Alzheimer's disease (AD) is a neurodegenerative disease dominated by a severe breakdown in memory processes. The hallmark symptom in AD is autobiographical memory deterioration, characterized by an impaired ability to encode and recall the events in one’s life and their spatiotemporal context. However, semantic knowledge impairments are also widely reported, with subtle deficits being observed even in the early stages of the disease (e.g., Barbeau et al., 2012; Chertkow & Bub, 1990; Hodges & Patterson, 1995). Moreover, there is evidence that AD patients are impaired on certain aspects of script knowledge, particularly the sequential organization of events, while findings are more mixed with regard to the semantic content of script knowledge. For instance, Rusted and Sheppard (2002) reported that AD patients retained knowledge for the central script actions of a routine event, indicating some preservation of semantic content, while Allain et al. (2008) found impaired semantic knowledge for all types of actions. So far, the few studies examining script knowledge in AD have focused on scripts for simple actions and routine events (e.g., making a cup of tea) (Allain et al., 2008; Cosentino et al., 2006; Grafman et al., 1991; Rusted & Sheppard, 2002; Weingartner et al., 1983). Surprisingly, little is known of script knowledge important to autobiographical memory in AD, despite autobiographical memory deficits being a prominent feature of the disease. In the present study, we aimed to address this gap in the literature by examining a specific type of a culturally transmitted schema representation, the cultural life script, which has been shown to be important in structuring retrieval from autobiographical memory.

The cultural life script

Cultural life scripts refer to culturally shared expectations concerning the order and timing of major life events and transitions (Berntsen & Rubin, 2004; Rubin & Berntsen, 2003) and relates to Habermas and Bluck’s (2000) notion of a cultural concept of a biography. The conceptualization of the cultural life script (Berntsen & Rubin, 2004) integrates the concept of scripts, as developed by Schank and Abelson (1977), with sociological and anthropological research on culturally sanctioned age norms (Neugarten et al., 1965). Berntsen and Rubin (2004) proposed that, similar to scripts, which outline the timing and order of events in a stereotypical episode (Schank & Abelson, 1977), the life script represents knowledge about the timing and order in which important life events and transitions are expected to take place in a prototypical life within a given culture. However, in contrast to Schank and Abelson’s (1977) script knowledge, which is acquired through personal experience in recurrent contexts, life scripts are culturally transmitted knowledge, learned independent of personal experience. This is consistent with children and younger adults easily generating life-script events for experiences they have not yet encountered (e.g., Berntsen & Rubin, 2004; Bohn & Berntsen, 2008, 2013; Umanath & Berntsen, 2020). Instead, life scripts are semantic representations of how life should ideally be lived within a given culture, and favors positive transitional events, most of which are set to occur in young adulthood (Berntsen & Rubin, 2004).

Berntsen and Rubin (2004) initially established the life script by asking a sample of Danish university students to nominate the seven most important events that were likely to take place in a prototypical infant’s life. Participants were also asked to rate the nominated events on several characteristics, including the age of occurrence and the emotional valence. Consistent with their conception of a shared cognitive structure, Berntsen and Rubin (2004) found a high overlap among the events nominated by the participants. The life script consisted predominantly of emotionally positive transitional events, having children, marriage, begin school, and college being the most frequently mentioned events. There was also a high degree of agreement across participants about the age of occurrence of these events, most of which were estimated to occur between the ages of 15 and 30 years. In contrast, emotionally negative events were less frequent, more evenly distributed across the life span, and had no prescribed time slot for their expected occurrence. Furthermore, consistent with life-script events being organized chronologically, Berntsen and Rubin (2004) reported a positive correlation between the order in which participants generated the events and the estimated ages for when these events were to occur in the life course. Berntsen and Rubin’s initial findings have since been replicated in different age groups (Bohn, 2010; Bohn & Berntsen, 2011, 2013; Janssen & Rubin, 2011; Janssen et al., 2014; Tekcan et al., 2012) as well as across multiple cultures (e.g., Erdoğan et al., 2008; Habermas, 2007; Janssen et al., 2014; Ottsen & Berntsen, 2014; Rubin et al., 2009; Zaragoza Scherman et al., 2017).

The importance of the life script stems from its assumed role in structuring recall from autobiographical memory (Berntsen & Rubin, 2004). Life-script theory posits that people use the life script as a template, when having to search for important and/or emotionally positive autobiographical memories, thus making events falling within the life script more accessible (Berntsen & Rubin, 2004; Rubin & Berntsen, 2003). This is consistent with a high percentage of people’s life story memories being of life-script events (e.g., Bohn, 2010; Rubin et al., 2009; Thomsen & Berntsen, 2008), while such an overlap is not found for word-cued recall (Berntsen & Bohn, 2010; Bohn & Berntsen, 2011; Koppel & Berntsen, 2016). It also agrees with evidence that the temporal distribution of life-script events peaks in young adulthood, paralleling the distribution of autobiographical memories in middle-aged and older adults showing more memories from the second and third decade of life than from adjacent life periods, commonly known as the reminiscence bump (e.g., Rubin et al., 1998; Rubin et al., 1986). Further highlighting the importance of the life script for autobiographical memory, Bohn and Berntsen (2008, 2013) found that the acquisition of a normative life script in adolescence seems to be an important prerequisite for the ability to construct a coherent life story. Evidence also suggests that the life script not only helps structure recall from the personal past but also influences recall of fictional life stories (Koppel & Berntsen, 2014) and is used as a scaffold when people imagine their personal future (Berntsen & Jacobsen, 2008; Bohn & Berntsen, 2013).

