IDEAL AFFECT, EMOTIONAL MEMORY, AND WELL-BEING: A CROSS- CULTRAL STUDY A Thesis Presented to the Faculty of the Graduate School of Cornell University In Partial Fulfillment of the Requirements for the Degree of Master of Arts by Renbo Zheng December 2019 © 2019 Renbo Zheng ABSTRACT People differ in the type of ideal emotional state they want to feel. While European Americans ideally want to feel high arousal positive affect, East Asian prefers low arousal positive affect. This study investigated what role the recall of emotional memory (i.e. memory of ideal affect) plays in personal well-being between East Asian and European American cultures. Our results showed that both high arousal positive (HAP) and low arousal positive (LAP) consistency between ideal affect and actual affect predicted positive measure of personal well-being. Compared with East Asians, ideal-actual discrepancy scores (general discrepancy scores, HAP discrepancy scores and LAP discrepancy scores) had a greater predictive effect on European Americans’ well-being. Moreover, we found that for East Asians, the amount of low arousal positive events they recall has a slightly positive predictive effect on their level of wellbeing. The result also showed that European Americans may generally retrieve more HAP memories than East Asians. BIOGRAPHICAL SKETCH Renbo Zheng received his B.S. in Psychology from UC San Diego. He has been doing research at Culture and Social Cognition Lab in the Department of Human Development at Cornell University from August 2018. His research interest is in emotion and how emotion functions in social situations. iii ACKNOWLEDGMENTS I would like to render my warmest thanks to Professor Qi Wang for her friendly guidance and valuable advice throughout this master thesis and my graduate studies. I would also like to express my gratitude to Professor Corinna Löckenhoff, who provided me with valuable suggestions for this thesis and my graduate studies as well. I would finally like to thank the members of the Cornell Culture & Social Cognition Lab, Cornell Statistical Consulting Unit, and all my dear friends at Cornell University. Statement: Data presented in this thesis are part of a larger on-going study on memory and well-being in cultural context. Therefore, the findings are incomplete and should not be used beyond the thesis. iv Introduction When you go to a party of European American college students, the atmosphere is usually very excited. People are drinking heavily, dancing wildly, and talking loudly. However, East Asian students may find it a little bit difficult to be comfortable in such ‘high arousal’ situations. Instead, East Asians tend to enjoy activities that are less exciting such as karaoke and card games. According to Jeanne Tsai (2006), this cultural difference can be explained by the term ideal affect. Ideal Affect and Culture Ideal Affect is a state that people value, desire, and ideally want to feel. By affect, we are referring to emotional states that vary in two dimensions. The dimension of valence is from positive to negative, and the dimension of arousal is from high to low (Russell & Feldman, 1999). For instance, there are positive states such as enthusiasm, elation, excitement and euphoria, which are all different in certain ways. What they share is that they are positive states that are highly arousing. Similarly, there are states such as relaxation, calmness, peacefulness, and serenity, which are equally positive but are low arousal. Moreover, these affective states have similar meaning across cultures and therefore can be compared across cultures. Most people want to feel good. But people really differ in the type of ideal emotional state they want to feel. For a prototypical European American college student, the response is “I just want to be happy. Normally for me that means I would be doing something exciting. I just want to be entertained ... I just like excitement.” In contrast, a typical 1 Chinese Hong Kong college student response is “"My ideal state is to be quiet, serene, happy and positive." (Tsai et al, 2006). In the first response, feeling happy means being excited. In the second response, feeling happy means feeling calm. In a sense, we can see the cultural difference in ideal affect. According to Tsai’s study in 2006, she found that European Americans value excitement states (HAP states) more than their Hong Kong Chinese counterparts, and Hong Kong Chinese students value the calm states (LAP states) more than their European American counterparts (Tsai et al, 2006). This difference in ideal affect not only emerges when we ask people how they want to feel, but also emerges when we compare interpersonal communication, religious traditions and cultural products in European American and East Asian contexts (Tsai et al, 2007). A few studies had compared the ideal affect of peer interactions in different cultures. One study by Farver, Kim, and Lee (1995) discovered that European American preschool- age playmates expressed more HAP states like laughing than did Korean American preschool-age playmates. Although there is no clear coding of the LAP status, the results did show that Korean Americans expressed more neutral effects such as emotional loss and unresponsiveness than European Americans (Farver, Kim, & Lee, 1995). According to their results, Tsai suggestesd that since LAP expressions look like neutral expressions, it is conceivable that Korean Americans were showing LAP states as well as neutral states (Tsai, 2007). Tsai also found that European American dating couples expressed more HAP states than their Chinese counterparts when the dating couples talking about the disagreement in their relationships (Tsai, Levenson, & McCoy, 2006). Studies on parent-child interactions across cultures showed similar results. American mothers are inclined to stimulate their 2 babies, whereas Japanese mothers tend to soothe and quiet their babies (Caudill & Weinstein, 1969). For instance, the vocalization of American mothers are usually high in level and range of fundamental frequency to make their babies to be active, whereas the vocalization of Chinese mothers are more soothing and reduced in level and range of fundamental frequency to make their