Adult memory for specific instances of a repeated event: a preliminary review

Abstract

In cases of repeated victimisation, a complainant’s statement of abuse, and therefore memory, is often critical evidence for forensic investigations and legal proceedings. It is therefore important to understand the functioning of adults’ memory for repeated events. As such, the purpose of this paper was to review the extant literature on adult memory for instances of a repeated event. The results of the review revealed a small number of heterogeneous studies on adult repeated-event memory (N = 12). The literature so far shows that while adults might have difficulty in recalling information specific to instances (narrow accuracy), they are capable of remembering information across multiple instances (broad accuracy). It was also found that several factors may impact recall of instances including age, the number of experienced instances, rehearsing an event, repeated retrieval and event distinctiveness. The discussion highlights the forensic implications of this research and future research directions.

Repeated events – a series of similar types of experiences that all share a common underlying structure or theme – make up a large proportion of our autobiographical memories. Despite the pervasiveness of repeated events in daily life, adults’ ability to accurately recall specific instances of a repeated event has received little research attention. In fact, much of what is known about memory accuracy in adults comes from research with single events. However, the question of how accurately an individual can recall specific instances of a repeated event has important forensic implications. That is because repeated victimisation, such as domestic violence, is unfortunately very common. For example, a recent national report in Australia showed that approximately 54% of those who have experienced domestic violence have so on more than one occasion (Australian Institute of Health and Welfare, 2018). Further, many legal responses to such abuse follow a ‘violent-incident’ model, meaning that charges of abuse are laid based on discrete acts of violence, where complainants are required to recall details that are specific to one or more separate incidents (Stark, 2012).

Given that understanding memory for specific instances has important forensic implications, our goal in this paper is to review the extant research on adults’ memory for specific instances of a repeated event. Since much of what is known about repeated-event memory comes from child samples, we begin by briefly reviewing research on children’s memory for repeated events, and relevant theoretical frameworks. After which, we review and discuss the extant literature on adults’ memory for repeated events. In the discussion, we consider the forensic implications of this research for complainants of repeated abuse, and possible pathways for future research in adults’ memory for repeated events.

Children’s memory for repeated events

Much of what is known about memory for repeated events currently comes from research in children (e.g. for review see Roberts, 2002; Roberts & Powell, 2001; Woiwod et al., 2019). This line of research has been influential in building foundational knowledge about how accurately children remember repeated incidents of sexual abuse. In an attempt to understand how children’s memory is organised for repeated events, researchers will typically follow an experimental paradigm where a group of children experience a repeated event (e.g. Connolly & Lindsay, 2001; Connolly & Price, 2006; Farrar & Boyer-Pennington, 1999; Powell & Thomson, 1996; Price et al., 2006; Price & Connolly, 2004). For example, children will participate in four to six play sessions that are similar in nature. During a specific instance, children might encounter details that are experienced in the same way each time (i.e. fixed details). For example, the same individual facilitates each instance (Connolly & Lindsay, 2001; Powell & Thomson, 1996). They might also encounter details that vary predictably that have associated options across each instance (i.e. variable details; Woiwod et al., 2019). For example, children might play a different ball game in each session. Lastly, they might experience details that change in an unpredictable or unexpected manner in one instance (i.e. deviations details; Brubacher et al., 2011; Connolly et al., 2016). That is, during one instance, someone interrupts the magic show, causing the rest of the show to play out differently compared to the other instances (that did not have an interruption; Connolly et al., 2016). Following the experienced events, children’s memory is tested for those details present during either one instance (i.e. the target instance) or all instances. In some studies, repeated-event children’s memory may also be compared against that of a control group of children who have only experienced a single event (e.g. Woiwod et al., 2019).

