The literature on gender differences in autobiographical memory is very consistent. Relative to men, women show greater fluency in retrieving autobiographical memories (Davis, 1999; Pillemer, Wink, DiDonato, & Sanborn, 2003; Seidlitz & Diener, 1998) and construct autobiographical narratives that are longer, more detailed, more emotional, and more coherent than those produced by men (Buckner & Fivush, 1998; Niedzwienska, 2003; Thorne, 1995). They also take a more active role than do men when engaged in joint reminiscence (Ross & Holmberg, 1992) and are more likely to mention other people and the thoughts and feelings experienced by others in their narratives (Buckner & Fivush, 1998). Nelson and Fivush (2004) have argued that these gender differences can be traced back to the way parents speak to their children about shared experiences. According to their view, gender differences arise because parents tend to use a more elaborative conversational style when speaking with their daughters and a more pragmatic conversational style when speaking with their sons (Reese, Haden, & Fivush, 1993; Sales, Fivush, & Peterson, 2003). Elaborative conversations are characterized by relatively long, embellished conversations in which parents use leading questions, new details, and positive feedback to extend the conversation (Boland, Haden, & Ornstein, 2003). In contrast, pragmatic conversations are characterized by a focus on “W” questions (who, what, where, when, etc.). Engaging in elaborative conversations with children increases the number and specificity of details included in their subsequent recollections of the past (Boland et al., 2003; Haden, Haine, & Fivush, 1997; Reese et al., 1993).

Although the socialization view provides a compelling explanation for the existence of gender differences and how they develop across the lifespan, questions remain regarding how socialization processes affect memory across retention intervals. Current models suggest that autobiographical events are not stored in memory like files on a computer. Rather, narratives are reconstructed “on the fly” at the time of retrieval to satisfy both short- and long-term goals of the working self (Conway, Singer, & Tagini, 2004). Narratives are cobbled together from fragments of autobiographical information stored in what Conway et al. referred to as the autobiographical knowledge base. According to this view, one explanation for gender differences in autobiographical narratives is that gender socialization influences what kinds of information wind up in the autobiographical knowledge base. Relative to women, men might pay more attention to factual information (i.e., who, what, where, and when) as events unfold or might reflect more on factual information as events recede into the past. Either way, this hypothesis predicts that the autobiographical knowledge base for men would include more factual information than would the autobiographical knowledge base for women. This will be referred to as the storage hypothesis, because it implies that men and women store different kinds of information in the autobiographical knowledge base. However, it is possible that gender differences in personal narratives exist even if the contents of the autobiographical knowledge base are largely the same for men and women, if one argues that men and women select different kinds of information to include in their narratives. Whereas for men, facts may be the essential components of personal narratives, for women, factual content may be the means for constructing interpretive narratives and giving the meaning of an event, rather than something that is essential for “knowing” or “retelling” the past. Thus, women may have more factual information stored in memory than they choose to include in their narratives.

Although the storage and selection hypotheses are not mutually exclusive, the focus of most empirical and theoretical work in the field has been on what occurs at the time of narrative construction. Most autobiographical memory experiments –particularly those in adults—require the subjects to recall an event from the past; no information is collected at or close to the time that the event occurred. Two exceptions to this general pattern are diary studies (e.g., Kristo, Janssen, & Murre, 2009) and flashbulb memory studies (e.g., Talarico, LaBar, & Rubin, 2004). However, these studies typically do not or cannot evaluate the accuracy of initial reports of the event (but see Nachson & Zelig, 2003). As such, researchers typically are unable to ascertain what information was part of the autobiographical knowledge base early in the retention interval and cannot compare initial memory against subsequent factual memory and narrative reports. Collecting factual memory data at two time points would provide some insight regarding gender differences in autobiographical narratives, particularly whether gender differences exist in storage, selection, or both. If men are taught that factual information is essential to personal narratives, men should store more factual information than should women and should reflect on factual information more as events recede into the past. If this hypothesis is true, not only should men remember more factual information at short retention intervals, but also the magnitude of the gender difference in factual memory should increase at longer retention intervals.