babies quiet (Papousek, Papousek, & Symmes, 1991). These studies suggest that interpersonal interactions in European American culture involve more HAP states and less LAP states than interpersonal interactions in East Asian culture. Tsai also conducted a series of studies to compare the endorsement of different ideal affects in religious cultures (Tsai, Miao, & Seppala, 2007). According to the prevalence of Christianity in Western countries, and the prevalenve of Buddhism in many East Asian cultures, Tsai predicted that Christianity endorses high arousal states more than low arousal states and Buddhism endorses low arousal states more than high arousal states. To test this hypothesis, Tsai compared affective content of classic Christian and Buddhist texts and contemporary self-help books. The results supported her predictions, which suggests that the difference in ideal affect can be traced back to to religious traditions (Tsai, Miao, & Seppala, 2007, Study 2 & Study 3). When it comes to cultural products like popular media, we can also see disparities in the affective content of mainstream media such as magazines and children’s literature. To test if there is a difference in ideal affect in magazine photographs, Tsai compared the facial expressions in mainstream women’s magazines in the United States and Hong Kong using an adapted version of the FACS, a technique to measure facial movement (Ekman & Friesen, 1978). In this coding scheme, calm smiles was expressed by a wrinkling of the eyes and a spreading of the mouth, and excited smiles was expressed by extra muscle 3 movement to separate the lips and open the jaw(Ekman & Friesen, 1978). As predicted, American women’s magazines demonstrated more excited smiles and less calm smiles than did their Chinese counterparts (Tsai et al, 2007). Moreover, in another study comparing best-selling stories in US and in Taiwan, results indicated that characters in American stories show more excited smiles and fewer calm smiles than the characters in the best- selling Taiwanese stories (Tsai et al, 2007). In another recent article, Jeanne Tsai analyzed three studies investigating if cultural differences in ideal affect exist in common official photos of well-known public figures (Tsai et al., 2016). Results showed that prominent American leaders showed more excited smiles than their Chinese counterparts, and that the national differences in leaders’ smiles hold in different occupations and ranks (Tsai et al., 2016, Study 1-2). It also suggested that a nation’s ideal affect is reflected in their prominent leaders’ facial expressions. For nations that value high arousal positive states, their leaders tended to exhibit excited smiles. The opposite held true for nations that valued low arousal positive states. Their leaders were more likely to show calm smiles (Tsai et al., 2016, Study 3). Ideal Affect/Actual Affect consistency and the Relation to Well-Being Before Tsai, most researchers, however, did not discriminate ideal affect and actual affect and implicitly considered ideal affect and actual affect as a single construct. Tsai conceptually thought that they are totally different. To test this idea, she conducted a study in which people were asked what they actually feel on average over the course of a typical week, and then what they would ideally like to feel on average during the same period (Tsai et al, 2006). Results from different samples show that people across cultures report wanting 4 to feel more positive than negative, and they want to feel more positive than they actually feel and less negative than they actually feel (2006). These results provide empirical support for the idea that ideal affect and actual affect are different things. So what would happen when people have discrepancies between ideal and actual affect? According to Self-discrepancy theory, people inherently want to achieve consistency between actual experience and their perception of ideal experience. (Higgins, 1987; Diener, Suh, Lucas, & Smith, 1999). Inconsistencies between actual self and ideal self (idealized version of yourself created from life experiences) can contribute to emotional discomforts (Higgins, Klein, & Strauman, 1985). Self-discrepancy theory may be useful to help us understand actual & ideal discrepancies in positive affect (A. P. Buunk & Gibbons, 2007). Since positive affect is a desirable, valued emotional state, discrepancies between one’s actual positive affect and ideal positive affect may predict emotional distress (Diener et al., 1999). For instance, Tsai’s study indicated that the inconsistency between actual affect and ideal affect appreciated by the culture a person is in is related to the person’s mental health (2006). She found that greater discrepancies between ideal and actual affect were linked with greater depression in people from different cultural backgrounds. For the European American group, discrepancies in ideal and actual HAP were correlated with depression. In contrast, for Chinese group, discrepancies in ideal and actual LAP were correlated with depression. Thus, Tsai concluded that this type of ideal-actual incongruity that was linked to depression in a particular culture depended on the culture’s ideal affect. (Tsai et al., 2006). 5 Several studies have found that actual and ideal affect discrepancies are associated with other indicators of health, such as physical symptoms and life satisfaction in different cultures (Tsai, 2017). Scheibe and her colleagues found that wider discrepancy between actual and ideal LAP was linked to more somatic health issues. On the other hand, the difference between actual and ideal HAP was not associated with somatic health issues (Scheibe et al., 2013). Tamir’s research showed that happier people were those who more often experienced emotions they desire across cultures (Tamir et al., 2017). Researchers also showed that LAP affect has significantly greater predictive power than HAP affect for anxiety and stress. And for life satisfaction, depression feeling good and mindfulness, the predictive power of LAP affect and HAP affect are about the same (McManus et al., 2018). However, there are cross-cultural studies showing that greater ideal and actual affect discrepancies