General Psychology Article Critique VII

Emotion Accentuate the Positive, Eliminate the Negative: Reducing Ambivalence Through Instructed Emotion Regulation Catherine J. Norris and Emily Wu Online First Publication, January 23, 2020. http://dx.doi.org/10.1037/emo0000716

CITATION Norris, C. J., & Wu, E. (2020, January 23). Accentuate the Positive, Eliminate the Negative: Reducing Ambivalence Through Instructed Emotion Regulation. Emotion. Advance online publication. http://dx.doi.org/10.1037/emo0000716

Accentuate the Positive, Eliminate the Negative: Reducing Ambivalence Through Instructed Emotion Regulation

Catherine J. Norris and Emily Wu Swarthmore College

Ambivalence, the simultaneous experience of positivity and negativity, is a conflicting, uncomfortable, arousing state but is a necessary catalyst for behavior change. We sought to examine whether feelings of ambivalence can be reduced using instructed emotion regulation of positive and negative affect, the components of subjective ambivalence. Event-related brain potentials (ERPs) were collected while participants played 3 blocks of mixed gambles, in which each trial involved losing or winning the lesser or greater of 2 amounts. In the 1st block participants responded naturally, and in the 2nd and 3rd blocks they were instructed to focus on either the positive or negative aspects of the outcome. Disappointing wins (e.g., winning $5 instead of winning $12) and relieving losses (losing $5 instead of losing $12) reliably elicited ambivalence; focusing on either the negative or positive aspects of the outcome reduced ambivalence as well as the magnitude of the late positive potential (LPP), indicating successful regulation. Both self-reported affect and ERPs indicated that emotional responses to losses were more difficult to regulate than responses to wins, consistent with a negativity bias in affective processing. Results are interpreted in the framework of theories of affect, and implications for changing behavioral motivation to support healthy behaviors are discussed.

Keywords: mixed emotions, LPP, FRN, negativity bias, affect

Supplemental materials: http://dx.doi.org/10.1037/emo0000716.supp

Ambivalence, or the co-occurrence of positive and negative affect (i.e., mixed feelings), is rare in human experience (Berrios, Totterdell, & Kellett, 2015; Larsen, Coles, & Jordan, 2017; Larsen, McGraw, & Cacioppo, 2001). As the core function of affect is to guide motivation and behavior in a bidirectional manner, to en- courage approach toward appetitive stimuli (i.e., as is the case with positive affect) and avoidance or withdrawal from aversive stimuli (negative affect), the simultaneous experience of both is theoreti- cally considered to be highly conflictual, uncomfortable, and arousing (Cacioppo & Berntson, 1994; Cacioppo, Gardner, & Berntson, 1997, 1999; Norris, Gollan, Berntson, & Cacioppo, 2010). Yet, ambivalence also has been found to increase during crucial stages of behavior change, suggesting that the experience of mixed emotions may at the least accompany decisions to change

behavior (Armitage, Povey, & Arden, 2003; Lipkus, Green, Fea- ganes, & Sedikides, 2001). Individuals attempting to make critical life changes, such as dieting or quitting smoking (Lipkus et al., 2001), may often exist in a constant state of ambivalence regarding their intended behavior. For example, for a smoker who is trying to quit, a cigarette is a source of both positive feelings (e.g., the satisfaction of a craving) and negative feelings (awareness of adverse effects, such as increased cancer risk), resulting in con- flicting approach and avoidance urges when presented with smoking-related stimuli (Armitage et al., 2003; Lipkus et al., 2001). A chronic state of ambivalence may have deleterious emo- tional and behavioral consequences for individuals; thus, using instructed emotion regulation to reduce feelings of ambivalence may be beneficial.

A Brief Review of Theoretical Perspectives on Ambivalence

Two dominant models of the structure of affect have conflicting theoretical perspectives regarding the existence of mixed emo- tions, or ambivalence. Bipolar models of affect argue that positiv- ity and negativity are inseparable in experience and that emotional valence (which ranges from unpleasant/negative affect to pleasant/ positive affect) is irreducible (Barrett, 2006; Barrett & Bliss- Moreau, 2009; Russell & Carroll, 1999; Titchener, 1908; Wundt, 1896). The circumplex model of affect, which is perhaps the most prevalent bipolar model, posits that valence constitutes one of the two primary dimensions of emotional experience, with activation (or arousal) constituting the second dimension, and that affective states fall along a circular perimeter defined by these two dimen-

X Catherine J. Norris and X Emily Wu, Department of Psychology, Swarthmore College.

We thank the members of the Norris Social Neuroscience Lab (nSNL) and the ERP Lab at Swarthmore College for their help with data collection, processing, and analysis. In addition, both Frank Durgin at Swarthmore College and Jeff T. Larsen at the University of Tennessee provided constructive comments. Some of the reported data were included as part of Emily Wu’s undergraduate Psychology thesis and have been presented at annual meetings of the Society for Psychophysiological Research and the Social and Affective Neuroscience Society.

Correspondence concerning this article should be addressed to Catherine J. Norris, Department of Psychology, Swarthmore College, 500 College Avenue, Swarthmore, PA 19081. E-mail: [email protected]

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

Emotion © 2020 American Psychological Association 2020, Vol. 1, No. 999, 000 ISSN: 1528-3542 http://dx.doi.org/10.1037/emo0000716

1

sions. Fundamental to bipolar models of affect is the assumption that affective experience cannot simultaneously encompass both ends of the valence dimension—in other words, that individuals cannot feel both unpleasant (negative) and pleasant (positive) at the same time.

In contradiction to bipolar models of affect, bivariate models argue that negativity and positivity are separable dimensions of emotional experience. The model of evaluative space (ESM; Ca- cioppo & Berntson, 1994; Cacioppo et al., 1997, 1999; Norris et al., 2010), for example, proposes that negativity and positivity are separable independent dimensions of affective experience. When combined (i.e., positivity–negativity), they produce a third dimen- sion, behavioral motivation, that ranges from avoidance to ap- proach. The ESM makes a number of unique theoretical predic- tions regarding affective experience, two of which are particularly relevant here. First, negativity and positivity can function in a reciprocal manner (such that an increase in one may be accompa- nied by a decrease in the other), but need not: They may function independently (e.g., an increase in negativity may not be accom- panied by any change in positivity) or coactively (such that both negativity and positivity may increase or decrease together). Im- portantly, this feature allows for the simultaneous co-occurrence of negativity and positivity (i.e., ambivalence), which produces a conflicting motivational state in which both avoidance and ap- proach are equally likely. Second, if negativity and positivity are independent dimensions of affective experience, they may function differently. One functional asymmetry that the ESM predicts is that, all else being equal, negativity has a stronger impact on behavior than positivity, given evolutionary pressures to prioritize survival over fitness. This negativity bias has been noted in diverse research areas, as evidenced by multiple literature reviews (Taylor, 1991; Rozin & Royzman, 2001; Baumeister, Bratslavsky, Finke- nauer, & Vohs, 2001), and may have implications for research on ambivalence.1

Research suggests that (strong) feelings of ambivalence are likely rare in daily emotional experience (cf. Larsen et al., 2001). Indeed, both narrative reviews (Larsen et al., 2017) and meta- analysis (Berrios et al., 2015) suggest that the range of stimuli that can elicit mixed feelings may be limited to exceptional events (e.g., tragicomic films, life transitions, cigarette cravings, thoughts of suicide). To better understand the nature of ambivalence, Larsen and his colleagues (Larsen et al., 2001; Larsen & McGraw, 2011) have designed careful studies to measure mixed emotions in re- sponse to major life events and to elicit ambivalence in a con- trolled laboratory environment. Results have shown that individ- uals report feelings of mixed emotions when moving out of their freshman dorm room (Larsen et al., 2001), graduating from college (Larsen et al., 2001), leaving a job (LaFarge, 1994), and starting a family (Premberg, Hellström, & Berg, 2008). In the laboratory, ambivalence can be elicited using mixed gamble outcomes (Larsen, McGraw, Mellers, & Cacioppo, 2004), music (Hunter, Schellenberg, & Schimmack, 2010; Larsen & Stastny, 2011), pictures (Schimmack, 2005), and movie clips (Larsen, To, & Fireman, 2007; Larsen et al., 2001).

Although ambivalence is thought to be rare in human experience (Berrios et al., 2015; Larsen et al., 2001, 2017) and has been studied primarily in response to major life events (e.g., Larsen et al., 2001) and controlled laboratory experiments (e.g., Larsen et al., 2004; Larsen & McGraw, 2011), there is cause to believe that

the regulation of mixed emotions may be beneficial. First, bivari- ate models of emotion predict that when ambivalence occurs, it is likely to be conflicting, uncomfortable, and arousing, due to its lack of motivational guidance (Cacioppo & Berntson, 1994; Ca- cioppo et al., 1997, 1999; Norris et al., 2010). In the absence of clear input to avoid or approach a stimulus, one might oscillate between the two states or simply not take action at all. Reducing ambivalence may produce clear behavioral motivation. Second, research in health psychology suggests that ambivalence may be a necessary catalyst for behavioral change. According to the transtheoretical model of change (TTM; Prochaska & DiCle- mente, 1982; Armitage et al., 2003), there exists a quadratic relationship between ambivalence and the five2 stages of change: ambivalence is low during precontemplation (i.e., the earliest stage); highest during the three intermediate stages, which include contemplation, preparation, and action; and low again during maintenance (the last stage). Importantly, the TTM predicts that for successful behavior change, pros and cons must first reach a balance (i.e., resulting in high ambivalence) and then pros must outweigh cons during the three middle stages. Take for example the cessation-motivated smoker, who may feel no ambivalence before contemplating quitting, but who must begin to simultane- ously experience negativity and positivity toward smoking before committing to and ultimately achieving their goal. If ambivalence toward smoking cues results in behavioral immobilization, no change may occur. If, however, our smoker can successfully regulate their mixed feelings by focusing on the negative conse- quences of smoking instead of the positive reinforcement of the act, their cessation efforts may be more effective. Thus, the current study sought to investigate whether emotion regulation can be effective in decreasing ambivalence, consequently reducing the associated conflict, discomfort, and arousal, and providing clear behavioral motivation.

