Abstract
The relationship between culture and autobiographical memory has become an important research topic since research over the past three decades has shown (cross)-cultural variations in the content and style of autobiographical memories that are related to differences in how people from diverse cultures describe themselves and others. However, most research has compared North-American to Asian cultures, leaving a need to address the relationship in others. Our study examines differences in characteristics of autobiographical memory in three countries: Mexico, Spain and Denmark, using Qi Wang’s methodology and categories of analysis. College students from these three countries were asked to describe their earliest memory. Results confirm the expected differences, with Danish and Spanish participants’ earliest memories dating significantly earlier than that of the Mexican participants. At the same time, Danish participants were less likely to spontaneously mention emotions and feeling states in their earliest memories. On the other hand, Mexicans demonstrated a higher level of autonomous orientation than the participants in the other two groups. Gender was found to interact with several indicators, and even displayed a cross-over effect in emotion. The findings are interpreted as supporting the view of cultured self-construals, while also suggesting the need for more attention to the situated character of cultural differences.
Autobiographical memory can be defined as memory for facts and events of one’s life (Conway, 2005; Conway and Rubin, 1993; Rubin, 1986). This form of memory is closely related to self. For many authors in fact, self and autobiographical memory constitute each other since on the one hand, personal memories are cognitively organised around the self (Conway and Pleydell-Pearce, 2000; Nelson, 2003), and on the other, the self cannot constitute itself and exist through time without the existence of personal memories in subjective time. We can say that autobiographical memory allows us to exist through time.
Research over the past two decades has shown important (cross)-cultural variations in the content and style of autobiographical memories that are related to differences in how people from diverse cultures and countries describe themselves and others. These studies have analysed the way in which predominant cultural values in different cultures and countries shape the autobiographical memories of their members. Such values are transmitted implicitly, especially through socialisation practices in which parents and children talk about the past (Demuth et al., 2007; Fivush, 2007; Wang, 2006; Wang and Fivush, 2005).
The number of studies on the relationship between autobiographical memory and culture, however, is limited. In general, they have focused on American and Asian (Chinese, Korean, Japanese) cultures (exceptions are Rubin et al., 2007, who included a Turkish sample; MacDonald et al., 2000, who compared three ethnic groups in New Zealand, and Harpaz-Rotem and Hirst, 2005, who compared adults from three kinds of Israeli upbringing). Thus, a comparison of a wider range of cultures and countries is needed for further understanding of this relationship. In this sense, there are very few studies about Scandinavian, Mediterranean, Latin-American or East European countries.
Cultural self-construals and autobiographical memory
Psychological and anthropological research has evidenced that conceptual representations of the self vary across cultures as a function of differing values and social orientations (Hui and Triandis, 1986; Kagitçibasi, 1997, 2005, 2007; Shweder and Bourne, 1984; Triandis, 1995). Cultural self-construal serves as a kind of filter for individuals’ personal memories, and these memories can help to develop a particular cultural self-construal.
Traditionally, most analyses of the relationship between culture and self-construal have been based on a classical distinction between individualistic and collectivistic cultures. The cultural value tradition of Hofstede (1980, 2001), for instance, has distinguished individualism–collectivism (I-C; among others) as a core dimension on which to model cultures. Scores on these dimensions were obtained for 50 countries on the basis of the IBM large-scale survey. Briefly, collectivism prevails in less industrialised and Eastern countries, while individualism prevails in European American countries. In a similar vein, Triandis (1989) has defended a dichotomic perspective about cultures when he has referred to individualism and collectivism (for a review and some theoretical extensions, see Brewer and Chen, 2007; Morling and Lamoreaux, 2008; Oyserman et al., 2002, etc.).
Markus and Kitayama’s (1991) psychological theory of self-construal is built upon the I-C dimension of cultural values. They distinguish between independent and interdependent self-construal. According to Markus and Kitayama, the independent self-construal is characteristic of Anglo-American, Northern and Central European cultures. This independent self is conceived as a separate and autonomous entity, defined by a unique repertoire of traits, capacities, thoughts and feelings. The focus tends to be on asserting one’s needs, as well as preserving individuality, uniqueness and independence. Thus, the self tends to remain relatively consistent across situations. In contrast, the interdependent self is characteristic of non-Western cultures (Asian, African and Latin-American cultures). In this case, the self is experienced as part of a social web (Cross and Markus, 1999; Kagitçibasi, 1997, 2005; Markus and Kitayama, 1991; Triandis, 1995). For a person with a highly interdependent self-construal, the boundaries of the self tend to be more fluid and include significant others in specific situations; the concept of the self tends to be less static and more contextualised and situated. These differences in the emphasis on independence versus interdependence are evidenced in diverse aspects of the self such as the experience and expression of emotions, cognitive processes, attributions and moral reasoning or achievement motivation, among others (Markus and Kitayama, 1991).