predicted increased psychological distress and lower life satisfaction for both European Americans and East Asians (Tsai et al., 2006; Tran et al., 2017), and we cannot clearly tell whether there are cultural differences in actual and ideal affect discrepancies that matter for health and well-being(Tsai, 2017). Emotional memories and well-being In a sense, ideal affect is more like a current or future emotional state. However, emotional experiences in the past also have a strong influence on individual’s well-being. Self-determination theory posits that humans are inherently motivated to pursue growth. This is expressed through the daily search to fulfill three psychological needs: autonomy, competence, and relatedness (Deci & Ryan, 2000). Thus, it can be inferred that a condition that elicits a need-satisfying memory should affect a person’s well-being in a positive 6 direction. In contrast, a condition that elicits a need-thwarting memory should affect a person’s well-being in the opposite direction (Philippe et al. 2012). Moreover, Helen G. Jing and her colleagues also conducted a study examining imagined constructive behavioral responses to hypothetical worrisome future experiences. (Jing et al.,2016). One aim of their study is to recognize the contribution of episodic memory and episodic simulation to psychological well-being. They directly manipulated episodic detail by using the specificity induction method. Then, based on this selective increase in episodic detail, they measured changes in subjective well-being (Jing et al., 2016). The data from their experiments supported the prediction that increasing the level of episodic detail in memory when imagining constructive behaviors about worrisome events is associated with a higher level of psychological well-being toward those events. So what role does culture play when we study the association between memory and well-being? According to Wang’s recent research, culture moderated the relation between episodic memory to a variety of aspects of psychological health and personal well-being (Wang et al., 2018). Specifically, Wang indicated that increased detailed recall showed marginally significant association with more depressive symptoms among East Asian children, for whom detailed recall of one’s own experiences is discouraged in their culture (Wang et al., 2018). Overall, these studies showed empirical evidence that need satisfaction in episodic memories is correlated with well-being. Current Study The findings of previous studies give rise to an interesting research question: how does the recall of ideal affect relate to well-being and what role does culture play in it? In the 7 present study, we investigated what role the recall of emotional memory (i.e. memory of ideal affect) plays in personal well-being in East Asian and European American cultures. We used online surveys to collect emotional memory recalls of Asians and European Americans. Specifically, participants were required to recall as many high/low arousal emotional events as they could in limited time. We then used SPSS to analyze the data to examine the cultural differences. The result allows us to infer the association between people’s high/low arousal emotional memory and their well-being when they come from different cultural backgrounds. Students at Cornell University participated in the study on SONA, an online system of the psychology department. Participants signed a consent form and were randomly assigned to two different conditions: to recall high or low arousal events. In the high arousal condition, participants were instructed to write down memories of events in which they felt excited, enthusiastic, or elated with short words or phrases. In the low arousal condition, they were instructed to write down memories of events in which they felt peaceful, calm, or relaxed with short words or phrases. Participants in both conditions were asked to recall as many such events as they can within a limited time. After recalling their respective high arousal or low arousal events, participants from different cultural backgrounds in both the high arousal condition and low arousal condition were given the Affect Valuation Index (Tsai & Knutson, 2006) to record their ideal affect and actual affect over the course of a typical week. At the end of the survey, they were asked to fill out the Flourishing Scale (Diener et al., 2009) to measure their psychological well-being in a more thorough way. 8 The goal of this study was to figure out the association between people’s high/low arousal emotional memory and their well-being in different cultures. Specifically, participants’ well-being was measured by comparing the discrepancies in their ideal and actual affect. The less the discrepancies, the higher their level of well-being will be. Based on previous studies, we proposed three hypotheses which are associated to our research question. First, both cultural groups may retrieve more memories of their own ideal affect, which means Asians may retrieve more memories of LAP and European Americans may retrieve more memories of HAP. Second, the consistency between ideal affect and actual affect predicts well-being regardless of the participant’s culture. Third, within each group, we propose that for East Asians, the more LAP events they recall, the higher level of well- being they have. For European Americans, the more HAP events they recall, the higher level of well-being they have. Method Participants A sample of 203 undergraduate students at Cornell University participated the study. Participants are recruited from SONA system of Cornell Psychology Department, and they are from different cultural background. All participants are required to complete an online survey which asked them to recall events from their life and answer some questions about the events. They were given one bonus point on their course grade. All subjects were required to sign a consent form, which gives them a brief explanation of this study. The between-subject factor in this study was cultural group (European Americans as culture 1 9 and Asians or Asian Americans as culture 2). Data of 61 participants were thrown because they came from other cultural background me or not finishing