Seeing the Whole Picture? Avoided Negative Affect and Processing of Others’ Suffering

Abstract

Noticing someone’s pain is the first step to a compassionate response. While past research suggests that the degree to which people want to avoid feeling negative (“avoided negative affect”; ANA) shapes how people respond to someone’s suffering, the present research investigates whether ANA also predicts how people process others’ suffering. In two studies, using complex photographs containing negative aspects (i.e., suffering), we found that the higher people’s ANA, the fewer details of negative aspects they correctly recognized, and the fewer negative words they used in their image descriptions. However, when asked to process negative content, the higher people’s ANA, the more negatively they rated that content. In Study 3, we report cultural differences in people’s sensitivity to notice suffering in an ambiguous image. ANA mediated these cultural differences. Implications for research on compassion are discussed.

Keywords

emotion avoided negative affect information processing suffering culture

A compassionate response consists of two steps: first, the sensitivity to another’s suffering and second, the motivation to alleviate that suffering (Goetz et al., 2010). Previous research found the degree to which people want to avoid feeling negative (“avoided negative affect”; ANA) shapes the second step of a compassionate response: The more people want to avoid feeling negative, the less comfortable they are sending sympathy cards that focus on the negative (e.g., “Words will not lighten a heavy heart”) versus the positive (e.g., “Memories will bring comfort”) (Koopmann-Holm & Tsai, 2014). The present article examines whether ANA also predicts the first step of a compassionate response, namely noticing and processing others’ suffering, and how negatively that suffering is perceived.

While past research has examined how people process negative information that is threatening to themselves (e.g., snakes), our work focuses on processing negative information related to the suffering of others, which is important given the large number of people in need of society’s compassion. We predict high ANA will motivate people to process negative aspects of scenes less as people want to avoid feeling negative. However, when people are asked to process negative aspects, we predict the more people want to avoid feeling negative, the more negatively they will judge those aspects (because they are forced to look at what they want to avoid). Specifically, we hypothesize the more people want to avoid feeling negative, the more negatively they would rate negative aspects of scenes, the fewer negative details they would remember, the fewer negative words they would use to describe scenes, and the less likely they would notice something negative in an ambiguous image.

We tested these hypotheses in three studies. Before describing the studies, we propose how ANA might predict processing of others’ suffering above and beyond how people actually feel during a typical week. Because previous work (Koopmann-Holm & Tsai, 2014) has demonstrated cultural differences in ANA, we conclude by hypothesizing cultural differences in the processing of others’ suffering, which, we predict, can be explained by ANA.

Previous Research

How we feel in the moment shapes not only our judgments (Schwartz & Clore, 1983, 2007), but also what we focus on, perceive, encode, and recall (e.g., Eich & Forgas, 2003; Siegel et al., 2018; Tamir & Robinson, 2007). People are better at processing information that is congruent with how they are currently feeling. For example, happy participants are faster at identifying happy words as words rather than nonwords, and sad participants make quicker decisions about sad words (Niedenthal & Setterlund, 1994).

Furthermore, people focus on affective stimuli before they focus on neutral stimuli (Calvo & Lang, 2004). When comparing negative with positive stimuli, some studies find people are more likely to recall negative than positive stimuli (e.g., Keightley et al., 2011), but other studies find the opposite (e.g., Matt et al., 1992). Kensinger and Schacter (2016) propose whether people remember positive or negative information better might be shaped by their goals (Lazarus, 1991; LeDoux, 1996). For example, older adults, who seek positive goal states, pay more attention to and better remember positive stimuli compared with negative ones (Mather & Carstensen, 2005). We turn to these affective goals now.

Beyond “Actual Affect”

Previous research has focused on how “actual” affective states influence judgments and information processing. However, there is more to people’s emotional lives than actual affect (e.g., Harmon-Jones et al., 2011; Larsen, 2000). Affect Valuation Theory (AVT; Tsai et al., 2006) posits that in addition to “actual affect” (the affective states people are actually feeling), “ideal affect” (the affective states people ideally want to feel) and “avoided affect” (the affective states people want to avoid feeling) are important predictors of behavior (e.g., Koopmann-Holm & Tsai, 2014; Tsai, 2007; Tsai et al., 2006). For example, above and beyond how people actually feel during a typical week (“global actual affect”), the way people would ideally like to feel predicts what type of vacation and music they prefer (Tsai, 2007; Tsai et al., 2015). Furthermore, the more people value exciting states, the more they prefer a physician who promotes a dynamic lifestyle, whereas the more people value calm states, the more they prefer a physician who promotes a relaxed lifestyle (Sims et al., 2014; Sims & Tsai, 2015). Importantly, people’s global actual affect did not predict their preferences for physicians, suggesting people respond more positively to physicians whose affective focus matches their ideal, not actual affect.

Similarly, ANA shapes people’s responses to another’s suffering above and beyond how negatively they actually feel (Koopmann-Holm & Tsai, 2014). The higher people’s ANA, the less they focus on the negative versus positive when responding to others’ suffering (Koopmann-Holm & Tsai, 2014).

Whereas past research has examined how different views of negative emotions such as social pressure not to feel negative (Bastian et al., 2012, 2017), beliefs about negative emotions (Dennis & Halberstadt, 2013), and disliking negative emotions (Harmon-Jones et al., 2011) are related to negativity ratings and information processing, we focus on how much people want to avoid feeling negative.

The importance of an “ideal affect match.”

Previous research suggests if activities or others’ facial expressions match people’s own ideal affect (“ideal affect match”; Chim et al., 2018), people respond more positively. For example, the more people ideally wanted to feel calm, the more they enjoyed calming versus exciting rides in amusement parks (Chim et al., 2018). These findings suggest an ideal affect match influences judgments as people derive more enjoyment from activities that match their ideal affect. Furthermore, an ideal affect match can also enhance information recall. Asian Americans, who wanted to feel calm more than did European Americans, recalled health information delivered with excitement less well than did European Americans (Sims et al., 2018). What happens if people who strongly want to avoid feeling negative are faced with negative information such as others’ suffering?

The Present Research: “Avoided Affect Mismatch”

We hypothesize ANA will predict both judgment (e.g., negativity rating) and processing (e.g., perception and recall) of negative information. Whereas people respond more positively and show enhanced recall in case of an ideal affect match (when they get closer to what they want to approach), we predict people will respond more negatively and show reduced recall in case of an avoided affect mismatch (when people get closer to what they want to avoid). Past research (e.g., Koopmann-Holm & Tsai, 2014; Sims et al., 2018) demonstrates ideal and avoided affect predict judgments and behavior above and beyond global actual affect. Ideal and avoided affect are what people use to evaluate their actual affect and predict what people do to feel good and to avoid feeling bad (Tsai et al., 2006). Based on this, we predict ANA will be associated with judgment and processing of suffering even after controlling for global actual negative affect.

Valence Judgment of Negative Information

When presented with negative information, the more people want to avoid feeling negative, the greater their avoided affect mismatch: People get closer to what they want to stay away from or avoid. While an ideal affect match is associated with responding more positively to events that get people closer to what they want to approach, we predict an avoided affect mismatch will be associated with responding more negatively to what people want to avoid. Specifically, we predict when people are presented with negative aspects in images, the more they want to avoid feeling negative, the more negatively they will rate these aspects. In fact, previous data suggest ANA might be associated with feeling more negative when exposed to negative stimuli (Koopmann-Holm & Tsai, 2014). It is possible that when people are not successful at avoiding stimuli that elicit what they want to avoid, they feel bad (Carver & Scheier, 1998). Alternatively, negative stimuli might be more negative for people with higher compared to those with lower ANA. We aim to test the latter possibility in the present article.

Processing of Negative Information

If negative information is more negative for people with high ANA, why would they process that information in the first place? One way to avoid experiencing this elevated negativity would be to not process negative information deeply. Hence, we hypothesize ANA predicts how people process negative information such as suffering. This is in line with the idea that in case of a discrepancy between someone’s current mood and their desired state, people can regulate their mood by focusing (or not focusing) on affect-relevant stimuli in their environment (Larsen, 2000). While an ideal affect match is associated with deeper processing, we hypothesize an avoided affect mismatch will be related to less processing such that the higher people’s ANA, the fewer negative details they will remember, and the fewer negative words they will use to describe images containing negative aspects.

