Interpersonal Emotion Regulation in Dementia: Diagnostic Differences and Associations with Caregiver Mental Health

Abstract

Close relational partners play a critical role in shaping each other’s emotions during social interactions, a process known as interpersonal emotion regulation. We examined how different forms of dementia (i.e., Alzheimer’s disease [AD], behavioral variant frontotemporal dementia [bvFTD], primary progressive aphasia [PPA]) affect interpersonal emotion regulation. We assessed how caregivers perceived their partners’ prosocial regulatory efforts and tracked real-time changes in caregivers’ affect during a dyadic interaction. We also investigated how these processes relate to caregivers’ mental health. Informal caregivers (N = 62) reported on their own mental health symptoms (depression, anxiety) and how their care recipients, who had different forms of dementia, use prosocial interpersonal emotion regulation (i.e., efforts to make their partner feel better). Dyads then engaged in a 10-minute unrehearsed discussion about a conflict in their relationship. Following this, caregivers viewed a video recording of the interaction and used a rating dial to provide continuous ratings of their own affective valence (negative-neutral-positive) during the conversation. Results revealed that among care recipients with AD, PPA, and bvFTD, those with bvFTD showed the lowest use of prosocial interpersonal emotion regulation, and their caregivers experienced the greatest increases in negative affect during the interaction. Across diagnoses, greater increases in caregiver negativity during the interaction were associated with greater caregiver depression (but not anxiety). These findings provide new information about interpersonal emotion regulation in dementia and highlight how the emotional changes that occur during dyadic interactions are tied to caregiver depressive symptoms.

Social interactions have emotional consequences that affect mental health (Gross & Muñoz, 1995). During interactions, close relational partners—most notably romantic partners— play a critical role in shaping each other’s emotions, both positively and negatively (Bloch et al., 2014; Niven et al., 2009). This process of changing another person’s emotions through social interaction or using another person to regulate your emotions through social interaction is often referred to as interpersonal emotion regulation (Jurkiewicz et al., 2023; Tran et al., 2022; Zaki and Williams, 2013). Interpersonal emotion regulation is thought to involve deliberate efforts to modulate another person’s emotional state (Nozaki & Mikolajczak, 2020; Zaki & Williams, 2013). Engaging in prosocial forms of interpersonal emotion regulation, such as trying to make others feel good or alleviating their suffering, may be particularly important for promoting and sustaining the mental health of close relational partners. However, individuals differ in their tendencies to regulate others’ emotions (Niven et al., 2024; Tran et al., 2023), which may shape their partners’ affective valence during social interactions as well as their partners’ mental health.

Interpersonal emotion regulation may be especially critical for mental health in older adulthood. Theories of adult socioemotional development suggest that, with age, people increasingly rely on social relationships as a means for regulating emotions (Carstensen et al., 1999). With age, we winnow our social relationships, primarily maintaining our closest relationships, especially those that bring us the most positive and least negative emotional experiences (Carstensen et al., 1999; English & Carstensen, 2014; Fredrickson & Carstensen, 1990). This shift may contribute to better mental health. Consistent with this, there are lower rates of clinical disorders such as Major Depressive Disorder and Generalized Anxiety Disorder in older adults compared to younger populations (Collier Villaume et al., 2023; Rook & Charles, 2017). However, older adulthood is also characterized by higher levels of subclinical mental health symptoms of depression and anxiety compared to younger adulthood (Geiselmann et al., 2001; Kvaal et al., 2008; Meeks et al., 2011).

Older age is also associated with higher rates of neurodegenerative diseases, such as Alzheimer’s disease (AD) and frontotemporal dementias (FTD), which ultimately cause dementia (Hou et al., 2019). AD, the most common form of dementia, is primarily characterized by cognitive impairment in the domain of episodic memory (Boxer et al., 2017). In contrast, frontotemporal dementia, the second most common form of dementia in individuals under the age of 65, encompasses three clinical subtypes including the behavioral variant of frontotemporal dementia (bvFTD) and two forms of primary progressive aphasias (PPA). BvFTD is marked by socioemotional and behavioral changes, including apathy, disinhibition, and lack of empathy (O’Connor et al., 2017), while the PPA syndromes primarily affect language and speech and are characterized by deficits in semantic knowledge or speech production (Bang et al., 2015; Gorno-Tempini et al., 2011).