Emotion Regulation

Research on emotion regulation has flourished in the past 30 years, due in part to thoughtful, thorough reviews of the existing literature (e.g., Gross, 1998); theoretical models that have guided empirical work (e.g., Gross, 2002; Ochsner, Silvers, & Buhle, 2012); and the development and refinement of paradigms and

1 It is worth noting that bipolar models of affect focus on (hedonic) valence as one of the descriptive features of core affect and differentiate between description and process (Barrett & Bliss-Moreau, 2009). In this way, although valence may be seen as irreducible, and thus feeling both positive and negative at the same time may be theoretically impossible, bipolar models do allow that people may alternate quickly between feelings of positivity and negativity (Barrett & Bliss-Moreau, 2009). Larsen and McGraw (2011) have countered this argument by showing that (a) partic- ipants report simultaneously experienced mixed emotions using continuous measures with millisecond accuracy, and (b) participants spontaneously report mixed emotions even when not cued by measures explicitly men- tioning positive and negative affect. Thus, individuals report the experience of simultaneous positivity and negativity, regardless of the underlying process at work. The current study was not designed to directly test hypotheses regarding the process underlying ambivalence (i.e., simultane- ity versus alternation) and instead focuses on the reported experience of mixed emotions (i.e., ambivalence).

2 The TTM has been modified by different researchers over the past 50 years; some versions currently include as many as 7 stages of change. We focus on the core stages here.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

2 NORRIS AND WU

psychophysiological tools (e.g., functional MRI [fMRI], eye track- ing, event-related brain potentials [ERPs]) necessary for investi- gating the processes underlying regulation. An exhaustive review of the literature on regulation is beyond the scope of the current paper; thus, we focus on research most relevant to our questions of interest.

Much research on emotion regulation has focused on reap- praisal, in which individuals are instructed to (a) reappraise the stimulus (e.g., movie, picture) in such a way as to feel nothing (Gross, 1998), (b) attempt to feel less bad or negative about the stimulus (Ochsner, Bunge, Gross, & Gabrieli, 2002), (c) distance themselves from the stimulus (Koenigsberg et al., 2010; McRae, Hughes, et al., 2010), or (d) reinterpret the stimulus to feel good or positive (Doré et al., 2017). This focus on reappraisal is driven by classic work that has demonstrated its efficacy in the parallel reductions of emotional experience, expression, and physiology (Gross, 1998). In addition, reappraisal is antecedent-based, mean- ing that regulation occurs early in the emotion-generative process, and generally is implemented as an instructed strategy for process- ing subsequent emotional stimuli (Gross, 2002). Taken together, these characteristics provide additional support for the efficacy of reappraisal, as it is implemented early on and can be trained through instruction.

The majority of studies examining reappraisal as an emotion regulation strategy have focused on the down-regulation of nega- tive feelings. Although this approach arguably has the strongest external validity, as reduction of negativity may have implications for well-being and mental health, it represents only one approach to studying emotion regulation. A subset of studies has investi- gated both the up- (increase) and down- (decrease) regulation of emotional responses toward unpleasant stimuli (Ochsner et al., 2004; Moser, Hajcak, Bukay, & Simons, 2006; van Reekum et al., 2007; Ray, McRae, Ochsner, & Gross, 2010). Their findings have begun to delineate mechanisms that are commonly shared versus unique to increasing and decreasing (negative) feelings. Additional research has explored the up-regulation of positive feelings (cf. Tugade & Fredrickson, 2004), with a particular focus on its con- sequences for well-being and resilience. Little to no research, however, has addressed in a single study the up- and down- regulation of both positive and negative affect, let alone investi- gating the regulation of mixed emotions.

Measurement of Emotion Regulation

Different measures have been used to evaluate reappraisal suc- cess. Often, participants are asked to self-report either their nega- tive affect during passive viewing and following reappraisal (e.g., Ochsner et al., 2002; McRae, Ciesielski, & Gross, 2012) or their success at implementing reappraisal (e.g., Eippert et al., 2007). These measures are, however, prone to demand characteristics—as participants likely are aware that the experimenter expects them to feel less negative and to report successful regulation during in- structed reappraisal of responses to unpleasant stimuli. Thus, re- searchers have turned to neural and psychophysiological measures to study reappraisal success. Ochsner and his colleagues have conducted extensive research utilizing fMRI to examine the neural networks of emotion regulation (for reviews see Ochsner et al., 2012; Buhle et al., 2014). Results have consistently shown that instructed down-regulation (using reappraisal) is associated with

decreased amygdala and increased prefrontal cortex (PFC; dorso- lateral and ventromedial regions, in particular) activation (Ochsner et al., 2002), whereas instructed up-regulation of negative affect is associated with increased amygdala and increased PFC activation (Ochsner et al., 2004). Together, Ochsner and his colleagues (2002, 2004) suggest that these patterns indicate that the amygdala, which is involved in emotion processing, is affected bidirectionally by up- and down-regulation and that regions of the PFC are implicated in the cognitive control required by reappraisal.

Other research has focused on ERPs as an indirect measure of emotion regulation, given that emotional processes unfold over time and that ERPs have superior temporal resolution compared to fMRI. These studies have focused on the late positive potential (LPP), a positive-going component maximal over centro-parietal midline sites occurring approximately 300 –1000 ms poststimulus. The LPP is larger to emotional than to neutral stimuli (e.g., arousing pictures; Hajcak & Olvet, 2008; Ito, Larsen, Smith, & Cacioppo, 1998; Schupp et al., 2000) and is reduced during in- structed emotion regulation (Hajcak & Nieuwenhuis, 2006; Haj- cak, Dunning, & Foti, 2009; Foti & Hajcak, 2008; for a review see Hajcak, MacNamara, & Olvet, 2010). In sum, the inclusion of an indirect measure (e.g., ERPs) may thus be helpful in providing convergent evidence (along with self-reported emotional re- sponses) of the successful regulation of mixed feelings.

In the current study, we sought to examine the effects of instructed regulation on emotional responses to mixed gamble outcomes. Given the nature of mixed gamble outcomes, which elicit both positive and negative affect, we were able to design a single study to address the following three research questions:

Question 1a. Can instructed regulation decrease ambivalence experienced when winning (i.e., disappointing win) or losing (i.e., relieving loss) the lesser of two amounts? Previous research has demonstrated that individuals reliably report mixed emotions when, for example, winning $5 instead of winning $12 (Larsen et al., 2004; Larsen, Norris, McGraw, Hawkley, & Cacioppo, 2009). We sought to determine whether instructions to focus on either the positive or negative aspect of such mixed gamble outcomes would reduce ambivalence.

Question 1b. Given the potential for experimental demands to impact self-reports of emotional responses, we also included ERPs to provide an indirect and convergent measure of successful reg- ulation. We predicted that, if instructed regulation is effective in reducing ambivalence to mixed gamble outcomes, the LPP would reflect this reduction (Hajcak & Nieuwenhuis, 2006).

Question 2. If instructed regulation is effective in decreasing ambivalence, what are the psychological mechanisms underlying this reduction? Specifically, we examined how positive and neg- ative affect (which combined produce ambivalence) are affected by focusing on the positive versus negative aspects of a mixed gamble outcome. Specifically, regulation involves a change in emotion. Focusing on one aspect of a mixed gamble outcome may result in reciprocal activation of positive and negative affect (i.e., an equivalent increase in one and decrease in the other), uncoupled activation (an increase/decrease in one with no change in the other), or even asymmetrical patterns of activation (e.g., a larger increase in one and smaller decrease in the other).

Question 3. Given research demonstrating a negativity bias in affective responses, such that negativity often outweighs positiv- ity, we investigated asymmetries in the regulation of ambivalence.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

3REGULATION OF AMBIVALENCE

In particular, we examined the relative efficacy of (a) focusing on the positive versus negative aspects of mixed gamble outcomes in response to (b) wins versus losses, as measured by (c) positive and negative affect. Asymmetries may arise in any of these domains.

Method

Participants

Sixty-four (34 female) native English-speaking Swarthmore College undergraduates between the ages of 18 and 21 (M � 19.12, SD � 1.02) were recruited through Sona (an online plat- form), fliers, and word-of-mouth. Behavioral data were excluded from two participants (one did not follow instructions; one did not finish the study) and ERP data were not collected for four partic- ipants (due to experimenter error). Sample size was set based on previous studies to allow for enough power to test predicted 3-way interactions, as well as moderation by individual difference factors (which are beyond the scope of the current study and will not be discussed here). Arguably, the most relevant studies for determin- ing an appropriate sample size for the current come from the literature on emotion regulation using ERPs. Using a design very similar to the current study, Hajcak and Nieuwenhuis (2006) had 14 participants passively view pleasant, unpleasant, and neutral IAPS images in an initial block and then use reappraisal during a subsequent regulation block consisting of only unpleasant IAPS images and reported a reliable reduction of the LPP during reap- praisal. Foti and Hajcak (2008) examined the LPP in response to IAPS images that were preceded by a statement designed to manipulate the interpretation of the emotional content. They re- ported a reliable effect of condition (neutral vs. negative state- ments preceding unpleasant pictures) from a sample size of 26 participants, with condition manipulated between participants. Hajcak et al. (2009) had 32 participants view neutral and unpleas- ant IAPS images and directed their attention to a more or less arousing portion of the picture; they reported a reliable reduction of the LPP when attention was directed to a less arousing portion of the picture. Notably, each of these studies used approximately the same number of trials per condition as the current study (�20), meaning that our LPP amplitudes should be equally reliable. Thus, we believe that our sample size (�60) is justified (and, in fact, more than sufficient) by the previous literature examining emotion regulation via ERPs. The Swarthmore College Institutional Re- view Board supervised the study, and participants received either course credit or $20 as compensation.