the survey. Procedure and Measures In the survey, participants were randomly assigned to recall their memory in two conditions: high arousal condition (condition 1) and low arousal condition (condition 2). In high arousal condition, participants were asked to recall memories of events in which they felt excited, enthusiastic, or elated. In low arousal condition, they were asked to recall memories of events in which they felt peaceful, calm, or relaxed. Participants in both conditions were asked to recall as many such events as they can within 4 minutes (20 entries in total). There was no force response and the survey will advance automatically when time is up. Both high arousal events and low arousal events were measured by the variable “amount of recalls” (the amount of memory that participants recalled). Participants’ sample responses in high arousal condition included “Finishing my first marathon” and “College acceptance”, while sample responses in low arousal condition included “Reading a book” and “Laying on the sofa”. Figure 1 is a screenshot of the instruction in high arousal condition. Fig 1. Instruction in high arousal condition 1 0 After recalling, participant in both conditions were given the Affect Valuation Index to write down their ideal affect and actual affect over the course of a typical week (Tsai & Knutson, 2006). For example, they were asked how often they would ideally want to have a specific type of feeling and how often they actually have a specific type of feeling. The affective states were answered on a 1–5 point scale, with 1 indicating never and 5 indicating all the time. By calculating the difference score between ideal affect and actual affect for each participant, we can measure their ideal-actual affect consistency, which is considered as a predictor of participant’s well-being in our study. We also calculated the HAP ideal- actual affect difference and LAP ideal-actual affect difference respectively. At the end of the survey, participants fill out a scale to measure their psychological well-being ---The Flourishing Scale (Diener et al., 2009). This is an 8-item list that summarizes the participant's self-perceived success in salient areas such as relationships, self-esteem, purpose, and optimism. Participants rate themselves on these four areas using the 1–7 scale, with 1 indicating strongly disagree and 7 indicating strongly agree. We used the sum of the score to measure participants’ well-being. Data is validated through double checking for errors, outliers, wrong or missing values. Statistical analysis is employed in this process to perform 2(culture) x 2(condition) between subjects ANOVAs and hierarchical regressions. 1 1 Data Analyses and Results Hypothesis 1 Two-way analysis of variance was used to verify the effects of condition and culture on amount of recalls. The results are shown in Table 1-2 and Figure 2. It can be seen from Table 2 that condition (F=0.281, P=0.597), culture (F=0.015, P=0.903), and interaction term Condition* Culture (F=0.218, P=0.641) are not significant. The above results indicate that condition, culture, and the interaction between condition and culture have not been found to have a significant impact on Amount of Recall. Table1 Descriptive Statistics European Americans Asian or Asian Americans High arousal Low arousal High arousal Low arousal item condition condition condition condition Mean Std. Std. Std. Std. Deviation Mean Deviation Mean Deviation Mean Deviation IA DIFFERENC 1.32 0.41 1.41 0.51 1.24 0.50 1.35 0.63 E HAP difference 1.37 0.79 1.48 0.76 1.22 0.68 1.40 0.78 LAP difference 1.53 0.89 1.78 0.85 1.58 1.05 1.63 0.87 HAP Ideal Affect 4.22 0.67 4.12 0.54 3.80 0.72 3.85 0.84 LAP Ideal Affect 4.28 0.61 4.31 0.51 4.28 0.59 4.27 0.74 HAP Actual Affect 2.85 0.78 2.64 0.74 2.58 0.75 2.45 0.49 LAP Actual Affect 2.74 0.66 2.53 0.66 2.70 0.82 2.64 0.57 Wellbeing 46.59 7.65 45.94 6.51 43.41 6.27 40.33 7.01 Number of Recall 13.33 5.94 13.39 5.09 13.03 5.06 13.91 4.70 1 2 Table2 Tests of Between-Subjects Effects Dependent Variable: Amount of Recalls Source Type III Sum of Squares df Mean Square F Sig. Corrected Model 13.390a 3 4.463 0.162 0.922 Intercept 25451.974 1 25451.974 923.599 0.000 Condition 7.731 1 7.731 0.281 0.597 Culture .413 1 .413 0.015 0.903 Condition * Culture 6.003 1 6.003 0.218 0.641 Error 3802.920 138 27.557 Total 29346.000 142 Corrected Total 3816.310 141 a. R Squared =0.004 (Adjusted R Squared = -0.018) Figure 2. Estimated Marginal Means of Amount of Recalls The two-way ANOVA results show that hypothesis 1 does not hold. However, considering the way in which we investigate the design to limit the time and number of recalls, there may be situations where the participants have more than 20 memories to recall but have no place to answer. Therefore, we use exploratory analysis to understand the 1 3 distribution of amount of recalls. The results are shown in Figure 3. From Figure 3, it can be seen that the ratio of the amount of 20 in the different conditions and different cultures is the highest. In view of the above situation, a better way to see the cultural difference might be analyzing the percentage of participants that hit the maximum 20 recalls under different conditions and cultures, instead of analyzing the quantity of recalls. In this way, we perform two-class encoding on the variable “amount of recalls”, 20 encoding as 1, and 0 through 19 encoding as 0, and then performing a layered chi-square test on the encoded new variable “Amount of Recalls Binary” with condition and culture. The results are shown in Table 3-4. Show. The percentage distribution of “Amount of Recalls Binary” is shown in Figure 4. Figure 3. Amount of recalls in each condition 1 4 Figure 4. Percentage of “Amount of Recalls Binary” in each condition As can be seen from Table 3, there are 20 cases of “amount of recalls” = 20 under European Americans (culture 1), of which 14 cases (70%) occur under high arousal condition, and only 6 cases (30%) occur under Low arousal condition; There were 17 cases of “amount of recall” = 20 under Asians (culture 2), of which 9 (52.9%) occurred under low arousal condition and 8 (47.1%) occurred under High arousal condition. As can be seen from Table 4, the chi-square test results under European American are significant at 10% significance level (chiq=3.540, sig=0.060), which means the hypothesis that European Americans (culture 1) may retrieve more memories in high arousal condition (condition 1) is assumed to be true at 10% of the level of significance. 