Dementia is difficult not only for those with symptoms (care recipients) but also for close relational partners who provide care (caregivers). Although caring for a loved one with dementia can be a highly meaningful experience (Lee & Li, 2022), many studies have documented the heightened risk for depression and anxiety among dementia caregivers (for reviews, see: Pinquart and Sörensen, 2003; Sallim et al., 2015). Caregivers of individuals with bvFTD are particularly vulnerable to mental health difficulties (Wong et al., 2012), likely reflecting the challenges associated with socioemotional decline and behavioral changes occurring in a close relational partner (Givens et al., 2014; Hughes et al., 2014; Pinquart & Sörensen, 2003b). Research has examined socioemotional functioning in people with dementia, such as their emotional reactivity (i.e., the ability to generate emotional responses; Brown et al., 2020; Chen et al., 2017; Hua et al., 2020; Sturm et al., 2015), emotion recognition (i.e., the ability to identify emotions in others; Brown et al., 2018; Brown et al., 2020; Oliver et al., 2014), and intrapersonal emotion regulation (i.e., the ability to control one’s own emotions; Muhtadie et al., 2021; Otero and Levenson, 2017). However, interpersonal emotion regulation remains largely unexplored. Given the profound socioemotional changes in bvFTD, individuals with this condition may exhibit impairments in interpersonal emotion regulation, which, in turn, may lead to greater negative affect and greater symptoms of depression and anxiety in their caregivers.

The Current Study

The current study is the first to our knowledge to examine interpersonal emotion regulation in care recipients with different kinds of dementia (bvFTD, PPA, AD) and their informal caregivers, and to investigate how these regulatory processes and real-time changes in caregiver affect relate to caregiver mental health. In this study, we assess interpersonal emotion regulation as caregivers’ reports of care recipients’ prosocial interpersonal emotion regulation tendencies (i.e., care recipients’ extrinsic efforts to make others feel better). Additionally, we utilize a dyadic behavioral measure that assesses changes in caregivers’ affective valence during an unrehearsed conversation on a topic of disagreement with their care recipient. Although each of these measures has its strengths and weaknesses, we believe that together they provide a good starting point for examining diagnostic differences in both perceived interpersonal emotion regulation and the modulation of caregivers’ affect, as well as how each of these processes relates to caregiver depression and anxiety. By examining care recipients’ interpersonal emotion regulation tendencies and caregivers’ changing affective valence, we can better understand how the erosion of socioemotional functioning in diseases like bvFTD serves as a key pathway through which diagnostic deficits impact caregiver mental health.

We tested two main hypotheses. First, we hypothesized that individuals with bvFTD (compared to individuals with AD and PPA) would evidence impairments in interpersonal emotion regulation as indicated by less prosocial emotion regulation and more negative modulation of caregivers’ affective valence during a conflict conversation (i.e., the extent to which caregivers’ affect ratings become more negative at the end of the conversation compared to the beginning of the conversation). Second, we hypothesized that greater impairments in interpersonal emotion regulation and more negative modulation of caregivers’ affective valence during a conflict conversation would be associated with worse caregiver mental health (i.e., greater symptoms of depression and anxiety).

Method

Participants

Care recipients and their informal (i.e., unpaid) caregivers (N = 62 dyads, 94% spouses) were recruited through the Memory and Aging Center at the University of California, San Francisco (UCSF). Care recipients were evaluated by a multidisciplinary team at UCSF and completed a clinical interview, neurological examination, neuropsychological assessment, and neuroimaging. They received a consensus diagnosis based on established research criteria (Gorno-Tempini et al., 2011; McKhann et al., 2011; Rascovsky et al., 2011). The 62 care recipients included: (a) 22 with bvFTD; (b) 20 with AD; and (c) 20 with PPA.¹ For participants’ demographic and clinical characteristics, see Table 1.

Table 1.

Descriptive statistics for demographic characteristics, questionnaires, and rating dial measures for caregivers and care recipients.

	Caregivers
	Total	bvFTD	AD	PPA
% Female	60%	64%	60%	55%
% European American	74%	77%	80%	65%
% Spouses of Care Recipient	94%	91%	95%	95%
Age	64.29 (8.59)	63.32 (6.68)	64 (9.21)	65.50 (9.80)
Caregiver Depressive Symptoms (CESD)	0.60 (0.35)	0.80 (0.39)	0.50 (0.26)	0.53 (0.33)
Caregiver Anxiety (BAI)	0.30 (0.28)	0.29 (0.20)	0.27 (0.26)	0.35 (0.37)
Initial Affective Valence	4.87 (0.67)	4.57 (0.73)	5.04 (0.65)	5.05 (0.48)
Final Affective Valence	4.59 (1.55)	3.85 (1.57)	4.67 (1.48)	5.32 (1.25)
Modulation of Affective Valence	−0.27 (1.31)	−0.72 (1.40)	−0.35 (1.14)	0.33 (1.16)
	Care Recipients
	Total	bvFTD	AD	PPA
% Female	40%	32%	45%	45%
% European American	69%	68%	80%	60%
Age	65.87 (8.14)	64.95 (7.63)	65.75 (8.72)	67.00 (8.37)
Cognitive Impairment (MMSE)	25.00 (3.97)	26.14 (3.11)	22.94 (4.68)	25.60 (3.56)
Prosocial Interpersonal Emotion Regulation	2.35 (1.33)	1.67 (1.38)	2.50 (1.30)	2.87 (1.07)