When researchers examine children’s memory narratives, they may assess two types of memory accuracy – narrow and broad (Price et al., 2006; Woiwod et al., 2019). A narrow definition of accuracy refers to children’s ability to attribute event details to the correct instance (i.e. particularise individual instances). For example, narrow accuracy is examined by coding the number of correct details (i.e. details reported to have occurred during a target instance) and internal intrusion errors (i.e. details reported to have occurred during a target instance, but had occurred during non-target instances) reported in the memory narrative. Thus, an ideal level of narrow memory accuracy would be recalling a high number of correct details and a low number of internal intrusion errors. Conversely, a broad definition of accuracy refers to children’s ability to recall all experienced details, regardless of during which instance they occurred [i.e. correct details (defined narrowly) + internal intrusion errors]. Thus, the only difference between the two definitions is whether or not children are expected to attribute event details to their specific instance.

Results from previous studies indicate that children’s recall of details that remain the same during repeated events (i.e. fixed details) is quite accurate¹ (e.g. Pearse et al., 2003; Powell et al., 1999; Powell & Thomson, 1996). In contrast, these same children find it more challenging to recall details that vary predictably across instances when compared to single-event children, specifically, when accuracy is defined narrowly (e.g. Woiwod et al., 2019). That is, compared to single-event children, repeated-event children struggle to attribute correct variable details to the target instance (e.g. 50% versus 29% correct, respectively). In contrast, when accuracy is defined broadly, repeated- and single-event children show comparable levels of accuracy; both groups report an equivalent amount of experienced details (e.g. 66% versus 57% correct, respectively; Woiwod et al., 2019). Together, these findings indicate that when children experience a series of similar instances, they are capable of remembering what they have experienced, but they may struggle to determine during which instance the specific variable details were originally encountered. Thus, it appears that memory for what happened is not necessarily impaired.

Research has also investigated how deviations introduced during one instance effects children’s memory for repeated events. As mentioned above, recall of deviation details refers to a situation where something unpredictable occurs during one instance when compared to what typically happens in each instance (Connolly et al., 2016). Some studies have examined memory for the deviation details compared to variable details (i.e. typically occurring, but slightly differentiating, event details). These studies show that, perhaps unsurprisingly, the unpredictable details (i.e. deviation details) are better remembered (Brubacher et al., 2011; Connolly et al., 2016). Interestingly, research has also shown that the presence of deviation details during one instance may have implications for how children remember the other typically occurring instances (i.e. non-deviation instances; Connolly et al., 2016). For example, Connolly et al. (2016) found that compared to the absence of deviation details, the presence of deviation details during one instance led children to recall more variable details across all instances, even those instances not containing the deviation details. Overall, it appears that memory ought to be more accurate when deviation details are present during one instance in the repeated-event series.

Theoretical framework for repeated-event memory

When accounting for memory for specific instances of repeated events, researchers have referred to three main theories: fuzzy trace theory (Brainerd & Reyna, 1990, 2002, 2004; Reyna et al., 2016), script theory (Hudson et al., 1992; Hudson & Mayhew, 2009; Schank & Abelson, 1977), and the source monitoring framework (Johnson et al., 1993; Mitchell & Johnson, 2009). Fuzzy trace theory proposes that instances of a repeated event are encoded, stored and retrieved as separate memory traces. Specifically, each time an instance is encountered, two parallel yet independent memory traces are encoded about that instance (Brainerd & Reyna, 1990, 2002, 2004; Reyna et al., 2016). Verbatim traces consist of the exact surface details and source information of an event, and gist traces consist of the underlying structure or ‘general meaning’ of the event. Each time a similar event is encountered, the same gist trace is activated and strengthened for that routine, and, simultaneously, a new verbatim trace is formed for that occurrence. Over time, the relative strength of the gist and verbatim traces start to differ, which then influences the degree to which the traces are accessible at the time of memory retrieval (Brainerd & Reyna, 2004). Gist traces are slower to decay than verbatim traces, given the former are strengthened more than the latter over time. Therefore, this theory predicts that when recalling a specific instance of a repeated event, individuals will be more likely to retrieve the general meaning of that event (i.e. gist trace) rather than the specific features of that instance (i.e. verbatim traces).