1 5 Table3 AmountofRecallsBinary * Condition * Culture Crosstabulation Condition Culture High Low arousal arousal Total condition condition Count 25 30 55 .00 % within AmountofRecallsBinary AmountofRecallsBinary 45.5% 54.5% 100.0% European Count 14 6 20 American 1.00 % within AmountofRecallsBinary 70.0% 30.0% 100.0% Count 39 36 75 Total % within AmountofRecallsBinary 52.0% 48.0% 100.0% Count 26 24 50 0.00 % within AmountofRecallsBinary 52.0% 48.0% 100.0% Asian or AmountofRecallsBinary Count 8 9 17 Asian 1.00 % within American AmountofRecallsBinary 47.1% 52.9% 100.0% Count 34 33 67 Total % within AmountofRecallsBinary 50.7% 49.3% 100.0% Count 51 54 105 .00 % within 48.6% 51.4% 100.0% AmountofRecallsBinary AmountofRecallsBinary Count 22 15 37 Total 1.00 % within AmountofRecallsBinary 59.5% 40.5% 100.0% Count 73 69 142 Total % within AmountofRecallsBinary 51.4% 48.6% 100.0% 1 6 Table4 Chi-Square Tests Asymp. Sig. (2- Exact Sig. Exact Sig. Culture Value df sided) (2-sided) (1-sided) Pearson Chi-Square 3.540c 1 .060 Continuity Correctionb 2.625 1 .105 European American Likelihood Ratio 3.626 1 .057 Fisher's Exact Test .072 .052 Linear-by-Linear Association 3.493 1 .062 N of Valid Cases 75 Pearson Chi-Square .124d 1 .725 Continuity Correctionb .005 1 .943 Asian or Asian Likelihood Ratio .124 1 .725 American Fisher's Exact Test .784 .471 Linear-by-Linear Association .122 1 .727 N of Valid Cases 67 Pearson Chi-Square 1.298a 1 .255 Continuity Correctionb .899 1 .343 Total Likelihood Ratio 1.305 1 .253 Fisher's Exact Test .339 .172 Linear-by-Linear Association 1.289 1 .256 N of Valid Cases 142 a. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 17.98. b. Computed only for a 2x2 table c. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 9.60. d. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 8.37. Hypothesis 2 Hierarchical regression was used to verify the predictive effect of difference scores between ideal affect and actual affect on well-being and whether there was culture by 1 7 different scores interaction effect. The first-level regression was included in culture and difference of ideal affect and actual affect, and the second-level regression was included in the interaction between culture and difference scores of ideal affect and actual affect to see if the change in fitness was significant. The results were shown in Table 5-7. The first-order regression model had a fit of 0.249, F =23.043, and the significance was less than 0.001, indicating that the first model had a significant regression relationship. From the regression equation coefficient (Table5), difference between ideal affect and actual affect had a significantly negative prediction for well-being. In the second-level regression model R- square increased by 2.5%, and the change was significant (F=4.693, P=0.032), indicating that culture had significantly interaction effect in difference of ideal affect and actual affect and well-being. Moreover, we can see that the slope of difference for European American is -8.19 while the slope of difference for East Asians is -3.7 from the Figure 5, suggesting there are cultural difference in actual and ideal affect discrepancies that matter for well- being. Table5 Model Summaryc Std. Error Change Statistics R Adjusted R of the R Square F Sig. F Model R Square Square Estimate Change Change df1 df2 Change 1 .499a .249 .238 6.333 .249 23.043 2 139 .000 2 .523b .274 .258 6.251 .025 4.693 1 138 .032 a. Predictors: (Constant), Difference between IA and AA, Culture b. Predictors: (Constant), Difference between IA and AA, Culture, culture_Difference c. Dependent Variable: Wellbeing 1 8 Table6 ANOVAa Model Sum of Squares df Mean Square F Sig. 1 Regression 1848.500 2 924.250 23.043 0.000b Residual 5575.162 139 40.109 Total 7423.662 141 2 Regression 2031.861 3 677.287 17.335 0.000c Residual 5391.801 138 39.071 Total 7423.662 141 a. Dependent Variable: Wellbeing b. Predictors: (Constant), Difference between IA and AA, Culture c. Predictors: (Constant), Difference between IA and AA, Culture, culture_Difference Table7 Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 58.644 2.219 26.429 0.000 Culture -4.743 1.067 -.327 -4.446 0.000 Difference between IA and AA -5.603 1.038 -.397 -5.397 0.000 2 (Constant) 68.144 4.902 13.902 0.000 Culture -10.729 2.957 -.741 -3.628 0.000 Difference between IA and AA -12.678 3.423 -.899 -3.704 0.000 Culture*Difference 4.491 2.073 .649 2.166 0.032 a. Dependent Variable: Wellbeing 1 9 Figure 5. Scatterplot of ideal-actual difference scores and well-being Hierarchical regression was used to verify the predictive effect of LAP ideal-actual affect difference on well-being and whether there was culture by the LAP different scores interaction effect. The first-level regression consisted of culture and LAP difference of ideal affect and actual affect, and the second-level regression consisted of the interaction factor to see if the change in fitness was significant. The results were shown in Table 8-10. The first- order regression model has a fit of 0.172, F =14.486, and the significance is less than 0.001, indicating that the first model had a significant regression relationship. From the regression equation coefficient (Table10), LAP difference score between ideal affect and actual affect had significantly negative predictive effect for well-being. In the second-level regression model R-square increased by 1.5%, but the change was not significant (F=2.492，P=0.117), 2 0 indicating that culture had no significantly interaction effect in LAP difference of ideal affect and actual affect and well-being. Table8 Model Summary Std. Error Change Statistics R Adjusted R of the R Square F Sig. F Model R Square Square Estimate Change Change df1 df2 Change 1 .415a .172 .161 6.648 .172 14.486 2 139 .000 2 .433b .187 .169 6.613 .015 2.492 1 138 .117 a. Predictors: (Constant), Culture, LAPdifference b. Predictors: (Constant), Culture, LAPdifference, LAP_culture