RESEARCH ARTICLE
Disentangling cognition and emotion in older adults: the role of cognitive control and mental health in emotional conflict adaptation Nathan C. Hantke1,2,3, Anett Gyurak1,2, Katie Van Moorleghem4, Jill D. Waring1,2,5, Maheen M. Adamson1,3, Ruth O’Hara1,2,6,† and Sherry A. Beaudreau1,2,6,† 1
Department of Psychiatry and Behavioral Sciences, Stanford University School of Medicine, Stanford, CA, USA Sierra Pacific Mental Illness Research Education and Clinical Center (MIRECC), Palo Alto VA Health Care System, Palo Alto, CA, USA 3 War Related Illness and Injury Study Center (WRIISC), VA Palo Alto Health Care System, Palo Alto, CA, USA 4 Department of Psychology, Palo Alto University, Palo Alto, CA, USA 5 Department of Psychology, Saint Louis University, Saint Louis, MO, USA 6 University of Queensland, Brisbane, Australia Correspondence to: Sherry A. Beaudreau, PhD, ABPP and Ruth O’Hara, PhD, E-mail:
[email protected];
[email protected] 2
†
S. B. and R. O H. contributed equally to this work.
Objective: Recent research suggests cognition has a bidirectional relationship with emotional processing in older adults, yet the relationship is still poorly understood. We aimed to examine a potential relationship between late-life cognitive function, mental health symptoms, and emotional conflict adaptation. We hypothesized that worse cognitive control abilities would be associated with poorer emotional conflict adaptation. We further hypothesized that a higher severity of mental health symptoms would be associated with poorer emotional conflict adaptation. Methods: Participants included 83 cognitively normal community-dwelling older adults who completed a targeted mental health and cognitive battery, and emotion and gender conflict-adaptation tasks. Results: Consistent with our hypothesis, poorer performance on components of cognitive control, specifically attention and working memory, was associated with poorer emotional conflict adaptation. This association with attention and working memory was not observed in the non-affective-based gender conflict adaptation task. Mental health symptoms did not predict emotional conflict adaptation, nor did performance on other cognitive measures. Conclusion: Our findings suggest that emotion conflict adaptation is disrupted in older individuals who have poorer attention and working memory. Components of cognitive control may therefore be an important potential source of inter-individual differences in late-life emotion regulation and cognitive affective deficits. Copyright # 2016 John Wiley & Sons, Ltd. Key words: emotion regulation; older adults; cognition; cognitive control; working memory; emotion conflict adaptation History: Received 06 November 2015; Revised 31 May 2016; Accepted 02 June 2016; Published online 22 July 2016 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/gps.4535
Introduction An individual’s ability to automatically regulate emotions (i.e. implicit emotion regulation) is utilized repeatedly over the course of daily activities and is involved in determining the duration, intensity, and type of emotion experienced (Gross & Thompson, Copyright # 2016 John Wiley & Sons, Ltd.
2007; Gyurak, Gross, & Etkin, 2011). Processing emotionally conflicting information, where there are incongruent stimuli characteristics, measures implicit emotion regulation and is dependent upon a dynamic relationship between both cognitive and emotion regulation abilities (Gyurak et al., 2011; Pessoa, 2008). The emotional conflict task is one method used to Int J Geriatr Psychiatry 2017; 32: 840–848
Emotion conflict adaptation in older adults
tap in to neural circuitry important in implicit emotion regulation, highlighting a regulatory interplay between the anterior cingulate cortex, prefrontal cortex, and the limbic system (Etkin et al., 2006; Gyurak et al., 2011). Previous studies using the emotional conflict task have consistently demonstrated an adaptation effect (faster reaction time) on repeat trials of incongruent stimuli (e.g. a happy facial expression overlaid with the word “fear”), suggesting that on conflict trials, regulatory mechanisms reduce emotional conflict, possibly through cognitive control (Egner et al., 2008; Etkin et al., 2006; Gratton et al., 1992). The validity of the task has been supported by prior studies with patient samples showing reduced emotional conflict adaptation in younger and middle-aged individuals with anxiety and depressive disorders and in adults with a history of childhood trauma as compared with asymptomatic controls, suggesting less emotional conflict adaptation is associated with deficits in implicit emotion regulation (Etkin et al., 2010; Etkin and Schatzberg, 2011; Gyurak et al., 2011; Marusak et al., 2015). Few studies have examined emotional conflict adaptation in older adults despite growing evidence that disordered emotional processing is associated with cognitive deficits in older adults (Beaudreau and O’Hara, 2008; Butters et al., 2008). One of the only studies to examine emotional conflict adaptation in an aging sample found similar task performance between younger and older adults during the emotional conflict task, yet older adults showed less adaptation during a non-affective-based gender conflict task compared with young adults (Monti et al., 2010). This finding suggests emotional conflict adaptation is selectively preserved in the healthy aging process, given the observed age-related cognitive decrements on the gender conflict task. However, this study did not directly measure cognition, or the impact of cognition on emotional processing. The relationship between cognition and emotional conflict adaptation is poorly understood. The process of regulating emotions requires cognitive control abilities that often decline in normal aging due to ageassociated changes in brain function, particularly those functions associated with the prefrontal cortex, such as attention, working memory, and inhibition (Drag and Bieliauskas, 2010; Egner and Hirsch, 2005; ElderkinThompson et al., 2008; Etkin et al., 2013). Attention, working memory, and inhibition are understood to fall within the broad construct of cognitive control, the mechanism thought to be responsible for goal-relevant behavior (Miller, 2000). Recent research suggests cognitive control has a bidirectional relationship with Copyright # 2016 John Wiley & Sons, Ltd.