The present experiment was designed to assess two possible mechanisms for the gender differences typically observed in autobiographical narratives that might occur either alone or in tandem. First, men and women may store different kinds of information in the autobiographical knowledge base. Second, men and women may sample information from that knowledge base differently when they construct a narrative. Male and female hockey players completed online questionnaires about games that they participated in during the first 3 weeks of their hockey season. The questionnaires asked the subjects to provide factual information about the events; the same questions were asked about each event. Most questionnaires were completed online within 12–48 h of each event. After a retention interval of approximately 6 weeks, the subjects came into the laboratory and provided narrative accounts for two of the games. After writing their narratives, the subjects completed the same questionnaires, again—one for each game from the initial test phase. Narrative accounts of games were not collected during the initial test phase because we did not want the narratives provided initially to influence the narratives that were generated after the retention interval.

Athletic contests were selected as the object of study because they provided a common event class for both males and females. The winter sports season was selected because there was a natural break for final exams and winter vacation that created a period during the retention interval when no games were scheduled. Hockey games were selected because the rosters on hockey teams allowed for a reasonably large sample of men and women (basketball rosters, for example, are much smaller). Collecting memories for athletic contests also allowed us to see whether men and women use different narrative styles to cope with success and failure. Previous research suggests that narratives from winning team members tend to be more cohesive and are more likely to include factual rather than evaluative information (Baker-Ward, Eaton, & Banks, 2005; Fivush, Hazard, Sales, Sarfati, & Brown, 2003). Because previous experiments have focused on children’s narratives, it is not clear whether these patterns would persist in a more verbally skilled young adult population.

To summarize, the main hypothesis in the present experiment was that men and women focus on different classes of information (either as events unfold or as they recede into the past) and that these differences would be reflected in subsequent narrative accounts of those same events. Two specific predictions were assessed. First, men should be able to answer more objective questions about an event both initially and after a retention interval. Second, males’ narratives should display a stronger factual orientation than females’ narratives by including a higher proportion of factual (rather than evaluative) statements in their narratives. If both of these results obtain, it would indicate that socialization influences not only the selection stage of autobiographical memory production, but also the composition of the underlying autobiographical knowledge base from which narratives are constructed.

Method

Subjects

Forty-five members of the men’s (n = 25) and women’s intercollegiate hockey teams at a small liberal arts college participated in the experiment. Two male and 2 female subjects did not complete at least one part of testing (one of the two questionnaires or the narrative responses). These subjects’ data were excluded from analyses for which they did not have the required data. A donation was made to each team in return for the players’ participation.

Materials

The objective memory questionnaire included questions about the number of goals and power-play goals scored, the number of penalties incurred, and the number of shots taken by each team. The subjects named the starting players for their team and identified which team wore dark-colored jerseys and which player scored the game-winning goal. Seven-point rating scales of emotional intensity (1 = not at all intense–7 = extremely intense), emotional valence (1 = extremely negative—7 = extremely positive), and predicted importance of the game in the context of the season (1 = not at all important–7 = extremely important) were also included to determine whether these subjective ratings influenced either objective memory performance or narrative structure during the second phase of the experiment. The males played seven games during the testing period: three wins, two losses, and two ties. The females played eight games during the test period: five wins, one loss, and two ties.

Procedure

Within 24 h of each of the first seven games of the season (eight for the female subjects), every subject received an e-mail reminder to complete the online memory questionnaire for that game. Of the forms, 97.4% were completed within 2 days of the respective game; the remaining forms were completed within 4 days of the respective game. The subjects were told at the time of recruitment and reminded throughout the experiment not to consult online sources or their teammates for help completing the questionnaire. Postexperiment interviews and the moderate performance levels on the questionnaire suggest that the subjects largely complied with this request. All tested games took place within a 23-day period. Although the season resumed after a break for the winter holidays, no subjects completed online forms for any subsequent games. This suggests that they had stopped thinking about the experiment.

The second phase of the experiment took place approximately 6 weeks after the last tested game. This retention interval was selected because it allowed for some fading of the to-be-remembered events but allowed us to complete data collection before classes for the spring semester resumed. This allowed the subjects to be run over the course of a 5-day period, keeping the retention interval for all subjects nearly the same. The subjects knew that they were participating in a second phase of the “questionnaire” study, but they were not told what they would be asked to do until they arrived at the lab. The subjects wrote narrative accounts of two games (one win and one loss) on separate lined sheets of paper with the name of each opponent printed at the top. These games were selected on the basis of the subjective valence and intensity ratings from the initial memory questionnaires. Games were selected that had the high average intensity ratings and extreme valence ratings (negative for losses, positive for wins). The subjects were instructed to write their narratives as though they were describing each game to a friend or family member who was knowledgeable about hockey but had not been at the game. The subjects were not told why these two games were selected. The order in which the narratives were collected was counterbalanced such that half of the subjects wrote about the negative game first. After turning in the second narrative, the subjects completed paper-and-pencil versions of the original memory questionnaire for each of the games from the first phase of the experiment. The narratives were collected first so that the content of the objective memory questionnaires would not influence the form or content of the subjects’ narratives.