Table9 ANOVAa Model Sum of Squares df Mean Square F Sig. Regression 1280.478 2 640.239 14.486 .000b 1 Residual 6143.184 139 44.196 Total 7423.662 141 Regression 1389.437 3 463.146 10.592 .000c 2 Residual 6034.225 138 43.726 Total 7423.662 141 a. Dependent Variable: Wellbeing b. Predictors: (Constant), Culture, LAPdifference c. Predictors: (Constant), Culture, LAPdifference, LAP_culture Table10 Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 54.519 2.027 26.890 .000 LAPdifference -2.262 .614 -.284 -3.685 .000 Culture -4.498 1.118 -.311 -4.024 .000 2 (Constant) 59.314 3.646 16.267 .000 LAPdifference -5.191 1.953 -.653 -2.658 .009 Culture -7.641 2.280 -.528 -3.351 .001 LAP_culture 1.929 1.222 .439 1.579 .117 a. Dependent Variable: Wellbeing 2 1 Figure 6. Scatterplot of LAP ideal-actual difference scores and well-being We used the same method to see the predictive effect of HAP ideal-actual affect difference on well-being and whether there was culture by HAP different scores interaction effect. The first-level regression consisted of culture and HAP difference of ideal affect and actual affect, and the second-level regression consisted of the interaction factor to see if the change in fitness was significant. The results were shown in Table 11-13. The first-order regression model has a fit of 0.155, F =12.749, and the significance is less than 0.001, indicating that the first model had a significant regression relationship. From the regression equation coefficient (Table13), HAP difference score between ideal affect and actual affect had significantly negative predictive effect for well-being. In the second-level regression model R-square increased by 0.7%, but the change was not significant (F=1.16，P=0.283), 2 2 indicating that culture had no significantly interaction effect in HAP difference of ideal affect and actual affect and well-being. Table11 Model Summary Std. Error Change Statistics R Adjusted R of the R Square F Sig. F Model R Square Square Estimate Change Change df1 df2 Change 1 .394a .155 .143 6.718 .155 12.749 2 139 .000 2 .403b .162 .144 6.714 .007 1.160 1 138 .283 a. Predictors: (Constant), HAPdifference, Culture b. Predictors: (Constant), HAPdifference, Culture, HAP_culture Table12 ANOVAa Model Sum of Squares df Mean Square F Sig. 1 Regression 1150.742 2 575.371 12.749 .000b Residual 6272.920 139 45.129 Total 7423.662 141 2 Regression 1203.011 3 401.004 8.896 .000c Residual 6220.651 138 45.077 Total 7423.662 141 a. Dependent Variable: Wellbeing b. Predictors: (Constant), HAPdifference, Culture c. Predictors: (Constant), HAPdifference, Culture, HAP_culture Table13 Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 54.413 2.104 25.857 .000 Culture -4.668 1.133 -.322 -4.121 .000 HAPdifference -2.433 .754 -.253 -3.229 .002 2 (Constant) 57.670 3.684 15.655 .000 Culture -6.886 2.350 -.475 -2.930 .004 HAPdifference -4.794 2.318 -.497 -2.068 .041 HAP_culture 1.631 1.515 .291 1.077 .283 a. Dependent Variable: Wellbeing 2 3 Figure 7. Scatterplot of HAP ideal-actual difference scores and well-being In addition, we employed independent samples T-Test to see culture effect in HAP ideal affect, LAP ideal affect, HAP actual affect and LAP actual affect. The results in Table 14-15 showed that The HAP ideal affect of European Americans was higher than that of Asians or Asian Americans, and the difference was statistically significant; there was no significant cultural difference in LAP ideal affect, HAP actual affect and LAP actual affect. Table14 Group Statistics Culture N Mean Std. Deviation Std. Error Mean HAP Ideal 1 75 4.171 .6113 .0706 Affect 2 67 3.824 .7750 .0947 LAP Ideal 1 75 4.293 .5602 .0647 Affect 2 67 4.275 .6623 .0809 1 75 2.747 .7615 .0879 HAP Actual Affect 2 67 2.516 .6355 .0776 1 75 2.640 .6641 .0767 LAP Actual Affect 2 67 2.672 .7030 .0859 2 4 Table15 Independent Samples Test Levene's Test for Equality of Variances t-test for Equality of Means 95% Confidence Sig. Std. Interval of the (2- Mean Error Difference taile Differe Differe Low F Sig. t df d) nce nce er Upper Equal HAP variances 1.48 9 .224 2.97 5 140 .003 .3468 .1165 .116 .5772 Ideal assumed 4 Affect Equal variances not 2.93 125.2 .004 .3468 .1181 .113 .5805 assumed 6 49 1 Equal variances 1.26 - LAP 5 .263 .182 140 .856 .0187 .1026 .184 .2216 Ideal assumed 2 Affect Equal -variances not .181 129.9 .857 .0187 .1036 .186 .2236 assumed 82 2 Equal - HAP variances 4.54 7 .035 1.94 3 140 .054 .2302 .1185 .004 .4645 Actual assumed 0 Affect Equal -variances not 1.96 139.3 .052 .2302 .1173 .001 .4622 assumed 3 78 7 Equal - LAP variances .063 .802 - .276 140 .783 -.0316 .1148 .258 .1953 Actual assumed 5 Affect Equal -variances not - 136.0.275 60 .784 -.0316 .1151 .259 .1961 assumed 3 Hypothesis 3 The hierarchical regression method is used to verify the effect of amount of recall on level of wellbeing and the effect of culture between amount of recall and level of wellbeing under different conditions. The results are shown in Table 16-21. Table 16 - Table 18 are the hierarchical regression results under Condition 1; Table 19 - Table 21 are the 2 5 hierarchical regression results under Condition 2; Table 16 shows the first-level regression model fit under Condition 1 is 0.051, F = 1.895, and sig=0.158, indicating that the regression relationship of the first model is not significant. The amount of recall under condition1 has no significant predictive effect on level of wellbeing. The fit of the second- order regression model under condition=1 is 0.061. F is 1.497 and sig = 0.223, indicating that the regression relationship of the second model is not significant, that is, the interaction between culture and amount of recall is not significant for the prediction of level of wellbeing. Table16 Model Summaryc Std. Error Change Statistics R Adjusted R of the R Square F Sig. F Model R Square Square Estimate Change Change df1 df2 Change 1 .227a .051 .024 7.088 .051 1.895 2 70 0.158 2 .247b .061 .020 7.102 .010 .719 1 69 0.400 a. Predictors: (Constant), Amount of Recalls, Culture b. Predictors: (Constant), Amount of Recalls, Culture, culture_amountofrecall c. Dependent Variable: Wellbeing Table17 ANOVAa Model Sum of Squares df Mean Square F Sig. Regression 190.359 2 95.179 1.895 .158b 1 Residual 