Note. On average, caregivers are 1.40 years younger (SD = 5.06) than care recipients, and approximately 6% of dyads are same sex dyads. See Supplemental Section 3 for additional details on relationship type. We conducted one-way ANOVAs to examine differences in all caregiver and care recipient characteristics reported in this table across diagnostic groups. The following variables differed significantly across diagnostic groups: caregiver CESD (F(2, 51) = 4.32, p = .019), caregiver initial (F(2, 55) = 3.89, p = .026) and final (F(2, 59) = 5.47, p = .007) affective valence, caregiver modulation of affective valence (F(2, 55) = 3.60, p = .034), care recipient MMSE (F(2, 57) = 3.90, p = .026), and care recipient prosocial interpersonal emotion regulation (F(2, 51) = 4.43, p = .017). Center for Epidemiological Studies Depression Scale (CESD). Beck Anxiety Inventory (BAI). Mini-Mental Status Examination (MMSE). Table values reflect “mean (standard deviation)” unless otherwise indicated.

Procedure

Caregivers completed an online questionnaire that included measures of the care recipient’s interpersonal emotion regulation tendencies as well as the caregiver’s own symptoms of anxiety and depression.

Caregivers and care recipients also participated in a comprehensive day-long laboratory assessment of emotional functioning at the Berkeley Psychophysiology Laboratory (Levenson, 2007). For the present study, we utilized data from an unrehearsed 10-minute conversation between the caregiver and the care recipient about an area of disagreement in their relationship. A topic of disagreement was used because emotions that arise during conflict are especially relevant for relationship satisfaction and mental health (Gottman et al., 1998; Gottman & Krokoff, 1989; Gottman & Levenson, 1992; Karney & Bradbury, 1997). Immediately following the conversation, caregivers watched the video recording of their conversation and provided continuous ratings of the valence of their own affect during the conversation using a rating dial device (Levenson & Gottman, 1983). The dial pointer moved over a 9-point scale ranging from “very bad” at the far left (shown with a schematic frowning face) to “very good” (shown with a schematic smiling face) at the far right, with “neutral” in the middle (shown with a schematic neutral face). Caregivers were instructed to adjust the dial position as often as necessary so that it always reflected the way they had felt during the conversation. The computer sampled the dial position every 3 milliseconds and averaged these readings into second-long time periods. This resulted in a second-by-second time series reflecting the caregiver’s affective valence throughout the conversation. This method for obtaining continuous self-reported affect is well-validated (Gottman & Levenson, 1985; Ruef & Levenson, 2007). All procedures were approved by the UC Berkeley Committee for the Protection of Human Subjects.

Measures

Interpersonal Emotional Processes

Care Recipients’ Prosocial Interpersonal Emotion Regulation.

Caregivers completed two items from the Caregiver Assessment of Socioemotional Functioning (CASEF; Ascher et al., 2007) that assessed the care recipient’s prosocial emotion regulation tendencies over the past month. The first item asked caregivers to rate the extent to which their care recipients “try to make other people feel good,” and the second item asked caregivers to rate the extent to which their care recipients “try to help others who are suffering.” Caregivers rated these two items on a scale ranging from 0 (not at all) to 4 (a lot). Items were averaged (α = .89), with higher scores indicating greater prosocial emotion regulation.

Interpersonal Modulation of Caregivers’ Affective Valence.

Using the second-by-second time series of the rating dial data, we derived the following variables: (a) Caregiver’s initial affective valence was determined by calculating the caregiver’s average rating dial score across the first 30 seconds of the conversation; (b) Caregiver’s final affective valence was determined by calculating the caregiver’s average rating dial score across the last 30 seconds of the conversation; and (c) modulation of caregivers’ affective valence was determined by subtracting the caregiver’s initial affective valence from their final affective valence.

Because the modulation of a caregiver’s affective valence could be quantified in a number of ways, we tested our primary hypotheses using several alternative quantitative approaches including: (a) z-scoring the affective valence time-series within subject (such that the initial and final affective valences would reflect the caregiver’s valence level relative to their valence across the entire conversation); (b) using the initial and final 60 seconds of the conversation (rather than 30 seconds); and (c) using the linear slope of the caregiver’s affect ratings across the entire conversation. All of these quantification methods yielded largely similar patterns of effect sizes (see Supplemental Material Section 1).

Caregiver Mental Health

Caregiver Depressive Symptoms.

Caregivers completed the 20-item version of the Center for Epidemiological Studies Depression Scale (CESD; Radloff, 1977). Caregivers rated their depressive symptoms (e.g., depressed mood, feelings of worthlessness, loss of appetite) on a scale ranging from 0 (rarely or none of the time) to 3 (most or all of the time) during the past month. Items were averaged, with higher scores indicating greater depressive symptoms (α = .84).

Caregiver Anxiety.

Caregivers completed the 21-item Beck Anxiety Inventory (BAI; Steer and Beck, 1997). Caregivers rated their cognitive and somatic symptoms of anxiety (e.g., heart pounding, difficulty breathing, fear of the worst) on a scale ranging from 0 (not at all) to 3 (severely) during the past month. Items were averaged, with higher scores indicating greater anxiety symptoms (α = .83).