In summary, the literature supports the existence of the life script as a cognitive schema with a key set of characteristics. So far, most research on life scripts have focused on cognitively healthy individuals and no studies have examined life-script knowledge in patients with AD. However, examining life-script knowledge in AD may enhance our understanding of patients’ autobiographical memory deficits. If life-script knowledge is impaired in AD, patients may lack the semantic scaffold provided by the cultural life script to support autobiographical retrieval processes. Furthermore, the life script has generally been treated as an integral knowledge structure. However, prior work suggests that script knowledge is not a unitary construct, but rather relies on multiple component processes (i.e., semantic and sequential aspects) (e.g., Cosentino et al., 2006). Thus, examining how AD affects the different key characteristics of the life script may provide new insights into the organization of life-script knowledge. Critically, the life script differs in important ways from that of scripts for simple actions and routine events. The life script represents culturally derived abstract knowledge as opposed to more experience-near scripts that unfold in a spatial setting. Lynch et al. (2020) showed that amnesia patients were selectively impaired on scripts that involve spatial context, leaving open the possibility that more abstract and conceptual script knowledge may not be as affected in AD. Moreover, the life script not only entails knowledge concerning culturally salient events and their sequential order but also the normative timing and emotional valence of these events, as such it represents a more complex type of script knowledge. In the present study, we examine the key characteristics of the life-script schema in older adults diagnosed with AD compared with healthy age-matched controls.

Cultural life scripts in Alzheimer’s disease: Research questions and hypotheses

Several components of life-script knowledge can be assumed to be affected in AD. First, since the life script is part of semantic knowledge, we want to establish to what extent AD patients have retained knowledge of which events constitute normative transitional life events. Previous studies have provided only indirect assessment of the semantic contents of life-script knowledge, by examining the extent to which life-script events are included in the personal life story of AD patients compared with healthy older adults (Fromholt & Larsen, 1991; Tippett et al., 2018; Usita et al., 1998). Findings from these studies are mixed. In contrast, here, we directly examine the semantic content and structure of cultural life scripts in AD, simply by asking older adults with AD and healthy controls to solve the standard life-script task (Berntsen & Rubin, 2004). Consistent with AD being associated with semantic knowledge deficits, we expect patients’ knowledge of cultural life-script events to be less fluent and adhere less to cultural norms, compared with healthy controls.

Second, the life script is not only about normative events in a prototypical life, but also about the timing of events—that is, when in the life course the event is expected to take place (e.g., marriage around age 25, not age 15, in our culture). AD has generally been linked to distortions in time perception even in the early stages of the disease (see El Haj & Kapogiannis, 2016, for a review). For instance, AD patients are impaired in their ability to make accurate time estimates (e.g., El Haj et al., 2013; Rueda & Schmitter-Edgecombe, 2009) and in making temporal judgments about the duration of common events (e.g., how long does a dental appointment last?) (Brand et al., 2003; Espinosa et al., 2009). A substantial body of work has also demonstrated semantic memory problems in AD when it comes to dating public events (e.g., the fall of the Berlin Wall) (e.g., El Haj et al., 2017; Muller et al., 2014; Sagar et al., 1988). We therefore expected AD patients to provide age estimates that were less consistent with cultural norms than those provided by healthy controls.

Third, because the life script is biased towards positive events and that disproportionally many of these events are expected to occur during adolescence and early adulthood, a related question concerns whether AD patients’ life scripts show the typical youth bump for positive events? Prior work has found that AD patients show a peak in the recall of autobiographical events from middle childhood and early adulthood, consistent with a reminiscence bump (e.g., Barnabe et al., 2012; Fromholt & Larsen, 1991; Fromholt et al., 2003; Kirk & Berntsen, 2018). These findings may indicate some retained knowledge for the timing of important events across the life span. Findings on a positivity bias in AD have produced mixed results. For example, Bohn et al. (2016) found that AD patients showed a memory advantage for positive over negative material, while others have reported findings in the opposite direction (Fleming et al., 2003). It has been argued that a decline in cognitive resources advantages the processing of positive information, as negative information is more cognitively taxing (Labouvie-Vief et al., 2007). Following this view, AD patients may demonstrate a greater positivity bias than older adult. However, whether AD patients’ life scripts show the typical reminiscence bump for positive events remains to be examined.