841
emotional processing (Beaudreau et al., 2013), but to our knowledge, prior studies have not examined either the influence of cognitive control or mental health symptoms, on emotional conflict adaptation in older adults. Thus, the goal of the current investigation was to examine if emotional conflict adaptation is related to aspects of cognitive function and self-reported mental health symptom severity in a sample of communitydwelling older adults, using an emotional conflict task and a non-emotional gender task (Egner et al., 2008; Etkin et al., 2010; Monti et al., 2010). We hypothesized that worse cognitive control abilities would be associated with poorer emotional conflict adaptation. We measured three components of cognitive control: attention, working memory, and inhibition (Miller, 2000; Miyake et al., 2000). We further hypothesized that poorer emotional conflict adaptation would be associated with greater self-reported severity of anxiety, depressive, and worry symptoms, which has not been the subject of previous investigations. In secondary analyses, adjusted for multiple testing, we also examined if the presence of a mental health diagnosis in older adults would be associated with poorer emotional conflict adaptation, as has been shown in studies with younger adults but not older adults (Etkin et al., 2010; Etkin and Schatzberg, 2011; Marusak et al., 2015). Methods Participants
Participants were recruited as part of a larger study exploring the relationship of cognition and mental health in older adults. Participants were excluded if they had a previous diagnosis of dementia, scored outside normal limits on a cognitive screen for dementia (Katzman et al., 1983), or endorsed current psychotic symptoms or psychotic disorders. Enrollees received $50 compensation for their baseline participation. Informed consent was obtained consistent with the Declaration of Helsinki and institutional guidelines established by the VA Palo Alto Human Subjects Review Committee. The study was approved by both the VA Palo Alto and Stanford School of Medicine Institutional Review Boards. Of the 121 English speaking community-dwelling older adults enrolled in the larger study, 91 participants were asked to complete the emotional conflict task and a gender conflict task. Participants who performed at or below chance on the practice trials Int J Geriatr Psychiatry 2017; 32: 840–848
N. C. Hantke et al.
842
(n = 1) or on either tasks (n = 6) were excluded from the study. One participant was excluded from the study because of the presence of psychotic symptoms not detected until the day of the assessment, leaving the final sample at 83. Participant characteristics are further described in Table 1. Procedure
Participants completed questionnaires about their medical history, demographic characteristics, and mental health on the day of the evaluation. Trained masters or doctoral level examiners administered structured clinical interviews and then administered the cognitive battery. Lastly, participants completed the computerized emotional and gender conflict tasks.
Emotional conflict task
The emotional conflict task (Etkin et al., 2006; Etkin et al., 2010; Monti et al., 2010) included 148 black and white photographs of happy or fearful facial expressions. All faces were overlaid with either the word “FEAR” or “HAPPY” in red letters, which resulted in facial expressions congruent with or incongruent with the word. Participants were instructed to press corresponding response buttons as fast and accurately as possible to indicate the facial affect (right index finger for “fear” or right middle finger for “happy”) and to ignore the overlaid emotional distractor word. Stimuli were presented for 1000 ms in a pseudorandom order using the EPRIME software package (Psychology Software Tools; http://www.pstnet.com/eprime.cfm), with a variable inter-trial interval of 4000 ms. Participants
Table 1 Demographic characteristics and conflict task performance for sample group Sample characteristics (n = 83) Age Gender (frequency) Years of education Presence of current Axis I diagnosis Presence of current anxiety disorder diagnosis Psychiatric symptom questionnaires BDI-II HAM-D total BAI PSWQ SIGH-A total Cognitive measures WAIS-IV Digit Span total Forward total Backward total WAIS-IV Coding total correct D-KEFS CWIT Color Naming Time D-KEFS CWIT Word Reading Time D-KEFS CWIT Inhibition Completion Time D-KEFS CWIT Inhibition/Switching Completion Time RAVLT trial 1 list recall RAVLT sum of trials 1–5 RAVLT delayed recall Emotional conflict task Emotional adaptation effect (ms) Task performance accuracy Overall reaction time (ms) Incongruent trial reaction time (ms) Congruent trial reaction time (ms) Non-emotional gender conflict task Gender adaptation effect (ms) Task performance accuracy Overall reaction time (ms) Incongruent trial reaction time (ms) Congruent trial reaction time (ms)
Mean
SD
72.6 48 F/35 M 16.67 n = 18 n = 10 Mean 4.38 4.49 3.40 36.76 9.28 Mean 28.08 10.31 9.07 60.25 31.35 22.88 64.16 69.59 5.99 46.87 9.58 Mean 26.95 95% 901 947 854 Mean 5.86 98% 842 871 818
6.50 N/A 2.45 N/A N/A SD 5.39 5.19 4.42 12.37 6.58 SD 5.98 2.55 2.60 14.15 6.16 4.84 14.72 19.53 2.19 3.39 3.07 SD 54.25 0.04 175 173 178 SD 50.60 0.03 202.52 208 201
Range 65–91 N/A 10–20 N/A N/A Min–Max 0–23 0–30 0–29 18–76 0–30 Min–Max 14–43 5–16 2–16 28–91 19–46 14–35 38–121 39–173 2–12 29–50 3–15 Min–Max 161–77 81–100% 581–1445 601–1444 581–1445 Min–Max 134–106 82–100% 551–1538 562–1539 540–1538
F, female; M, male; BDI-II, Beck Depression Inventory-II; HAM-D, Hamilton Depression Rating Scale; BAI, Beck Anxiety Inventory; PSWQ, The Penn State Worry Questionnaire; SIGH-A, The Structured Interview Guide for the Hamilton Anxiety Rating Scale; WAIS-IV, Wechsler Adults Intelligence Scale – Fourth Edition; D-KEFS CWIT, Delis–Kaplan Executive Function System Color–Word Interference Test; RAVLT, Rey Auditory Verbal Learning Test; ms, milliseconds.