Covariates

Caregivers’ age (in years) and gender were used as covariates along with care recipients’ cognitive impairment, given prior research linking these factors to caregiver mental health. To assess cognitive impairment, UCSF clinicians administered the Mini-Mental Status Examination to care recipients (MMSE; Folstein et al., 1975). The MMSE assesses several domains of cognitive functioning (e.g., recall, orientation, language). Scores were summed and ranged from 0–30, with lower scores indicating greater cognitive impairment.

Statistical Analyses

Preliminary Analyses

In addition to descriptive statistics for key measures, we examined whether care recipients who employ less prosocial interpersonal emotion regulation have caregivers with greater increases in negative affect across their conversation by regressing caregivers’ emotion modulation onto care recipients’ prosocial interpersonal emotion regulation tendencies.

Diagnostic Differences

To test our first hypothesis—that individuals with bvFTD (compared to individuals with AD and PPA) would evidence greater impairments in interpersonal emotion regulation as indicated by less prosocial interpersonal emotion regulation and more negative modulation of caregivers’ affective valence during a conflict conversation (i.e., greater increases in negative affect from the beginning to the end of the conversation)—we conducted a series of analyses.

First, we examined whether care recipients’ prosocial interpersonal emotion regulation tendencies differed across diagnostic groups. We conducted a one-way Analysis of Variance (ANOVA) with diagnosis (bvFTD, AD, PPA) as the independent variable and prosocial interpersonal emotion regulation as the dependent variable.

Next, we examined whether the modulation of caregivers’ affective valence differed across diagnostic groups using a repeated-measures ANOVA, with time (caregivers’ initial vs. final affective valence) as a within-subject factor and diagnosis (bvFTD, AD, PPA) as a between-subject factor to test for an interaction effect.

For these analyses, Bonferroni-corrected post hoc comparisons were used, and we re-ran each model with covariates (caregivers’ age and gender, care recipients’ cognitive impairment) to assess whether observed differences remained consistent after accounting for these factors. Additionally, we conducted sensitivity analyses using alternative quantification methods for modulation of caregivers’ affective valence, including (a) z-scoring affective valence within participants, (b) using the first and last 60 seconds of the conversation instead of the first and last 30 seconds, and (c) calculating the linear slope of caregivers’ affect ratings across the conversation. The overall pattern of results remained consistent across these approaches (see Supplemental Material section 1).

Associations with Caregivers’ Mental Health

To test our second hypothesis—that impairments in interpersonal emotion regulation and more negative modulation of caregivers’ affective valence during the conflict conversation would be associated with greater caregiver mental health symptoms (depression and anxiety)—we conducted a series of linear regression analyses.

First, we examined whether care recipients’ prosocial interpersonal emotion regulation was associated with caregivers’ depressive symptoms and anxiety. Prosocial interpersonal emotion regulation was entered as a predictor, with caregivers’ depressive symptoms as the dependent variable in one model and caregivers’ anxiety symptoms as the dependent variable in another model.

Next, we examined whether the modulation of caregivers’ affective valence was associated with caregivers’ mental health symptoms. The modulation of caregivers’ affective valence (difference score determined by subtracting caregivers’ initial affective valence from their final affective valence) was used as a predictor, with caregivers’ depressive symptoms as the dependent variable in one model and caregivers’ anxiety symptoms as the dependent variable in another model. We then added caregivers’ initial affective valence as a covariate for each model.

We re-ran each of these regression models including covariates (caregivers’ age and gender, care recipients’ cognitive impairment and diagnosis) to assess whether the observed associations, or lack thereof, remained consistent after accounting for these factors.

Transparency and openness

Sample size was determined by the number of participants in each diagnostic category that could be recruited from 2013 to 2020. Post-hoc power analyses were conducted for the present sample (N = 62) using G*Power with a medium effect size of .25 and an alpha of .05.

For Hypothesis 1, to detect diagnostic differences in care recipients’ prosocial interpersonal emotion regulation using a one-way ANOVA with three groups, we are underpowered (40%). However, for detecting diagnostic differences in care recipients’ interpersonal emotion modulation using a repeated-measures ANOVA, with a within-between interaction effect, three groups with two measurements, and a correlation of .53 among measures, there is adequate power (95%).

For Hypothesis 2, to detect associations between interpersonal emotion regulation and caregiver mental health using linear regression analyses with five predictor variables, there is adequate power (84%).

De-identified data used for the current analyses, along with syntax for statistical analyses can be found on the Open Science Framework (Brown, 2025).

Results

Preliminary Analyses

Table 1 displays descriptive statistics for measures of caregivers’ affect ratings during the conversation. Caregivers’ initial affective valence was just below the neutral midpoint of 5.0 (M = 4.87, SD = 0.67) and modulation of caregivers’ affective valence from the beginning to the end of the conversation was negative on average (M = −0.27, SD = 1.31), which is consistent with conversations about relationship conflicts becoming more negative over time (Gottman & Levenson, 1992). Table 2 presents correlations among key study variables.

Table 2.