Lastly, life scripts have a temporal structure (Berntsen & Rubin, 2004). We therefore examine the chronological order of life-script events across both groups. Deficient memory for temporal order has commonly been reported in AD (Fama et al., 2001; Hampstead et al., 2010; Johnson & Kesner, 1997; Madsen & Kesner, 1995). Similarly, AD patients consistently show impaired knowledge for sequential order of familiar actions and routines (e.g., Allain et al., 2008; Cosentino et al., 2006; Weingartner et al., 1983). These sequencing errors have been linked to prefrontal dysfunction (Grafman, 1995, 2002). This is in line with observations that patients with frontal damage display intact knowledge for the semantic content of script events, but an impaired ability to organize these events in a sequential order (e.g., Sirigu et al., 1995; Zanini et al., 2002) and neuroimaging evidence for prefrontal involvement during script sequencing tasks (Crozier et al., 1999; Knutson et al., 2004). Consistent with prior findings of deficits in the sequencing dimension of general script knowledge in AD patients (e.g., Weingartner et al., 1983), we expected errors in the temporal ordering of life-script events compared with healthy controls.

Method

Participants

Forty-three participants were included in the study: 21 participants diagnosed with AD (10 females, 11 males)Footnote 1 and 22 non-cognitively impaired elderly controls (13 females, 9 males). AD participants were recruited via regional dementia workers and all met the clinical criteria for probable AD dementia according to the international guidelines by the National Institute on Aging and Alzheimer’s Association clinical criteria (McKhann et al., 2011). Disease severity was assessed on the Mini-Mental State Examination (MMSE; Folstein et al., 1975). According to the MMSE score, 12 participants had mild (MMSE = 21–26) and nine participants had moderate (MMSE =10–20) dementia. Exclusion criteria for participating in the study included not being a native Danish speaker, substantial vision or hearing impairments or the presence of other conditions that could interfere with cognition, such as significant head injury, substance abuse, and primary psychiatric illness. Participants in the control group were additionally required to have a MMSE score of 25 or above, to exclude the possibility of undiagnosed dementia (Folstein et al., 1975). Healthy older participants were recruited through the Center on Autobiographical Memory Research participant database and local advertising. The study obtained approval by the Central Denmark Region Committees on Health Research Ethics. All participants provided informed consent to participate in the study, for participants with AD consent was obtained in the presence of a primary caretaker.

Table 1 summarizes the demographic and clinical characteristics of the study participants. No differences were found between the AD group and healthy controls in terms of age, years of education (Table 1), or gender, χ2(1) = 0.57, p = .45.

Table 1 Demographic and clinical characteristics of participants
Full size table

Materials

Neuropsychological measures

All participants were screened on a number of tasks to assess their cognitive status and well-being. These tests included the Addenbrooke Cognitive Examination (ACE; Mathuranath et al., 2000; with scores ranging from 0 to 100), which includes the Mini-Mental State Examination (MMSE; Folstein et al., 1975; with scores ranging from 0 to 30), as measures of global cognitive function. Lower scores on both measures indicate more impaired cognition. Executive functioning was assessed with the phonemic (letter S) and semantic (animal category) verbal fluency task (Lezak et al., 2012). Participants were screened for depression using the Geriatric Depression Scale (GDS; Brink et al., 1982) with scores ranging from 0 to 15, and a score below 5 being considered within the normal range.

The life-script task

To assess life-script knowledge, participants answered the standard cultural life-script task (Berntsen & Rubin, 2004). Participants were instructed to imagine a quite ordinary infant of their own gender and to name the seven most important events that they thought would happen in the infant’s life from birth to death. After participants had generated all seven events (or fewer, could the participants not think of seven events) they were further asked to estimate at what ages these events most likely would occur. The interviewer made it clear to the participants that there were no wrong answers in the task, but that we were simply interested in their opinion. The task differed from previous work, in that participants provided their responses orally and not in a written format, in order to take into account potential difficulties in the patient group with providing written answers. No time limit was set for completing the task, and the experimenter repeated both the task instructions and the events mentioned by the participants when necessary, to lower demands on working memory. Moreover, events were generally written down in the order in which participants generated them. However, if participants themselves made remarks that the order was not as it should have been (for instance if “getting married” came with a remark that this event should have been before “having children” already mentioned), then the order was changed accordingly.

Procedure and coding

The study was part of a larger data collection that included a more extensive procedure. The AD participants were tested in their own home and completed the neuropsychological battery and the life scripts task in two sessions. The first session included AD participants providing informed consent, a brief background interview, assessment of cognitive function (MMSE and ACE) and the GDS. The second session included a life story interview in addition to the life-script task. The order of the two tasks were fixed, to prevent the life-script task influencing life story retrieval (here we only report findings from the life-script task). Participants in the control group were all tested at the university, and all measures were obtained in one test session.

Data coding procedure

All nominated life-script events were categorized according to the 35 categories of the Danish life script established by Berntsen and Rubin (2004, Table 3). Events that did not fit these categories were scored as non-life-script events. Two independent raters blind to group membership and the hypotheses of the study scored 30% of the life scripts. Interrater agreement was high, with the two raters agreeing on 92% of all event categories. Disagreements was resolved by discussion, and one rater scored the remaining life scripts.