3516.764 70 50.239 Total 3707.123 72 Regression 226.609 3 75.536 1.497 .223c 2 Residual 3480.515 69 50.442 Total 3707.123 72 a. Dependent Variable: Wellbeing b. Predictors: (Constant), Amount of Recalls, Culture c. Predictors: (Constant), Amount of Recalls, Culture, culture_amountofrecall 2 6 Table18 Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 49.001 3.302 14.841 0.000 Culture -3.161 1.664 -0.221 -1.900 0.062 Amount of Recalls 0.056 .152 0.043 0.371 0.712 2 (Constant) 53.839 6.596 8.162 0.000 Culture -6.637 4.426 -0.465 -1.499 0.138 Amount of Recalls -0.310 .458 -0.238 -0.677 0.501 Culture*amountofrecall 0.264 .312 0.380 0.848 0.400 a. Dependent Variable: Wellbeing It can be seen from Table 19 that the first-order regression model fit under condition 2 is 0.177, F = 7.081, and sig = 0.002, indicating that the first model has a significant regression relationship; as shown in Table 21, the regression coefficient of amount of recall is positive, but t=1.441, sig=0.154, indicating that under Conditon 2, the amount of recall has a slight trend toward significance. The second-level regression model fit under Condition 2 is 0.177, F = 4.675, sig = 0.005, indicating that the regression relationship of the second model is significant; from Table 21, the regression coefficient of amount of recall*culture is not significant (P<0.05), indicating that under condition 2, amount of recall*culture is not significant for the prediction of level of wellbeing. Table19 Model Summaryc Std. Error Change Statistics R Adjusted R of the R Square F Sig. F Model R Square Square Estimate Change Change df1 df2 Change 1 .420a .177 .152 6.697 .177 7.081 2 66 0.002 2 .421b .177 .139 6.745 .001 .063 1 65 0.803 a. Predictors: (Constant), Amount of Recalls, Culture b. Predictors: (Constant), Amount of Recalls, Culture, culture_amountofrecall c. Dependent Variable: Wellbeing 2 7 Table20 ANOVAa Model Sum of Squares df Mean Square F Sig. 1 Regression 635.175 2 317.588 7.081 0.002b Residual 2960.129 66 44.850 Total 3595.304 68 2 Regression 638.026 3 212.675 4.675 0.005c Residual 2957.279 65 45.497 Total 3595.304 68 a. Dependent Variable: Wellbeing b. Predictors: (Constant), Amount of Recalls, Culture c. Predictors: (Constant), Amount of Recalls, Culture, culture_amountofrecall Table21 Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 48.466 3.308 14.653 0.000 Culture -5.736 1.616 -0.397 -3.549 0.001 Amount of Recalls 0.240 .167 0.161 1.441 0.154 2 (Constant) 46.812 7.402 6.324 0.000 Culture -4.577 4.906 -0.317 -0.933 0.354 Amount of Recalls .362 0.514 0.243 0.703 0.484 Culture*amountofrecall -.085 0.338 -0.121 -0.250 0.803 a. Dependent Variable: Wellbeing Discussion In this study, we investigated how the recalls of emotional memory (i.e. memory of ideal affect) were associated with personal well-being between East Asian and European American cultures. Extending to previous findings by Jeanne Tsai that examined depressive symptoms (2006), our results showed that both HAP and LAP consistency between ideal affect and actual affect also predicts positive measure of personal well-being, and the distinct finding in our study is that we found there did exist cultural difference in actual and ideal affect consistency that matter for health and well-being. Meanwhile, we also 2 8 discovered that European Americans may retrieve more memories in high arousal condition at 10% of the level of significance. The results for each hypothesis are explained in more detail, respectively, in the following sections. Cultural and condition impact on amount of recalls The initial ANOVA results showed that under the two the conditions, there is no significant difference in the amount of recalls by European Americans and East Asians. However, it can be seen that the proportion of the participants who hit the amount of 20 is the highest among in both cultures and both conditions. Considering that our survey design has a four minutes timer to limit the answer time and maximum number of recalls is 20, there may be situations where the participants have more than 20 memories to recall but have no place or no time to answer. Regarding the above situation, we used another method to see the cultural difference in the data. Instead of analyzing the quantity of recalls, we analyze the percentage of participants that hit the maximum 20 recalls under different conditions and cultures. Specifically, we created a new variable “Amount of Recalls Binary” by performing two- class encoding on the variable “amount of recalls”, 20 encoding as 1, and 0 through 19 encoding as 0. Then we performed a chi-square test on the encoded new variable “Amount of Recalls Binary” with condition and culture. The chi-square test results by European American were marginal significant (sig=0.060), which means the hypothesis that European Americans may retrieve more memories in high arousal condition is assumed to be true at 10% of the level of significance. In contrast, there was still no significant effect on East 2 9 Asians under high arousal and low arousal conditions (sig=0.725), which means that part of the hypothesis one should be rejected. Consistency between ideal affect and actual affect predicts well-being As we expected, hierarchical regression showed that difference between ideal affect and actual affect has a significant negative predictive effect on well-being. In other words, for both cultural groups, the smaller the difference between ideal affect and actual affect, the higher the well-being level, which supported the hypothesis two. We also separated HAP and LAP ideal-actual affect discrepancies and applied same regression models to analyze the data, and the results indicated that both HAP and LAP difference scores between ideal affect and actual affect had significantly negative predictive effect for well- being. Moreover, we can see that the slopes of difference scores (the general ideal-actual affect discrepancy scores, HAP ideal-actual affect discrepancy scores and LAP ideal-actual discrepancy scores showed similar results) for European American are twice as much as the slope of difference for East Asians from the scatterplot. This suggested that compared with Asians, ideal-actual discrepancy scores had a greater predictive effect on European Americans’ well-being. Finally, the independent samples T-Test showed HAP ideal affect of European Americans was higher than that of Asians or Asian Americans, which is consistent with Tsai’s research. Culture and condition impact on Well-being To test our third hypothesis, we performed a hierarchical regression for each condition. 