Intercorrelations between demographic characteristics, questionnaires, and rating dial measures for caregivers and care recipients.

	1	2	3	4	5	6	7
1. Caregiver Initial Affective Valence	1.00
2. Caregiver Final Affective Valence	0.53***	1.00
3. Modulation of Caregiver Affective Valence	0.11	0.90***	1.00
4. Care Recipient Cognitive Impairment (MMSE)	−0.18	−0.10	−0.01	1.00
5. Care Recipient Prosocial Interpersonal Emotion Regulation	0.31*	0.42**	0.33*	−0.13	1.00
6. Caregiver Depressive Symptoms (CESD)	−0.30*	−0.37**	−0.34*	0.14	−0.21	1.00
7. Caregiver Anxiety (BAI)	−0.14	0.05	0.00	−0.12	0.10	0.45***	1.00

Note.

^*p < .05;

^**p < .01;

^***p < .001

Caregivers of care recipients with greater prosocial interpersonal regulation tendencies experienced more positive (and less negative) affective valence from the beginning to the end of the conversation (β = 0.33, t(48) = 2.40, p = .021). This association remained significant even after accounting for caregivers’ initial affective valence in the conversation (β = 0.32, t(47) = 2.23, p = .030). However, although the size and direction of the effect remained similar after controlling for caregiver age and gender, as well as care recipient diagnosis and cognitive impairment, the effect was no longer significant (β = 0.28, t(43) = 1.81, p = .078).

Diagnostic Differences

Care Recipients’ Prosocial Interpersonal Emotion Regulation

Care recipients’ prosocial interpersonal emotion regulation differed significantly as a function of their diagnosis, F(2, 51) = 4.43, p = .017, see Figure 1, Panel A. Bonferroni-corrected post hoc comparisons indicated that care recipients with bvFTD (M = 1.67, SD = 1.38) had significantly less prosocial emotion regulation tendencies (M_diff = −1.20, SE = .41, p = .016) compared to care recipients with PPA (M = 2.87, SD = 1.07). There was no significant difference between AD care recipients (M = 2.50, SD = 1.30) and PPA care recipients (M_diff = −0.37, SE = .42, p = 1.00) or AD care recipients and bvFTD care recipients (M_diff = −0.83, SE = .42, p = .165). Diagnostic differences in care recipients’ prosocial interpersonal emotion regulation remained significant even after controlling for covariates, including caregiver age and gender, as well as care recipients’ cognitive impairment, F(2, 47) = 3.99, p = .025.

Figure 1.

Diagnostic differences in care recipients’ prosocial interpersonal emotion regulation tendencies (left panel) and modulation of caregivers’ affective valence (right panel).

Modulation of Caregivers’ Affective Valence

Modulation of caregivers’ affective valence differed significantly as a function of their care recipient’s diagnosis. In a repeated-measures ANOVA, we found a significant interaction effect between time (initial and final affective valence) and diagnosis (bvFTD, AD, PPA), F(2, 55) = 3.60, p = .034, partial η2 = 0.12, see Figure 1, Panel B. BvFTD caregivers reported more negative (and less positive) affective valence from the beginning to the end of the conversation (M_diff = −0.72, SE = 0.27, p = .009). In contrast, PPA caregivers reported relatively less negative (and more positive) affective valence across the conversation, though this change was not statistically significant (M_diff = 0.33, SE = 0.29, p = .261). AD caregivers experienced relatively more negative (and less positive) affective valence, and this change was also not significant (M_diff = −0.35, SE = .30, p = .251). Bonferroni-corrected post hoc comparisons revealed that initial affective valence did not differ significantly between any groups, including bvFTD and AD caregivers (M_diff = −0.47, SE = 0.21, p = .074), bvFTD and PPA caregivers (M_diff = −0.49, SE = 0.20, p = .054), or AD and PPA caregivers (M_diff = −0.01, SE = 0.21, p = 1.000). However, the final affective valence of bvFTD caregivers was significantly more negative (and less positive) than that of PPA caregivers (M_diff = −1.53, SE = 0.44, p = .003). There were no significant differences between bvFTD and AD caregivers (M_diff = −0.84, SE = 0.46, p = .211), or AD and PPA caregivers (M_diff = −0.69, SE = 0.47, p = .446) in final affective valence.

The interaction effect between time and diagnosis remained significant after covarying caregiver age and gender, as well as care recipient cognitive impairment (F(2, 48) = 3.33, p = .044, partial η2 = 0.12, see Supplemental Material 2 for Bonferroni-corrected post hoc comparisons). We also conducted analyses using alternative quantification methods for modulation of caregiver’s affective valence, and the pattern of findings using these various methods are similar to our main analyses, although we note that results using 60 second window led to the diagnosis X affect interaction becoming non-significant according to conventional standards (p = .078; see Supplemental Material 1).