For each participant, the total number of events provided (ranging from 0 to 7) and the number of events that could be categorized according to the established life script categories (ranging from 0 to 7) was calculated. An additional score was generated to assess life-script normativity based on previous work (Bohn & Berntsen, 2008; Hatiboğlu & Habermas, 2016; Rubin et al., 2009). A typicality score was calculated by summing up the cultural life-script events generated by the participant, each event weighted by its relative frequency in the normative sample (Berntsen & Rubin, 2004) divided by the number of events generated by the participant (up to 7). Scores ranged from 0 to 100, with higher scores indicating more normative life scripts. Following the procedure in Bohn (2010), the valence of the life-script events was scored as either positive, neutral, or negative based on the valence ratings obtained from Berntsen and Rubin (2004).

In order to assess the chronological structure of the life script, successive pairs of events (excluding undated events) were classified as either forward, simultaneous, or backward in time (see Fromholt & Larsen, 1991, for a similar procedure).

Data analysis

For demographic variables and standardized neuropsychological tests, statistical comparisons between groups were carried out using independent-samples t tests. For the life-script task, between-group comparisons also included t tests; however, as some variables deviated from normality, significant effects were controlled with nonparametric Mann–Whitney’s U tests. When nonparametric analyses yielded a different result, this is reported. Pearson’s chi-squared and Fisher’s exact tests were computed to compare frequency data across the AD participants and healthy controls. Correlational analyses were conducted to assess how performances in the two groups related to the normative data from Berntsen and Rubin (2004, Table 3). Fisher’s r-to-z transformation was performed to test for potential group differences between correlations. For all tests, the level of statistical significance was set to p < .05. Effect sizes (Cohen’s d) were also calculated to indicate the relative strength of significant group differences (0.2 = small, 0.5 = medium, 0.8 = large; Cohen, 1988).

Results

The results are organized as follows. First, AD and control participants are compared on measures obtained from the neuropsychological assessment. Second, performances on the life-script task are examined for the two groups with regard to the content of the life script events, and the timing and temporal order of these events. Third, exploratory correlational analyses examining the relationship between life script scores and performance on verbal fluency measures in the AD group are presented.

Neuropsychological and clinical measures

The results from the neuropsychological assessment are summarized in Table 1. As expected, the AD participants and the healthy controls differed significantly from each other on measures of global cognitive function (MMSE and ACE scores) and executive functioning, as measured by the semantic and phonemic fluency tasks. Further analysis using standard z-scores (i.e., patients’ fluency scores expressed in terms of standard deviations from the controls’ mean performance score) indicated that AD patients were more impaired on semantic fluency (Mz = −2.31, SD = .95) than on phonemic fluency (Mz = −1.59, SD = .69), t(20) = −3.47, p = .002, d = .76. A pattern that has been linked to degradation of semantic knowledge (Henry et al., 2004). The AD group scored significantly higher than the healthy controls on the Geriatric Depression Scale, indicating a higher incidence of depressive symptoms in the AD group. However, the mean score of the patients (M = 2.00) was well within the normal range (Brink et al., 1982).

Content of the life script

Number of life-script events

The performance of AD and control participants on the life-script task is shown in Table 2. The AD group generated significantly fewer events on the life-script task compared with the controls, with three AD patients unable to provide any events at all. No easily identifiable factor discriminated these three participants from the rest of the AD group. Their MMSE scores ranged from 24 to 25, placing them in the mild dementia range, suggesting that disease severity in itself could not account for the lack of ability to generate events.

Table 2 Means (and standard deviations) of scores on the standard life script task
Full size table

The number of events that could be classified according to the life scripts categories, established by Berntsen and Rubin (2004), showed the same pattern as the overall number of generated events. Again, AD participants produced significantly fewer life-script events compared with the controls. In order to control for differences in fluency, we reran the analysis based on the percentage of life-script events out of the total number of events generated. The differences between the two groups—with AD participants generating a lower percentage of life-script events compared with healthy controls—remained significant. However, nonparametric test, U = 166.5, p = .096, showed only a trend in this direction.

Content of life-script events and agreement with cultural norms

Table 3 shows the frequencies of mentioned life-script events (in percentage) and the estimated mean ages of occurrence (and standard deviations) for the AD and control participants. The life-script events are listed in the rank order of the frequencies established by Berntsen and Rubin (2004) in their normative study for the Danish life script. In both the AD and the control group, the seven most frequently mentioned events were having children, marriage, begin school, college, fall in love, begin daycare, and confirmation. Except for the latter two events, these findings replicate the top five event categories reported by Berntsen and Rubin (2004). Both begin daycare and confirmation attained higher recording frequencies in the present study than in Berntsen and Rubin’s study. However, the relative frequencies (percentages) for confirmation are similar to those reported in later studies on the Danish life script (Bohn & Berntsen, 2008; Bohn, 2010). Similarly, Bohn (2010) reported begin daycare as the sixth most frequently reported life script event in a group of older Danes, suggesting a high overlap between the events reported in the present study and prior findings.

Table 3 Frequencies (percentages) and mean estimated age at event for life script event categories
Full size table

This was further corroborated by the recording frequencies correlating highly with the ones reported by Berntsen and Rubin (2004, Table 3) in both the AD and the control group, rs = .85 and .87, respectively, ps < .001 (under the assumption that the 35 event categories can be treated as independent observations). No group differences were found in these correlations, z = 0.31, p = .38, using Fisher’s r-to-z transformation. We also ran the same correlation again on only the top 10 event categories found in Berntsen and Rubin (2004), as the big differences between high and low frequency categories may have boosted the correlations found. The correlation coefficient for AD participants and healthy controls remained high, rs = .88 and .93, respectively.