Our hypotheses are consistent with previous research showing that motivational states like hunger can impact how people attend to visual stimuli; for example, hungry people pay more attention to food-related stimuli (Mogg et al., 1998) and remember them better (Morris & Dolan, 2001). They want to approach food and show enhanced attention and memory for what they want to get closer to. We argue ANA is a motivational state just like hunger. However, people who want to avoid feeling negative are motivated to avoid stimuli that would lead to negative feelings. Therefore, they should show reduced attention and memory for what they want to get away from.

The Role of Culture

While individual differences related to temperamental factors exist in people’s desired affective states (e.g., Rusting & Larsen, 1995), past research has also reported cultural differences in ideal and avoided affect (Koopmann-Holm & Tsai, 2014; Tsai et al., 2006). In fact, AVT proposes ideal and avoided affect are shaped by culture more than is actual affect (Koopmann-Holm & Tsai, 2014; Tsai et al., 2006). Empirical findings support this premise. For example, European Americans want to avoid feeling negative affect more than Germans do, while the cultural differences were less pronounced for actual negative affect (Koopmann-Holm & Tsai, 2014). These cultural differences in ANA are associated with how people respond to others’ suffering. European Americans feel less comfortable sending sympathy cards that focus on the negative versus the positive partly because they want to avoid feeling negative more than Germans do (Koopmann-Holm & Tsai, 2014). Building on this work, we predict there will be cultural differences in how people process others’ suffering and that ANA can explain these differences.

Overview of Studies

We tested our predictions (summarized in Figure 1) in three studies. Our work focuses on how ANA is related to noticing other people’s suffering rather than negative information important for one’s own survival. Because negative stimuli used in past research mainly depict information that is threatening to participants themselves, we developed new stimuli (i.e., complex images) in which the portrayal of other people’s suffering is embedded within other emotional and neutral information (see the supplemental appendix for stimuli, methods, and results of our pretest). In all studies, we operationalized others’ suffering as negatively rated aspects of images. We pretested these images to establish which aspects are perceived to be negative by most people (see the supplemental appendix). In Study 1, using new participants, we assessed whether ANA predicts the negativity ratings of the negative aspects of the images and the number of negative details participants correctly recognize seeing. In Study 2, we examined whether ANA predicts how participants describe what they remember seeing in these pretested images using a free-recall paradigm. Study 3 examines whether the previously reported cultural differences in ANA between Americans and Germans (Koopmann-Holm & Tsai, 2014) are associated with cultural differences in perceiving suffering in an ambiguous image. We also tested whether these cultural differences in perceiving suffering would be mediated by ANA. The datasets for all three studies are available via the following link: https://doi.org/10.6084/m9.figshare.11340863.v1.

Figure 1.

Overview of studies and predictions.

Study 1: Does ANA Predict Negativity Ratings of Negative Aspects and Recognition of Negative Details?

Based on our avoided affect mismatch approach, we predicted the higher people’s ANA, the more negatively they will rate negative aspects of images (Hypothesis 1) and the fewer details of negative aspects they will correctly recognize as having seen before (Hypothesis 2).

Method

Participants

Sixty U.S. undergraduate students (79.31% female; 41.67% Caucasian, 20.00% Asian American, 10.00% Hispanic, 28.33% other; mean age = 18.57 years; SD = 0.86) participated in an online “Emotion and Information Processing” study in exchange for course credit. Sample size was determined using G*Power software: Ensuring a power of .80 to detect medium size effects (f² = .25) using regression with three to six predictors and an alpha error probability of .05, we needed 48 to 62 participants.

Materials and procedure

Affect Valuation Index (AVI)

To assess actual, ideal, and avoided affect, participants first completed the extended version (as described in Koopmann-Holm & Tsai, 2014) of the AVI (Tsai et al., 2006). Participants rated how often they actually felt, ideally wanted to feel, and how often they wanted to avoid feeling 37 affective states during a typical week on a 5-point scale ranging from 1 (“never”) to 5 (“all the time”). Given the lack of order effects in previous research, all participants received the same order (actual, ideal, avoided affect). For the purpose of this study and in line with Koopmann-Holm and Tsai (2014), we created aggregate scores for negative and positive affect. For negative affect, we combined sad, unhappy, lonely, fearful, hostile, nervous, dull, sleepy, and sluggish (Cronbach’s alphas were .77 for actual affect and .83 for avoided affect). For positive affect, we combined enthusiastic, excited, elated, happy, content, satisfied, calm, relaxed, and serene (Cronbach’s alphas were .90 for actual affect and .78 for ideal affect).

Complex images

After completing the AVI, participants viewed the six pretested images in one standard order. After viewing an image for 5 s, participants answered the question “What do you remember seeing in the image on the last page?” Participants were presented with a number of possible details, and they selected what they thought they saw. Half of the negative and other details presented were accurate (e.g., man wearing a tan shirt), and the other half were inaccurate (e.g., man wearing a blue shirt). The same procedure was repeated for all images.

In the next block, participants viewed the same six images again. This time, each image stayed on the page while participants were asked to look at the image and indicate how positive/negative the different negative and other aspects (as outlined in Table 1) were to them (3 = extremely positive, 0 = neither positive nor negative, −3 = extremely negative).

Table 1.

Mean Ratings of the Negative and Other Aspects of the Images and Standard Deviations (Study 1).

Aspect of Image	M	SD
Image 1
Red car	−2.29	1.06
Truck	−2.36	1.10
Debris	−1.67	1.24
Fire truck	1.35	1.40
Fireman	1.60	1.27
Pool	0.21	0.79
Background	−0.21	0.91
Barn	0.24	0.84
Image 2
Homeless man	−1.64	1.22
Women	0.36	1.75
People walking	0.05	0.87
Ground	−0.43	0.98
Stairs	0.05	0.90
Building	0.12	0.90
Image 3
Poor man	−2.14	1.14
Sports car	0.38	1.80
People	−0.36	1.07
Ground	−0.19	0.76
Background	−0.24	1.19
Car in background	−0.17	0.75
Image 4
Man wrapped in blanket	−2.14	1.08
People with umbrellas	−0.19	1.15
Background	0.21	1.02
Ground	0.07	0.95
Image 5
Robbery	−2.79	0.83
Background	−1.10	0.99
Wall	−0.38	0.95
Image 6
Poor people	−1.78	1.19
Rich woman	−0.29	1.31
Pole	−0.55	0.78
Ground	−0.40	0.65
Walls	−0.67	0.83
Background	−0.69	0.88

Demographics questionnaire

Participants completed a demographics questionnaire which assessed their gender, age, and ethnicity.

Results

Analysis strategy for all studies

In line with past research on AVT (Tsai et al., 2006), we control for actual negative affect when examining ANA to determine their independent effects. We also initially controlled for ideal positive and actual positive affect to ensure the results are not due to these two types of affect. When ideal and actual positive affect were not significant, we dropped both from the final analyses. We also statistically held constant how much participants remembered overall to examine memory for negativity specifically. The pattern of findings did not change when we excluded these covariates (see Supplementary File “Results Without Covariates”). In all regressions, we entered all variables in one step and report unstandardized B coefficients.

Furthermore, in line with Koopmann-Holm and Tsai (2014), we used mean-deviated scores for all avoided, actual, and ideal affect variables because previous research suggests (a) cultural differences in response style for these variables (e.g., Koopmann-Holm and Tsai, 2014) and (b) that people differ in how much they want to avoid feeling emotions in general (e.g., Maio & Esses, 2001). Hence, we subtracted each participant’s overall mean response to all 37 avoided affect items from the raw score for each avoided affect item (e.g., avoided sad). We followed the same procedure for actual and ideal affect in that we subtracted each participant’s mean to all 37 actual affect items from the raw score for each actual affect item, and each participant’s mean to all 37 ideal affect items from the raw score of each ideal affect item. We computed the aggregates for avoided, actual, and ideal affect using these mean-deviated scores.