Notwithstanding, authors like Kagitçibasi (1997, 2005, 2007) have rejected the traditional consideration of autonomy as opposed to relatedness, which is characteristic of Western psychology. For Kagitçibasi, the Western individualistic concept of autonomy connotes two different meaning dimensions. One is related to personal separateness–relatedness (the degree of distancing the self from others). She calls this the ‘interpersonal distance’ dimension, with two poles (separatedness and relatedness). The other is the ‘agency’ dimension, which extends from agency (autonomy) to dependency (heteronomy). The distinction between autonomous and heteronomous morality suggested by Piaget is mentioned as an example of this dimension. The main thesis defended by Kagitçibasi is that these two dimensions are independent, so that agency does not necessarily mean separateness, contrary to what is usually believed in Western psychology (and Western thinking in general).
Kagitçibasi proposes the construct of autonomous-related self that has both logical and psychological validity, but it is not sufficiently recognised in psychology. This construct is based on a reconceptualisation of autonomy, evoking its meaning of agency and untangling it from personal distance. The autonomous-related self-model would be prevalent in urban contexts of traditionally collectivistic cultures in the ‘Majority world’. It is emerging from socioeconomic development and urbanisation in these societies (i.e. Mexico).
In demonstrating cultural and country differences in autobiographical memory in research, one of the aspects that has received the most attention is the age at the earliest memory (Wang, 2001b, 2006; Wang and Conway, 2004). Evidence shows that first memories of individuals from cultural groups that emphasise the importance of the personal past are earlier than in other cultural groups (MacDonald et al., 2000; Mullen, 1994; with additional support in Harpaz-Rotem and Hirst, 2005). When asked to recall their earliest childhood memory, Europeans and Caucasian Americans remember events back to, on average, age 3.5 years, whereas Asian adults’ first memories are dated about 6 months later (MacDonald et al., 2000; Mullen, 1994; Wang, 2001a; Wang and Ross, 2005). The cultural variation in the age at the earliest memory is thought to stem from different cultural conceptions of selfhood (Markus and Kitayama, 1991; Mullen, 1994; Pillemer, 1998; Wang, 2001a, 2003). The autonomous self-construal characteristic of Western cultures may drive the early emergence of an organised, articulated, durable memory system for events that happened to ‘me’. In contrast, a relational self-construal that focuses on community rather than agency may de-emphasise individuality and promote social integration and dependence. In addition, earliest memories reported by Caucasian Americans are more elaborated, specific, self-focused, emotionally elaborate and less socially oriented than memories of Asian people (Han et al., 1998).
With regard to self-construal in Denmark, Spain and Mexico, there is some evidence of differences between these countries in terms of individualism versus collectivism (Hofstede, 1980, 2001; Oyserman et al., 2002). Hofstede’s (2001) data show a high score in Denmark (79, with only a few countries scoring higher), a relatively low score in Mexico (30, which is low, although slightly higher than other Latin-American countries, with an average of 21) and an intermediate level in Spain (51, which is lower than Central and Northern European countries, although higher than Portugal and Greece, for instance).
The emotional content of personal memories is of particular interest. Cross-cultural studies have indicated that for independently oriented selves, emotions are often regarded as a direct expression of the self and as an affirmation of the importance of the individual. Research on cultural differences in emotion has increased considerably in the past two decades. Underlying much of this work has been an assumption that cultural differences in emotion result from or reflect differences in cultural values along the I-C dimension, and related constructions of the self (Nezlek et al., 2008).
As Leichtman et al. (2003) note, understanding which emotions are appropriate reactions to particular situations is an important precursor to effective remembering of personal events. Some studies analysing the relationship between culture, emotion situation knowledge and autobiographical memory have reported cultural variations in emotion situation knowledge and autobiographical memory. In general, American children and adults make more references to internal states (including emotions), rate emotional situations as more intense (especially for negative emotions such as sadness and fear) and show a greater understanding of emotional situations than Asian (Chinese and Korean) children and adults (Han et al., 1998; Wang, 2001b, 2003, 2004). These differences in emotion situation knowledge are correlated with differences in autobiographical remembering (Wang, 2006).
