hajcak, 2008; Hajcak & Nieuwenhuis, 2006; Hajcak et al., 2006; Moran et al., 2013; cf., Bauer et al., 2015; Langeslag & Surti, 2017; Langeslag & Van Strien, 2010; Myurski et al., 2019; Pedersen & Larson, 2016), these findings indicate that participants were able to successfully utilize the cognitive reappraisal strategy on command to reduce the LPP.

Whereas some studies have speculated that older adults may have difficulty using cognitive reappraisal due to the high cognitive demand (Liang et al., 2017; Opitz et al., 2012; Shiota & Levenson, 2009), we found that older adults successfully utilized the cognitive reappraisal instructions. These results are consistent with previous findings by Langeslag and van Strien (2010), who similarly found that older adults were able to successfully modulate LPP amplitude when instructed to increase or decrease the feelings that the image elicited.

It is important to note that we did not find evidence that individuals with high and low emotional well-being differed in their ability to modulate the LPP. That is, although a group difference in the ability to regulate negative emotion in response to unpleasant images was evident according to self-reported negative emotion during the task, groups did not differ in their ability to modulate the LPP. Across both high and low WB groups, we found a reduction in LPP amplitude following reappraisal instructions for both unpleasant and neutral images. These data are consistent with work by MacNamara and colleagues (2009), who found that pre-appraisals (i.e., negative or neutral descriptions of upcoming stimuli) could modulate the LPP to both unpleasant and neutral stimuli – consistent with the notion that ER processes impact the LPP, regardless of stimulus valence.

Finally, the low WB group was characterized by a blunted post-instruction LPP that was evident across image type and instruction type. That is, whereas the low WB group did demonstrate the expected effect of instruction type on the LPP (i.e., LPP was attenuated after reappraise instructions), they also showed a reduction of the LPP response overall compared to the high WB group. Given literature demonstrating a reduced LPP among depressed individuals (Foti et al., 2010; Hill, South et al., 2019; Klawohn et al., 2020; MacNamara et al., 2016; Proudfit et al., 2015; Weinberg et al., 2016), we speculated whether this overall blunted LPP for the low WB group may be related to level of depressive symptoms. Indeed, in the entire sample lower LPP amplitude was related to higher depressive symptoms across conditions and when examining LPP amplitude to unpleasant images only (see footnote 7). This finding is generally consistent with the emotion-context insensitivity (ECI) view of depression (Rottenberg, 2007; Rottenberg & Hindash, 2015), which posits that depressed individuals demonstrate blunted reactivity to all emotional stimuli (both positive and negative; Bylsma, 2021). The blunted LPP amplitude in the current study provides further evidence for this view of how depression affects an individual’s emotional processing of negative stimuli and demonstrates the first electrophysiological evidence for this phenomenon in older adults.

It should be noted that although comparable effects of depression were observed on the pre-instruction LPP (i.e., the low WB group demonstrated smaller LPP amplitude compared to the high WB group and higher depressive symptoms were related to lower LPP amplitudes), neither of these effects were statistically significant. One possibility is that the absence of an effect of WB and depressive symptoms during the initial image viewing may be due to the anticipation of subsequent ER instructions—it is possible that participants were more engaged throughout this time period relative to studies that have employed passive viewing paradigms to examine the LPP in relation to depression and depressive symptoms.

When considering both self-report and LPP data, the findings of overall higher levels of negative emotion and overall blunted LPP amplitude in the low WB group compared to the high WB group are at odds with one another. This was not an effect we predicted, and it seems to suggest that group-related differences in emotional well-being influence different response modalities in different ways (i.e., the low WB group reported higher negative emotion, but were also characterized by reduced LPP amplitude). Future work might further examine whether this divergence between self-report and the LPP is important as an individual different measure. Further, the low WB group demonstrated worse reappraisal to unpleasant images according to self-reported negative emotion, but this was not evidenced in ability to modulate the LPP, suggesting individual differences in ER ability were evident in terms of self-reported negative emotion, but were not reflected in the LPP. It is possible that due to high variability in LPP data, subtle differences in the ability to modulate emotional response are more difficult to detect than when analyzing self-reported level of negative emotion. It is also possible that the inclusion of multiple selection criteria (i.e., habitual ER use, resiliency, and depressive symptoms) obscured group differences in ability to modulate the LPP. Further, self-report ratings share method variance with ER measures. Finally, it is possible that while this laboratory-based ER task is useful for assessing modulation of the LPP, it may not be optimal for studying individual differences in the habitual use of ER.

4.1. Limitations & Future Directions

There are several limitations of the current study that should be considered. First, the instructions provided for cognitive reappraisal during the ER task described different methods of cognitive reappraisal (i.e., reinterpreting the image in a less negative way - positive reappraisal or viewing the picture from a more detached, objective point of view - detached reappraisal). Indeed, some work has shown differences in the LPP based on ER strategy such that distraction reduced LPP amplitude earlier compared to cognitive reappraisal (Thiruchselvam et al., 2011). Future studies might examine whether different types of cognitive reappraisal are better or worse than others at attenuating emotional reactivity among older adults, as some forms of cognitive reappraisal may be preferred over others in this population (e.g., positive reappraisal vs detached reappraisal; Scheibe et al., 2015; Shiota & Levenson, 2009). Such differences might potentially be due to differences in cognitive load required to engage in various ER strategies. Further work in this area may be especially important for developing treatment strategies that are specific to older adults who may be experiencing cognitive decline.