Preliminary analysis: Are the negative aspects rated more negatively than the other aspects?

We examined whether the aspects of the images that were identified as negative by at least 90% of participants in the pretest were rated as more negative compared with all other aspects in the same image by this new set of participants. The means are depicted in Table 1. We conducted pairwise t tests on the most negative aspect(s) of one image with the second most negatively rated aspect. All means statistically differed from each other, p < .001.

Does ANA predict how negatively negative aspects are rated?

To test Hypothesis 1, we averaged the ratings of all six negative aspects (two cars in an accident, a homeless person, poor man, man wrapped in a blanket, robbery, and poor people) of the six pretested images (Cronbach’s alpha = .87). We regressed this average rating of the negative aspects onto ANA, controlling for actual negative affect. To control for rating any aspect positively or negatively, we also added each participant’s average rating of all remaining aspects of all six images (Cronbach’s alpha = .78) into the regression as a covariate. As predicted, the higher people’s ANA, the more negatively they rated the negative aspects of the images, B = −.70 (95% confidence interval [CI]: [−1.08, −0.32]), SE = 0.19, t(54) = −3.68, p = .001, Cohen’s f² = .35 for the whole model. Each participant’s rating of the other aspects also significantly predicted the mean ratings of the negative aspects, B = .51 (95% CI: [0.04, 0.97]), SE = 0.23, t(54) = 2.18, p = .03. Actual negative affect was not significant, B = −.04 (95% CI: [−0.46, 0.37]), SE = 0.21, t(54) = −.22, p = .83.

To test whether ANA is only related to negativity ratings of the negative aspects (and not to negativity ratings of the other aspects), we regressed the average rating of the other aspects onto ANA, controlling for actual negative affect, and the average rating of all negative aspects. No significant predictors emerged (all ps > .44) except for the mean rating of the negative aspects, B = .16 (95% CI: [0.01, 0.31]), SE = 0.07, t(54) = 2.18, p = .03, Cohen’s f² = .09 for the whole model.

Does ANA predict recognition of negative details?

To test Hypothesis 2, we created a sum score of all negative details participants correctly recognized (negative details participants selected that were actually part of the image). We also created a sum score of all negative details participants falsely recognized (negative details participants selected that were not part of the image). We also created sum scores of the other details participants correctly recognized and of the other details participants falsely recognized.

We regressed the number of the correctly recognized negative details onto ANA, controlling for actual negative affect, ideal positive affect, and actual positive affect. In addition, we controlled for each participant’s number of falsely recognized negative details to control for participants’ tendencies to report any negative information. As predicted, the higher people’s ANA, the fewer negative details they correctly recognized, B = −2.13 (95% CI: [−4.09, −0.17]), SE = 0.98, t(47) = −2.18, p = .03, Cohen’s f² = .20 for the whole model. Furthermore, the more people ideally wanted to feel positive, the more negative details they correctly recognized, B = 2.44 (95% CI: [0.03, 4.86]), SE = 1.20, t(47) = 2.03, p = .048. All other predictors were not significant (all ps > .14).

It is possible that people with a high ANA rate the negative aspects of the images more negatively because they look at them less during the first viewing (hence the worse recognition), and get used to them less. To rule out this alternative explanation, we ran a mediation analysis using Hayes’s PROCESS Macro (Model 4) for SPSS (Hayes, 2013). ANA was entered as independent variable; the rating of negative aspects was entered as outcome variable, and number of correctly recognized negative details was entered as mediator. We also entered actual negative, actual positive, ideal positive affect, and number of falsely recognized negative details into the model as control variables. While the direct effect of ANA on the rating of negative aspects was significant, B = −.60 (95% CI: [−1.14, −0.06]), SE = 0.27, t(46) = −2.25, p = .03, the indirect effect of ANA on the rating of negative aspects through number of correctly recognized negative details was not; the effect was estimated to lie between −.10 and .20 with 95% CI using Hayes’s (2013) bootstrapping macro with 5,000 bootstrap samples. This finding suggests less processing (indirectly assessed by number of negative details correctly recognized) of negative aspects cannot explain the association between ANA and negativity ratings of negative aspects.

To examine whether ANA is only related to number of negative details participants correctly remember, and not to negative details they falsely remember, we regressed number of falsely recognized negative details onto ANA, controlling for actual negative affect, and number of correctly recognized negative details. No significant predictors emerged (all ps > .05).

Finally, we also tested whether ANA is only related to correctly recognizing fewer negative details, not fewer details in general. We regressed number of correctly recognized other details onto ANA, controlling for actual negative affect, and number of the falsely recognized other details. No significant predictors emerged (all ps > .36) except for number of falsely recognized other details, B = .79 (95% CI: [0.44, 1.15]), SE = 0.18, t(49) = 4.51, p < .001, Cohen’s f² = .52 for the whole model. We also regressed number of falsely recognized other details onto ANA, controlling for actual negative affect, and number of correctly recognized other details. Interestingly, the higher people’s ANA, the fewer other details they falsely recognized, B = −2.57 (95% CI: [−4.29, −0.84]), SE = 0.86, t(49) = −2.99, p = .004, Cohen’s f² = .77 for the whole model. Furthermore, the more people correctly recognized other details, the more other details they falsely recognized, B = .37 (95% CI: [0.21, 0.53]), SE = 0.08, t(49) = 4.51, p < .001. Actual negative affect was not significant, B = −.43 (95% CI: [−2.28, 1.42]), SE = 0.92, t(49) = −.47, p = .64. See Table 2 for means and correlations between all variables.

Table 2.

Means and Standard Deviations (in Parentheses) of Variables and Pairwise Correlations (Study 1).

		R
Variable	M (SD)	RNA	ROA	CRND	CROD	FRND	FROD	ANA	AN	IP	AP
Rating of negative aspects (RNA)	−2.16 (0.82)
Rating of other aspects (ROA)	−0.11 (0.42)	.27*
Correctly recognized negative details (CRND)	12.86 (3.21)	.02	−.01
Correctly recognized other details (CROD)	6.75 (2.13)	.05	.11	.62**
Falsely recognized negative details (FRND)	3.70 (1.94)	.01	−.03	.19	.22
Falzely recognized other details (FROD)	2.99 (1.83)	.15	−.04	.23	.44**	.62**
Avoided negative affect (ANA)	1.27 (0.52)	−.44**	−.03	−.18	−.07	−.28*	−.38**
Actual negative affect (AN)	0.11 (0.48)	.06	.01	.04	.01	−.07	.04	−.18
Ideal positive affect (IP)	1.70 (0.42)	−.42**	−.09	.01	.01	−.25	−.21	.54**	−.19
Actual positive affect (AP)	0.29 (0.60)	−.07	−.06	−.09	−.08	.03	.00	.10	−.85**	.22

p < .05. ** p < .01.

Discussion

As predicted, the higher people’s ANA, the more negatively they rated negative aspects of an image. This finding was specific to only negative and not to the other aspects (positive or neutral) of an image, suggesting that exposure to what people want to avoid feeling makes experiencing it even worse. Importantly, this could not be explained by the number of negative details correctly remembered, an indirect measure of participants not processing and hence not getting used to the negative aspects in the first viewing: In a meditation analysis, the indirect effect of ANA on the rating of negative aspects through number of negative details remembered was not significant. This is in line with the idea that emotional goals, not actual emotion regulation such as attentional deployment, predict negativity ratings of negative aspects.

Actual negative affect was not associated with how negatively negative aspects were rated. Hence, people’s affective goals seem more important than people’s global actual affect in determining the valence of a stimulus. Previous research has reported individual differences in affective responses to different stimuli (e.g., Gross et al., 1998), but our study is the first to find that ANA can predict systematic individual differences in valence ratings of negative aspects of images.