Associations with Caregivers’ Mental Health

Care Recipients’ Prosocial Interpersonal Emotion Regulation

Care recipients’ prosocial interpersonal emotion regulation was not associated with their caregivers’ symptoms of depression (β = −0.21, t(49) = −1.47, p = .149) or anxiety (β = 0.10, t(48) = 0.69, p = .493). Effects for depression (β = −0.02, t(43) = −0.14, p = .892) and for anxiety (β = 0.19, t(42) = 1.32, p = .193) remained non-significant after accounting for covariates. See Table 3 for full regression model statistics.

Table 3.

Linear regressions examining associations between care recipients’ prosocial interpersonal emotion regulation tendencies and caregivers’ mental health.

Caregiver Depressive Symptoms (CESD)	b (SE)	95% C.I.	p
Caregiver Age	−0.11 (0.01)	[−0.02, 0.01]	0.439
Caregiver Sex	0.25 (0.10)	[−0.03, 0.38]	0.086
Care Recipient Cognitive Impairment (MMSE)	0.17 (0.01)	[−0.01, 0.04]	0.268
Care Recipient AD Diagnosis	−0.38 (0.12)	[−0.53, −0.04]	0.025*
Care Recipient PPA Diagnosis	−0.34 (0.12)	[−0.49, −0.004]	0.046*
Care Recipient Prosocial Interpersonal Emotion Regulation	−0.02 (0.04)	[−0.08, 0.07]	0.892
Caregiver Anxiety (BAI)	b (SE)	95% C.I.	P
Caregiver Age	−0.28 (0.004)	[−0.02, 0.00]	0.048*
Caregiver Sex	0.40 (0.08)	[0.07, 0.38]	0.005**
Care Recipient Cognitive Impairment (MMSE)	−0.03 (0.01)	[−0.02, 0.02]	0.824
Care Recipient AD Diagnosis	−0.20 (0.09)	[−0.31, 0.07]	0.211
Care Recipient PPA Diagnosis	0.05 (0.09)	[−0.16, 0.22]	0.738
Care Recipient Prosocial Interpersonal Emotion Regulation	0.19 (0.03)	[−0.02, 0.10]	0.193

Note.

^*p < .05;

^**p < .01;

^***p< .001; BvFTD serves as the reference group for diagnosis

Modulation of Caregivers’ Affective valence

Caregivers who reported more negative (and less positive) affective valence from the beginning to the end of the conversation had greater depressive symptoms (β = −0.34, t(48) = −2.51, p = .015), even after accounting for caregivers’ initial affective valence as well as covariates (β = −0.33, t(42) = −2.62, p = .012). Modulation of caregivers’ affective valence was not associated with caregivers’ anxiety symptoms (β = 0.002, t(47) = 0.01, p = .989), even after accounting for caregivers’ initial affective valence as well as covariates (β = −0.11, t(41) = −0.80, p = .431). See Table 4 for full regression model statistics.

Table 4.

Linear regressions examining associations between the modulation of caregivers’ affective valence during an interaction with their care recipient and caregivers’ mental health.

Caregiver Depressive Symptoms (CESD)	b (SE)	95% C.I.	P
Caregiver Age	−0.13 (0.01)	[−0.02, 0.01]	0.328
Caregiver Sex	0.35 (0.09)	[0.07, 0.43]	0.007**
Care Recipient Cognitive Impairment (MMSE)	0.19 (0.01)	[−0.01, 0.04]	0.190
Care Recipient AD Diagnosis	−0.26 (0.11)	[−0.42, 0.04]	0.096
Care Recipient PPA Diagnosis	−0.15 (0.11)	[−0.33, 0.12]	0.339
Caregiver Initial Affective Valence	−0.19 (0.07)	[−0.26, 0.04]	0.141
Modulation of Caregiver Affective Valence	−0.33 (0.03)	[−0.16, −0.02]	0.012*
Caregiver Anxiety (BAI)	b (SE)	95% C.I.	p
Caregiver Age	−0.26 (0.00)	[−0.02, 0.00]	0.086
Caregiver Sex	0.40 (0.07)	[0.07, 0.37]	0.005**
Care Recipient Cognitive Impairment (MMSE)	−0.01 (0.01)	[−0.02, 0.02]	0.930
Care Recipient AD Diagnosis	−0.01 (0.10)	[−0.20, 0.19]	0.942
Care Recipient PPA Diagnosis	0.24 (0.10)	[−0.06, 0.33]	0.167
Caregiver Initial Affective Valence	−0.15 (0.06)	[−0.20, 0.06]	0.287
Modulation of Caregiver Affective Valence	−0.11 (0.03)	[−0.08, 0.04]	0.431

Note.

^*p < .05;

^**p < .01;

^***p< .001; BvFTD serves as the reference group for diagnosis

Discussion

We examined interpersonal emotion regulation in dyads affected by three kinds of dementia. We specifically focused on: (a) how dementia type (AD, bvFTD, PPA) affects both interpersonal emotion regulatory tendencies of care recipients and the modulation of caregiver affective valence during social interaction, and (b) how these interpersonal processes are associated with caregivers’ mental health (depression, anxiety). In terms of diagnostic differences, results revealed that individuals with bvFTD engaged in less prosocial interpersonal emotion regulation, and their caregivers’ affective valence was modulated more negatively across the course of a 10-minute conversation about a relationship conflict, even after accounting for care recipients’ level of cognitive impairment. We also found evidence that more negative modulation of caregivers’ affective valence during the conversation was associated with more depressive symptoms in caregivers, but not more anxiety symptoms. Caregivers’ reports of care recipients’ prosocial interpersonal emotion regulation were not associated with caregivers’ mental health.