Differences in emotion situation knowledge have been interpreted in relation to differences in the cultural values ascribed to the expression of emotions. Higher tendency to reflect on and express their own emotions and internal states in American children and adults may reflect the greater value that the American culture poses on individuality and autonomy. Conversely, Chinese and other East-Asian cultures give priority to the maintenance of harmony within social groups. The tendency to moderate and control the expression of emotions (especially those that might be potentially dangerous for social harmony) is consistent with this different cultural emphasis on relatedness (Kagitçibasi, 2005).
Autobiographical memory and gender
Cultural differences in autobiographical memory and self overlap with gender differences in both processes. Studies comparing autobiographical memory between males and females have evidenced differences in issues like the age at the earliest memory, the form and the content of autobiographical memories (Cala and De la Mata, 2010).
With regard to the age of a person’s earliest memory, some studies have shown that women’s first memories are earlier (Davis, 1999; Hayne and MacDonald, 2003), while other studies have found no gender difference (Wang, 2001a; Wang and Fivush, 2005).
With regard to the form and the content, females tend to report longer and more detailed memories than men (Buckner and Fivush, 1998; Fivush, 2007, 2011), especially for events that occurred after the age of seven (Bauer et al., 2003). Gender differences in specificity have also been found, with women showing a higher amount and density of specific memories (Pillemer et al., 2003). Self-themes vary as well, with predominance of themes related to communion or affiliation in females and of themes related to agency and autonomy in males (Buckner and Fivush, 1998; Niedzwienska, 2003; Pillemer et al., 2003). Regarding emotions and internal states, females show a greater variety of emotions and a higher tendency to provide explanations of emotional states than males. The above tendencies are augmented for sad events (Bauer et al., 2003; Bird and Reese, 2006; Buckner and Fivush, 1998). These gender differences, however, may vary across cultures (Han et al., 1998; Hayne and MacDonald, 2003; Wang, 2001a, 2003; Wang and Fivush, 2005), suggesting complexity of gender effects on autobiographical memory, and interrelations between gender and culture. Further research is needed to help us understand the construction of gendered autobiographical memory in context.
To summarise, our study tries to contribute to a better understanding of the relationship between autobiographical memory and self by comparing autobiographical memories in Mexican, Spanish and Danish college students. These countries are different in terms of their dominant self-construal: more oriented towards independence in the Danish participants, more oriented towards interdependence in Mexicans, with the Spaniards being in an intermediate position (Hofstede, 2001; Markus and Kitayama, 1991). In accordance with previous cross-cultural studies on this field, we address two main research questions. The first one is if there are differences in the characteristics of autobiographical memories (such as the age at the earliest memory, emotionality, specificity and focus on the self vs. others) between the countries investigated. The second is if there are also gender differences in these characteristics.
The study
Our study seeks to replicate Wang’s (2001a) findings on the relationship between cultural self-construal and autobiographical memory focusing on three countries, Denmark, Spain and Mexico, that vary in terms of individualism and collectivism on Hofstede’s (1980, 2001) scale. With this study, we attempt to extend findings from comparisons between North-American and Asian cultures to other cultures. For that reason, we have used Wang’s methodology and categories of analysis. We also pay special attention to possible gender differences and to the interaction between culture and gender.
More specifically, the aims of this study are the following: (1) to compare Mexican, Spanish and Danish students on the characteristics of their earliest memories, (2) to analyse gender differences in the earliest memory and (3) to study the interaction between country and gender and its influence on autobiographical memory.
In general, we expect to observe country differences in characteristics of autobiographical memory in the sense that Danish students’ autobiographical memories will be earlier, longer and more specific than those reported by Mexican participants, with the memories from the Spanish participants being at an intermediate level.
Method
Participants
A total of 232 college students from Aalborg University in Denmark (n = 37 females and 37 males), the University of Seville in Spain (n = 40 females and 38 males) and the Autonomous University of Tamaulipas in Mexico (n = 40 females and 40 males) participated in the study. They were all born in the country of their university (Denmark, Spain and Mexico). Their ages ranged from 18 to 25 years. The socioeconomic background of the participants was similar and equivalent in the three country sub-samples. None of them reported any sensory handicap that may affect their resolution of the tasks. The students were recruited in psychology classes. Participation was entirely voluntary, unpaid and based on informed consent.