Although only just reaching significance (p = .03) but in line with other research (Caballero, Goyal, Davidson, & Koopmann-Holm, 2020), we also found, as predicted, the higher people’s ANA, the fewer negative details they correctly remembered. ANA neither predicted number of negative details falsely remembered nor number of other details correctly remembered. However, it did predict number of falsely recognized other details. The higher people’s ANA, the fewer other details they falsely recognized. It is possible that the more people want to avoid feeling negative, the more attention they paid to the other details of the images. Although this did not make them better at correctly identifying them, they were less likely to falsely choose wrong distractors.

Interestingly, actual negative affect did not predict how many negative details people correctly remembered. This finding, however, is not necessarily at odds with the literature on mood congruency (e.g., Bower, 1981). We measured people’s global actual affect (i.e., affective states people feel over the course of a typical week), not momentary actual affect (i.e., affective states people feel in the moment). That global actual affect did not emerge as predictor is in line with AVT (Tsai et al., 2006), suggesting it is avoided, not actual negative affect that predicts what people do to avoid feeling bad.

Finally, in addition to ANA, we found the more people ideally wanted to feel positive, the more negative details they correctly remembered. How can this finding be explained? The more people ideally want to feel positive, the more they are guided by their goal to approach the positive. Focusing on approaching the positive might make negative information more unexpected. Because unexpected pictorial information is associated with greater recall compared to expected information (Heckler & Childers, 1992), people who focus on the positive might be more likely to remember negative details. These different correlations of avoided negative and ideal positive affect with recognition align with previous research demonstrating ANA and ideal positive affect are separate constructs (Koopmann-Holm & Tsai, 2014). One limitation of this study is that we only examined recognition performance. Would the results generalize to an open-ended free-recall paradigm?

Study 2: Does ANA Predict How People Describe Complex Images?

In this study, we examined whether people describe complex scenes differently depending on their ANA above and beyond actual negative affect. This study extends the previous study in that we are no longer examining recognition of negative details, but how participants report in an open-ended format what they remember having seen. We predicted the more people want to avoid feeling negative, the fewer details of negative aspects of images they will mention (Hypothesis 1) and the fewer negative words they will use in their descriptions of the images (Hypothesis 2).

Method

Participants

One hundred fifty-seven U.S. undergraduates (72.61% female; mean age = 18.93 years; SD = 0.97) participated in an “Emotion and Information Processing” study in exchange for course credit. Sample size was determined using G*Power software: Ensuring a power of .80 to detect small size effects (f² = .08; we expected smaller effect sizes for free recall compared with recognition) using regression with three to four predictors and an alpha error probability of .05, we needed 141 to 155 participants.

Materials and Procedure

AVI

As in Study 1, participants first completed the AVI to assess actual, ideal, and avoided affect. We calculated the same mean-deviated aggregates as in the first study. For actual and avoided negative affect, internal consistencies (Cronbach’s alphas) were .85 and .88, respectively. For actual and ideal positive affect, internal consistencies (Cronbach’s alphas) were .89 and .82, respectively.

Complex images

After participants completed the AVI, we presented them with the six pretested images in the same order as in Study 1 for 5 s. After the presentation of each image, participants answered the question “What did you see in the image that was just presented to you? Please describe in as much detail as you can” in an open-ended format.

Number of details described from negative and other aspects of the images

Three independent research assistants coded the number of details participants correctly mentioned about negative and other aspects of the images in their open-ended responses. The coders overlapped on at least 25% of responses for each image. To assess interrater reliability, we correlated the number of negative and other details from one coder with the number of negative and other details from the other coder. Pearson’s r ranged from .83 (p < .001) to .97 (p < .001) for the coding of negative details and from .83 (p < .001) to .98 (p < .001) for the coding of other details, suggesting high interrater reliability. Separately for negative and other aspects, we averaged the number of described details across all six images for each participant.

Percentage of words that were negative or positive

We used the Linguistic Inquiry and Word Count program (LIWC; Pennebaker et al., 2001) to code the text of the responses to the images. LIWC counts the total number of words, and then calculates the percentage of total words that fall into a given category. To assess percentage of negative words, we used the percentage of total words that fell into the established LIWC category of “Negative Emotions” (e.g., hate, worthless, enemy) of the internal dictionary. To assess percentage of positive words, we used the percentage of total words that fell into the established category “Positive Emotions” (e.g., happy, pretty, good) of the internal dictionary. We assessed the percentages for those two categories for all six images. For each category separately, we averaged the percentages across all six images. See Table 3 for means and correlations between all variables.

Table 3.

Means and Standard Deviations (in Parentheses) of Variables and Pairwise Correlations (Study 2).

		r
Variable	M (SD)	NND	NOD	WC	PNW	PPW	ANA	AN	IP	AP
Number of negative details (NND)	38.06 (19.61)
Number of other details (NOD)	36.44 (20.87)	.87**
LIWC word count (WC)	157.85 (129.23)	.93**	.93**
LIWC percentage of negative words (PNW)	0.25 (0.65)	−.04	−.02	.07
LIWC percentage of positive words (PPW)	0.40 (0.61)	.03	.05	.12	.05
Avoided negative affect (ANA)	1.36 (0.48)	.12	.08	.08	−.19*	−.10
Actual negative affect (AN)	−0.01 (0.54)	.04	−.08	−.05	−.03	−.04	.08
Ideal positive affect (IP)	1.75 (0.49)	.07	.06	.05	.06	−.12	.58**	.06
Actual positive affect (AP)	0.44 (0.63)	.04	.13	.11	.04	−.00	.05	−.89**	.17*

p < .05. **p < .01.

Demographics questionnaire

Participants completed a demographics questionnaire which assessed their gender, age, and ethnicity.

Results

When examining number of negative details recalled, we controlled for number of other details recalled (to hold recalling any information constant). However, for the analyses of frequencies of negative words, we did not control for total number of words since LIWC produces percentage scores, and therefore already takes number of words into account (to control for overall amount written).

Does ANA predict how many details people mention of the negative aspects of the images?

To test Hypothesis 1, we regressed number of negative details recalled onto ANA, controlling for actual negative affect, and number of other details recalled. Contrary to Hypothesis 1, ANA did not predict people’s tendency to mention negative details, B = 1.48 (95% CI: [−1.64, 4.61]), SE = 1.58, t(155) = .94, p = .35. However, the more people actually felt negative, the more negative details they mentioned, B = 3.70 (95% CI: [0.88, 6.51]), SE = 1.42, t(155) = 2.60, p = .01, Cohen’s f² = 3.35 for the whole model. Also, the more other details people recalled, the more negative details they mentioned, B = .82 (95% CI: [0.75, 0.90]), SE = 0.04, t(155) = 22.59, p < .001.

We also regressed number of other details recalled onto ANA, controlling for actual negative affect, and number of negative details recalled. ANA was not a significant predictor, B = −.46 (95% CI: [−3.80, 2.88]), SE = 1.69, t(155) = −.27, p = .79. However, the more people actually felt negative, the fewer other details they recalled, B = −4.25 (95% CI: [−7.23, −1.26]), SE = 1.51, t(155) = −2.81, p = .006, Cohen’s f² = 3.35 for the whole model. Finally, the more negative details people mentioned, the more other details they mentioned, B = .93 (95% CI: [0.85, 1.01]), SE = 0.04, t(155) = 22.59, p < .001.

Does ANA predict how many negative words people use when describing images?

To test Hypothesis 2, we regressed percentage of negative words as calculated by LIWC onto ANA, controlling for actual negative affect, ideal positive affect, and actual positive affect. The higher people’s ANA, the fewer negative emotion terms they used when describing complex scenes, B = −.47 (95% CI: [−0.74, −0.21]), SE = 0.13, t(157) = −3.53, p = .001, Cohen’s f² = .09 for the whole model. Actual negative and actual positive affect did not emerge as significant predictors—actual negative affect: B = −.04 (95% CI: [−0.51, 0.44]), SE = 0.24, t(157) = −.15, p = .88; actual positive affect: B = −.02 (95% CI: [−0.43, 0.38]), SE = 0.20, t(157) = −.11, p = .92. However, the more people ideally wanted to feel positive, the more negative emotion terms they used when describing complex scenes, B = .35 (95% CI: [0.07, 0.63]), SE = 0.14, t(157) = 2.45, p = .02.