From inspection of the frequencies in Table 3, it is clear that fewer AD participants mentioned the high frequency events than did the controls. For example, having children was mentioned by 86.4% in the control group compared with 55.6% in the AD group and marriage by 81.8% in the control group compared with 61.1% in the AD group. We therefore ran a series of chi-squared tests on the seven most frequently mentioned categories to examine frequencies across the two groups. Here, we found that AD participants reported the categories having children, χ2(1)= 4.71, p = .030, and begin daycare, χ2(1)= 4.31, p = .038, significantly less frequently than the controls, with a trend in the same direction for college, χ2(1)= 3.25, p = .071. Thus, although the most frequently mentioned events were the same across the two groups, lower agreement was seen within the responses of the AD group compared with the healthy older adults.

We also examined the content characteristics of the events included in the participants’ life script with regard to how well the life script followed cultural norms. Here, we found no significant difference between the two groups on life-script typicality, when typicality was corrected for total number of generated events. The typicality score is based on the frequencies in the normative sample (Berntsen & Rubin, 2004), which show a steep decline even across the top 10 events (from 90.3% for having children [rank 1] to 21.4% for first job [rank 10]). Given that the healthy participants generated significantly more events (M = 6.91) than did the AD participants (M = 4.67), they would have a greater chance of producing less frequent events, relative to the patient group. We therefore ran the same analyses restricted to the first five events mentioned, while still correcting for number of events generated (up to five). The results showed only a marginally significant difference between the two groups on life-script typicality, t(38) = −1.95, p = .058, d = 0.62, with AD participants’ life scripts being less normative (M = 33.80) compared with the healthy controls (M = 42.61).

Positivity bias

Consistent with the life script consisting predominately of positive events, we found that the majority of life-script events generated by both groups were positive—92.9% in the patient group (2.4% negative and 4.7% neutral events) and 80.7% in the control group (5.0% negative and 14.3% neutral events), respectively. A chi-squared test examining the proportion of positive, neutral, and negative life-script events across the two groups showed that AD participants’ life scripts were in fact even more positive than the healthy controls, χ2(2) = 6.40, p = .041.

Dating of events and the temporal structure of the life script

Dating of events

Of the 250 events generated, 13 events could not be dated. The AD group had 12 undated events out of 98 (12.2%), whereas the control group had one undated event out of 152 (0.7%), χ2(1) = 16.23, p < .001. Thus, the AD participants produced a significantly higher percentage of undated events compared with controls (see Table 2).

To examine the degree to which the estimated ages of events (excluding events that could not be classified according the life-scripts’ categories) in the two groups fitted the normative data, we ran correlations with the age estimates obtained from Berntsen and Rubin (2004) (under the assumption that these measures can be treated as independent observations). For the control group, the correlation was high, r(139) = .95, p < .001. There was also a strong correlation between the normative data and the AD group, r(80) = .77, p < .001, but these two correlation coefficients nonetheless differed significantly, z = 5.69, p < .001, using Fisher’s r-to-z transformation. This suggests that the AD group gave less normative estimates of the timing of life-script events compared with healthy controls.

Life-span distribution of life-script events

Figure 1 shows the age distribution of dated life-script events for both AD participants and controls, in 10-year bins. As illustrated by the figure and consistent with previous studies, a high percentage of events was dated during the reminiscence bump period (15–30 years of age) in both groups—49.3% in the controls and 57.0% in the AD group, respectively, with no significant difference between the two groups, χ2(1) = 1.28, p = .258. Furthermore, while the majority of generated life-script events were positive in both the patient and control group overall (92.9% and 80.7%, respectively), this positivity bias was even more pronounced in the bump period, with no neutral or negative events being dated between 15 and 30 years of age in either group (ps < .05; Fisher’s exact test). Again, this is consistent with previous work (Berntsen & Rubin, 2004; Bohn, 2010; Zaragoza Scherman et al., 2017), demonstrating that positive events in young adulthood are highly favored in the life script.

Fig. 1
figure 1

Life-span distribution of cultural life-script events for Alzheimer’s patients and controls

Full size image

In both groups, a high percentage of events were from the first decade of life (0–9 years) (31.4% for patients and 29.8% for controls). This replicates earlier findings that, compared with the reminiscence bump found in autobiographical memory, the bump in the cultural life script includes more events from the first decade of life (e.g., Bohn, 2010; Tekcan et al., 2012; Zaragoza Scherman et al., 2017). The difference is probably due to memories of personal experiences being affected by childhood amnesia, while the cultural life script is semantic knowledge, learned independent of personal experience.