When memory accuracy is defined narrowly, fuzzy trace theory appears to account for the results. Previously, we noted that repeated-event children are less capable of attributing variable details to the target instance than single-event children (e.g. Woiwod et al., 2019). These data are consistent with the notion that the specific verbatim trace for the target instance has been subject to decay, thus reducing the likelihood of recalling the correct specific details. However, when memory accuracy is defined broadly, fuzzy trace theory does not appear to fully account for the results. Specifically, a broad measure of accuracy classifies any reported detail as correct if it has been experienced in any of the instances, with results showing that repeated- and single-event children have comparable levels of broad accuracy. This indicates that repeated-event children are capable of retrieving specific event details – verbatim traces – that occurred in either the target or non-target instances (i.e. all experienced details). Thus, if repeated-event children are able to remember these specific event details then it suggests that instances are not necessarily organised as individual memory traces (i.e. verbatim memory traces). That is, if events were organised as individual memory traces, repeated-event children should have lower broad accuracy than single-event children. After all, the former should still find it more difficult to access specific details, given they have created multiple traces, than the latter. This is not the case (Price et al., 2006; Woiwod et al., 2019).

Fuzzy trace theory might also not account for the effect of deviation details on repeated-event memory. Compared to variable details (i.e. typically occurring details), deviation details during one instance might ‘stand out’ as an unusual case. Thus, during memory encoding more attention is directed to these deviation details that ‘stand out’. This increased attention could lead to a better consolidated memory for that instance (Davidson, 2006; Farrar & Goodman, 1992). Therefore, fuzzy trace theory would predict that the verbatim trace for the deviation instance should be better consolidated and more accessible than the verbatim traces for the typically occurring instances. However, this is not what the child literature indicates – the findings showed that the presence of a deviation compared to the absence appeared to enhance memory for all instances, including the typically occurring instances (the non-deviation instances). These findings together suggest that details from each instance, whether deviation or variable details, do not appear to be tied to specific memory traces as indicated by fuzzy trace theory.

Recently, Woiwod et al. (2019) proposed that script theory might offer a more comprehensive account of the memory processes involved in repeated events. Script theory proposes a schema-based account of remembering instances of repeated events (Hudson & Mayhew, 2009; Schank & Abelson, 1977). According to script theory (Schank & Abelson, 1977), humans organise experiences with similar instances (i.e. repeated event) as general event representations or scripts. Specifically, scripts are conceptualised as hierarchically organised knowledge structures, composed of general and specific levels. At the general level, a script outlines what usually happens for that type of event, including information about the typical people, actions and objects for that event, as well as the temporal order in which the event usually takes place (Bower et al., 1979; Hudson & Mayhew, 2009). For example, many people have a script that outlines what usually happens when visiting a restaurant, such as entering the restaurant, ordering food, eating and exiting (Bower et al., 1979). At the specific level, each component (i.e. eating at a restaurant) of a script contains a list of options that are associated with that component. For example, when attending an Italian restaurant, an individual knows that they have eaten (component) ravioli, gnocchi or pizza (options). These associated options are believed to be organised as a list (i.e. specific dishes) linked to the components of the script (i.e. eating at Italian restaurant; Woiwod et al., 2019).

During memory retrieval, a script can act as a schematic framework that aids an individual to reconstruct what occurred during a specific instance (Hudson & Mayhew, 2009; Woiwod et al., 2019). In a situation where an individual is remembering an instance of a repeated event, they might draw upon their script and recount what usually happens by providing information about the typical people, actions and objects for those events. In addition, an individual might then attempt to recall details that are specific to that event (i.e. associated options). For example, one might decide that they ate ravioli during their last visit to an Italian restaurant. The ability to attribute which specific option (e.g. ravioli) occurred in which instance (e.g. the last visit) might depend on the strength of the source memory or source monitoring ability (Johnson et al., 1993).