To examine whether the effect of ANA on word use is specific to negative words, we also regressed percentage of positive words onto ANA, controlling for actual negative affect. No significant predictors emerged (all ps > .21).

Discussion

We predicted the higher people’s ANA, the fewer details they would mention of negative aspects. We did not find support for this hypothesis. While ANA predicted recognition of negative details in Study 1 (albeit only just reaching significance but in line with other research; Caballero, Goyal, Davidson, & Koopmann-Holm, 2020), ANA was not associated with number of negative details remembered using a free-recall paradigm in Study 2. How can these seemingly discrepant findings be explained? Koriat and Goldsmith (1996) have argued standard free-recall instructions motivate people to report what they think is important to successfully complete the task. When we asked participants what they saw using the free-recall format, we asked them to describe in as much detail as they could. With this instruction, we might have increased participants’ motivation to report as much as they can remember. That latter motivation might have interfered with people’s motivation to avoid feeling negative.

Alternatively, the fact that participants can choose what to write about in the free-recall paradigm might explain our findings: People might use distraction to avoid feeling negative. They might focus on specific details (e.g., patterns of a shirt) of negative information to distract themselves from the actual content of the negative information (i.e., the suffering). Because they can freely write about these details, their number of details correctly remembered might be as high as for people who deeply process the negative information. However, a recognition test likely contains more details than just the ones they focused on to distract themselves. Hence, while being able to recognize a list of details is a measure of attention, writing about self-selected details might measure attention and/or distraction. The fact that number of recalled negative details and percentage of negative words did not correlate (see Table 3) supports the idea that number of recalled details might not have measured what we intended to measure. In that respect, the recognition task might produce a more valid measure than number of recalled details.

We did find the higher people’s actual negative affect, the more negative and fewer other details they mentioned. This is in line with mood congruence (Bower, 1981), assuming people felt similarly during the study as they felt over the course of a typical week, which is how we assessed actual affect.

Even though ANA did not predict number of negative details recalled, it did predict how people wrote about what they remembered seeing. As hypothesized, the higher people’s ANA, the fewer negative emotion terms they used when describing the images. Positive word use was not predicted by ANA, suggesting ANA is specifically related to using fewer negative words. ANA might predict word use (how people write about complex scenes) but not the number of negative details recalled (the content) because the negative words we assessed are emotional and therefore more closely related to ANA compared with number of details remembered. However, these discrepant findings might also be due to the possibility that we might not have measured what we intended to measure with number of recalled details, as outlined above.

Interestingly, in line with the findings from Study 1 that the more people wanted to feel positive, the more negative aspects they correctly remembered, in this study we found the more people wanted to feel positive, the more negative emotion terms they used when describing the images. Like in Study 1, for people who want to feel positive, negative information might be more unexpected. Therefore, they might focus on it more in their descriptions and because of that, use more negative emotion terms.

Our studies up to this point are limited in three ways: First, we used actual photographs with negative and other aspects. Hence, we juxtaposed negative with other (positive and neutral) aspects, not solely negative with positive aspects. Second, we only examined recognition and recall of negative aspects, but processing also consists of perceiving, not just remembering. Third, our samples were solely U.S. undergraduates. The last study addresses these limitations by presenting an ambiguous image consisting of negative and positive information to U.S. American and German community members. The advantage of using an ambiguous image is that people can perceive different things and hence, this last study is about perception, not just memory. We predicted the higher people’s ANA, the less likely they will perceive something negative in an ambiguous image. Finally, previous research suggests Americans want to avoid feeling negative more than Germans do (Koopmann-Holm & Tsai, 2014). Therefore, we also examined whether cultural differences exist in the tendency to report seeing something negative in an ambiguous image and whether ANA could mediate these cultural differences.

Study 3: Does ANA Mediate Cultural Differences in Perceiving Suffering?

Previous research suggests Americans want to avoid feeling negative more than Germans do (Koopmann-Holm & Tsai, 2014). Because of these cultural differences in ANA and the findings that the more people want to avoid feeling negative, the less they focus on suffering, we predicted there will be cultural differences in processing suffering.

People’s motivations are reflected in what they see in ambiguous visual information (see Dunning & Balcetis, 2013, for a review). Therefore, we hypothesized people’s motivation to avoid feeling negative predicts processing of an ambiguous image. Hence, instead of using the six images from Studies 1 and 2, we created an ambiguous image in which people can either see something positive (a happy face), something negative (a distressed face), neither, or both.

To summarize, in this study, we attempted to replicate our previous finding that Americans would want to avoid feeling negative more than Germans do (Koopmann-Holm & Tsai, 2014) (Hypothesis 1). We also predicted Americans would be less likely than Germans to report seeing suffering in an ambiguous image (Hypothesis 2). Finally, we predicted ANA would mediate cultural differences in perceiving suffering (Hypothesis 3).

Method

Participants

One hundred fifty-two Americans (64.47% female) and 315 Germans (66.35% female) participated in this study. Sample size was determined using G*Power software: Ensuring a power of .80 to detect small to medium size effects (f = .17) using an analysis of covariance (ANCOVA) with two groups, two covariates, and an alpha error probability of .05, we needed a total sample size of 274 participants. We aimed for 140 participants in each country; however, we did not close the survey in Germany after we reached our goal, which is why we ended up with a larger German sample. Americans grew up and reported having lived in the United States most of their lives. Germans grew up and reported having lived in Germany most of their lives. There were no group differences in gender, χ²(1, 467) = 0.16, ns. However, Americans were significantly older (mean age = 26.19, SD = 14.03) than Germans (mean age = 23.39, SD = 4.30), F(1, 461) = 10.45, p = .001, $η_{p}^{2}$ = .02. Therefore, we controlled for age in the analyses described below when it emerged as significant covariate.

Procedure

In the United States and Germany, participants were recruited by sending out emails to community members advertising the study. All measures were completed online in both countries. American participants completed all measures in English, and German participants in German. All measures were translated into German using standard translation–back-translation procedures. We included other measures as fillers. Neither American nor German participants received compensation for their participation.

Materials

AVI

Like in Studies 1 and 2, participants completed the AVI to assess actual, ideal, and avoided affect, and we used the same mean-deviated aggregates. For actual and avoided negative affect, internal consistencies (Cronbach’s alphas) were .85 and .86 for Americans and .81 and .90 for Germans, respectively. For actual and ideal positive affect, internal consistencies (Cronbach’s alphas) were .86 and .78 for Americans and .82 and .72 for Germans, respectively.

Ambiguous image

Participants viewed an ambiguous black and white image created by the research team that includes a happy face on one side and a distressed face on the other. This image is an altered version of one of Sébastien Thibault’s color illustrations (see Figure 2 for the image we used in this study; see Thibault, 2013, for the original image). Underneath the image, participants described in an open-ended format what they were seeing in the image while it was still visible. On the next page, participants indicated whether they saw any emotion(s) in the image on the previous page and if they saw any, they described which emotions they saw in an open-ended format. We pretested this image as well: The participants from Study 1 viewed this image for as long as they wanted and then indicated how positive or negative the happy and distressed face were to them (3 = extremely positive, 0 = neither positive nor negative, −3 = extremely negative). Participants rated the happy face as more positive (M = 1.72, SD = 1.14) than the negative face (M = −1.57, SD = 1.23), t(58) = −12.77, p < .001, Cohen’s d = 2.77.

Figure 2.

Ambiguous image used in Study 3.

Manual coding of perception of suffering

Two trained research assistants manually coded participants’ responses. For this purpose, we combined the responses to the questions “What do you see in the image above?” and “Did you see any emotion(s) in the image on the last page? If so, please describe below.” Before coding, the German responses were translated into English and then back-translated into German to ensure accuracy. Then, both coders coded the responses in English to ensure they did not know which response stemmed from which cultural group. The coders were also blind to participants’ scores on avoided, ideal, and actual affect. They coded the degree to which participants perceived suffering in the ambiguous image: If a participant mentioned the distressed face only, the response was coded as −2. If a participant mentioned both the distressed and happy face, the response was coded as −1. If a participant did not mention any face or reported not seeing any emotions, the response was coded as 0. Finally, if a participant mentioned the happy face only, the response was coded as 1. To assess reliability, coders overlapped on 65.52% of the responses. Inter-rater reliability was high (Pearson’s r = .76, p < .001; Spearman’s rho = .76, p < .001).