From inspection of the temporal distribution, the location of the bump appears to be located slightly earlier in the AD group than in the control group, with the bump peaking from 10–19 years in the AD participants compared with 20–29 in the controls (see Fig. 1). In order to address this possible group difference, we sought to estimate the central tendency of the age of generated events in the two groups by estimating the mean and median age of events. For the AD group, the mean age of events were 16.45 years (SD = 10.26), whereas the mean age of events generated by the control group was 20.23 years (SD = 17.29), reflecting a small, but significant difference, t(235) = 2.11, p = .036, d = 0.27. The groups also differed in the median age of generated events, with median age of events being lower in the AD groups (Mdn = 17.00) than in the control group (Mdn = 18.50), using an independent-samples Median test, χ2 (1) = 4.60, p = .032. To explore the difference in the temporal distribution of the life-script events further, we examined the frequencies of events across the two groups for the reminiscence bump periods covering the 10-to-19 and 20-to-29-years bins that showed a bump in the data. When comparing the frequency of events across the two groups in the 10-to-19-years bin, we found a significant difference, χ2(1) = 4.52, p = .034, with 36.0% of events falling in this bin in the AD group versus 23.2% in the controls. Thus, AD patients were 1.87 times more likely to generate an event with an age between 10 and 19 years than were control participants. We did not, however, find any significant group differences in the frequencies of events generated from the age period 20-to-29 years, χ2(1) = 1.18, p = .277.

These results suggest that the two distributions differ, with AD patients’ bump peaking earlier compared with the that of the healthy control participants. This finding may be due to patients dating events earlier in the life span than the controls. To explore this possibility, we used the normative age estimates for the generated event categories (Berntsen & Rubin, 2004) to estimate the temporal distribution of events, had our participants dated events according to these norms. Here, we found no group differences for the percentage of events falling in the 10-to-19-years bin (17.3% for patients vs. 11.8% for controls) or the 20-to-29-years bin (40.8% for patients vs. 44.7% for controls), ps > .22, suggesting that deviations in the dating of events may help explain the AD patients’ earlier peak. The small bump observed for events in the 40-to-50-years bin in the AD group also seemed to be caused by deviations in dating. The four life-script events that made up this small bump were “fall in love,” “marriage,” “having children,” and “the ‘right’ job”—that is, events which usually fall within the reminiscence bump period (15–30 years) (Berntsen & Rubin, 2004, Table 3).

Chronological structure of the life script

To examine whether AD participants had retained knowledge of the temporal, sequential order of life-script events, we analyzed the chronological relation between the generated events by classifying all successively mentioned pairs of events as either forward in time, simultaneous, or backward (Fromholt & Larsen, 1991). Six AD participants did not provide usable data, either because they did not generate any events or because they were unable to date more than one event. We found that both groups generated a high percentage of forward pairs (see Table 2), consistent with both groups relying on a chronological sequential structure when generating events. However, the AD group reported a significantly higher proportion of backward pairs and a lower proportion of forward pairs than did the control group, consistent with AD patients providing life scripts with a lower degree of chronological order than healthy controls.

To ensure that these findings were separate from deviations in estimating the timing of events in the patient group, we performed the same analyses, using the normative age estimates for event categories (Berntsen & Rubin, 2004) to classify the chronological order of events. Using this classification, we replicated our findings with AD patients generating a significantly lower proportion of forward pairs than healthy controls (59.1% vs. 79.9%, respectively), t(20.2) = −2.63, p = .016, d = 0.96, and a significantly higher proportion of backward pairs (39.4% vs. 13.0%), t(19.1) = 3.22, p = .004, d = 1.18. These findings suggest that the lower degree of chronology in AD patients’ life scripts could not be explained by differences in dating, but rather impaired knowledge concerning the sequential order of normative life events.

Correlations between life-script scores and verbal fluency measures in the AD group

We carried out complementary analyses in order to explore whether executive functioning influenced performance on the life-script task in the AD group. Spearman’s rho correlations were performed to assess the relationship between performance on the CLS task (percentage CLS events, undated events, and chronological organization) and verbal fluency measures (semantic and phonemic fluency). There were significant positive correlations between phonemic fluency and the percentage of CLS events, r(21) = .59, p = .005. Moreover, semantic fluency was significantly correlated with the percentage of forward, r(16) = .62, p = .010, and backward pairs, r(16) = −.64, p = .007, based on the normative age estimates (Berntsen & Rubin, 2004), suggesting that impairments in the temporal organization of the life script was associated with frontal dysfunction. There was also a significant negative correlation between percentage CLS events and percentage undated events, r(18) = −.64, p = .004, suggesting that more preserved knowledge of life-script events was associated with fewer problems dating events. No correlations were found between measures of chronological organization and percentage CLS events or percentage undated events.

Discussion

Semantic memory deficits and impaired memory for scripts are well-documented in AD. However, the cultural life scrip, which represents cultural norms for the content, order, and timing of important life events, has been surprisingly little studied in individuals with AD. The goal of the present study was to begin to fill this gap using the standard life-script task. The aim was to evaluate the effect of AD on key characteristics of the cultural life script, in particular, knowledge concerning the content and timing of life-script events and the temporal structure. Our results showed that AD participants and healthy controls both showed many of the key characteristics of the life script (i.e., the most frequent events being the same across the two groups, a clear positivity bias with events mostly from early adulthood), indicating that AD patients do retain some life-script knowledge. However, the two groups also differed in several ways. First, whereas the content of the generated events was quite consistent with those of the healthy controls and with cultural norms, the AD participants produced significantly fewer life-script events compared with controls. Second, their life scripts showed a more pronounced positivity bias. Third, the AD participants had problems estimating the timing of these events, and fourth, they provided life scripts with a lower degree of chronological order than healthy controls. Each of these results are discussed in turn below.