Postexperiment interviews revealed that none of the subjects expected to complete narrative accounts of any of the games. Only a handful of the subjects claimed that they had anticipated the questionnaire portion of the follow-up, but none—whether they had anticipated this portion of the experiment or not—reported that they used online or any other sources of information to review the details of the games that were tested. As well, initial analyses indicated that the subjects who anticipated the follow-up questionnaires did not perform significantly better than the remaining subjects.

Results

All reported analyses were conducted with α = .05. When appropriate, degrees of freedom were adjusted using the Greenhouse–Geisser correction; only the unadjusted values are reported. Because the subjects in the experiment were asked to remember different events (i.e., men’s games for the male subjects, women’s games for the female subjects), it is possible that gender differences in memory might have arisen from differences between the to-be-remembered events, rather than from gender differences in autobiographical memory/narrative construction. Several analyses were conducted to assess possible differences in the to-be-remembered events; the relevant data are presented in Table 1. First, the subjective ratings for men’s and women’s games were compared both for the initial test session and for the follow-up session. Significant gender differences for intensity at the initial test session were revealed by t-tests, t(42) = 2.10, SEM = 0.2, as well as significant differences for valence at both the initial sessions, t(42) = 3.06, SEM = 0.2, and follow-up sessions, t(42) = 3.82, SEM = 0.1. In all three cases, women’s ratings were higher than men’s ratings (see Table 1). Because of these differences, the emotional valence, emotional intensity, and predicted importance of each game were entered as covariates when appropriate in subsequent analyses. A second set of analyses examined whether there were differences between the objective events (goals scored and penalty minutes) for the males’ and females’ games (see Table 2); gender differences in both the means (independent samples t-tests) and variances (Levene’s test) were assessed. With respect to variability, the only significant difference was for goals scored by the subjects’ own team; the number of goals scored by the women’s team was more variable than the number of goals scored by the men’s team, F(1, 13) = 8.394, p < .05. It is unclear how this difference in variability would affect memory. One could argue that it would be easier to remember outliers in a more variable set of numbers or that it would be easier to guess the correct answer in a set of less variable numbers. Therefore, differences in variability will not be discussed further. With respect to means, the only significant difference was for penalties: The men’s team committed significantly more penalties than did the women’s team, t(13) = 2.73, SEM = 0.94. Because it is not clear whether having more or fewer penalties would facilitate/interfere with memory performance, this difference will not be discussed further. Finally, we examined whether there were differences in the predictability of the starting lineups. Neither team used the same starting lineup more than twice (the men used two different starting lineups twice; the women used only one starting lineup twice). Eleven different male players started at least one game; the corresponding value for the females was 17. The mean numbers of games started by males (M = 3.87) and females (M = 2.62) were not significantly different from one another.

Table 1 Mean (SE) ratings of intensity, valence, and importance as a function of gender and test session
Full size table
Table 2 Mean (SE) number of goals actually scored (own team and other team) and penalties actually incurred (own team and other) as a function of gender for all team games
Full size table

Memory questionnaires

The number of correct responses for men and women were tallied for both the initial and follow-up questionnaires. Data from four of the questions (“power play goals” and “shots on goal” for both own and the other team) were not entered into any analyses. Given that very few of the subjects (<30%) were able to answer these questions correctly on the initial questionnaire, we felt that these questions would not provide any insight regarding memory 6 weeks later. Preliminary analyses indicated that excluding these questions from further analysis did not change the overall pattern of results.