Percentage of words that were negative or positive

As in Study 2, we used LIWC’s established categories “Negative Emotions” and “Positive Emotions” to code the text of participants’ responses. For the responses from American participants, we used the English internal dictionary, and for the responses from German participants, we used the German internal dictionary. See Table 4 for means and correlations between all variables.

Table 4.

Means and Standard Deviations (in Parentheses) of Variables and Pairwise Correlations Across Americans and Germans (Study 3).

		r
Variable	M (SD)	CP	PNW	PPW	ANA	AN	IP	AP
Coding of perception (CP)	0.34 (0.91)
LIWC percentage of negative words (PNW)	3.75 (8.53)	−.63**
LIWC percentage of positive words (PPW)	21.26 (20.70)	.44**	−.21**
Avoided negative affect (ANA)	1.16 (0.43)	.14**	−.08	.15**
Actual negative affect (AN)	−0.16 (0.45)	−.05	.07	−.06	.04
Ideal positive affect (IP)	1.50 (0.43)	.14**	−.10*	.15**	.46**	−.20**
Actual positive affect (AP)	0.47 (0.55)	.12*	−.07	.09	.04	−.85**	.32**

p < .05. **p < .01.

Demographics questionnaire

Participants completed a demographics questionnaire, which assessed their gender, age, city, and country where they primarily grew up, and city and country where they lived most of their lives.

Results

Do Americans and Germans differ in the degree to which they want to avoid feeling negative?

To test Hypothesis 1, we conducted a 2 (Group: Americans, Germans) × 2 (Affect Type: Avoided Affect, Actual Affect) repeated-measures ANCOVA with Group as between-subjects factor, Affect Type as within-subject factor, and age as covariate. We found a significant Group × Affect Type interaction, F(1, 457) = 7.57, p = .006, $η_{p}^{2}$ = .02. Therefore, we conducted a univariate ANCOVA by Group for ANA, controlling for actual negative affect and age. As predicted and in line with Koopmann-Holm and Tsai (2014), Americans wanted to avoid feeling negative more than did Germans, American mean = 1.31, SE = 0.04; German mean = 1.09, SE = 0.02; F(1, 456) = 27.03, p < .001, $η_{p}^{2}$ = .06. We also conducted a univariate ANCOVA by Group for actual negative affect, controlling for ANA and age. In line with AVT, which proposes culture shapes avoided more than actual affect, Americans and Germans did not differ in how much they actually felt negative, American mean = −0.13, SE = 0.04; German mean = −0.18, SE = 0.03; F(1, 456) = 1.14, p = .29.

Do Americans and Germans differ in their perception of suffering?

To test Hypothesis 2, we examined the manual coding of perception of suffering and the percentage of words that were negative or positive as coded by LIWC.

Manual coding of perception of suffering

We conducted an ANOVA by Group on the coding of the degree to which participants perceived suffering in the ambiguous image. As predicted, Americans were more likely to only see the happy face (and hence, not the distressed face) than Germans were, American mean = .53, SE = 0.07; German mean = .25, SE = 0.05; F(1, 461) = 9.83, p = .002, $η_{p}^{2}$ = .02.

Percentage of words that were negative or positive

We conducted an ANOVA by Group on percentage of negative words as coded by LIWC. As predicted, Germans used more negative emotion terms when describing the ambiguous image than Americans did, German mean = 4.53, SE = 0.48; American mean = 2.14, SE = 0.69; F(1, 466) = 8.12, p = .005, $η_{p}^{2}$ = .02. We also conducted an ANCOVA by Group on percentage of positive words as coded by LIWC, controlling for age. Americans used more positive emotion terms when describing the ambiguous image than Germans did, American mean = 31.62, SE = 1.56; German mean = 16.05, SE = 1.08; F(1, 460) = 66.91, p < .001, $η_{p}^{2}$ = .13.

Does ANA mediate cultural differences in perceiving suffering?

To test Hypothesis 3, we examined the manual coding of perception of suffering as well as the percentage of words that were negative or positive as coded by LIWC.

Manual coding of perception of suffering

We examined whether ANA mediated the cultural difference between Americans and Germans in the degree to which people perceived suffering by using Hayes’s PROCESS Macro (Model 4) for SPSS (Hayes, 2013). Group (Americans = 1, Germans = 2) was entered as independent variable, degree of perceived suffering was entered as outcome variable, and ANA was entered as mediator. We also entered actual negative affect, ideal positive affect, and actual positive affect into the model as statistical controls.

The direct effect of Group on amount of perceived suffering was significant, B = −.21 (95% CI: [−0.39, −0.03]), SE = 0.09, t(456) = −2.34, p = .02, Cohen’s f² = .03 for the whole model. Furthermore, as hypothesized, a significant indirect effect of Group on amount of perceived suffering through ANA emerged. The effect was estimated to lie between −.07 and −.01 with 95% CI using Hayes’s (2013) bootstrapping macro with 5,000 bootstrap samples.

Percentage of words that were negative or positive

We ran the same mediation analysis as described above using Model 4 of Hayes’s PROCESS Macro (Hayes, 2013), but with percentage of negative words as outcome variable. Although the direct effect of Group on percentage of negative words was significant—B = 2.07 (95% CI: [0.38, 3.76]), SE = 0.86, t(461) = 2.41, p = .02, Cohen’s f² = .02 for the whole model—contrary to our hypothesis, the indirect effect of Group on percentage of negative words through ANA was not: The effect was estimated to lie between −.03 and .37 with 95% CI using Hayes’s (2013) bootstrapping macro with 5,000 bootstrap samples.

A post hoc explanation for this finding is that the content perceived in the ambiguous image predicts participants’ word use. For example, if participants saw the happy face only, they might not use negative words. Hence, we tested a sequential mediation using Model 6 of Hayes’s PROCESS Macro (Hayes, 2013), in which Group predicts ANA, which in turn predicts perception of suffering, which then predicts word usage (see Figure 3B). Our data are in line with this post hoc explanation: The direct effect of Group on percentage of negative words was not significant, B = .83 (95% CI: [−0.51, 2.18]), SE = 0.68, t(455) = 1.22, p = .22, but the indirect effect of Group through ANA and then through perceived suffering on percentage of negative words was: It was estimated to lie between .0001 and .38 with 95% CI using Hayes’s (2013) bootstrapping macro with 5,000 bootstrap samples.

Figure 3.

(A) Mediational model from Study 3 using the manual coding of perception of suffering and (B) sequential mediational model from Study 3 using percentage of words that were negative as coded by LIWC.

We reran these two meditation models for positive emotion words as well, but none of the indirect effects were significant, suggesting that the sequential effects are specific to negative emotions words.

Discussion

In this last study, we examined whether cultural differences in ANA would translate into cultural differences in perceiving suffering in an ambiguous image. We found support for this. First, using different samples (e.g., compared with previous work, in this study, age differences between Americans and Germans were reversed and participants were community members rather than undergraduates), we replicated Koopmann-Holm and Tsai’s (2014) finding that Americans want to avoid feeling negative more than Germans do, suggesting that these cultural differences are robust.

Second, we found that Americans perceived and talked about suffering less compared with Germans when describing an ambiguous image in which one can either see a happy face, a distressed face, neither, or both. We found this irrespective of whether we coded the English and German responses manually in the same language (English) or whether we used LIWC (Pennebaker et al., 2001) to calculate the percentage of positive and negative emotion words from the participants’ responses in their respective languages. Finally, as predicted, ANA mediated the cultural differences in perception of suffering using the manual coding. However, we did not find the same mediation using LIWC coding. Post hoc analyses revealed that cultural differences in how negatively people describe an ambiguous image are not mediated by ANA per se, but by ANA, shaping what people perceive in an ambiguous image, which then predicts usage of negative words.