Diagnostic Differences in Interpersonal Regulation and Modulation of Caregiver Affect

Our finding that caregivers of individuals with bvFTD perceive their partners as less likely to try and make others feel good and reduce others’ suffering suggests that less prosocial interpersonal emotion regulation is characteristic of bvFTD. There are a number of disease related factors that may contribute to caregivers’ beliefs regarding their care recipients’ intentional efforts to regulate others’ emotions. First, individuals with bvFTD typically exhibit higher levels of dysexecutive and neuropsychiatric symptoms compared to other diagnostic groups, which could make their caregivers believe they put less effort into influencing others’ emotions positively (Lanata & Miller, 2016). Individuals with bvFTD often have atypical emotional reactions and empathy deficits that could make them less responsive to others’ emotional needs (Brown et al., 2020). Further, individuals with bvFTD engage in less prosocial behavior towards strangers (Sturm et al., 2018), and this lack of prosociality could manifest as a lack of effort to regulate others’ emotions. Future research should explore the specific socioemotional changes (e.g., changes in emotion recognition or reactivity) in bvFTD that lead to the differences in interpersonal emotion regulation observed in the current study.

The pattern of diagnostic differences in caregivers’ perceptions of care recipients’ interpersonal emotion regulation was mirrored in caregivers’ affective valence during a conversational interaction with their care recipients. Caregivers of individuals with bvFTD felt more negative by the end of the conversation compared to PPA caregivers, who felt relatively more positive by the end of their conversation. The fact that any diagnostic group had caregivers who tended to feel more positive by the end of the conversation is surprising, given the topic of the conversation was an area of disagreement. It is interesting to consider why caregivers of individuals with PPA might experience relatively more positive affect by the end of their conversation compared to bvFTD. One possibility is that caregivers of individuals with language disorders tend to avoid complex conversations with their care recipients to avoid language difficulties that can be challenging and uncomfortable. However, when given the time and space to converse with their partners with no other commitments or interruptions, conversations might end up being a quite pleasant experience, providing these caregivers an opportunity to connect in a meaningful way with their close loved one. Another possibility is that deficits in language disrupt care recipients’ ability to effectively communicate their opposing views, creating a more positive tone of conversation than would be expected during typical disagreements. Alternatively, research suggests enhanced positive emotional reactivity in PPA, which could make conversations more enjoyable for caregivers (Shdo et al., 2022; Sturm et al., 2015). Future research should investigate these and other mechanisms through which language deficits impact conversational interactions between relational partners.

In preliminary analyses, although we found that care recipients’ interpersonal emotion regulation was significantly correlated with the modulation of their caregivers’ affective valence from the beginning to the end of the conflict conversation, this association was not robust following the inclusion of covariates (e.g., cognitive impairment, diagnosis). These findings lend some limited support to theories that suggest interpersonal emotion regulation (intentionally regulating others’ emotions) is a related, but distinct construct from interpersonal emotion modulation (modulation of affect during interactions, which may occur without deliberate intentions; Zaki & Williams, 2013). This study is among the first to examine associations between perceptions of a partner’s interpersonal emotion regulatory tendencies and dynamic emotional changes that occur during social interactions with that partner (Jurkiewicz et al., 2023). Perceiver reports offer valuable insight into dyadic emotional functioning, but they likely reflect a blend of the care recipient’s actual behavior and the caregiver’s interpretation of available cues. Future research utilizing both perceiver reports and objective behavioral coding will be essential to better understand the alignment between a care recipient’s regulatory intent and their partner’s perception of those efforts. Moreover, disorder-specific factors (e.g., social inappropriateness in bvFTD) may account for the link between caregivers’ perception of their care recipients’ prosocial interpersonal emotion regulation and their own emotional changes during conversations. Future research should continue to examine the extent to which emotional changes during interactions might inform or correspond with perceptions of a partner’s regulatory tendencies.

Caregivers’ Mental Health

Importantly, our findings suggest that the modulation of caregivers’ affective valence during conflict is associated with caregivers’ depressive symptoms, even after accounting for caregivers’ age, gender, and their care recipients’ diagnosis and cognitive impairments. However, caregivers’ perception of their care recipients’ interpersonal emotion regulation tendencies was not associated with caregivers’ depressive symptoms. These findings suggest that perceptions of a partner’s emotion regulation may not be as relevant to depressive symptoms as the actual ways that a partner modulates emotions during interactions. Notably, emotional changes observed during conflict may reflect a broader set of interpersonal processes influenced by the care recipient’s diagnosis.