Design
The independent variables of our study were country (Denmark, Spain and Mexico) and gender (males and females). The dependent measures were concerned with autobiographical memory (earliest memory), and largely based on Wang (2001, 2004). The different measures employed are presented below, along with definitions of the categories.
Materials and procedure
The materials were taken from Wang (2001a, 2004). A written questionnaire was used. In this questionnaire, participants wrote their earliest memory. They had as much space as they needed for their answers and no time limit was given to them. The participants were asked to estimate their age at the time of the remembered event (‘How old were you?’ in years and months) and to rate its emotionality (‘How did you feel in those moments?’ rated on a five-point scale from ‘very well’ to ‘very bad’). They were also asked about the kind of emotions they experienced (‘What specific emotions did you experience at the time?’) and whether they had ever talked or thought about this memory before. Finally, they rated their tendency to rehearse childhood memories (‘In general, how often have you thought or talked about your childhood memories?’ rated on a five-point scale from ‘never until now’ to ‘very often’).
The questionnaire was presented in the original language of the participants (Danish or Spanish). The questionnaires were translated and back translated between English and Spanish, between English and Danish and between Danish and Spanish.
Participants completed the questionnaires in their class settings. The questionnaire forms included instructions for carrying out the task. Participants were asked to take a moment to think about the earliest memory they could recall and then to describe this memory in detail, and it was emphasised that it must be what they remembered themselves, not something heard from parents or others. After they wrote the memory, participants answered the above-mentioned questions about (a) how old they were when the remembered event happened, (b) emotions and (c) prior recollection.
Coding
A manual for content coding of the memories was derived from Wang’s (2001a, 2004) studies of preschoolers and college students from China and the United States. All coding was performed on participants’ responses in their original language. The categories were the following:
Age at the earliest memory
The estimated age at the earliest memory reported by the participants was coded in months. When a participant gave an estimated span (3–4 years), the intermediate value (42 months) was coded.
Memory volume
The number of words was counted for each memory as an index of memory volume. Previous research indicates that the use of word count provides a proximal index of volume in each language (Wang, 2001a, 2004).
Memory content
The memory was coded into one of the following categories that reflect different context of the participant’s life (Waldfogel, 1948; in Wang, 2001a):
Individual. Things that concern the individual (success, frustration, dreams, nightmares). For example, ‘That day I felt very angry because I really behaved well’).
Family (parental discipline, family activities). For example, ‘Every day, my siblings and I went to visit my granddad, who was ill’.
Neighbourhood. Playing, arguing, and so on related to neighbours and friends. For example, ‘One day, my friends and I went fishing in the river…’.
School. Activities or situations related with school, such as teachers, subjects and classes. For example, ‘I liked my Maths teacher very much’.
Memory specificity
Each memory was coded as either specific or general. Specific memories refer to events that happened at one particular moment in time (‘When I was about 4, I got stung by a bee’), while general memories refer to events that took place regularly or on multiple occasions (‘My mother helped me to do my homework every day’) (Pillemer, 1998; Wang, 2001a).
Memory emotionality
Three different aspects were considered in relation to emotions:
Appearance and frequency of reference to emotions in the memory description (coded by researchers).
Emotion experienced during the remembered event (participant’s rating on a five-point scale from very unpleasant, 1, to very pleasant, 5).
Specific emotions reported in the questionnaire (answer to open-ended question, coded by researchers).
Autonomous orientation
The participants’ tendency to express autonomy and self-determination in their memories was coded according to the narrative content analysis developed by Wang and Leichtman (2000). For that purpose, the number of occurrences of the following instances was counted and summed up to produce a general score for autonomous orientation:
Reference to personal needs, desires or preferences (‘I really wanted to open my birthday gift’).
Reference to personal dislikes or avoidance (‘I didn’t like that walk’).
Reference to personal judgements, opinions or evaluations of people, objects or events (‘I loved my birthday gift’).
Reference to control over own actions, or to pressure of a social groups they belonged to ‘I wanted to be in the boy scouts group, but they asked so many things’.
Others/self ratio
The number of times the participants mentioned himself or herself and the number of times he or she mentioned other people was counted. The ratio others/self was calculated by dividing the number of mentions of others by the number of the mentions of the self.
Prior recollection
Two answers in the questionnaire addressed prior recollection:
Frequency of recollection of the earliest memory.
Frequency of talking and thinking about childhood memories in general, in a five-point scale, ranking from 1 (never before) to 5 (very often).