Copyright # 2016 John Wiley & Sons, Ltd.
Int J Geriatr Psychiatry 2017; 32: 840–848
Emotion conflict adaptation in older adults
encountered an average of 31 incongruent–incongruent pairings and 31 congruent–incongruent pairings during the task. Participants also completed a 2-minute practice version of the task. Refer to prior descriptions of the task for more detail (Egner et al., 2008; Etkin et al., 2006; Monti et al., 2010). Emotion Conflict Adaptation Effect values are calculated by subtracting the mean RT of incongruent trials preceded by a congruent trial from the mean RT of incongruent trials preceded by another incongruent trial (i.e. incongruent–incongruent mean RT —congruent–incongruent mean RT). The difference between these mean RT values indicates adaptation to emotion conflict. Faster responses for incongruent trials that immediately follow an incongruent trial calculate as negative RT values; thus, lower values reflect greater conflict adaptation. Gender conflict task
A gender conflict task served as a non-affectively focused control task. The gender conflict task employed the same facial stimuli as the emotional conflict task, except the distractor words are either “MALE” or “FEMALE” overlaid in red letters, producing gender congruent or incongruent stimuli (Egner, et al., 2008). Participants were instructed to press corresponding response buttons as quickly and accurately as possible to indicate the gender of the face (right index finger for male faces or right middle finger for female faces), and to ignore the superimposed word. Participants encountered an average of 32 incongruent–incongruent pairings and 34 congruent–incongruent pairings; during the task Gender Conflict Adaptation Effect, values are calculated by subtracting the mean RT of incongruent trials preceded by a congruent trial from the mean RT of incongruent trials preceded by another incongruent trial (i.e. incongruent–incongruent mean RT— congruent–incongruent mean RT). The difference between these mean RT values indicates adaptation to gender conflict.
843
Mental health and health measures
The Structured Clinical Interview for the DSM-IV-TR (SCID; First et al., 2007) was administered to participants to determine the presence of current or past psychiatric diagnoses, examined in our primary analyses. Secondary analyses examined mental health symptom severity measures. In particular, anxiety and worry symptom severity were assessed with The Beck Anxiety Inventory (BAI; Beck et al., 1988), The Structured Interview Guide for the Hamilton Anxiety Rating Scale (SIGH-A; Shear et al., 2001), and The Penn State Worry Questionnaire (PSWQ; Meyer et al., 1990). Depressive symptom severity was assessed with The Beck Depression Inventory-II (BDI-II; Beck et al., 1996) and the Hamilton Depression Rating Scale (HAM-D; Hamilton, 1960). Participants reported their subjective health in relation to peers on a scale of 1 (excellent) to 4 (poor), which was included as a covariate in the models. Cognitive tests
The cognitive battery included the three cognitive control measures examined in our primary analyses: inhibitory ability based on time in seconds to complete a Stroop task (Delis–Kaplan Executive Function System (D-KEFS); Color-Word Interference Test (CWIT); Delis et al., 2001), and attention and working memory for numbers (Wechsler Adult Intelligence Scale-IV (WAIS-IV) Digit Span Forward and Backward; Wechsler, 2008). Given the pervasive role of episodic memory and processing speed in predicting function in older adults, secondary analyses included episodic learning and memory ability as measured by the Rey Auditory Verbal Learning Test trial 1 list recall, sum of trials 1–5, and delayed recall performance (RAVLT; Rey, 1958), and speed of information processing based on the rapid transcription of symbols matched to numbers (WAIS-IV Digit Symbol Coding; Wechsler, 2008). Statistical analysis
Standard outlier removal for conflict tasks
For both the emotional conflict and gender conflict tasks, trials were excluded from the analysis for the following reasons: (a) if the RT was