3 0 In high arousal condition, the result showed that the regression model is not significant, and there was no significant interaction between culture and amount of recalls, indicating that the amount of recalls under high arousal condition had no significant predictive effect on level of wellbeing. In low arousal condition, the amount of recalls had a positive predictive effect on level of wellbeing and has slight trend toward significance. There was no significant interaction found between amount of recalls and culture wellbeing. Limitations and future directions The study has several limitations that should to be mentioned. First, sample size is not large enough, future studies may recruit more participants in each condition and each cultural group to perform a culture by condition study. Second, a proportion of the students are Asian Americans, which may have limited the potential variation in low arousal affect orientation to East Asian culture. Future researchers are suggested to collaborate with universities in East Asia to recruit indigenous participants. Third, the design of the questionnaire should also be adjusted to fit different participants and various situations. For example, it should allow unlimited number of entries in the memory recall section. Moreover, there might be potential influence of memory recall on the affect valuation section. When participants are assigned to recall certain affective states, participants might realize that the experimenter was just asking them to one specific type of memory and subconsciously change their response. In sum, we believed that our study could enlighten future researchers to further explore cultural difference in ideal affect, emotional memory, and their relation to health and well-being. 3 1 Reference Buunk, A. P., & Gibbons, F. X. (2007). Social comparison: The end of a theory and the emergence of a field. Organizational Behavior and Human Decision Processes, 102, 3-21. Caudill, W., & Weinstein, H. (1969). Maternal care and infant behavior in Japan and America. Psychiatry, 32(1), 12-43 Diener, E., Suh, E. M., Lucas, R. E., & Smith, H. L. (1999). Subjective well-being: Three decades of prog- ress. Psychological Bulletin, 125, 276-302. Diener, E., Wirtz, D., Tov, W., Kim-Prieto, C., Choi, D., Oishi, S., & Biswas-Diener, R. (2009). New measures of well-being: Flourishing and positive and negative feelings. Social Indicators Research, 39, 247-266. Ekman, P., & Friesen, W. (1978). Facial action coding system: A technique for the measurement of facial movement. Palo Alto, CA: Consulting Psychologists Press. Higgins, E. & Klein, Ruth & Strauman, Timothy. (1985). Self-Concept Discrepancy Theory: A Psychological Model for Distinguishing among Different Aspects of Depression and Anxiety. Social Cognition. 3. 51-76. 10.1521/soco.1985.3.1.51. Higgins, E. T. (1987). Self-discrepancy: A theory relating self and affect. Psychological Review, 94(3), 319–340. James A. Russell. (1991). Culture and the Categorization of Emotions. Psychol Bull : 110:426–50. Lisa A. Feldman. (1999). Variations in the Circumplex Structure of Mood. Personality and Social Psychology Bulletin, 21(8), 806–817. 3 2 McManus, Maria D., Jason T. Siegel, and Jeanne Nakamura. "The Predictive Power of Low-Arousal Positive Affect." Motivation and Emotion, 2018, pp. 1-15. Papousek, M., Papousek, H., & Symmes, D. (1991). The meanings of melodies in motherese in tone and stress languages. Infant Behavior and Development, 14, 415–440. Philippe, F. L., Koestner, R., Beaulieu-Pelletier, G., Lecours, S., & Lekes, N. (2012). The Role of Episodic Memories in Current and Future Well-Being. Personality and Social Psychology Bulletin, 38(4), 505–519. Ryan, R. M., & Deci, E. L. (2000). Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. American Psychologist, 55(1), 68. Scheibe, S., English, T., Tsai, J. L., & Carstensen, L. L. (2013). Striving to feel good: idea affect, actual affect, and their correspondence across adulthood. Psychology and aging, 28(1), 160. Tamir, Maya & Schwartz, Shalom & Oishi, Shigehiro & Kim, Min. (2017). The Secret to Happiness: Feeling Good or Feeling Right?. Journal of Experimental Psychology: General. 146. Tran, A. G. T. T., Su, J. C., Chong, S. S., & Woei-Haur, L. (2017). When the Grass Could Be Greener: Psychological Correlates of Positive Affect Discrepancies in White American and Taiwanese Samples. Journal of Cross-Cultural Psychology, 48(6), 931–949. Tsai, Jeanne L., Knutson, Rung. (2006). Cultural Variation in Affect Valuation. J Pers Soc Psychol: 288–307.15. Tsai, Jeanne L. "Ideal Affect: Cultural Causes and Behavioral Consequences." Perspectives on Psychological Science, vol. 2, no. 3, 2007, pp. 242-259. 3 3 Tsai, Jeanne.L., Ang, J., Blevins, E., Goernandt, J., Fung, H.H., Jiang, D., Elliott, J., Kölzer, A., Uchida, Y., Lee, Y.-C., Lin, Y., Zhang, X., Govindama, Y., & Haddouk, L. (2016). Leaders' smiles reflect cultural differences in ideal affect. Emotion, 16(2), 183-195. Tsai, Jeanne L., Miao, F.F., & Seppala, E. (2007). Good feelings in Chris- tianity and Buddhism: Religious differences in ideal affect. Personality and Social Psychology Bulletin, 33, 409–421 Tsai, Jeanne L., & Louie, Jennifer & E Chen, Eva & Uchida, Yukiko. (2007). Learning What Feelings to Desire: Socialization of Ideal Affect Through Children's Storybooks. Personality & social psychology bulletin. 33. 17-30. Tsai, Jeanne L. (2017). Ideal affect in daily life: implications for affective experience, health, and social behavior. Current opinion in psychology, 17, 118–128. Tsai, J. L., & Knutson, B. (2006). The Affect Valuation Index: Reliability and validity. Manuscript in preparation. Wang, Q., Hou, Y., Koh, J. B. K., Song, Q., & Yang, Y. (2018). Culturally Motivated Remembering: The Moderating Role of Culture for the Relation of Episodic Memory to Well-Being. Clinical Psychological Science, 6(6), 860–871. 3 4