Source monitoring, as outlined in the source monitoring framework, refers to the ability to determine the origin of memories (Johnson et al., 1993; Mitchell & Johnson, 2009). This theory suggests that memories are encoded based on the content, and not the source of the memory. As such, it is during memory retrieval that an individual must decide where they originally encountered that memory. When a memory is attributed to the wrong source, a source monitoring error has occurred (Johnson et al., 1993). In the context of repeated-event research this would be classified as an internal intrusion error – a memory, or detail thereof, is attributed to the wrong instance (source) from a series of instances (Woiwod et al., 2019). Attributing memories or specific details to the correct instance can depend on a number of factors. First, the memory characteristics or quality of the original memories (e.g. perceptual, contextual, semantic and affective information) can influence correct attribution (Johnson et al., 1993). Memories with richer memory characteristics would presumably elicit more accurate judgements because there are more cues available to discern when the memory was experienced. A second factor that might influence source attributions is how similar each of the experienced instances are to one another. The more similar the experienced instances are, the more likely incorrect source attribution will occur (Lindsay et al., 1991).

Together, script theory and the source monitoring framework indicate that remembering one instance of a repeated event will involve reconstructing what happened at a general level (what usually happens) and then attributing specific event details to that instance. Given details are not tied to instances (i.e. sources), an individual must engage in source monitoring to complete this task. This notion is supported by the fact that repeated-event children are able to remember experienced details, but struggle to attribute them to the correct occurrence (Woiwod et al., 2019). This is shown when considering differences in memory accuracy for narrow and broad definitions.

One question remaining concerns how deviation details impact upon scripts and source monitoring following repeated experiences. We mentioned previously that deviation details tend to capture attention as they stand out as a special case, and that this increased attention in turn might support better consolidation of the deviation instance. However, the data indicate that deviation details have a general effect on memory (memory is better for all instances). As such, it is possible that in order for an individual to recognise that a deviation instance stands out, they would need to compare it to other typically occurring instances. This ‘comparison process’ might lead one to rehearse and further consolidate all previously experienced details, thus strengthening memory for the details associated with all instances (Connolly et al., 2016). Consequently, with increased elaborate processing of all experienced details, an individual might be better able to accurately attribute (i.e. source monitoring) specific details to the correct instance.

Aim of review

In the previous section, we discussed the findings concerning children’s memory for instances of repeated events and the theories that can account for the memory processes underlying their performance. Surprisingly, less is known about the functioning of adults’ memory for repeated events. There are, however, two main reasons why adults’ memory deserves further research attention, and why it might not be appropriate to generalise the child literature to adult samples. We discuss these two reasons below.

It is noteworthy that the child repeated-event literature has focused on examining memory for children between the ages of around 6.4 and 10.0 years (Woiwod et al., 2019). As such, the findings within the child research might not be an appropriate estimate of adults’ memory performance.² That is, while we expect that the underlying memory processes involved in remembering repeated events (e.g. script memory, source memory) might be similar between children and adults, memory ability between the populations could be different. Indeed, previous research – into memory for single events – indicates that memory performance increases throughout childhood to adolescence and into adulthood (e.g. Cycowicz et al., 2001; Dirks & Neisser, 1977; Ofen et al., 2007). Further, these age-related differences in memory performance appear to correlate with maturation in brain regions that are associated with memory encoding and retrieval, such as the prefrontal cortex (Ofen et al., 2007, 2019; Sowell et al., 2001). Therefore, it appears that the maturation of the prefrontal cortex across development indicates that the development of other cognitive constructs such as attention and strategic control processes might be involved in predicting age-related differences in memory performance between children and adults (Ofen et al., 2019). Together, the findings above indicate that compared to children, adults might be expected to show higher estimates of memory accuracy (both narrowly and broadly defined).

In further support of these findings, fuzzy trace theory, script theory and the source monitoring framework all predict increases in memory performance across development. For example, fuzzy trace theory proposes that the ability to store, consolidate and retrieve gist and verbatim traces improves with age (Brainerd & Reyna, 2004; Brainerd et al., 2002). Stronger gist and verbatim traces in adults compared to children means that the former population might be more capable of recalling details specific to a target instance. In a similar way, compared to children, adults’ scripts are more elaborate and complex (Hudson et al., 1992). For example, adults are better than children at recognising predictable and unpredictable variations in instances, potentially increasing the number of specific details consolidated for each component of the script (Hudson et al., 1992). Adults also have better source monitoring ability than children. For example, adults experience less confusion between similar details than children because the former are better at discriminating between similar sources than the latter (Lindsay et al., 1991). As a result, when adults remember one instance of a repeated event, they might experience less confusion in memory and be more capable of associating correct event details with the correct instance (i.e. fewer source monitoring errors).