When interpreting these results, it is important to consider that individuals with depressive or anxious symptoms often exhibit a negativity bias, whereby they perceive neutral or ambiguous stimuli as more negative (Hamilton & Gotlib, 2008; Leppänen et al., 2004; Yoon & Zinbarg, 2008). Such biases could influence how caregivers rate their affective valence during a conflict discussion, as well as how they regulate and remember those emotions (Hayes et al., 2023). In the current study, we sought to minimize the influence of these baseline perceptual differences by including caregivers’ initial affective valence as a covariate in our analyses. By focusing on the shift in valence from the beginning to the end of the conversation, our findings specifically highlight the real-time modulation of affect rather than stable differences in emotional perception.

Our findings fit with theories that suggest interpersonal emotional processes are important for mental health as we age (Rook & Charles, 2017). Within a biopsychosocial framework, the frequent experience of modulating toward higher levels of negative affect during dyadic interactions may function as a repeated social stressor. Extensive research has demonstrated that such chronic interpersonal strain can lead to significant psychological distress and detrimental physical health outcomes, including impaired immune function, including in the context of caregiving (Christian et al., 2023; Fagundes et al., 2011; Kiecolt-Glaser et al., 1991; 2009). Findings may also reflect the importance of having a partner who can regulate our distress (Brown et al., 2017) and the effects of dyadic processes on caregiver mental health (Connelly et al., 2020).

Interestingly, although anxiety and depressive symptoms were significantly correlated in this sample, more negative modulation of caregivers’ affective valence during an interaction with their care recipient was associated with greater caregivers’ depressive symptoms but was not associated with caregivers’ anxiety symptoms. This joins a body of literature that suggests interpersonal processes may play a bigger role in depressive than anxiety symptoms (Brown et al., 2023; Joiner, 1994). Loneliness plays an important role in depression, and the loss of interpersonal emotional regulation from a close relational partner may contribute to loneliness that promotes depression in older adulthood (Cacioppo et al., 2006; Erzen & Çikrikci, 2018). Collectively, findings highlight the added value of affective science laboratory methods for continuous emotional observation. Utilizing such methods allows for deeper understanding of the mental health implications of emotional processes that may be missed by self-report questionnaires alone.

Strengths and Limitations

The current study had a number of strengths. First, we included care recipients with different kinds of dementia whose diagnoses had been confirmed carefully using clinical interviews, neuroimaging, neuropsychological batteries, and neurological exams. Second, we measured two factors potentially related to care recipients’ interpersonal emotion regulation, one based on informants’ evaluation of the care recipients and the other on self-reported affect during an actual interpersonal interaction. Third, we also examined associations between interpersonal emotion regulation and two well-validated measures of mental health. The study also had several limitations. First, affect ratings were made using a bipolar rating dial that includes positive and negative affect on a single continuum. Although this approach matches the way affective valence is characterized in circumplex models (Posner et al., 2005), it does not allow for assessing mixed affective states (e.g., increases in both positive and negative affect) or the role of discrete emotions (e.g., experienced sadness and fear). Second, our sample was relatively small for between group comparisons and did not allow us to consider separately individuals with the non-fluent versus semantic variants of PPA. Semantic and fluency problems may create different challenges for caregivers. Although power analyses suggest the sample size was adequate for regression analyses, power analyses were retrospective in nature, which has limitations (Hoenig et al., 2001). Third, the sample was predominately White and well-educated, and thus findings may not generalize to more diverse samples collected using alternative recruitment methods or containing alternative relationship types (e.g., adult-child, paid caregiver). Fourth, we did not systematically code the specific content, topic, themes, or intensity of the conflict discussions. The observed diagnostic differences and associations with caregiver mental health emerged despite the natural variation in conversation topics and intensity. Future research should investigate whether specific conflict themes vary with diagnosis to influence the dynamics of interpersonal emotion regulation and reveal different aspects of the associations between interpersonal emotion regulation and caregiver health. Finally, our research design was cross-sectional and we cannot draw any conclusions regarding directions of influence (e.g., the extent to which caregiver mental health influences care recipient and dyadic interpersonal emotion regulation or vice versa).

Conclusion

Close relational partners are typically motivated to influence each other’s emotions in ways that promote mental health. However, these interpersonal emotional regulatory processes can be impacted by dementia. Care recipients with bvFTD were seen as engaging in less prosocial interpersonal emotion regulation and their caregivers were more likely to experience increases in negative affect from the beginning to the end of a conflictual conversation compared to those with PPA. Importantly, even when accounting for diagnoses, interactions that led to more negative modulation of caregivers’ affective valence were associated with greater depressive symptoms in caregivers. Findings underscore the importance of characterizing both informant-reported interpersonal emotion regulation and the changes in caregivers’ affect across different types of dementia. Notably, our results suggest that the modulation of caregivers’ affect during social interactions—rather than perceptions of care recipients’ regulatory intent—is more closely linked to caregiver depressive symptoms, highlighting it as a possible target for interventions aimed at mitigating caregiver distress.