The number of questions answered correctly was tallied for each subject. These data were submitted to a 2 × 2 × 2 analysis of covariance (ANCOVA). The three independent variables were gender (male vs. female), test session (initial vs. follow-up), and game outcome (win vs. loss/tie). Ties and losses were condensed into a single category because the women’s team suffered only one defeat, but the analyses yielded the same pattern of significant results even when losses and ties were treated separately. The subjective rating scales (emotional intensity, emotional valence, and importance) were used as covariates to account for differences across the remembered events. Because each subject provided questionnaire data for multiple games, some of the variability in responses was due to differences between games and/or individuals. A game that was particularly exciting might be remembered better than a less exciting one; a subject with a good memory is likely to remember more game information than is a subject with a poor memory. To control for this variability, two “dummy” variables were created (one for the subjects, one for the games) and entered into the ANCOVA as covariates. Each subject was assigned a unique integer between 1 and 45; similarly, each game was assigned a unique integer between 1 and 15. Including these dummy variables in the ANCOVA eliminated any variability due to differences between subjects/games, which afforded the analysis more power to identify variability associated with our independent variables.

The analysis yielded main effects of gender, F(1, 258) = 61.27, MSE = 4.17, η 2 = .19, session, F(1, 258) = 6.19, MSE = 2.23, η 2 = .02, and game outcome, F(1, 258) = 5.91, MSE = 4.17, η 2 = .03. Men (M = 9.6, SE = 0.2) answered more questions correctly than did women (M = 6.8, SE = 0.2); performance was better at the initial test (M = 9.6, SE = 0.1) than at the follow-up (M = 6.8, SE = 0.1); and losses (M = 8.5, SE = 0.2) yielded more correct responses than did wins (M = 7.9, SE = 0.1). Two significant interaction effects were also observed. The gender × session interaction was significant, F(1, 258) = 36.82, MSE = 2.23, η 2 = .13; the difference between males and females was larger for the follow-up sessions than it was for the initial session (see Table 3). The gender × outcome interaction was also significant, F(1, 258) = 6.69, MSE = 4.17, η 2 = .03; whereas females performed similarly for wins and losses/ties, males performed better for losses/ties than they did for wins (see Table 3).

Table 3 Mean (SE) number of correct responses on the of objective memory questionnaire as a function of gender and test session (upper panel) and gender and game outcome (lower panel)
Full size table

Narratives

The length of each narrative (in words) was computed and submitted to a 2 (gender) × 2 (outcome) ANCOVA. The subjective rating scales (emotional intensity, emotional valence, and importance) and subject dummy variables were used as covariates. The ANCOVA revealed a significant main effect of gender, F(1, 63) = 8.87, MSE = 2,079.91, η 2 = .12; women (M = 123.9, SE = 7.6) reported longer narratives than did men (M = 90.8, SE = 7.9). Neither the main effect of outcome, F(1, 63) = 1.61, MSE = 2,079.91, p ≈ .40, nor the outcome × gender interaction, F(1, 63) < 1, was significant.

Narratives were broken down into idea units that were categorized as either factual or interpretive, using a method similar to the one described by Pasupathi and Hoyt (2009). Idea units were coded as facts if they included objective information that could be determined by reading a box score of the game, including goals, scores, and dates. Idea units were coded as interpretations if they included subjective information requiring the judgment of the narrator, including characterizations of the quality of play and/or the mental states of individuals or teams. Interpretations were further coded as to whether the information related to externally observable behavior (e.g., quality of play) or internal behavior (e.g., mental states of individuals/teams). Two independent coders blind to condition coded every narrative. Cronbach’s alpha for the two coders averaged over .98 for all coded variables; discrepancies were resolved via discussion. Because of the significant effects of gender and outcome reported for narrative length, two dependent measures were calculated: (1) the raw number of idea units per narrative and (2) the number of idea units as a proportion of the length of the corresponding narrative. These two analyses yielded the same pattern of results, so only the analyses of the raw numbers will be described below.

The mean number of idea units per narrative was submitted to a 2 (gender) × 2 (outcome) ANCOVA; the subjective rating scales (emotional intensity, emotional valence, and importance) and subject dummy variables were used as covariates. The ANCOVA revealed no significant main or interaction effects, all Fs < 2.19, all ps > .14. Paradoxically, although women produced narratives that were significantly longer than those written by men, females’ narratives did not include significantly more idea units. One explanation for this discrepancy is that women’s idea units were more complex (requiring more words to express them) than the idea units included in men’s narratives. A measure of words/idea unit was calculated by dividing the number of words for each narrative by the number of idea units included in that narrative. These data were submitted to the same gender × outcome ANCOVA. The analysis revealed a significant effect of gender, F(1, 63) = 11.94, MSE =1.41, η 2 = .16; women’s idea units (M = 7.1, SE = 0.2) were significantly longer than men’s idea units (M = 6.1, SE = 0.2). Neither the main effect of outcome nor the gender × outcome interaction was significant, all Fs < 1.68, all ps > .19.