General Discussion

In three studies, we found ANA predicts processing of others’ suffering above and beyond global actual negative affect. The higher people’s ANA, the more negatively they rate negative aspects of images (Study 1) and the less likely they recognize different negative details (Study 1). Furthermore, the higher people’s ANA, the fewer negative words they use to describe images that contain both negative and other aspects (Study 2). Finally, because Americans want to avoid feeling negative more than Germans do, they perceive less suffering in an ambiguous image and therefore, write about it less (Study 3). Taken together, these findings suggest ANA predicts valence ratings, recognition, and even perception of suffering, as well as the way people talk about others’ suffering.

ANA Predicts the Negativity of Negative Aspects

Our research suggests negative stimuli are not equally negative for different people. The higher people’s ANA, the more negatively they perceive negative aspects. This finding suggests not being able to distance oneself from what one tries to avoid (avoided affect mismatch) does not just make people feel bad, as suggested by Carver and Scheier (1998), but makes negative stimuli actually more negative when exposed to them. This finding has important implications for research using affective stimuli. For example, ANA might be able to explain why mood inductions with pretested stimuli might not be equally effective for different people.

Because our data are correlational, it is possible that ANA does not shape how negatively negative aspects are perceived, but rather the other way around: Maybe the more negatively people perceive negative aspects, the more they want to avoid feeling negative. However, research suggests that perceiving negativity (i.e., watching a negative rather than a neutral or positive film clip) does not motivate people to avoid feeling negative affect more, but less (Koopmann-Holm, 2020). Hence, elevated perceived negativity does not seem to be a precursor, but rather a consequence of higher ANA. The more frequently people feel negative affect, the more they might habituate to negativity, which might lead to lower ANA. Higher ANA, however, might be negatively reinforced by the actual avoidance of the negative just like in the maintenance of phobias (e.g., Skinner, 1958). Future studies should investigate this possibility.

ANA Predicts Recognition of Negative Details

As predicted, we found some support for our prediction that the higher people’s ANA, the less likely they correctly recognized negative details. Our work suggests that when presented with what people want to avoid feeling (avoided affect mismatch), people process it less effectively compared with people whose avoided affect mismatch is not as large. This is in line with other research suggesting that, under certain circumstances, society’s expectation not to feel negative is associated with avoiding negative information (Bastian et al., 2017). Future studies should examine whether, in the case of an avoided affect mismatch, the increased perceived negativity hinders information processing or whether judging something as very negative and not processing it deeply are independent processes.

ANA and Free Recall of Negative Details

We did not find support for the hypothesis that ANA would predict how many negative details people would mention in a free recall. Why would ANA predict recognition of negative details, but not the number of negative details recalled using an open-ended format?

As described earlier, it is possible our free-recall instruction to describe the image in as much detail as possible interfered with people’s motivation to avoid the negative. Future research should eliminate this instruction to compare the effects of ANA on recognition and free recall without this confounding variable. As outlined above, it is also possible recalling any negative details could measure distraction from rather than attention to negativity.

Interestingly, in Study 3, we also used a free-recall paradigm and found ANA was related to reporting suffering in an ambiguous image. Study 3 differed from Study 2 in two important ways: First, in Study 3, we did not ask participants to mention as many details as possible, so we might have eliminated a motivation that competed with the motivation to avoid feeling negative. Second, we used an ambiguous image instead of actual scenes. Hence, in Study 3, the free-recall instruction might have been more a measure of perception than memory. Future research should examine whether ANA shapes automatic processes (e.g., what people see in an ambiguous stimulus) more or processes that are under regulatory control (e.g., free recall).

ANA Predicts How People Talk About Complex Images

Even though we did not find an effect of ANA on amounts of freely recalled negative details in Study 2, we also examined whether participants’ descriptions of what they saw differed depending on their ANA. As predicted, we found the higher people’s ANA, the fewer negative words they used when describing images. This suggests the way people talk about what they have seen is predicted by ANA even if the actual content might not be.

Cultural Differences in ANA and Perception of Others’ Suffering

We found cultural differences in ANA translate into cultural differences in processing of suffering. As predicted, we found Americans wanted to avoid feeling negative more than Germans did, replicating earlier findings (Koopmann-Holm & Tsai, 2014). We also found, as predicted, Americans reported seeing suffering in an ambiguous image less often and used more positive and fewer negative emotion words in their image descriptions than Germans did. Finally, the cultural difference in perceiving suffering was mediated by ANA. Future research should examine whether these cultural differences in perception generalize to recognition and free recall of suffering and whether ANA can also mediate the latter differences.

Implications for Research on Compassion

Our work has important implications for compassion, which is defined as the sensitivity to the suffering of another person, coupled with a motivation to alleviate that suffering (Goetz et al., 2010). We found ANA predicts individual and cultural differences in the sensitivity to others’ suffering, the first step in compassionate responding: High ANA might hinder the initial sensitivity to suffering and compassionate responses when people can focus on something else. However, ANA also predicts how negatively suffering is perceived. The higher people’s ANA, the more negatively they rate suffering, which might be related to a higher motivation to alleviate suffering, the second step of a compassionate response. Hence, high ANA might increase actual helping when people cannot escape or distract themselves from processing suffering. Future research should test these predictions.

Limitations and Current Directions

Our findings raise important questions: Does ANA predict earlier stages of processing (e.g., perception, attention) or later stages (e.g., encoding, storage, retrieval) more? Does ANA predict only perceived negative valence of negative information or also arousal? What causes individual and cultural differences in ANA? Past research suggests cultural factors such as frontier spirit values (overcoming and mastering nature) can partly explain cultural differences in ANA (Koopmann-Holm & Tsai, 2014). Temperamental factors and past experiences with negative affect should also be investigated as sources of individual differences in ANA.

Conclusion

When looking at scenes containing negative (i.e., suffering) and other information, does everyone “see the whole picture?” As predicted, the higher people’s ANA, the fewer details of suffering they recognize seeing (despite perceiving others’ suffering as more negative), the fewer negative words they use when describing scenes, and the less likely they perceive suffering in an ambiguous image. Furthermore, compared with Germans, Americans are less likely to perceive suffering because they want to avoid feeling negative more.

Supplemental Material

Appendix_Pretest – Supplemental material for Seeing the Whole Picture? Avoided Negative Affect and Processing of Others’ Suffering

Supplemental material, Appendix_Pretest for Seeing the Whole Picture? Avoided Negative Affect and Processing of Others’ Suffering by Birgit Koopmann-Holm, Kathryn Bartel, Maryam Bin Meshar and Huiru Evangeline Yang in Personality and Social Psychology Bulletin

Supplemental Material

Koopman_Online_Appendix – Supplemental material for Seeing the Whole Picture? Avoided Negative Affect and Processing of Others’ Suffering

Supplemental material, Koopman_Online_Appendix for Seeing the Whole Picture? Avoided Negative Affect and Processing of Others’ Suffering by Birgit Koopmann-Holm, Kathryn Bartel, Maryam Bin Meshar and Huiru Evangeline Yang in Personality and Social Psychology Bulletin

Supplemental Material

Results_without_covariates – Supplemental material for Seeing the Whole Picture? Avoided Negative Affect and Processing of Others’ Suffering

Supplemental material, Results_without_covariates for Seeing the Whole Picture? Avoided Negative Affect and Processing of Others’ Suffering by Birgit Koopmann-Holm, Kathryn Bartel, Maryam Bin Meshar and Huiru Evangeline Yang in Personality and Social Psychology Bulletin

Footnotes

Acknowledgements

We would like to thank Kathryn Bruchmann, Tracey Kahan, Jeanne Tsai, and the Culture Impacts Emotion Laboratory for their invaluable feedback on earlier versions of this manuscript. We would also like to thank Anabel Homnack and Brian Holm for their help with the translations as well as Sébastien Thibault for allowing us to alter one of his illustrations and to use it in our research.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by Professional Development Grant 14015 awarded to the first author by the College of Arts and Sciences at Santa Clara University as well as Travel Grant 14015 awarded to the first author by the Dean of the College of Arts and Sciences at Santa Clara University.