Reduced access to life-script events

The finding that AD participants produced fewer events and that a lower percentage of these events could be classified as life-script events, compared with controls, can be seen as consistent with earlier studies showing deficient script knowledge for simple activities in AD (e.g., Allain et al., 2008; Cosentino et al., 2006; Grafman et al., 1991; Weingartner et al., 1983). For instance, Grafman et al. (1991) tested AD patients on a script generation task asking participants to list all the actions they could think of, that occurred between getting up in the morning and leaving the house for work. They found that patients gave fewer script actions compared with healthy controls, but also generated more events that fell outside of the script boundaries.

However, in the present study, the most frequently mentioned events were the same across the two groups, and we found no group differences in life-script typicality, suggesting that while there is a loss in script content, importantly, semantic knowledge of some key life-script events remains intact in mild to moderate AD. These findings accord with prior reports on routine scripts that AD is associated with omissions of script actions, but that some central script actions are retained (Grafman et al., 1991; Rusted & Sheppard, 2002). Impaired script production has previously been linked to semantic memory deterioration in AD (Grafman et al., 1991), suggesting that semantic deficits may help explain impaired performance on the life-script task. This is consistent with results from the neuropsychological assessment that the AD patients were disproportionately impaired on semantic compared with phonemic fluency, indicating a breakdown in semantic knowledge (Henry et al., 2004). However, whether these deficits reflect actual degradation of script knowledge or simply a reduced ability to access this information remains to be determined (e.g., Bayles et al., 1991; Chertkow & Bub, 1990; Corbett et al., 2012).

Enhanced positivity bias

The finding that AD patients showed a more pronounced positivity bias for life-script events compared with controls may be relevant to theories of emotional processing and cognitive aging. Older adults show a positivity bias whereby they prioritize processing positive over negative information (e.g., Charles et al., 2003; Knight et al., 2002; Thomas & Hasher, 2006). The socioemotional selectivity theory (Carstensen et al., 1999) proposes that as time perspective decreases, the motivation to optimize emotional satisfaction increases, which leads to changes in goal-directed attention favouring positive over negative stimuli. However, alternative accounts have also been proposed, suggesting that neural atrophy in the amygdala, a prominent feature in AD (Poulin et al., 2011), leads to decreased emotional response to negative, but not positive, stimuli (Cacioppo et al., 2011), and that age-related decline in cognitive resources advantage the processing of positive stimuli, thought to be less cognitively demanding (Labouvie-Vief et al., 2007). The fact that AD patients showed a more pronounced positivity bias compared with controls with arguably more cognitive resources is supportive of the alternative accounts proposing that the effect reflects neural or cognitive degradation. The increased access to positive life-script events experienced by the AD group may also have influenced the life-span distribution of events, in that positive events are predominately dated to occur during the period of the reminiscence bump (15–30 years of age), while negative and neutral events are more evenly spread across the life span (Berntsen & Rubin, 2004). Although not significant, AD patients did produce more events from this period than controls did (57.0% vs. 49.3%).

Timing and chronological order

Consistent with our prediction, AD patients showed impaired knowledge for the normative timing of culturally salient events. More specifically, patients were unable to provide time estimates for a higher proportion of life-script events, compared with controls, and the time estimates they did provide were less normative. These results are in line with previous findings of time estimation deficits in AD (e.g., El Haj et al., 2013; Rueda & Schmitter-Edgecombe, 2009). In their review of the literature, El Haj and Kapogiannis (2016) suggested that time distortions may impair episodic memory or, conversely, that degraded episodic memory may cause disturbances in time estimates. However, most studies rely on subjects having to judge time elapsed for shorter intervals (from seconds to minutes). In contrast, here we asked participants to estimate at what age culturally salient events were expected to happen, which likely requires access to information from semantic or episodic memory, namely, semantic knowledge of cultural age norms and/or episodic memory for specific episodes related to the life-script events (e.g., “My wife and I got married at age 25”).

Interestingly, AD patients tended to estimate the ages of life-script events as occurring earlier in the life span compared with controls. Similar findings have been reported for chronological age, with AD patients underestimating their own age (Bohn et al., 2016; Shomaker, 1989), which has been explained in terms of episodic memory deficits. The fact that a similar pattern was observed for knowledge concerning normative ages suggests that semantic deficits may also contribute to time distortions in AD. Deficient semantic memory and executive dysfunction have been linked to impaired performance on cognitive estimation tasks in AD (Brand et al., 2003), which shares several features with the life-script task, requiring people to provide answers with no exact solution readily available.