General Procedure

Upon arrival at the ERP Laboratory at Swarthmore College, participants were given an overview of the study and provided written and oral informed consent. Participants were seated in a comfortable chair, and a 64-channel electrode net (Electrical Geo- desics, Inc., www.egi.com) was placed on their scalp. Participants performed two tasks, a card-choice gambling task and a picture- viewing task, and then completed a series of personality surveys and a demographics form.3 At the end of the session, which lasted approximately 2 hrs, participants answered a series of questions and were debriefed.

Card-Choice Gambling Task

Participants were given a $10 endowment at the beginning of the gambling task and were informed that the outcome of each trial would increase or decrease that amount (i.e., a loss of $1 on the task would correspond to a loss of 10 cents from the real life endowment). Participants played a series of gambles in which they chose one of two facedown cards; both cards either corresponded to wins or losses (i.e., a win and a loss were never possible outcomes on the same trial), with each card containing a different amount. Four different gamble types, in which participants either won or lost a greater or lesser amount of money, were possible. Winning or losing a greater amount was considered an outright win (OW; e.g., winning $15 instead of $5) or outright loss (OL; losing $15 instead of $5), respectively. Winning or losing a lesser amount was considered a disappointing win (DW; winning $5 instead of $15) or relieving loss (RL; losing $5 instead of $15), respectively. Each gamble type (OW, DW, RL, OL) occurred for five different levels of dollar amounts: $5/$15, $10/$30, $15/$45, $20/$60, and $25/$75. For example, the set of OW outcomes included winning $15 instead of $5, $30 instead of $10, $45 instead of $15, $60 instead of $20, and $75 instead of $25. DW outcomes included winning $5 instead of $15, $10 instead of $30, $15 instead of $45, $20 instead of $60, and $25 instead of $75. The sets of RL and OL outcomes were identical, but involved losses instead of gains. Each gamble type and level was repeated four times during each block, comprising a total of 80 trials per block (4 Gamble Types � 5 Levels � 4 Repetitions � 80). The order of gambles was prerandomized such that the tally of money won or lost never exceeded $200 and that participants never experienced one gamble type more than three times in a row. Participants were randomly assigned to one of two prerandomized orders. Within each prerandomized order, gamble sequences for each block were identical for all participants to ensure that they would all experi- ence the same emotional stimuli at the same time and in the same order.

After each gamble outcome, participants rated their emotional responses using the evaluative space grid (ESG; Larsen et al., 2009), a 5x5 grid that allows for independent assessments of positivity (horizontal axis) and negativity (vertical axis). Partici- pants are instructed to move the mouse to one of the cells in the grid to indicate their positive and negative feelings. The x- and y-coordinates of responses provide scores on two 5-point unipolar scales, ranging from 0 (not at all positive/negative) to 4 (extremely positive/negative), which reflect independent measures of positiv- ity and negativity (Larsen et al., 2009). In addition to these two ratings, the ESG allows for a measure of ambivalence, which is simply the minimum of the positivity and negativity ratings (i.e., MIN[pos, neg]; Kaplan, 1972; Schimmack, 2001; Larsen et al., 2004). Finally, we also calculated a measure of valence (i.e., positivity—negativity; Larsen et al., 2009) to allow for compari- sons between bipolar and bivariate predictions.

Each gamble began with a fixation cross for 500 ms, followed by a blank screen for 250 ms, a screen indicating the gamble stakes (e.g., WIN $5, WIN $15) for 1500 ms, another blank screen for

3 Results from the picture-viewing task and the personality surveys are reported elsewhere, contact the first author for details.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

4 NORRIS AND WU

250 ms, and finally the two cards appearing face down. At this point, participants either clicked the left mouse button to choose the left card or clicked the right mouse button to choose the right card. After their decision, the gamble outcome was displayed for 1500 ms, and then participants rated their emotional responses using the ESG (see Figure 1 for a trial schema). Before beginning, participants read written instructions for the card-choice gambling task and performed 8 practice trials.

All participants completed three blocks (80 trials each) of the card-choice gambling task. Participants completed the first block with no emotion regulation instruction, responding naturally to each outcome. Before the second and third blocks, participants received written and verbal instructions to regulate their emotions by focusing on the positive or negative aspects of each trial, with the order of positive and negative regulation counterbalanced across participants. For example, in the positive regulation condi- tion participants were told: “If you won $5 instead of $15, focus on the positive fact that you won money instead of the negative fact that you could have won more money.”

EEG Data Collection & Processing

Continuous electroencephalographic (EEG) data were collected using an EGI standard amplifier and a Netstation system. Caps contained 64 silver chloride electrodes fitted to an elastic cap in compliance with the International 10 –20 system, and electrodes were referenced to the average of the two (independent) mastoid electrodes. Researchers adjusted electrodes to ensure none had impedances above 40 k�. EEG was subjected to an online high- pass filter at 0.1 Hz and was sampled at 500 Hz. In between blocks, researchers checked and fixed impedances.

Data preprocessing was performed using the EEGLAB package (Delorme & Makeig, 2004) within MATLAB. Continuous EEG was filtered through a high-pass filter of 0.5 Hz and a low-pass filter of 30 Hz then downsampled to 256 Hz. Researchers removed

portions of data collected during cap fitting and impedance checks by eye then interpolated bad channels. Ocular artifact was removed using independent components analysis (ICA), then data were rereferenced to an average. Data were segmented into 1500 ms epochs starting 500 ms before the outcome was revealed and ending 1000 ms after and were baseline corrected. Segments were sorted and averaged into groups according to emotion regulation instruction (none, negative focus, positive focus), outcome (loss, win), and comparison (negative, positive). Segments containing data greater than 75 uV or less than �75 uV were rejected to eliminate muscular or movement artifacts. Participants who had fewer than 15 (of 20 possible) good trials for any of the 12 conditions (3 Regulation � 2 Outcome � 2 Comparison Design) were dropped from ERP analyses. We set this as a strict criterion to ensure that there were no differences between conditions in the number of trials entered into averaged ERP waveforms. Out of a possible 240 trials, the average number of bad trials for excluded participants was 107 (SD � 42); the average number of bad trials for included participants was 19 (SD � 15). Eight participants were dropped due to having too many bad trials, six due to software issues, and three due to data collection problems, leaving a sample size of 43 participants for ERP analyses.

Demographics & Debriefing

Finally, participants completed a short demographics question- naire (to assess ethnicity, racial background, etc.) and answered a series of questions about the study, including how they performed the tasks and their perceptions about the study and our hypotheses. Afterward, the experimenter orally debriefed participants.

Results

The overall design of the study was a 3(regulation instruction: none, negative focus, positive focus) � 2(outcome: loss, win) �

Figure 1. Mixed Gamble Trial Schema. The fixation screen and stakes screen were followed by 250 ms of black screen to allow for the collection of ERPs. The card choice screen and Evaluative Space Grid (ESG) screen remained up until a response was made. See the online article for the color version of this figure.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

5REGULATION OF AMBIVALENCE

2(comparison: negative, positive)4 factorial, with all factors ma- nipulated within-participants. To streamline data reporting and more directly provide tests of our hypotheses, we systematically break down these analyses by (a) focusing on just the no regulation instruction condition to replicate past results and (b) focusing on regulation of mixed gamble outcomes (i.e., DW, RL). Full results from the omnibus model are presented in online supplementary materials. A strict alpha level of 0.05 was used throughout, as well as the Bonferroni correction for multiple comparisons when ad- dressing pairwise analyses, and exact p values are presented up to p � .001. Effect sizes (�p2) and 95% confidence intervals (CI [low, high]) are included; data are available through OSF (osf.io/4a6k7).

Results from the No Regulation Condition

We sought to replicate past research (e.g., Larsen et al., 2004) by demonstrating that our gambling paradigm elicited expected pat- terns of emotional responses. We limited these analyses to re- sponses from the first block of gambles, in which no emotion regulation instruction was provided, and participants responded naturally. We conducted 2(outcome: loss, win) � 2(comparison: negative, positive) general linear models (GLMs) separately on minimum scores (i.e., MIN[negativity, positivity]; Kaplan, 1972; Larsen et al., 2004, 2009), negativity ratings, and positivity ratings.

Minimum scores. The GLM conducted on minimum scores (i.e., ambivalence) revealed only one significant effect, the out- come x comparison interaction, F(1, 61) � 146.55, p � .001, �p2 � .71 (Figure 2a). As expected, pairwise tests indicated that RL elicited more ambivalence (M � 0.66, SE � .05) than OL (M � 0.07, SE � .02), 95% CIs [0.49, 0.70], and OW (M � 0.07, SE � .02), 95% CIs [0.48, 0.70], both ps � .001. Similarly, DW elicited more ambivalence (M � 0.60, SE � .05) than did OW, 95% CIs [0.44, 0.63], and OL, 95% CIs [0.44, 0.63], both ps � .001. This finding shows that participants did indeed report the experience of mixed emotions in response to relieving losses and disappointing wins.