Second, understanding adult memory for repeated events has important forensic implications. That is, just like children, adults can be repeatedly victimised, and are required to testify to their experiences of such abuse. For instance, adults are known to experience the following forms of repeated abuse: domestic and family violence, bullying and harassment, and sexual harassment (Australian Institute of Health & Welfare, 2018). Further, given the private and concealed nature of these types of abuse, there is often very little corroborating evidence. This means that within a legal context, a complainant’s statement of abuse, and thus memory, might become a crucial aspect of evidence in such cases.

In sum, understanding adult memory for repeated events is important from a theoretical and forensic perspective. Based on previous research it is likely that adults have better memory abilities than children. Regardless, adults might be expected to show a higher estimate of memory accuracy (narrow and broad) than children. If this is the case, the child repeated-event literature is not an appropriate estimate of adults’ memory performance. Additionally, like children, adults can be victims of repeated abuse who then are required to provide testimony as evidence of the abuse. Given the seriousness of these crimes, and the fact that the literature on adult memory is still in its infancy, we deemed a review appropriate in order to establish the current state of the literature and determine the appropriate next steps for this research. In the following, we identify and summarise the extant literature on adults’ memory for repeated events.

Literature search and identification

We obtained relevant research on adult memory for repeated events by conducting a systematic search of multiple scientific databases (i.e. PsycInfo, Medline, Web of Science and Scopus) using the following key terms: (memory OR eyewitness memory OR recall) AND (repeated event OR multiple event OR single event OR unique event OR recurring event) AND (adults). In addition, we searched the reference lists of articles identified using the search terms above, and we contacted key researchers in the field for any unpublished data. Our inclusion criteria for the review were liberal. We included any studies that recruited an adult population (i.e. 18 years or older), had a repeated-event group (participants experienced two or more instances) and investigated memory accuracy (narrow and/or broad definitions). Studies assessing memory for laboratory-based or personally experienced events (i.e. autobiographical memories) were included, under the condition that objective truth could be verified (i.e. memory accuracy). The final database search was completed in April 2020 and returned 141 papers; the abstracts and full texts of articles were screened with the aforementioned inclusion criteria, leaving 12 studies to be included in the review.

The systematic search of the literature on adult memory for repeated events returned a small heterogeneous research base (N = 12). It is notable from our search that the adult literature has received less systematic research attention than the child literature. In general, most studies investigated adults’ ability to particularise individual instances (i.e. report details that are specific to individual instances). However, the studies varied in what specific research questions were being investigated, the type of population sampled, the study methodology/procedures and the type of memory measures used. Table 1 outlines the characteristics of each of the reviewed studies. Below, we do not present an exhaustive summary of every finding from each study, rather we focus on summarising and synthesising the main findings (significant or not) to highlight what has been examined and to identify trends in the literature. We have organised the literature in three parts. We first review the research examining memory for single versus repeated events when all details vary in a predictable manner (i.e. variable details). We then review factors that might influence recall of repeated events. Then we review research examining the impact of event distinctiveness (e.g. deviation details) on repeated-event memory.

Table 1.

Characteristics of studies included in the review.