Results
Descriptive statistics for all memory measures according to country and gender are presented in Table 1. Before analysing the relationship between the two explanatory factors and the dependent variables, normality tests (Kolmogorv–Smirnov) were applied to the distribution of all memory measures.
Descriptive statistics for autobiographical memory variables by country and gender.
SD: standard deviation.
None of the memory variables fit into a normal distribution in relation to culture. The only group that did not significantly differ from this distribution was that of the Mexicans in the measure of the number of words in their memories. With regard to the second factor, gender, none variables fit into a normal distribution.
Since the results of the variables of our study did not fit into a normal distribution, non-parametric (Kruskal–Wallis and Mann–Whitney) tests were selected to determine the effect of the two independent factors, country and gender, on numerical variables, while chi-square tests were used for categorical variables, memory specificity and content. Tables 2 and 3 show the rank means obtained in the memory variables by country and gender.
Mean ranks for autobiographical memory variables by country.
Mean ranks for autobiographical memory variables by gender.
Analysis of memory volume by using non-parametric Kruskal–Wallis tests revealed no significant effect of country (p = .122, n.s.; η2 = .025) and gender (p = .066, n.s.; η2 < .001) on the number of words.
Results also revealed a significant effect of country on the age at the earliest memory (p < .001; η2 = .102). Multiple comparisons using Mann–Whitney U tests showed that Mexican participants reported significantly later memories than did the other two cultural groups (Mexicans–Spanish, Z = −4.209; p < .001; Mexicans–Danes, Z = −4.362; p < .001). No significant difference between Spanish and Danish participants in this variable was observed (Z = −.168; p = .867). Non-parametric Kruskal–Wallis tests revealed no significant effect of gender on the age at the earliest memory (p = .779).
Analysis of the average number of spontaneous mentions of emotions revealed a significant effect of country on participants’ spontaneous references to emotions in their earliest memories (p < .001; η2 = .105). Multiple comparisons using Mann–Whitney U tests showed that Mexican and Spaniards participants reported significantly more spontaneous emotional references than did Danes (Mexican–Danes, Z = −5.678; p < .001, Spaniards–Danes, Z = −4.558; p < .001). No significant differences were observed between Mexicans and Spaniards (Z = −1.530; p = .136). Results also revealed no significant effect of gender in this variable (p = .364).
With regard to autonomous orientation, results showed a significant main effect of country on this variable (p = .031; η2 = .033). Multiple comparisons using the Mann–Whitney test showed that Mexican participants reported significantly more autonomy and self-determination in their memories than did the Danes (Z = 2.409, p = .016). No significant differences were observed in the other two comparisons (Mexicans–Spaniards, Z = 1. 924, p = .54, Spaniards–Danes, Z = −.992, p = .321). Results also revealed a significant effect of gender on autonomous orientation (p = .026; η2 = .015). Women reported significantly more autonomy and self-determination in their memories than did men (women, M = 2.92, standard deviation (SD) = 2.12; men, M = 2.34, SD = 1.77).
With regard to the ratio others–self variable, no significant effect of country (p = .616, n.s.) or gender (p = .5.94, n.s.) was observed.
The last analyses were applied to the relationship between the explanatory factor and the likelihood of participants to have previously thought or talked about their childhood memories (prior recollection). With regard to this variable, Kruskal–Wallis non-parametric tests showed no significant effect of country (p = .773, n.s.). In contrast, results revealed a significant effect of gender on this variable (p = .001), with women reporting a higher frequency of prior recollections of their memories than did men (M = 3.35, SD = 0.99; M = 2.93, SD = .92).
The next analyses were applied to categorical variables, memory specificity and content. None of these analyses showed significant interaction between country and gender, and these variables. For memory specificity, no significant interaction was found between this variable and country (χ2 = 1.385; df = 2; p = .500) or gender (χ2 = .288; df = 1; p = .592). A similar pattern was observed for memory content. There was no significant interaction of this variable with country (χ2 = 10.080; df = 6; p = .121) or gender (χ2 = 3.711; df = 3; p = .294).
Relations between autobiographical memory variables
To analyse the relationship between characteristics of autobiographical memories at the individual level, we performed correlational analysis across the entire sample, with the effects of culture and gender partialed out. Partial correlations among all memory variables were calculated; they are depicted in Table 4.
Partial correlations between autobiographical memory variables with country and gender partialed out.
p < .01.
p < .05.