Negativity ratings. The main effect of outcome, F(1, 61) � 413.95, p � .001, �p2 � .87, indicated that losses were rated as more negative (M � 1.87, SE � .06) than wins (M � 0.56, SE � .04; Figure 2b), 95% CIs [1.18, 1.43]. Similarly, the main effect of comparison, F(1, 61) � 480.81, p � .001, �p2 � .89, indicated that negative comparisons were rated as more negative (M � 1.88, SE � .06) than positive comparisons (M � 0.55, SE � .04), 95% CIs [1.21, 1.45]. The outcome x comparison interaction, F(1, 61) � 75.32, p � .001, �p2 � .55, qualified both of the main effects. Pairwise comparisons showed that outright losses (OL) elicited more negativity (M � 2.71, SE � .07) than did relieving losses (RL; M � 1.02, SE � .07), p � .001, 95% CIs [1.54, 1.84], and (DW; M � 1.05, SE � .07), p � .001, 95% CIs [1.50, 1.83]. Outright wins (OW; M � 0.08, SE � .02) elicited less negativity than did DW, 95% CIs [0.83, 1.11], and RL, 95% CIs [0.80, 1.09], both ps � .001.

Positivity ratings. The main effect of outcome, F(1, 61) � 390.21, p � .001, �p2 � .87, indicated that wins were rated as more positive (M � 1.93, SE � .07) than losses (M � 0.72, SE � .05; Figure 2c), 95% CIs [1.08, 1.33]. The main effect of comparison, F(1, 61) � 572.16, p � .001, �p2 � .90, indicated that positive comparisons were rated as more positive (M � 2.02, SE � .07) than negative comparisons (M � 0.63, SE � .04), 95% CIs [1.28,

1.51]. The outcome x comparison interaction, F(1, 61) � 14.76, p � .001, �p2 � .20, qualified both of the main effects. Pairwise comparisons showed that OW elicited more positivity (M � 2.71, SE � .07) than did DW (M � 1.10, SE � .06), p � .001, 95% CIs [1.44, 1.68], and RL (M � 1.34, SE � .09), p � .001, 95% CIs [1.22, 1.52]. OL (M � 0.11, SE � .02) elicited less positivity than did DW, 95% CIs [0.89, 1.19], and RL, 95% CIs [1.06, 1.40], both ps � .001.

In sum, all analyses replicated past research and indicated that our gambling task reliably elicited ambivalence toward RL and DW gamble outcomes.

The Effects of Regulation on Ambivalence: Self-Report Measures

The first of our central research questions (Question 1a) con- cerned the efficacy of instructed regulation in reducing ambiva- lence to mixed gamble outcomes. To examine this question, we focused solely on emotional responses to relieving losses (RL) and disappointing wins (DW) and subjected minimum scores (i.e., ambivalence) to a 3(regulation condition: no instruction, negative focus, positive focus) � 2(gamble outcome: RL, DW) GLMs. Again, we included the order of regulation and the outcome level as variables of no interest.

Minimum scores. The gamble outcome main effect, F(1, 61) � 4.20, p � .045, �p2 � .06, indicated that RL elicited more ambivalence (M � 0.59, SE � .04) than did DW (M � 0.53, SE � .04), 95% CIs [0.001, 0.10]. Both the small effect size and the fact that this main effect collapses across the no regulation and regu- lation conditions make this pattern difficult to interpret. The reg- ulation main effect was not significant, p � .09, �p2 � .08. More importantly, the regulation x gamble outcome interaction, F(2, 60) � 13.54, p � .001, �p2 � .31, was significant (see Figure 3). Pairwise tests revealed that a negative focus decreased ambiva- lence for RL (M � 0.48, SE � .06) compared to no regulation (M � 0.66, SE � .05), p � .028, 95% CIs [0.02, 0.34]; whereas a positive focus had no effect on ambivalence toward RL (M � 0.62, SE � .06), p � 1.00, 95% CIs [�.12, .20]. Similarly, a positive focus decreased ambivalence for DW (M � 0.38, SE � .05) compared to no regulation (M � 0.60, SE � .05), p � .003, 95% CIs [0.06, 0.38]; whereas a negative focus had no effect on ambivalence toward DW (M � 0.62, SE � .06), p � 1.00, 95% CIs [�.16, .14]. In sum, regulation did decrease ambivalence, with a negative focus impacting responses to RL and a positive focus impacting responses to DW.

4 Note that including separate factors for outcomes (i.e., losses and wins) and comparisons (i.e., negative and positive) is both (a) critical for our question regarding the functioning of the negativity bias (i.e., asymmetries) in regulation of ambivalence and (b) a typical approach in studies of emotional responses to mixed gamble outcomes (cf. Larsen et al., 2004, 2009). Together, these two factors capture the four different gamble outcomes: a loss with a negative comparison is an outright loss, a loss with a positive comparison is a relieving loss, a win with a negative comparison is a disappointing win, and a win with a positive comparisons is an outright win.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

6 NORRIS AND WU

The Effects of Regulation on Positive and Negative Affect

Our second central research question (Question 2) concerned the psychological mechanisms by which instructed regulation reduced ambivalence. Specifically, we sought to examine how a negative versus a positive focus affected unipolar emotional responses (i.e., negativity, positivity) to mixed gamble outcomes. In addition, this set of analyses allowed us to also investigate asymmetries in the efficacy of a negative versus a positive focus on positive and

negative affect experienced in response to losses (RL) and wins (DW; Question 3). We conducted 3(regulation) � 2(gamble out- come) GLMs separately on negativity and positivity ratings to address these questions.

Negativity ratings. The regulation main effect, F(2, 60) � 41.56, p � .001, �p2 � .58, showed that instructions to focus on the negative elicited more negativity (M � 1.31, SE � .07) than did no regulation (M � 1.04, SE � .05), 95% CIs [0.11, 0.43], which in turn elicited more negativity than instructions to focus on the positive (M � 0.68, SE � .06), 95% CIs [0.20, 0.50], both ps � .001 (Figure 4a). The gamble outcome main effect was not signif- icant, p � .18, �p2 � .04. The regulation x gamble outcome interaction, F(2, 60) � 20.66, p � .001, �p2 � .41, however, was significant (Figure 4a). Pairwise tests revealed that a negative focus increased negativity for RL (M � 1.31, SE � .08) compared to no regulation (M � 1.02, SE � .07), p � .001, 95% CIs [0.13, 0.45]; whereas a positive focus had no effect on negativity toward RL (M � 0.89, SE � .08), p � .14, 95% CIs [�.03, .30]. Similarly, a positive focus decreased negativity for DW (M � 0.48, SE � .06) compared to no regulation (M � 1.05, SE � .07), p � .001, 95% CIs [0.38, 0.76]; whereas a negative focus had a marginal effect on negativity toward DW (M � 1.30, SE � .09), p � .05, 95% CIs [.00, .51].

Positivity ratings. The regulation main effect, F(2, 60) � 42.23, p � .001, �p2 � .59, showed that instructions to focus on the positive elicited more positivity (M � 1.43, SE � .07) than did no regulation (M � 1.24, SE � .06), p � .008, 95% CIs [0.04, 0.33], which in turn elicited more positivity than instructions to focus on the negative (M � 0.86, SE � .06), 95% CIs [0.23, 0.54], p � .001. The gamble outcome main effect was not significant, p � .67, �p2 � .003. The regulation x gamble outcome interaction, F(2, 60) � 6.43, p � .003, �p2 � .18, however, was significant

a

b c

Figure 2. Results from the 2(outcome: loss, win) � 2(comparison: negative, positive) GLMs conducted on minimum scores (i.e., ambivalence; a), negativity ratings (b), and positivity ratings (c) in the no regulation condition. Error bars represent 1 SE.

Figure 3. Results from the 3(regulation: no instructions, negative focus, positive focus) � 2(gamble outcome: relieving loss, disappointing win) interaction for minimum scores (i.e., ambivalence). A negative focus decreased ambivalence toward RL; a positive focus decreased ambivalence for DW.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

7REGULATION OF AMBIVALENCE

(Figure 4b). Pairwise tests revealed that a negative focus decreased positivity for RL (M � 0.77, SE � .09) compared to no regulation (M � 1.34, SE � .09), p � .001, 95% CIs [0.32, 0.81]; whereas a positive focus had no effect on positivity toward RL (M � 1.47, SE � .10), p � .53, 95% CIs [�.11, .37]. A positive focus increased positivity for DW (M � 1.39, SE � .07) compared to no regulation (M � 1.15, SE � .07), p � .001, 95% CIs [0.09, 0.38], and a negative focus decreased positivity toward DW (M � 0.94, SE � .08), p � .004, 95% CIs [.06, .37].

In sum, instructed regulation decreased ambivalence toward mixed gamble outcomes. Instructions to focus on the negative both increased negativity and decreased positivity to RL; instructions to focus on the positive both decreased negativity and increased positivity to DW. In addition, focusing on the negative also de- creased positivity to DW, suggesting that a negative focus may have a stronger impact than a positive focus and providing some evidence for a negativity bias in regulation of emotional responses to mixed gamble outcomes.