Study	Participants	Manipulated factors, experiment type	Event type	Memory report type	Main findingsb
Deck and Paterson (2020)	Students N = 41 Mage = 19.89 63% female	Event frequency (repeat vs. single) Controlled experiment	Healthy lifestyle sessions	Free recall Cued recall	Event frequency effects: The repeat group attributed less correct (narrow) information to the target instance than did the single group. Repeat and single groups report a similar number of correct (broad) information.
Dilevski et al. (2020)	Students N = 122 Mage = 20.94 100% female	Event frequency (repeat vs. single) Emotional stress (stressful vs. non-stressful) Controlled experiment	Imagined experiences of abusive and non-abusive relationship encounters	Free recall	Event frequency effects: The repeat group attributed less correct (narrow) information to the target instance than did the single group (regardless of emotional stress group). Repeat and single groups report a similar number of correct (broad) information.
van Golde et al. (2017)	Students N = 42 Mage = 19.52 63% female	Immediate recall vs. no immediate recall Controlled experiment	Workplace bullying videos	Guided free recall	Immediate recall effects: Immediate recall group and no immediate recall group report a similar amount of correct (broad) information. Immediate recall group reports fewer errors than no recall group; thus, the former group produces a higher proportion of correct (broad) information than the latter.
Kontogianni (2018)	Students N = 150 Mage = 21.26 81% female	Content deviation (present vs. absent) Interviewing format (timeline vs. MMIF vs. free recall)a Controlled experiment	Videos depicting terrorist plot meeting	See manipulated factors	Content deviation effects: No significant content deviation effects
Means and Loftus (1991) Experiment 3	Community N = 27 Age: 24–62 67% female	Event frequency (repeat vs. single) Event seriousness (serious vs. minor) Quasi-experiment	Autobiographical memories about health care visits	Free recall	Event frequency and seriousness effects: Repeat and single groups report a similar number of correct (broad) information. Repeat minor recalled less correct information than single minor. No difference in correct recall between repeat and single serious groups.
MacLean et al. (2018) Experiment 1	Students N = 73	Instances (5 instances) Content deviation (present vs. absent) Post-event rehearsal (present vs. absent) Controlled experiment	Food tasting sessions	Cued recall	Instance effects: More correct (narrow) information was attributed to Instance 1 than the other instances. Post-event rehearsal and instance effects: When rehearsal was present, less information was misattributed to Instance 1 than the middle instances, and to Instance 5 than 4. Deviation effects: More information was correctly (narrow) attributed to all instances when deviation was present than when absent.
MacLean et al. (2018) Experiment 2	Students N = 108 Mage = 19.66 64% female	Instances (5 instances) Content deviation (present vs. absent) Post-event rehearsal (present vs. absent) Controlled experiment	Food tasting sessions	Cued recall	Deviation, post-event rehearsal, and instance effects: When content deviation and rehearsal were present, more correct (narrow) information was attributed to Instance 1 and Instance 3 (deviation instance). When content deviation was absent and rehearsal present, more correct information was attributed to Instance 1 than other instances. When rehearsal was absent, more correct information was attributed to Instance 1 than other instances.
Rubínová, Blank, Ost, et al. (2020) Experiment 1	Students N = 96 Age: 18–40 53% female	Instances (4 instances) Content deviation (present vs. absent) Order deviation (present vs. absent) Controlled experiment	Categorised word lists	Free recall	Instance effects: More correct (narrow) information attributed to Instances 1 and 4 than to the middle instances. Content deviation effects: More correct (narrow) information was attributed to all instances when content deviation was present than when absent. This difference, however, was descriptive rather than statistically significant. Order deviation effects: Less correct (narrow) information was attributed to all instances when order deviation was present than when absent.
Rubínová, Blank, Ost, et al. (2020) Experiment 2	Students N = 80 Age: 18–45 68% female	Instances (4 instances) Content deviation (present vs. absent) Order deviation (present vs. absent) Controlled experiment	Categorised word lists	Free recall	Content deviation effects: More correct (narrow) information was attributed to all instances when content deviation was present than when absent. Order deviation effects: Less correct (narrow) information was attributed to all instances when order deviation was present than when absent. This difference, however, was descriptive rather than statistically significant.
Rubínová, Blank, Koppel, et al. (2020)	Students N = 149 Mage = 20.15 75% female	Instances (4 instances) Content deviation (present vs. absent) Order deviation (present vs. absent) Controlled experiment	Unfamiliar wedding ceremony stories	Free recall	Content deviation and instance effects: Less correct (narrow) information was attributed to Instance 4 (deviation instance) when content deviation was present than when absent. Order deviation and instance effects: Less correct (narrow) information was attributed to Instance 4 (deviation instance) when order deviation was present than when absent.
Theunissen et al. (2017)	Students N = 65 Mage = 19.5 78% female	Event frequency (repeat vs. single) Report session (Time 1 vs. Time 2) Controlled experiment	Traumatic car accident videos	Self-Administered Interview©	Event frequency effects: The repeat group attributed less correct (narrow) information to the target instance than the single group. Repeat group attributed more correct (narrow) information in Report 2 than in Report 1, while Reports 1 and 2 were equivalent for single group.
Willén et al. (2016)	Community N = 95 Mage = 43.33 75% female	Predictor variables: age, number of dental visits, interviewer, unpleasantness of dental visits & rehearsal of dental visits Quasi-experiment	Autobiographical memories about dental visits	Free recall + cued recall	Age, number of dental visits, self-reported unpleasantness, talking about the visits positively predicted the number of dental visits referred to. Age negatively predicted the number of specific details recalled for each visit. No other factors were significant predictors.