As demonstrated in Table 4, the analysis of the correlations between memory variables, with country and gender effects controlled, showed that the number of spontaneous emotions mentioned by the participants in their memories was positively correlated with autonomous orientation. Other–self ratio was negatively correlated with the spontaneous mention of emotions and with autonomous orientation.
Discussion
The present study has examined differences in autobiographical memory between participants from three diverse countries, Denmark, Spain and Mexico, which vary in terms of individualism and collectivism (Hofstede, 1980, 2001). Specifically, we have compared Mexican, Spanish and Danish students on characteristics of their earliest memories. At the same time, we have studied the role of gender and its influence on autobiographical memories. This study intends to replicate previous findings from comparisons between Western and East-Asian adults and children (MacDonald et al., 2000; Mullen, 1994; Wang, 2001a, 2006; Wang and Conway, 2004; Wang and Ross, 2005). In particular, we have replicated Wang’s (2001a) methodology and analyses, focusing on both the cultural (country) and the individual level.
Findings are discussed as follows. First, we review the results on autobiographical memories at the country level, considering the relationship between these variables with country and gender, and we conclude by reviewing the individual level in terms of correlations between memory variables.
The results of this study evidenced some country differences in autobiographical memory. In regard to the age at the earliest memory, both Danish and Spanish students dated their first memories to an earlier age than did Mexicans. This result is consistent with other studies which show that in cultures more oriented towards independence (Markus and Kitayama, 1991), the reported age of the earliest memory is lower than in cultures of interdependence (MacDonald et al., 2000; Mullen, 1994; Wang, 2001a, 2006; Wang and Conway, 2004; Wang and Ross, 2005). In this respect, although there are not many studies with Danish, Spanish and Mexican participants, Hofstede’s measures of I-C cover these countries and show a higher score of individualism in Danes, followed by Spaniards and with Mexicans, considered to be more collectivistic than the others (Hofstede, 2001). Thus, differences in the age at the earliest memory are, in general, consistent with results from previous studies.
The results regarding emotions are very interesting, as they do not fit with an interpretation in terms of independent versus interdependent self-construal. In other studies, differences in emotion situation knowledge have been interpreted in relation to differences in cultural values ascribed to the expression of emotions. A higher tendency to reflect on and express their own emotions and internal states in American children and adults may reflect the greater value that American culture poses on individuality and autonomy. Conversely, Chinese and other East-Asian cultures give priority to the maintenance of harmony within social groups. Our data, however, suggest that country preferences for showing emotional states may be unrelated to independence–interdependence.
Some authors have stressed the need to analyse differences in the ‘ecology of emotions’ (Mesquita et al., 1997), that is, the social and cultural practices in which emotions are displayed, as well as the meaning of emotional experiences (Lutz and White, 1986; Mesquita and Leu, 2007; Shweder and Haidt, 2000). Both practices and meanings vary extensively across cultures. From our perspective, these data point to the need to consider other aspects and dimensions than independence–interdependence in order to understand cultural differences in emotion situation knowledge.
The differences in autonomous orientation, with Mexicans scoring higher on this variable, seem to be in contrast with previous studies. In general, studies on autobiographical memory and culture have reported that individuals from cultures of independence show a greater degree of autonomous orientation than do individuals from cultures of interdependence. The fact that differences in autonomous orientations resemble those observed in emotionality lead us to think that it was the spontaneous mention of emotions that made the difference in autonomous orientation. It is necessary not to forget that this variable included, among other categories of mental states, spontaneous mentioning of preferences, likes and dislikes and evaluations, all related with feelings and emotions. From a more theoretical–methodological perspective, this may also reflect some ambiguities or inconsistencies in the way of defining autonomy. While theoretical proposals as those by Kagitçibasi (2005) define autonomy in terms of agency, others, such as Markus and Kitayama (1991) or Wang (2001a, 2004), Wang and Leichtman (2000), include in its definition the expression of mental states. Although both aspects are probably connected, a theoretical and methodological refinement of the notion is, in our opinion, necessary to disentangle the different dimensions that the concept may be encompassing.