Electrocortical Mechanisms of the Regulation of Mixed Gamble Outcomes

In addition to self-report measures, we sought to use ERPs to examine evidence for the regulation of ambivalence in response to mixed gamble outcomes (Question 1b). To examine this question, we focused on two ERP components. First, the feedback related negativity (FRN) is a component that has previously been shown to be sensitive to both bad outcomes (e.g., losses; Hajcak, Moser, Holroyd, & Simons, 2006) and to unexpected negative outcomes (Holroyd, Nieuwenhuis, Yeung, & Cohen, 2003). We anticipated that the FRN might be sensitive to negative comparisons, based on the fact that in the card choice gambling paradigm, participants are informed early on in a trial whether they have won or lost money and only find out later if they won or lost the greater or lesser of two amounts. If individuals anticipate or hope for the best possible outcome, a negative comparison (i.e., losing the greater or winning the lesser of two amounts) is an unexpected negative outcome. We measured the FRN as the peak negative amplitude at Fz between 200 and 400 ms after the outcome was revealed.

Second, the LPP has been shown to be (a) larger to emotional versus neutral stimuli (e.g., arousing pictures; Hajcak & Olvet, 2008) and (b) reduced during instructed emotion regulation (Haj- cak & Nieuwenhuis, 2006). Thus, we anticipated that the LPP would be reduced when participants were instructed to either focus on the negative or on the positive aspects of the gamble outcomes. We measured the LPP as the average amplitude at Pz between 300 and 750 ms after the outcome was revealed. Note that windows for both components were determined by examining average group waveforms collapsed across all regulation conditions in order to not allow our analyses to be biased by the effects of regulation (Luck, 2012), and we limit analyses to sites of primary interest, where the FRN (Fz) and LPP (Pz) are maximal, to prevent Type I error.

FRN peak amplitudes. FRN peak amplitudes were subjected to a 3(regulation: no instruction, negative focus, positive focus) � 2(outcome: loss, win) � 2(comparison: negative, positive) GLM. The main effect of comparison, F(1, 42) � 10.76, p � .002, �p2 � .20, showed that, as predicted, negative comparisons elicited larger FRN amplitudes (M � �2.11, SE � .26) than did positive com- parisons (M � �1.71, SE � .25), 95% CIs [.15, .65]. In addition, regulation had a significant linear effect, F(1, 42) � 4.41, p � .042, �p2 � .10, such that FRN amplitudes were largest during a negative focus (M � �2.12, SE � .28), middling during no regulation (M � �1.86, SE � .25), and smallest during a positive focus (M � �1.73, SE � .27). Although none of the pairwise tests were significant, the linear main effect indicates a significant linear trend across conditions.

FRN latencies. We subjected FRN latencies to a parallel GLM. The comparison main effect, F(1, 42) � 17.85, p � .001, �p

2 � .30, indicated that the FRN showed a later peak during a negative (M � 319.60 ms, SE � 4.83) versus a positive (M � 299.98, SE � 6.17) comparison, 95% CIs [10.25, 29.00]. Regula- tion had a significant quadratic effect on FRN latencies, F(1, 42) � 4.92, p � .033, �p2 � .11, such that the FRN was delayed during a negative focus (M � 316.86, SE � 6.03) as compared to no regulation (M � 300.55, SE � 6.92), p � .05, 95% CIs [0.003, 32.61]. Neither the negative focus, p � 1.00, 95% CIs [�8.95, 18.76], nor the no regulation condition, p � .35, 95% CIs [�6.36,

a

b

Figure 4. Results from the 3(regulation: no instructions, negative focus, positive focus) � 2(gamble outcome: relieving loss, disappointing win) interactions for (a) negativity and (b) positivity ratings. Generally, a negative focus affected responses to RL; a positive focus affected re- sponses to DW. In addition, a negative focus decreased positive affect toward DW.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

8 NORRIS AND WU

29.16], differed from the positive focus condition (M � 311.96, SE � 5.75).

LPP area. LPP areas were subjected to a 3(regulation: no instruction, negative focus, positive focus) � 2(outcome: loss, win) � 2(comparison: negative, positive) GLM. As expected (and replicating past research; Hajcak & Nieuwenhuis, 2006), the reg- ulation main effect, F(2, 39) � 10.85, p � .001, �p2 � .36, showed that the LPP was larger during no regulation (M � 0.94, SE � .08) than during either the negative focus (M � 0.75, SE � .07), p � .001, 95% CIs [.11, .27], or the positive focus (M � 0.79, SE � .06), p � .007, 95% CIs [.04, .25] conditions. The negative and positive focus conditions did not differ, p � .30, 95% CIs [�.04, .12]. The outcome main effect, F(1, 40) � 6.33, p � .016, �p2 � .14, showed that the LPP was larger for wins (M � 0.86, SE � .07) than for losses (M � 0.80, SE � .07), 95% CIs [.01, .11]. This was qualified by the outcome x comparison interaction, F(1, 40) � 17.08, p � .001, �p2 � .30. Pairwise tests revealed that OW elicited

larger LPP areas (M � 0.91, SE � .07) than RL (M � 0.74, SE � .07), p � .001, 95% CIs [.09, .24]; but there was no significant difference between DW (M � 0.80, SE � .06) and OL (M � 0.85, SE � .07), p � .18, 95% CIs [�.02, .11]. In addition, OL elicited larger LPP areas than did RL, p � .001, 95% CIs [.05, .17]; and OW elicited alrger LPP areas than did DW, p � .006, 95% CIs [.03, .17].

Finally, there was a significant effect for the regulation x outcome x comparison interaction (see Figure 5), F(1, 40) � 5.17, p � .028, �p

2 � .11. For OL, both a negative (M � 0.75, SE � .07), p � .001, 95% CIs [.15, .37], and positive (M � 0.80, SE � .07), p � .019, 95% CIs [.04, .38], focus decreased the LPP as compared to no regulation (M � 1.00, SE � .09). For RL, only a negative focus (M � 0.67, SE � .07), p � .021, 95% CIs [.02, .25], decreased the LPP compared to no regulation (M � 0.81, SE � .08); a positive focus (M � 0.75, SE � .08) had no impact on LPP areas, p � .47, 95% CIs [�.10, .22]. For DW, a negative focus (M � 0.75, SE �

Figure 5. Group averaged waveforms at Pz for (a) RL and (c) DW, as well as effects of instructed regulation on LPP areas for (b) RL and (d) DW. A negative focus was effective at reducing the LPP for both gamble outcomes; a positive focus was only marginally effective at reducing the LPP for just DW. � p � .05.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

9REGULATION OF AMBIVALENCE

.07), p � .045, 95% CIs [.003, .26], decreased the LPP compared to the no regulation condition (M � 0.88, SE � .07); a positive focus (M � 0.79, SE � .06), one-tailed p � .07, 95% CIs [�.03, .26], marginally decreased the LPP. For OW, both a negative focus (M � 0.84, SE � .07), p � .002, 95% CIs [.09, .35], and a positive focus (M � 0.83, SE � .07), p � .002, 95% CIs [.09, .37], decreased the LPP compared to no regulation (M � 1.06, SE � .10).

In sum, our hypotheses regarding both ERP components of interest were supported. First, the FRN was larger to negative versus positive comparisons, consistent with our prediction that a negative comparison (e.g., discovering that one had lost the greater or won the lesser of two amounts) functioned as negative feedback. Second, the LPP was reduced under instructed regulation versus during no regulation, consistent with behavioral results indicating that participants were successful at reducing their emotional re- sponses. There were, however, asymmetries in the efficacy of regulation, such that a negative focus was effective at reducing LPP areas for both RL and DW; whereas a positive focus was only marginally effective at reducing LPP areas for DW and had no impact on LPP areas for RL.

Discussion

First, we examined whether instructed regulation could reduce ambivalence experienced in response to mixed gamble outcomes. Relieving losses and disappointing wins reliably elicited ambiva- lence in our participants. More importantly, instructions to focus on either the positive or negative aspects of those mixed gamble outcomes reduced ambivalence (Question 1a). In other words, when individuals lost the lesser of two amounts, they reported feeling both happy and sad. But when they were instructed to focus on either the negative (i.e., losing money) or the positive (i.e., the fact that they could have lost more money) aspects of that out- come, ambivalence was reduced. The reduction in ambivalence was seen for both relieving losses and disappointing wins, and when focusing on either the negative or positive aspects of the gamble outcome. ERPs showed that, as expected, regulation de- creased the magnitude of the LPP (Question 1b), consistent with past research (Hajcak & Nieuwenhuis, 2006). In sum, ambivalence can be regulated and one potential electrocortical mechanism im- plicated in this reduction is the amplitude of the LPP.

We also sought to examine the psychological mechanisms by which instructed regulation impacts ambivalence (Question 2). Regulation instructions asked participants to focus on one aspect of their emotional experience: either positive or negative. Different theories of emotion predict different patterns of effects. Bipolar models of emotion (e.g., the circumplex; Barrett, 2006; Barrett & Russell, 1999) that rely on a single continuum of valence ranging from unpleasant to pleasant predict that focusing on positive aspects of a mixed gamble outcome would have equal and oppos- ing effects on positivity and negativity: as positivity increases, negativity decreases (and vice versa for a negative focus). Bivari- ate models of emotion (e.g., the ESM; Cacioppo & Berntson, 1994; Cacioppo et al., 1997, 1999; Norris et al., 2010) that allow for independence of positive and negative affect predict that fo- cusing on positive aspects of a mixed gamble outcome may result in either increased positivity, decreased negativity, or some com- bination of the two (and, again, vice versa for a negative focus).