ORCID iD

Birgit Koopmann-Holm

Supplementary Material

Supplemental material is available online with this article.

References

Bastian

Kuppens

Hornsey

M. J.

Park

Koval

Uchida

(2012). Feeling bad about being sad: The role of social expectancies in amplifying negative mood. Emotion, 12(1), 69–80.

Bastian

M. L.

Kuppens

(2017). Perceived social pressure not to experience negative emotion is linked to selective attention for negative information. Cognition and Emotion, 31(2), 261–268.

Bower

G. H.

(1981). Mood and memory. American Psychologist, 36(2), 129–148.

Caballero

Koopmann-Holm

(2020). Looking away from other people’s pain? Predictors and consequences. Manuscript in preparation.

Calvo

M. G.

Lang

P. J.

(2004). Gaze patterns when looking at emotional pictures: Motivationally biased attention. Motivation and Emotion, 28(3), 221–243.

Carver

Scheier

(1998). On the self-regulation of behavior. Cambridge University Press.

Chim

Hogan

C. L.

Fung

H. H.

Tsai

J. L.

(2018). Valuing calm enhances enjoyment of calming (vs. exciting) amusement park rides and exercise. Emotion, 18(6), 805–818.

Dennis

P. A.

Halberstadt

A. G.

(2013). Is believing seeing? The role of emotion-related beliefs in selective attention to affective cues. Cognition and Emotion, 27(1), 3–20.

Dunning

Balcetis

(2013). Wishful seeing: How preferences shape visual perception. Current Directions in Psychological Science, 22(1), 33–37.

10.

Eich

Forgas

J. P.

(2003). Mood, cognition, and memory. In Healy

A. F.

Proctor

R. W.

(Eds.), Handbook of psychology: Experimental psychology (Vol. 4, pp. 61–83). John Wiley.

11.

Goetz

J. L.

Keltner

Simon-Thomas

(2010). Compassion: An evolutionary analysis and empirical review. Psychological Bulletin, 136(3), 351–374.

12.

Gross

J. J.

Sutton

S. K.

Ketelaar

(1998). Relations between affect and personality: Support for the affect-level and affective-reactivity views. Personality and Social Psychology Bulletin, 24(3), 279–288.

13.

Harmon-Jones

Amodio

D. M.

Gable

P. A.

(2011). Attitudes toward emotions. Journal of Personality and Social Psychology, 101(6), 1332–1350.

14.

Hayes

A. F.

(2013). Introduction to mediation, moderation, and conditional analysis: A regression-based approach. Guilford Press.

15.

Heckler

S. E.

Childers

T. L.

(1992). The role of expectancy and relevancy in memory for verbal and visual information: What is incongruency? Journal of Consumer Research, 18(4), 475–492.

16.

Keightley

M. L.

Chiew

K. S.

Anderson

J. A.

Grady

C. L.

(2011). Neural correlates of recognition memory for emotional faces and scenes. Social Cognitive and Affective Neuroscience, 6(1), 24–37.

17.

Kensinger

E. A.

Schacter

D. L.

(2016). Memory and emotion. In Barrett

L. F.

Lewis

Haviland-Jones

(Eds.), Handbook of emotions (4th ed., pp. 564–573). Guilford Press.

18.

Koopmann-Holm

(2020). Perceiving negativity shapes avoided negative affect. Manuscript in preparation.

19.

Koopmann-Holm

Tsai

J. L.

(2014). Focusing on the negative: Cultural differences in expressions of sympathy. Journal of Personality and Social Psychology, 107(6), 1092–1115.

20.

Koriat

Goldsmith

(1996). Monitoring and control processes in the strategic regulation of memory accuracy. Psychological Review, 103(3), 490–517.

21.

Larsen

R. J.

(2000). Toward a science of mood regulation. Psychological Inquiry, 11(3), 129–141.

22.

Lazarus

R. S.

(1991). Emotion and adaptation. Oxford University Press.

23.

LeDoux

J. E.

(1996). The emotional brain: The mysterious underpinnings of emotional life. Simon & Schuster.

24.

Maio

G. R.

Esses

V. M.

(2001). The need for affect: Individual differences in the motivation to approach or avoid emotions. Journal of Personality, 69(4), 583–615.

25.

Mather

Carstensen

L. L.

(2005). Aging and motivated cognition: The positivity effect in attention and memory. Trends in Cognitive Science, 9(10), 496–502.

26.

Matt

G. E.

Vázquez

Campbell

W. K.

(1992). Mood-congruent recall of affectively toned stimuli: A meta-analytic review. Clinical Psychology Review, 12(2), 227–255.

27.

Mogg

Bradley

B. P.

Hyare

Lee

(1998). Selective attention to food-related stimuli in hunger: Are attentional biases specific to emotional and psychopathological states, or are they also found in normal drive states? Behaviour Research and Therapy, 36(2), 227–237.

28.

Morris

J. S.

Dolan

R. J.

(2001). Involvement of human amygdala and orbitofrontal cortex in hunger-enhanced memory for food stimuli. Journal of Neuroscience, 21(14), 5304–5310.

29.

Niedenthal

P. M.

Setterlund

M. B.

(1994). Emotion congruence in perception. Personality and Social Psychology Bulletin, 20(4), 401–411.

30.

Pennebaker

J. W.

Francis

M. E.

Booth

R. J.

(2001). Linguistic Inquiry and Word Count: LIWC 2001. Lawrence Erlbaum.

31.

Rusting

C. L.

Larsen

R. J.

(1995). Moods as sources of stimulation: Relationships between personality and desired mood states. Personality and Individual Differences, 18(3), 321–329.

32.

Schwartz

S. H.

Clore

G. L.

(1983). Mood, misattribution, and judgments of well-being: Informative and directive functions of affective states. Journal of Personality and Social Psychology, 45(3), 513–523.

33.

Schwartz

S. H.

Clore

G. L.

(2007). Feelings and phenomenal experiences. In Kruglanski

A. W.

Higgins

E. T.

(Eds.), Social psychology: Handbook of basic principles (2nd ed., pp. 385–407). Guilford Press.

34.

Siegel

E. H.

Wormwood

J. B.

Quigley

K. S.

Barrett

L. F.

(2018). Seeing what you feel: Affect drives visual perception of structurally neutral faces. Psychological Science, 29(4), 496–503.

35.

Sims

Koopmann-Holm

Young

H. R.

Jiang

Fung

Tsai

J. L.

(2018). Asian Americans respond less favorably to excitement (vs. calm)-focused physicians compared to European Americans. Cultural Diversity and Ethnic Minority Psychology, 24(1), 1–14.

36.

Sims

Tsai

J. L.

(2015). Patients respond more positively to physicians who focus on their ideal affect. Emotion, 15(3), 303–318.

37.

Sims

Tsai

J. L.

Koopmann-Holm

Thomas

E. A.

Goldstein

M. K.

(2014). Choosing a physician depends on how you want to feel: The role of ideal affect in health-related decision making. Emotion, 14(1), 187–192.

38.

Skinner

B. F.

(1958). Reinforcement today. American Psychologist, 13(3), 94–99.

39.

Tamir

Robinson

M. D.

(2007). The happy spotlight: Positive mood and selective attention to rewarding information. Personality and Social Psychology Bulletin, 33(8), 1124–1136.

40.

Thibault

. (2013). Negative emotions are key to well-being. Retrieved from http://sebastienthibault.com/Negative-Emotions-are-Key-to-Well-Being-Scientific-American.

41.

Tsai

J. L.

(2007). Ideal affect: Cultural causes and behavioral consequences. Perspectives on Psychological Science, 2(3), 242–259.

42.

Tsai

J. L.

Chim

Sims

(2015). Consumer behavior, culture, and emotion. In Ng

Lee

A. Y.

(Eds.), Handbook of culture and consumer behavior (pp. 68–98). Oxford University Press.

43.

Tsai

J. L.

Knutson

Fung

H. H.

(2006). Cultural variation in affect valuation. Journal of Personality and Social Psychology, 90(2), 288–307.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB

3.75 MB

0.06 MB