Gender was considered to be another explanatory factor. However, in general, the results did not show strong effects of gender on memory variables. The only main effects of gender were on autonomous orientation and the frequency of prior recollections. As expected, women reported having recalled their earliest memories more frequently than did men in all groups. This result replicates Wang’s findings (Wang, 2001a) and is consistent with the results obtained by Davis (1999), who reported that females (both girls and women) evidenced a greater accessibility to their autobiographical memories than males. At the same time, our data showed a significant effect of gender on autonomous orientation. This effect is not easy to interpret, as some previous studies have shown that males tend to include more references to agency–autonomy in their memories than did females (Bauer et al., 2003; Bird and Reese, 2006; Buckner and Fivush, 1998), although this tendency may vary across cultures (Han et al., 1998; Hayne and MacDonald, 2003; Wang, 2001a, 2003; Wang and Fivush, 2005). As we said for country differences, inconsistencies in the definition of autonomy (agency versus the expression of internal states) may account for these seemingly contradictory results.
Although we did not find a main effect of country on specificity as Wang did, a significant country by gender interaction was observed in this variable. When comparing groups, a contrary tendency was observed in the Danes and in the Mexicans. While Danish women tended to be more specific than men, the opposite trend was found in the Mexican sample, with males being more specific than females. These results are different from Wang’s (2001a), who observed country and gender differences in specificity. In her study, American college students reported more specific memories than their Chinese counterparts. In addition, women in both groups tended to be more specific than men. In another study comparing American women and men from two different age ranges (68–71 and 76–79 years), Pillemer et al. (2003) analysed specific episodes included in autobiographical narratives. The aim was to find out whether women’s autobiographical memories focused on specific instances while those of men included, to a greater extent than women’s, general experiences (problems with parents, work). The results showed that women’s memories were more specific than men’s.
Contrary to this, our results suggest that gender differences in specificity must be considered within specific cultural contexts. Again, Danish and Mexican participants showed opposite gender differences, with women being more specific in the Danish sample, but men being more specific in the Mexican group; the Spanish group was in between, with no gender differences in these aspects. More studies considering gender comparisons in relation to cultural contexts are needed.
The analysis of correlations between memory variables, with country and gender effects controlled, may also shed additional light on some of the relationships evidenced by the previous analyses. In general, the correlational analyses showed that individuals who reported earlier first memories were also more autonomous, more emotional and more self-focused (most rarely mentioned other people) than those who reported later first memories. The latter were less autonomous and more other-focused. The negative correlation between autonomous orientation and other/self ratio provides additional support for this interpretation. These data support the analyses of the age at and the content of the earliest memory in relation with the characteristics of self-construal not only at the macro (country), but also at the individual level, as in Wang’s (2001a) study.
Conclusion
In this study, we have compared characteristics of the earliest memories of college students (male and female) from three countries (Denmark, Spain and Mexico). These countries differ in their values of I-C according to Hofstede’s assessment. At a first glance, the results obtained seem to provide some support for the link between I-C (and independent vs. interdependent self-construal) and characteristics of the earliest memories of the participants (age at the earliest memory, specificity). The differences observed were not very strong, however. Furthermore, a number of unexpected data was found. These data are concerned with country differences in emotionality and autonomous orientation (showing that Mexicans tended to be more emotional and autonomous than Spaniards and Danes) and with gender by country interactions in specificity. With regard to country, we have pointed to differences in the way of conceiving autonomy between authors such as Kagitçibasi, on the one hand, and Markus and Kitayama and Wang, on the other. In our view, it is necessary to refine the conceptualisation of autonomy, both in the theoretical and methodological domain to clarify this issue. In the case of emotionality, the observed differences may be related to differences in emotion situation knowledge that go beyond independence–interdependence.
With regard to specificity, more research is also needed to clarify why in some cases (in some countries, for instance), women tend to be more specific than men. In general, these unexpected gender differences and gender by country interaction, point to the need to deepen the study of gender construction in cultural contexts (Cala and De la Mata, 2010).
From a general theoretical perspective, the inconsistencies observed in our results stress the need to refine (both in the theoretical and in the methodological domain) the analyses of self-construal to account for, on the one hand, the complexity of the notion, and on the other, for the variations associated with personal and contextual factors (Brockmeier, 2012). We agree with (cross)-cultural researchers that have criticised Hofstede’s and Markus and Kitayama’s approaches to I-C and to independence–interdependence for their limitations in both the theoretical and empirical spheres. In the theoretical domain, the model is criticised for being too dichotomous and vague, as it confounds meanings of how (and how much) individualistic–collectivistic different countries are (Harb and Smith, 2008; Hardin et al., 2004; Kashima and Hardie, 2000; Omi, 2012). At the same time, authors such as Matsumoto (1999) claim that there is not a robust cross-national distinction in I-C that would allow us to accept the necessary premise of the self-construal theory. Recent research has cast doubts over some of the basic assumptions of Markus and Kitayama’s theory (notably that culture determines self-construal and that individuals from different cultures display differences in self-construal (see Matsumoto, 1999; etc.).