Our results indicate support for both bipolar and bivariate models: First, instructed regulation did have opposing effects on positivity and negativity, supporting a bipolar perspective; but second, those effects were clearly not symmetrical or equivalent, supporting a bivariate perspective. Instructions to focus on positive aspects were more effective in reducing negativity (vs. increasing positiv- ity) to disappointing wins, and instructions to focus on negative aspects were more effective in reducing positivity (vs. increasing negativity) to relieving losses. Thus, focusing on one aspect of a mixed gamble outcome (e.g., negativity) had a stronger impact on reducing the other aspect (positivity) — and this was particularly noticeable on gambles in which the dominant affect (dictated by whether the outcome was a win or a loss) was consistent with regulation instructions.

It is worth noting at this point that our emotion regulation manipulation did not directly instruct participants to decrease their ambivalence; rather, we instructed individuals to focus on one aspect (either positive or negative) of the outcome, with the as- sumption that affecting the component affective responses would ultimately decrease ambivalence. We implemented regulation in this way for the following reasons. First, asking participants to decrease their mixed emotions could drive self-reports of positive and negative affect, as participants would know that we expected them to feel and report both. Second, much as asking individuals to “reappraise” their emotional responses may result in the use of multiple strategies that could differ across individuals and even across trials, instructions to decrease mixed emotions may produce different strategies that would ultimately make results difficult to interpret. Third, and most importantly, we believe that the most effective method of decreasing ambivalence in real life scenarios (e.g., quitting smoking) may be focusing on one aspect of the behavior (the positive consequences of quitting), with the ultimate goal of allowing the pros to eventually outweigh the cons (per TTM; Armitage et al., 2003). Thus, we aimed to implement the most direct, effective, and clear method available to ultimately decrease ambivalence. Finally, the use of instructions to focus on the positive versus negative aspects of a mixed gamble outcome allowed for an examination of multiple asymmetries in affective processing.

Furthermore, our instructions to focus on one aspect (either positive or negative) of the outcome may arguably be more con- sistent with methods of regulation that rely on attentional deploy- ment rather than reappraisal, as we did not explicitly instruct participants to reappraise the gamble outcome in such a way as to feel nothing, feel less negative, or feel more positive; nor to distance themselves from the outcome. In addition to reappraisal, researchers have studied the effects of attentional deployment on emotional responses (cf. Gross, 2002 for a model of emotion regulation strategies). van Reekum and her colleagues (2007) tracked eye movements while participants were instructed to re- appraise their responses to affective images and found that (a) patterns of gaze fixation indicated that individuals paid attention to different aspects of the pictures when instructed to up- and down- regulate their emotional responses as compared to a control con- dition and that (b) gaze fixation patterns in these conditions ac- counted for a significant portion of the variance in brain activation associated with reappraisal, including regions associated with visu- ospatial attention. In a partial replication of this study, Manera, Samson, Pehrs, Lee, and Gross (2014) found that participants spent

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

10 NORRIS AND WU

more time looking at emotional regions in videos of faces when asked to up-regulate their responses and less time when asked to down-regulate their responses, but also found that reappraisal and attentional deployment are “. . . distinct components that uniquely influence negative emotions” (p. 833). Finally, in the absence of any regulation instructions, Dunning and Hajcak (2009) directed participants’ attention to either low- or high-arousing regions of affective pictures and found that reduced LPP amplitudes when attention was deployed to low-arousing areas. These few examples demonstrate that attentional deployment may be used spontane- ously during reappraisal, may contribute to effective reappraisal, and — even in the absence of regulation instructions — may affect emotional responses. Given that we did not directly instruct our participants to manipulate or control their emotional responses to gamble outcomes, but rather asked them to focus on one aspect of each outcome, our manipulation is more similar to methods of attentional deployment than to cognitive reappraisal. Thus, simply changing one’s attentional focus when confronting a complex, mixed emotional stimulus may be effective in reducing ambiva- lence.

Although in the current study we have shown that focusing on one or the other aspect of a mixed emotional response is effective in reducing experienced ambivalence, the effect of cognitive re- appraisal on ambivalence remains unknown. Traditional reap- praisal instructions might ask participants to decrease their ambiv- alence, to distance themselves from the stimulus, or to try to feel nothing. Each set of instructions might be effective in decreasing ambivalence, but their impact on the underlying mechanisms (i.e., positive and negative affect) would be expected to greatly differ. Given that we did not directly ask participants to report on their ambivalence (i.e., using a measure of subjective ambivalence; “How mixed do you feel about this gamble outcome?”), a reduc- tion in ambivalence would be observed via reductions in positive and/or negative affect. How individuals choose to regulate their mixed emotional responses would therefore be informative regard- ing the relative ease or difficulty of regulating positivity versus negativity, as well as shedding light on individual differences in the recruitment of these processes. Even though we have demon- strated that instructed attentional deployment does impact negative affect, positive affect, and experienced ambivalence (i.e., their combination), many questions remain regarding the underlying mechanisms or the effects of traditional cognitive reappraisal on the regulation of ambivalence. We look forward to exploring these questions in future research and further elucidating the regulation of complex, mixed emotional states.

Asymmetries in Instructed Regulation of Mixed Gamble Outcomes

We investigated asymmetries in the efficacy of regulating re- sponses to wins versus losses, changing positive versus negative affect, and focusing on the positive versus negative aspects of a gamble outcome (Question 3). Across all of our analyses, it is clear that focusing on the positive aspects of a relieving loss is simply not as effective as other forms of regulation: Ambivalence was not as reduced (Figure 3a), positivity and negativity were not as impacted (Figure 4a, 4b), and the magnitude of the LPP was not as decreased (Figure 5a). This pattern could be due to an asymmetry in regulation efficacy (i.e., a positive focus is not as effective as a

negative focus), an asymmetry to regulating losses versus wins (i.e., losses are harder to regulate than wins), or some combination of the two. Addressing these possibilities, our results indicate that a positive focus is at least moderately effective in decreasing ambivalence (via a reduction in negativity and in the LPP) for disappointing wins. Thus, it appears likely that responses to losses are simply more difficult to regulate than responses to wins; or, given our discussion above regarding attentional deployment, it may be more difficult to shift attention away from the negative aspect of a stimulus. Even so, our data suggest that both positive and negative affect can be regulated to equivalent degrees and that both a positive and a negative focus may be effective in reducing ambivalence, depending on the context. In sum, there is evidence for a negativity bias in the regulation of responses to gamble outcomes, such that responses to losses are more difficult to change than responses to wins; and although both negative and positive regulation of negativity and positivity seem to be effec- tive, a negative focus may be more so.

The FRN and Negative Comparisons

There is one additional ERP finding that is worth noting. We hypothesized that the FRN, a component sensitive to both bad outcomes (e.g., losses; Hajcak et al., 2006) and to unexpected negative outcomes (Holroyd et al., 2003), would be larger for negative comparisons (i.e., OL, DW). This hypothesis was based on the fact that participants discovered early in the trial whether they had won or lost, and only later whether the final outcome was for better or worse. As predicted, we observed larger FRNs to negative comparisons, which is consistent with the idea that indi- viduals anticipate or hope for the best possible outcome, and a negative comparison is therefore both unexpected and unpleasant. However, it is worth noting that the observed pattern of larger FRNs to negative versus positive comparisons is also consistent with predictions made from rational choice theory (cf. Walsh & Anderson, 2012 for an excellent review). Specifically, in the current study, negative comparisons (i.e., outright losses and dis- appointing wins) always had a lower expected value (EV; which is calculated as the sum of the probabilities of each possible out- come: e.g., for a win trial with possible $5 and $15 outcomes, the EV � $5 � 50% $15 � 50% � $10) than positive comparisons did. Rational choice theory would predict that the FRN would be larger for outcomes smaller than the EV (i.e., negative compari- sons). Our results are clearly also consistent with this hypothesis, and future studies should consider addressing these competing perspectives: Are FRNs larger to negative comparisons because the obtained outcome is smaller than the expected value of the gamble, or because individuals are, indeed, hoping for or antici- pating the best possible outcome? In general, we acknowledge that the field of literature in economic decision-making is extremely relevant for the current study and recognize that these different perspectives may make contradictory predictions regarding our results.

Limitations

There are a number of potential limitations to the current study, including both theoretical interpretations and methodological de- tails. First, as mentioned above, the observed pattern of larger FRN

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

11REGULATION OF AMBIVALENCE

amplitudes to negative versus positive comparisons may be inter- preted in different ways: For example, the FRN may be sensitive to either unexpected negative feedback or to obtained outcomes smaller than the expected value for a given gamble. Notably, the FRN was not sensitive to outcome (i.e., was not larger to losses vs. gains); this provides some evidence to suggest that — at least in our dataset — the FRN does not seem to be solely sensitive to EV (as the EV for losses will always be lower than the EV for wins). However, the question remains open, and future studies should consider these multiple perspectives.

In addition, the interpretation of the pattern of results for the LPP may also be debatable, for two reasons. First, as P300 am- plitudes (a component related to the LPP) are known to habituate over the course of many trials (Polich & Kok, 1995), it is possible that lower amplitudes during regulation blocks (2 and 3) than during the no regulation block (1) could reflect habituation and not regulation. If the LPP does habituate over time, resulting in de- creased amplitudes in later blocks, then we should see an interac- tion between regulation instruction and order of regulation instruc- tion, such that the LPP should be smaller during the last regulation condition (as order was counterbalanced across participants). To examine the LPP data for evidence of this Order � Regulation interaction, we conducted an analysis using 2(order: negative first, positive first) as an additional factor. None of the interactions involving the order and regulation factors were significant: Most critically, results for the Order � Regulation interaction were: F(2, 40) � 0.79, p � .46. These results suggest that the decreased LPPs observed during regulation were not due to habituation over time. Second, one could argue that a negative focus should only up- regulate negative affect, therefore increasing (not decreasing) the magnitude of the LPP. Emotional responses to the gamble out- comes argue against this interpretation, however, and for our conclusion that decreased LPPs did, in fact, reflect regulation. Specifically, a negative focus had an even stronger effect on down-regulating positive affect than it did on up-regulating neg- ative affect, and these effects together ultimately decreased am- bivalence. Thus, we argue that the overall summed effect of a negative focus was to decrease emotional responses, given a large decrease in positive affect, a small increase in negative affect, and ultimately, a decrease in ambivalence (and the opposite is true for a positive focus).