Note: MMIF= Multi-method interviewing format. ^aThe findings for this variable are not shown as we are not presenting research relating to interviewing techniques in this review. ^bThis paper does not present an exhaustive summary of every finding from each study; rather we focus on summarising and synthesising the main findings across studies to highlight what has been examined and to identify trends in the literature. Across the studies, researchers have operationalised their memory outcome measures in different ways. As such, when reporting the main findings across the studies, we have focused on conveying the results regarding ‘memory accuracy’ (narrow and broad definitions). A narrow definition of memory accuracy captures the ability to correctly remember information that was experienced during specific instances. A broad definition of memory accuracy captures the ability to correctly remember information that was experienced during any instance, regardless of when it occurred.

Adult memory for single and repeated events

Earlier, we noted that when children experience repeated events that vary predictably across each event, they show difficulty in particularising the target instance compared to children that have experienced a single event. At present, three controlled studies have compared adults’ ability to particularise the target instance after experiencing a repeated event or a single event (Deck & Paterson, 2020; Dilevski et al., 2020; Theunissen et al., 2017). In a sample of young university students, all studies compared memory accuracy (defined narrowly) for a target instance between a repeated-event group and a single-event group. The type of to-be-remembered events used across the studies varied, including healthy lifestyle sessions (Deck & Paterson, 2020), traumatic car accident videos (Theunissen et al., 2017) and ‘imagined experiences’ of abusive and non-abusive relationship encounters (Dilevski et al., 2020). Across studies, memory was always assessed for the last experienced instance between five to nine days after the last instance. The results across the three studies indicated that the repeated-event group was less capable of attributing correct information to the target instance than the single-event group (Deck & Paterson, 2020; Dilevski et al., 2020; Theunissen et al., 2017).

Two of the above studies also examined the difference in repeated- and single-event memory when accuracy was defined broadly (i.e. experienced details across all instances). Deck and Paterson (2020) and Dilevski et al. (2020) both found that the repeated- and single-event groups recalled a similar number of experienced details. This same pattern was also found in a quasi-experiment conducted by Means and Loftus (1991). The researchers assessed autobiographical memories for health care visits in a sample of community members that occurred either on multiple occasions (i.e. repeated event) or on one occasion (i.e. single event). Similar to the controlled experiments, both groups recalled an equivalent amount of correct information when using a broad definition of accuracy.

Although limited by a handful of studies, it appears that when all instances in a series follow a predictable manner, repeated-event participants are capable of remembering experienced variable details, but may struggle to attribute these details to the appropriate instance (i.e. source) when compared to single-event participants. Similar to children, adults’ organisation and retrieval of instances of repeated events are not organised around separate memory traces as predicted by fuzzy trace theory (Brainerd & Reyna, 1990, 2002, 2004; Reyna et al., 2016). Rather, as argued in Woiwod et al. (2019), it appears that individuals might have a list of specific experienced details that have been accumulated over repeated experiences that are not tied to individual instances (i.e. a script). Thus, remembering an instance of a repeated event will necessarily involve a reconstructive process, whereby individuals might use what usually happens as a framework and then attribute the specific details (i.e. options) to when they occurred. Attributing correct details to an instance will involve source monitoring ability.