Second, there was a lack of assessment about what participants did during the ER task to regulate emotions. That is, we did not collect verbal report from participants about what they actually did while engaging with the task. Such information would allow us to further assess potential differences (Ericsson & Simon, 1984). Whereas participants completed practice trials during which they completed reappraise trials out loud so that the experimenter could confirm appropriate use of ER, it is possible that participants used other strategies to disengage with unpleasant stimuli (e.g., looked away or engaged in a different type of ER) while viewing images during the actual trials of the task.

Third, we cannot draw conclusions as to whether the results are specific to older age groups due to the lack of a younger adult comparison group. In addition to exploring efficacy of different types of cognitive reappraisal as described above, future work could include a broader age range to allow for more direct comparisons between younger and older adults. While some work has examined the LPP during an ER task in both younger and older adults (Langeslag & Van Strien, 2010) and found that ER ability did not differ based on age, this comparison has not been studied in the context of individual differences in emotional well-being.

Fourth, although the LPP demonstrated moderate-to-excellent reliability in the current sample, the post-instruction LPP to unpleasant images demonstrated the lowest reliability. We hypothesize that individual differences in the level of unpleasantness perceived by the images may have contributed to the lower reliability values. Further, engagement in react or reappraisal instructions may produce more variability for emotional content due to individual differences in perceived emotional intensity and ability to modulate it. Other trial-to-trial variability factors could also be contributing such as image-specific processes, strategies for engagement with the instructions, fatigue, or fluctuations in attention (see Macatee et al., 2021 and Moran et al., 2013 for in depth analysis of the reliability of the LPP).

Fifth, due to the sample selection methods used, it is difficult to conclude whether the observed group differences were due to habitual ER use, depressive symptoms, or resiliency, as all three were utilized as selection criteria for the low WB group, and two of the three (i.e., habitual ER use & resiliency) used for selection of the high WB group in the current study. Based on high intercorrelations (see Table 2), it is likely these variables reflect a similar construct in the current sample, that which we are referring to as emotional well-being. Further, submitting these three variables to multiple regressions predicting current negative emotion ratings (i.e., probing the main effect of group; see footnote 4) and retrospective negative emotion ratings with respect to reappraise instructions for unpleasant image blocks (i.e., probing the three-way interaction effect; see footnote 6), resulted in significant relationships overall, but no individual predictor was significant over the others in either analysis. Thus, it is likely that the shared variance among these measures is driving the difference in ratings between groups.

Ratings of negative emotion overall were significantly related to depressive symptoms and resiliency, but not habitual ER use (albeit in the expected direction and trending toward significance, see Table S1 in the Supplement). Thus, it is possible that depressive symptoms played a larger role in the main effect of higher self-reported negative emotion in the low WB group compared to the high WB group. However, negative emotion ratings with respect to reappraise instructions for unpleasant image blocks (i.e., the interaction effect) were significantly related to ER, depressive symptoms, and resiliency (see Table S1 in the Supplement). Thus, as described above, it is likely that shared variance among these factors is driving the difference in ratings between groups. This was to be expected as adaptive ER use is thought to increase resiliency (Tugade & Frederickson, 2007) and higher levels of ER are related to lower levels of depression (see review, Fiske et al., 2009; Kraaij et al., 2002).

Finally, due to limited funding, the sample size of the current study was relatively small and may have limited our ability to detect small effects. Future work should seek to replicate the current findings in a larger sample of older adults. This could allow for self-reported ER use, depressive symptoms, and resiliency to be treated as continuous variables rather than used as selection criteria for extreme groups. In addition to suggestions described above, future work might try to improve the ability to utilize the LPP to investigate individual differences in habitual ER use. One way to achieve this could be to adapt ER tasks to include stimuli based on personal experience (e.g., Speed et al., 2017), or to use stimuli that might elicit greater distress—both approaches might be more akin to situations in which ER would be needed during daily life.

5. Conclusion

In summary, our findings corroborate previous work and extend several findings to older adults. Specifically, we demonstrate 1) enhanced LPP amplitude to emotionally salient unpleasant images compared to neutral images, 2) attenuation of the LPP following instructions to reappraise emotional response to images, and 3) blunted LPP amplitude overall for individuals with increased depressive symptoms. Adding to the existing literature, we have demonstrated that older adults selected for self-reported low and high emotion well-being (as measured by habitual ER use, resiliency, and depressive symptoms) are able to utilize reappraisal instructions to modulate emotional reactivity, and that according to self-reported ratings of negative emotion, those low on emotional well-being are less successful at doing so despite an apparent similar ability to modulate the LPP.

Taken together, the current results suggest the LPP is sensitive to emotional stimuli and ER instruction in older adults – although the LPP did not differentiate high versus low WB groups. On the other hand, older adults who reported low habitual ER use were worse at engaging in reappraisal according to self-reported ratings of negative emotion during the ER task when compared to older adults who reported high habitual ER use. Collectively, these data suggest that laboratory-based ER tasks might be used to understand abnormal ER use—though the LPP may not be a sensitive measure for investigating individual differences in ER ability. Further, the presence of depressive symptoms may complicate findings and should be considered when investigating ER using the LPP.

Highlights.

• Older adults high & low on well-being can use reappraisal to reduce LPP reactivity

• Those low on well-being are less successful at reappraisal based on self-report

• Older adults with increased depressive symptoms demonstrate blunted LPP overall