Another potential limitation that concerns both the FRN and the LPP is that, although we collected ERPs from 64 sites on the scalp, we only report results from one (i.e., Fz for the FRN and Pz for the LPP). We acknowledge that there is an ongoing argument in the field of electrophysiology regarding the appropriate method of reporting ERP results: Some researchers focus on a priori compo- nents and sites of interest (as in the current study), others average across sites to examine activity at regions and include additional factors (e.g., anterior/central/posterior and left/midline/right) in analyses, yet others include multiple individual sites (e.g., midline; Fz, Cz, Pz, Oz). Arguments can be made to support each of these approaches; we chose to limit analyses to sites of primary interest, where the FRN (Fz) and LPP (Pz) are maximal, to prevent Type I error (Luck, 2012). The primary limitation of this approach, of course, is that we have not examined the spatial distributions of the FRN and LPP across the scalp. Future studies may want to con- sider this approach.

Conclusion

The current study is unique in that it explores both positive and negative regulation of both positive and negative affect (as well as ambivalence), whereas most studies of emotion regulation concern solely the down-regulation (i.e., reduction) of negative affect. More often than not, emotion regulation is adaptive when used to down-regulate negativity, and this form of regulation is likely the most commonly used. Ambivalence, however, is a complicated and unique emotional state that may benefit from both up- and down-regulation of either positive or negative affect. First, ambiv- alence provides little to no behavioral guidance or motivation, as individuals feel both positive (appetitive) and negative (avoidant) at the same time. Second, ambivalence can be experienced toward things that are overall either good for us (e.g., establishing a regular exercise regime) or bad for us (e.g., smoking cigarettes). It is therefore to our benefit to understand how focusing on the positive or negative aspects of an outcome affect our ambivalence, our positive and negative affect, and ultimately our behavior. Take, for example, receiving a revise-and-resubmit review on the initial submission of a article. Most academics would see this as a primarily positive event but would experience some ambiva- lence— clearly, the editor and reviewers found enough substance worthy to consider a revision, but critical comments can be plen- tiful, time-consuming to address, and disheartening. Models of ambivalence would predict that this combination of positivity and negativity might stall (i.e., produce equal approach and avoidance motivation) revisions necessary for publication. Our results ad- dress a potential solution to this problem: for an overall positive event (such as a revise-and-resubmit decision), the most effective manner of reducing ambivalence (and ultimately guiding behavior toward approaching the task of revising) is to focus on the positive, which ultimately should decrease negative affect (and avoidant behavior) while increasing positive affect (and approach behavior). Such an approach should decrease ambivalence and focus moti- vation on facing the reviews. A similar perspective may be helpful when encountering decisions regarding our health. Our results suggest that ambivalence is not uniform across outcomes, and considering the overall affective context, the method of regulation and the mechanisms behind it may ultimately make our regulatory efforts more effective, motivated, and productive.

References

Armitage, C. J., Povey, R., & Arden, M. A. (2003). Evidence for discon- tinuity patterns across the stages of change: A role for attitudinal ambivalence. Psychology & Health, 18, 373–386. http://dx.doi.org/10 .1080/0887044031000066553

Barrett, L. F. (2006). Solving the emotion paradox: Categorization and the experience of emotion. Personality and Social Psychology Review, 10, 20 – 46. http://dx.doi.org/10.1207/s15327957pspr1001_2

Barrett, L. F., & Bliss-Moreau, E. (2009). Affect as a psychological primitive. Advances in experimental social psychology, 41, 167–218. http://dx.doi.org/10.1016/S0065-2601(08)00404-8

Barrett, L. F., & Russell, J. A. (1999). The structure of current affect: Controversies and emerging consensus. Current Directions in Psycho- logical Science, 8, 10 –14. http://dx.doi.org/10.1111/1467-8721.00003

Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohs, K. D. (2001). Bad is stronger than good. Review of General Psychology, 5, 323–370. http://dx.doi.org/10.1037/1089-2680.5.4.323

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

12 NORRIS AND WU

Berrios, R., Totterdell, P., & Kellett, S. (2015). Eliciting mixed emotions: a meta-analysis comparing models, types, and measures. Frontiers in Psychology, 6, 428. http://dx.doi.org/10.3389/fpsyg.2015.00428

Buhle, J. T., Silvers, J. A., Wager, T. D., Lopez, R., Onyemekwu, C., Kober, H., . . . Ochsner, K. N. (2014). Cognitive reappraisal of emotion: A meta-analysis of human neuroimaging studies. Cerebral Cortex, 24, 2981–2990. http://dx.doi.org/10.1093/cercor/bht154

Cacioppo, J. T., & Berntson, G. G. (1994). Relationship between attitudes and evaluative space: A critical review, with emphasis on the separa- bility of positive and negative substrates. Psychological Bulletin, 115, 401– 423. http://dx.doi.org/10.1037/0033-2909.115.3.401

Cacioppo, J. T., Gardner, W. L., & Berntson, G. G. (1997). Beyond bipolar conceptualizations and measures: The case of attitudes and evaluative space. Personality and Social Psychology Review, 1, 3–25. http://dx.doi .org/10.1207/s15327957pspr0101_2

Cacioppo, J. T., Gardner, W. L., & Berntson, G. G. (1999). The affect system has parallel and integrative processing components: Form fol- lows function. Journal of Personality and Social Psychology, 76, 839 – 855. http://dx.doi.org/10.1037/0022-3514.76.5.839

Delorme, A., & Makeig, S. (2004). EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. Journal of Neuroscience Methods, 134, 9 –21.

Doré, B. P., Boccagno, C., Burr, D., Hubbard, A., Long, K., Weber, J., . . . Ochsner, K. N. (2017). Finding positive meaning in negative experi- ences engages ventral striatal and ventromedial prefrontal regions asso- ciated with reward valuation. Journal of Cognitive Neuroscience, 29, 235–244. http://dx.doi.org/10.1162/jocn_a_01041

Dunning, J. P., & Hajcak, G. (2009). See no evil: Directing visual attention within unpleasant images modulates the electrocortical response. Psy- chophysiology, 46, 28 –33. http://dx.doi.org/10.1111/j.1469-8986.2008 .00723.x

Eippert, F., Veit, R., Weiskopf, N., Erb, M., Birbaumer, N., & Anders, S. (2007). Regulation of emotional responses elicited by threat-related stimuli. Human Brain Mapping, 28, 409 – 423. http://dx.doi.org/10.1002/ hbm.20291

Foti, D., & Hajcak, G. (2008). Deconstructing reappraisal: Descriptions preceding arousing pictures modulate the subsequent neural response. Journal of Cognitive Neuroscience, 20, 977–988. http://dx.doi.org/10 .1162/jocn.2008.20066

Gross, J. J. (1998). Antecedent-and response-focused emotion regulation: Divergent consequences for experience, expression, and physiology. Journal of Personality and Social Psychology, 74, 224 –237. http://dx .doi.org/10.1037/0022-3514.74.1.224

Gross, J. J. (2002). Emotion regulation: Affective, cognitive, and social consequences. Psychophysiology, 39, 281–291. http://dx.doi.org/10 .1017/S0048577201393198

Hajcak, G., Dunning, J. P., & Foti, D. (2009). Motivated and controlled attention to emotion: Time-course of the late positive potential. Clinical Neurophysiology, 120, 505–510. http://dx.doi.org/10.1016/j.clinph.2008 .11.028

Hajcak, G., MacNamara, A., & Olvet, D. M. (2010). Event-related poten- tials, emotion, and emotion regulation: An integrative review. Develop- mental Neuropsychology, 35, 129 –155. http://dx.doi.org/10.1080/ 87565640903526504

Hajcak, G., Moser, J. S., Holroyd, C. B., & Simons, R. F. (2006). The feedback-related negativity reflects the binary evaluation of good versus bad outcomes. Biological Psychology, 71, 148 –154. http://dx.doi.org/10 .1016/j.biopsycho.2005.04.001

Hajcak, G., & Nieuwenhuis, S. (2006). Reappraisal modulates the electro- cortical response to unpleasant pictures. Cognitive, Affective & Behav- ioral Neuroscience, 6, 291–297. http://dx.doi.org/10.3758/CABN.6.4 .291

Hajcak, G., & Olvet, D. M. (2008). The persistence of attention to emotion: Brain potentials during and after picture presentation. Emotion, 8, 250 – 255. http://dx.doi.org/10.1037/1528-3542.8.2.250

Holroyd, C. B., Nieuwenhuis, S., Yeung, N., & Cohen, J. D. (2003). Errors in reward prediction are reflected in the event-related brain potential. Neuroreport, 14, 2481–2484. http://dx.doi.org/10.1097/00001756- 200312190-00037

Hunter, P. G., Schellenberg, E. G., & Schimmack, U. (2010). Feelings and perceptions of happiness and sadness induced by music: Similarities, differences, and mixed emotions. Psychology of Aesthetics, Creativity, and the Arts, 4, 47–56. http://dx.doi.org/10.1037/a0016873