The idea units were categorized as either interpretations or facts. The proportion of idea units (e.g., interpretations divided by idea units) that fell into each category was analyzed because the proportional data are more meaningful (e.g., 50% of the narrative was composed of facts vs. four idea units were facts). The proportion of interpretations data was submitted to the 2 × 2 ANCOVA. Gender and game outcome were the independent variables; valence, intensity, and importance were added as covariates. The analysis revealed a significant main effect of gender, F(1, 63) = 25.75, MSE = 0.03, η 2 = .29. As is displayed in Table 4, women reported a higher proportion of interpretations than did men. The analysis also revealed a marginal gender × outcome interaction, F(1, 63) = 2.98, MSE = 0.03, η 2 = .05, p < .10; whereas the proportion of interpretations in women’s narratives was relatively unaffected by outcome, men reported a slightly higher proportion of interpretations in wins relative to losses. The main effect of outcome was not significant, F(1, 63) = 2.41, p > .24. The proportion of interpretations was further broken down into those that reflected internal states versus external behavior. The data for internal and external interpretations were submitted to separate 2 × 2 ANCOVAs. Once again, gender and game outcome were the independent variables; valence, intensity, and importance were added as covariates. For internal interpretations, the main effect of gender was significant, F(1, 63) = 16.12, MSE = 0.01, η 2 = .20, but neither the main effect of outcome nor the gender × outcome interaction was significant, both Fs (1, 63) < 2.71, both ps > .10. As is displayed in Table 4, women reported a higher proportion of internal interpretations, relative to men. For external interpretations, the main effect of outcome was not significant, F(1, 63) < 1. However, the main effect of gender was significant, F(1, 63) = 8.49, MSE = 0.02, η 2 = .12. As is displayed in Table 4, women reported a higher proportion of external interpretations than did men. The gender × outcome interaction was also significant, F(1, 63) = 11.57, MSE = 0.02, η 2 = .16. Whereas women reported more external interpretations for losses than for wins, men reported a much higher proportion of external interpretations for wins than for losses.

Table 4 Mean (SE) proportion of facts and interpretations (internal and external) as a function of gender and game outcome
Full size table

The proportion of facts was also submitted to the same 2 (gender) × 2 (outcome) ANCOVA described previously (valence, intensity, and importance were the covariates). Although, the main effect of outcome was not significant, F(1, 63) = 1.44, p > .23, the main effect of gender was significant, F(1, 63) = 25.01, MSE = 0.03, η 2 = .28. As is displayed in Table 4, women reported a lower proportion of facts, relative to men. The gender × outcome interaction was marginal, F(1, 63) = 3.28, MSE = 0.02, η 2 = .05. Whereas women reported similar proportions of facts for losses and wins, men reported a slightly higher proportion of facts for losses.

Discussion

The consistent gender differences observed in autobiographical memory experiments are often attributed to differences in the way men and women are taught to reflect on the past (Nelson & Fivush, 2004). The present experiment assessed two other mechanisms for explaining gender differences in autobiographical narratives. First, men and women may store different kinds of information in the autobiographical knowledge base. Second, there may be gender differences in selection; that is, men and women choose to report different kinds of information in their narratives to satisfy different expectations about what makes for a “good” autobiographical narrative. To be clear, these mechanisms are not mutually exclusive; differences in storage and selection could operate simultaneously. Memory for objective information about a series of athletic contests was collected soon after each event and again 6 weeks later; during the second session, narrative accounts of two of the games (one win and one loss) were also obtained. If males exhibit a factual orientation either at encoding or early in the retention process and, therefore, store or retain more factual information, one would expect males to perform better than females on the objective questionnaires at both test sessions. If, on the other hand, gender differences arise because women (men) choose to include more evaluative (factual) information in their narratives, one would expect males and females to perform comparably on the objective questionnaires, despite gender differences emerging in their personal narratives of the same events.