Rather than considering them as characteristic of countries or cultures in general, some authors claim that individualistic and collectivistic attitudes are not mutually exclusive; rather, they can coexist at the individual level. Individuals, from this perspective, may be characterised by specific combinations of individualistic and collectivistic attitudes (Green et al., 2005).
In a similar vein, the notions of agency and relatedness should not be regarded as opposite dimensions. On the contrary, they can be combined to define self-construal as a multi-dimensional construct (Brewer and Chen, 2007; Harb and Smith, 2008; Santamaría et al., 2010). From our perspective, Kagitçibasi’s (2005, 2007) criticism of the classical dichotomy between autonomy and relatedness as too simplistic and her advocacy for an autonomous-related self must be taken into account and further developed in studies about cultural self-construal and autobiographical memory.
In our view, the notion of situatedness may capture that complexity and provide new insights to the analyses of the relationship between autobiographical memory, self and culture (Oyserman and Lee, 2007, 2008). But what does situatedness as a characteristic of self-construal designate? For us, situatedness is related to both between-cultures and within-cultures variations.
First, we must not forget that cultures/societies are dynamic, typically moving towards more individualism (e.g. Greenfield, 2009). This seems to be, for instance, the case of Spain, moving away from a more collectivistic to a more individualistic country, and from traditional gender role ideals to becoming more like Scandinavia in this respect (Royo-Vela et al., 2008).
In regard to within-countries variations, we must recognise that our study does not analyse these variations. As said before, and given our purpose to extend cross-cultural studies of autobiographical memory and self beyond the comparison between East-Asian and North-American cultures, we have used country as a proxy for culture and added data from other countries (Denmark, Mexico and Spain). But, of course, there is much heterogeneity within any given country. Sociodemographic factors such as socioeconomic status, residence (urban vs. rural) and so on, and especially, the participation in different sociocultural activities/practices may lead to differences in autobiographical memory and self-construal. Moreover, the participants in our study were college students. As some theorists argue, formal schooling is a cultural experience that promotes a ‘developmental pathway’ to autonomy and thus, a cultural notion of self that enhances the individual’s autonomy (Greenfield, 2009; Kagitçibasi, 2005, 2007; Keller, 2007). In the specific case of autobiographical memory, formal schooling seems to promote ways of remembering that resembles those reported in the cultures of independence (De la Mata et al., 2011; Santamaría et al., 2012).
The fact that participants were college students may also explain, at least in part, the small gender differences found in our results. The educational background of the participants could be a factor that in some way modulates gender differences. For all these reasons, more research is needed to deepen the relationship between (gendered) cultural practices and activities such as formal schooling and the development of different self-construals.
Situatedness may also imply within-individual variations, in the sense that any individual participates in diverse cultural practices (family, school, work…). If we assume that self-construal is situated in relation to culture, the fact that every individual in any culture participates in different activities in which the relative importance of issues concerned with autonomy/agency and relation/communion, power and other basic human needs (Kagitçibasi, 2005) vary, leads us to assume that self-construal may also vary across cultural practices. As shown by Wang and Ross (2005), contextual demands can ‘prime’ independent and interdependent dimensions of self-construal to a different degree. In this vein, a recent study by Antalíková et al. (2011) has evidenced differences in the predominant dimensions of self-construal (autonomy vs. relatedness) in adolescents’ memories, depending on the domain of experience (family, school or friends) to which these memories refer. Again, more research is needed to document variations in dimensions such as agency and relatedness in memories from different cultural life domains.
As a general way of posing the question, we think that culture (and gender) differences in self-construal should not be conceived in terms of the existence of an independent versus an interdependent self-construal in one culture (or gender) or another, but in the ‘ecology’ of self-construal, that is, in the extent to which cultural activities may promote and activate the agency and/or relatedness dimensions of self-construal. From that perspective, differences in self-construal are not located in the culture or the individual, but in the activities carried out by individuals in sociocultural settings. This does not deny the existence of differences between groups of individuals from different cultures (or between females and males), when comparing them. However, this perspective provides us with conceptual tools to understand why such differences do not always appear and how contextual factors and cultural practices may help to interpret both differences and similarities.