Ito, T. A., Larsen, J. T., Smith, N. K., & Cacioppo, J. T. (1998). Negative information weighs more heavily on the brain: The negativity bias in evaluative categorizations. Journal of Personality and Social Psychol- ogy, 75, 887–900. http://dx.doi.org/10.1037/0022-3514.75.4.887

Kaplan, K. J. (1972). On the ambivalence-indifference problem in attitude theory and measurement: A suggested modification of the semantic differential technique. Psychological Bulletin, 77, 361–372. http://dx.doi .org/10.1037/h0032590

Koenigsberg, H. W., Fan, J., Ochsner, K. N., Liu, X., Guise, K., Pizzarello, S., . . . New, A. (2010). Neural correlates of using distancing to regulate emotional responses to social situations. Neuropsychologia, 48, 1813– 1822. http://dx.doi.org/10.1016/j.neuropsychologia.2010.03.002

LaFarge, V. V. S. (1994). The ambivalence of departing employees: Reactions of involuntary and voluntary exiters. The Journal of Applied Behavioral Science, 30, 175–197. http://dx.doi.org/10.1177/0021886394302003

Larsen, J. T., Coles, N. A., & Jordan, D. K. (2017). Varieties of mixed emotional experience. Current Opinion in Behavioral Sciences, 15, 72–76. http://dx.doi.org/10.1016/j.cobeha.2017.05.021

Larsen, J. T., & McGraw, A. P. (2011). Further evidence for mixed emotions. Journal of Personality and Social Psychology, 100, 1095– 1110. http://dx.doi.org/10.1037/a0021846

Larsen, J. T., McGraw, A. P., & Cacioppo, J. T. (2001). Can people feel happy and sad at the same time? Journal of Personality and Social Psychology, 81, 684 – 696. http://dx.doi.org/10.1037/0022-3514.81.4 .684

Larsen, J. T., McGraw, P. A., Mellers, B. A., & Cacioppo, J. T. (2004). The agony of victory and thrill of defeat: Mixed emotional reactions to disappointing wins and relieving losses. Psychological Science, 15, 325–330. http://dx.doi.org/10.1111/j.0956-7976.2004.00677.x

Larsen, J. T., Norris, C. J., McGraw, A. P., Hawkley, L. C., & Cacioppo, J. T. (2009). The evaluative space grid: A single-item measure of positivity and negativity. Cognition and Emotion, 23, 453– 480. http:// dx.doi.org/10.1080/02699930801994054

Larsen, J. T., & Stastny, B. J. (2011). It’s a bittersweet symphony: Simul- taneously mixed emotional responses to music with conflicting cues. Emotion, 11, 1469 –1473. http://dx.doi.org/10.1037/a0024081

Larsen, J. T., To, Y. M., & Fireman, G. (2007). Children’s understanding and experience of mixed emotions. Psychological Science, 18, 186 –191. http://dx.doi.org/10.1111/j.1467-9280.2007.01870.x

Lipkus, I. M., Green, J. D., Feaganes, J. R., & Sedikides, C. (2001). The relationship between attitudinal ambivalence and desire to quit smoking among college smokers. Journal of Applied Social Psychology, 31, 113–133. http://dx.doi.org/10.1111/j.1559-1816.2001.tb02485.x

Luck, S. J. (2012). Event-related potentials. In H. Cooper, P. M. Camic, D. L. Long, A. T. Panter, D. Rindskopf, and K. J. Sher (Eds.), APA handbook of research methods in psychology: Vol. 1, Foundations, planning, measures, and psychometrics (pp. 523–546). Washington, DC: American Psychological Association. http://dx.doi.org/10.1037/ 13619-028

Manera, V., Samson, A. C., Pehrs, C., Lee, I. A., & Gross, J. J. (2014). The eyes have it: The role of attention in cognitive reappraisal of social stimuli. Emotion, 14, 833. http://dx.doi.org/10.1037/a0037350

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

13REGULATION OF AMBIVALENCE

McRae, K., Ciesielski, B., & Gross, J. J. (2012). Unpacking cognitive reappraisal: Goals, tactics, and outcomes. Emotion, 12, 250 –255. http:// dx.doi.org/10.1037/a0026351

McRae, K., Hughes, B., Chopra, S., Gabrieli, J. D., Gross, J. J., & Ochsner, K. N. (2010). The neural bases of distraction and reappraisal. Journal of Cognitive Neuroscience, 22, 248 –262. http://dx.doi.org/10.1162/jocn .2009.21243

Moser, J. S., Hajcak, G., Bukay, E., & Simons, R. F. (2006). Intentional modulation of emotional responding to unpleasant pictures: An ERP study. Psychophysiology, 43, 292–296. http://dx.doi.org/10.1111/j.1469- 8986.2006.00402.x

Norris, C. J., Gollan, J., Berntson, G. G., & Cacioppo, J. T. (2010). The current status of research on the structure of evaluative space. Biological Psychology, 84, 422– 436. http://dx.doi.org/10.1016/j.biopsycho.2010 .03.011

Ochsner, K. N., Bunge, S. A., Gross, J. J., & Gabrieli, J. D. (2002). Rethinking feelings: An FMRI study of the cognitive regulation of emotion. Journal of Cognitive Neuroscience, 14, 1215–1229. http://dx .doi.org/10.4324/9780203496190

Ochsner, K. N., Ray, R. D., Cooper, J. C., Robertson, E. R., Chopra, S., Gabrieli, J. D., & Gross, J. J. (2004). For better or for worse: Neural systems supporting the cognitive down-and up-regulation of negative emotion. NeuroImage, 23, 483– 499. http://dx.doi.org/10.1016/j .neuroimage.2004.06.030

Ochsner, K. N., Silvers, J. A., & Buhle, J. T. (2012). Functional imaging studies of emotion regulation: A synthetic review and evolving model of the cognitive control of emotion. Annals of the New York Academy of Sciences, 1251(1), E1–E24. http://dx.doi.org/10.1111/j.1749-6632.2012 .06751.x

Premberg, Å., Hellström, A. L., & Berg, M. (2008). Experiences of the first year as father. Scandinavian Journal of Caring Sciences, 22, 56 – 63. http://dx.doi.org/10.1111/j.1471-6712.2007.00584.x

Polich, J., & Kok, A. (1995). Cognitive and biological determinants of P300: an integrative review. Biological Psychology, 41, 103–146. http:// dx.doi.org/10.1016/0301-0511(95)05130-9

Prochaska, J. O., & DiClemente, C. C. (1982). Transtheoretical therapy: Toward a more integrative model of change. Psychotherapy: Theory, Research, & Practice, 19, 276 –288. http://dx.doi.org/10.1037/h0088437

Ray, R. D., McRae, K., Ochsner, K. N., & Gross, J. J. (2010). Cognitive reappraisal of negative affect: Converging evidence from EMG and self-report. Emotion, 10, 587–592. http://dx.doi.org/10.1037/a0019015

Rozin, P., & Royzman, E. B. (2001). Negativity bias, negativity domi- nance, and contagion. Personality and Social Psychology Review, 5, 296 –320. http://dx.doi.org/10.1207/S15327957PSPR0504_2

Russell, J. A., & Carroll, J. M. (1999). On the bipolarity of positive and negative affect. Psychological Bulletin, 125, 3–30. http://dx.doi.org/10 .1037/0033-2909.125.1.3

Schimmack, U. (2001). Pleasure, displeasure, and mixed feelings: Are semantic opposites mutually exclusive? Cognition and Emotion, 15, 81–97. http://dx.doi.org/10.1080/0269993004200123

Schimmack, U. (2005). Response latencies of pleasure and displeasure ratings: Further evidence for mixed feelings. Cognition and Emotion, 19, 671– 691. http://dx.doi.org/10.1080/02699930541000020

Schupp, H. T., Cuthbert, B. N., Bradley, M. M., Cacioppo, J. T., Ito, T., & Lang, P. J. (2000). Affective picture processing: The late positive potential is modulated by motivational relevance. Psychophysiology, 37, 257–261. http://dx.doi.org/10.1017/S0048577200001530

Taylor, S. E. (1991). Asymmetrical effects of positive and negative events: The mobilization-minimization hypothesis. Psychological Bulletin, 110, 67– 85. http://dx.doi.org/10.1037/0033-2909.110.1.67

Titchener, E. B. (1908). The tridimensional theory of feeling. The Ameri- can Journal of Psychology, 19, 213–231. http://dx.doi.org/10.1037/ 10867-004

Tugade, M. M., & Fredrickson, B. L. (2004). Resilient individuals use positive emotions to bounce back from negative emotional experiences. Journal of Personality and Social Psychology, 86, 320 –333. http://dx .doi.org/10.1037/0022-3514.86.2.320

van Reekum, C. M., Johnstone, T., Urry, H. L., Thurow, M. E., Schaefer, H. S., Alexander, A. L., & Davidson, R. J. (2007). Gaze fixations predict brain activation during the voluntary regulation of picture-induced neg- ative affect. NeuroImage, 36, 1041–1055. http://dx.doi.org/10.1016/j .neuroimage.2007.03.052

Walsh, M. M., & Anderson, J. R. (2012). Learning from experience: event-related potential correlates of reward processing, neural adapta- tion, and behavioral choice. Neuroscience & Biobehavioral Reviews, 36, 1870 –1884.

Wundt, W. M. (1896). Lectures on human and animal psychology. New York, NY: George Allen & Company.

Received June 20, 2018 Revision received August 20, 2019

Accepted November 17, 2019 �

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

14 NORRIS AND WU