Consistent with the first of these predictions, males performed better than females on the objective questionnaires. This suggests that they paid more attention to game facts at the time the events took place. Moreover, the magnitude of the gender difference on the objective questionnaires increased at the second retention interval. This suggests that men continued to exhibit a stronger factual orientation than did women as time passed. There are a number of plausible explanations for these data: Men might have paid more attention to game facts as the events unfolded; men might be better at retrieving factual information from memory; men might have talked more about game facts with teammates, friends, or family; men might have thought about game facts more than did women. Although the present experiment did not provide data that could be used to choose among these hypotheses, it is clear that men retained more factual information than women did, both initially and over the course of a 6-week retention interval.

Consistent with the second prediction, gender differences in the importance of factual information were also reflected in the subjects’ narratives. Whereas women’s narratives included roughly equal numbers of facts and interpretations, men’s narratives included almost twice as many facts as interpretations. These data make it clear that men demonstrated a more intense factual orientation when constructing their narratives than women did. Evidence of an increased factual orientation for males has been observed outside the autobiographical memory literature, as well. Although women tend to have better episodic memories, in general, (Herlitz & Yonker, 2002), men tend to perform better than women on some but not all tests of general knowledge (see Rubin, Schulkind, & Rahhal, 1999, for a review).

The present experiment also yielded interesting data regarding how males and females relate personal successes and failures. For men, losses elicited more correct responses on the objective memory questionnaires and yielded narratives that included a higher proportion of facts than wins did; women showed negligible differences between wins and losses on both of these measures. These results conflict with previous research showing that children’s “win narratives” tend to be more factual than their “loss narratives” (Baker-Ward et al., 2005; Fivush et al., 2003). This difference might reflect the fact that the young adult subjects in the present experiment were more concerned about personal failures and how they reflect on identity formation than were the children in previous work (Bohn & Berntsen, 2007; Levine & Bluck, 2004; Schwartz & Clore, 1983). More interesting than the discrepancies with past work, however, is the fact that men’s behavior was more affected by game outcome than women’s behavior was. In particular, when faced with a negative event that challenged their identity, men reverted to a more factual orientation than they demonstrated when recalling positive events. This supports the idea that men, more than women, view factual information as essential to knowing, understanding, and being able to talk about the past.

Some might question whether athletic contests are the best event type to test for gender differences in autobiographical narratives. Athletic contests generally have an achievement theme, and men are more likely to spontaneously generate personal narratives with achievement themes (Niedzwienska, 2003; Thorne, 1995). However, in spite of the achievement theme, significant gender differences were still observed. Women told longer, more evaluative narratives than did men. They also were more likely to include internal state language reflecting not just what was happening in front of them, but what they and others involved in the event were thinking and feeling. Finally, women’s idea units were longer, suggesting that they expressed more complex ideas. In these ways, the narratives produced in the present experiment reflect the same patterns of gender differences observed elsewhere in the literature. In fact, one might argue that the gender differences observed in the present experiment would have been even larger if men and women had been free to choose the events that they narrated. The design of the present experiment could not allow the subjects to select their own events to narrate because of the need to have some objective record of the to-be-remembered event. Thus, the conclusions drawn from this experiment cannot be dismissed because of the event type selected for study. Others might question why evaluative information in the form of narrative accounts of the games was not collected at the initial test session and compared with narratives at the follow-up. Evaluative information was not collected at the initial test session because most autobiographical narrative experiments have focused on evaluative information to the exclusion of factual information and have consistently demonstrated that females incorporate more evaluative information into their narratives. Thus, collecting evaluative information at both test sessions would have served only to replicate findings that have been firmly established by previous work.

The goal of the present experiment was to better understand how gender differences in autobiographical memory are formed. The present data demonstrate that gender differences in autobiographical narratives emerge from at least two sources. Given that narratives appear to be reconstructed at the time of retrieval, there appear to be significant gender differences in terms of both the kind of information that is selected for inclusion and the contents of the autobiographical knowledge base. Thus, although some of the gender differences observed in past literature may have been due to choices made by the narrator, some were also likely due to differences in available information. The data also demonstrate that men display a factual orientation—that is, an increased interest in and facility for recalling factual details—not only at the time of retrieval, but much earlier during encoding and/or the retention interval. Furthermore, gender differences in factual orientation may contribute directly to the narrative differences observed both in the present experiment and throughout the literature on autobiographical narratives.