The finding that AD patients generated a significantly lower proportion of forward pairs and a higher proportion of backward pairs than the healthy controls is consistent with our prediction that AD patients’ life scripts would show a lower degree of chronological order. This is in line with prior findings of impaired sequential organization of script knowledge for familiar actions in AD (e.g., Allain et al., 2008; Grafman et al., 1991; Weingartner et al., 1983). More generally, it relates to findings of impaired memory for temporal order in AD patients across a range of different domains (e.g., Hampstead et al., 2010; Madsen & Kesner, 1995). Impairments related to the sequential structure of scripts for actions and everyday tasks have been linked to compromised prefrontal functioning (Grafman, 1995; Sirigu et al., 1995). There is some evidence to suggest that this is also the case for AD patients (Allain et al., 2008; Cosentino et al., 2006). Allain et al. (2008), for instance, reported that AD patients’ performance on measures of executive functioning correlated with the number of sequencing errors on their script task. Similarly, we found a significant relationship between executive measures and the percentage of forward and backward pairs, indicating that impairments in the sequential organization of the life script were linked to frontal dysfunction.

The present findings may have implications for our understanding of the organization of life-script knowledge. The cultural life script has generally been treated as a unitary construct. In support of this, developmental studies have shown that life-script events and their normative timing are acquired simultaneously (Bohn & Berntsen, 2008; Habermas, 2007; Saraiva et al., 2021), suggesting that the cultural life script is learned as an entity. We found that while the AD patients produced fewer life script event, the content of the actually generated events was quite consistent with cultural norms. Patients showed greatest deficits regarding the order and timing of the events, suggesting that these components of the cultural life script are more vulnerable to cognitive impairments. These findings may be explained by different component processes contributing to different aspects of life-script knowledge, in line with the broader script literature, which dissociates sequential processes reliant on frontal-based processes from the semantic content of scripts (e.g., Cosentino et al., 2006).

Limitations and conclusion

The present study holds a number of limitations. First, the use of a generation task to assess life-script knowledge makes it difficult to determine the relative contribution of executive and semantic impairments to script performance. The life-script task shares several features with verbal fluency tests, which tap both the ability to access semantic memory as well as aspects of executive functioning (Lezak et al., 2012). Hence, performance on the semantic aspects of the life script may have been confounded by task-related executive demands. However, while the association found between phonemic fluency and life script content may be due to material-based similarities, Grafman et al. (1991) argued that the many similarities observed between AD patients’ breakdown in script production and verbal fluency performance reflected that the internal structure of a script is similar to that of a lexical network. We used a generation task here because this particular task is the most widely used task in the life-script literature, and using it thereby rendered our study comparable with previous work. However, future research might aim to assess life-script knowledge independently of verbal production—for example, by asking participants to decide whether events belong to the life script.

Second, although the life-script task explicitly probed semantic knowledge, we cannot exclude the possibility that some participants relied on autobiographical memory to generate their responses. If, indeed, participants relied on autobiographical or episodic memory to generate events, this would have put the AD group at a disadvantage. Asking participants to indicate whether the nominated life-script events referred to events from their own life might help address this concern.

In spite of these limitations, our findings are of relevance to studies that have examined the inclusions of normative life events (i.e., life-script events) in personal life narratives of patients with AD. These studies have reported contradictory findings with regard to whether AD patients include a higher or lower proportion of normative transitional events in their life story compared with healthy older adults (Fromholt & Larsen, 1991; Usita et al., 1998), while one study failed to find any group differences (Tippett et al., 2018). The present study assessed life-script knowledge using the standard life-script task. Using this strategy, we showed that although AD patients’ overall production curtailed, semantic knowledge of some key life-script events remained intact. Importantly, the study adds to previous findings by demonstrating that AD patients have deficient knowledge of the timing and temporal order of life-script events.

The present findings may also be important to autobiographical memory in AD more generally. Life-script knowledge may allow for better memory for events across the life span in a similar way to which other types of semantic knowledge and scripts can support episodic memory in cognitively healthy individuals (e.g., Anderson, 1984; Bartlett, 1932; Bower et al., 1979; Brewer & Nakamura, 1984) and in AD (Johnson & Smith, 1998; Rusted et al., 2000). For example, some work suggests that life scripts support memory retrieval, making events falling within the script more accessible (e.g., Berntsen & Rubin, 2002; Koppel & Berntsen, 2014). Life-script knowledge has also been found to be a prerequisite for the ability to narrate a coherent life story (e.g., Bohn & Berntsen, 2008; Köber et al., 2015), and life-script events are used as landmark events for dating autobiographical memories (Bohn & Habermas, 2016). Future studies might examine whether interventions aiming at restoring participants’ knowledge of the life script would help to improve autobiographical memory and life narratives in AD.

Only a handful of studies have examined script representations in patients with AD, and all of them have addressed scripts for simple action sequences or events, such as familiar actions or routines. In contrast, we examined a different type of script, the cultural life script, which is a culturally transmitted knowledge structure serving as an important mnemonic template when people recall important events from their personal past or narrate their life story. Our results provide evidence that life-script knowledge is impaired in AD patients, especially regarding the timing and temporal order of life-script events. These impairments may diminish the influence of top-down semantic representations during memory retrieval in AD, thus affecting patients’ ability to structure autobiographical recall and their ability to organize life events into a temporally coherent narrative.