Factors that impact recall of repeated events

Researchers have also examined factors that might interact with event frequency in their effect on memory. In a controlled experiment, Dilevski et al. (2020) examined the role of emotional stress at encoding on repeated- and single-event memory. A sample of female university students engaged in ‘imagined experiences’ of abusive or non-abusive relationship encounters on either one occasion or multiple occasions. One week after the last encounter, memory was assessed for the target instance (i.e. the last instance). The researchers found that the repeated event group were less accurate (defined narrowly) in their memory for the target instance than the single group, regardless of whether the events were emotionally stressful (i.e. abusive encounter) or not (i.e. non-abusive encounter). Similarly, Means and Loftus (1991) examined whether the seriousness of an event (i.e. minor versus serious event) differentially impacted recall of single and repeated health care visits. A serious event was defined as ‘involving a problem that would have a high probability of resulting in a major infection, debility, or death if not treated by a medical professional’ (Means & Loftus, 1991, p. 307). Results showed that participants were less accurate (broadly defined) in their recall of memories of repeated minor health care visits than of a single minor health visit. Recall of memories for repeated and single serious health visits did not differ. Therefore, it appears that the frequency of an event will impact memory only when the experiences are perceived as inconsequential and not life-threatening.

In a controlled experiment, Theunissen et al. (2017) examined the accuracy (narrowly defined) of memory reports across two interviews for a single event versus an instance of a repeated event. Although not significant, there was a trend towards the repeated-event group recalling more correct information in their second report than in their first, while the two reports were similarly accurate for the single-event group. This suggests that a second retrieval attempt had a beneficial impact on the memory accounts of repeated-event participants.

In a quasi-experiment, Willén et al. (2016) recruited dental care patients to recall memories about dental visits that occurred over a 10-year period. They specifically examined whether five different factors (i.e. age, the number of dental visits experienced, the interviewer, the self-reported unpleasantness of the visits, amount of time spent rehearsing the visits) predicted the number of visits referred to and the number of correct details recalled about the visits. They found that the number of referred-to visits increased as the total number of visits increased, the more unpleasant they reported the visits to be, and the more they spoke about the visits to other people. They also found that age (M = 43.33 years, SD = 15.17) was a significant negative predictor of the number of specific correct details recalled for each visit. This suggests that as age increased the number of correct details recalled decreased. No other factors were significant predictors.

In a sample of university students, van Golde et al. (2017) assessed the effect of immediately recalling each instance on memory for repeated events. Using a controlled design, they compared the final recall of participants who either previously noted down their recollection of workplace bullying videos immediately after they watched them or did not do this. One week after watching the final video, they found that the amount of correct details recalled between both groups did not differ when using a broad accuracy measure. However, compared to the no immediate recall group, those that engaged in immediate recall after each video were less likely to include non-experienced details, and, thus, their memory report included a higher proportion of accurate information (defined broadly). This suggests that the act of initially writing down what happened during each instance of a repeated event might increase overall recall accuracy in a subsequent interview.

Distinctiveness and recall of repeated events

Our search also found that six adult studies have examined what impact deviations details during one instance of a repeated event has on memory for each instance (accuracy – defined narrowly; Kontogianni, 2018; MacLean et al., 2018; Rubínová, Blank, Koppel, et al., 2020; Rubínová, Blank, Ost, et al., 2020 ). Across the six controlled studies, university students were allocated to either a deviation group (i.e. deviation details occurred during one instance) or a no deviation group. Similar to the child literature, all studies implemented a deviation that changed the content of one instance in the series, which involved changes to what happens during the instance. For example, the environmental setting of the study changed during the deviation instance (i.e. MacLean et al., 2018), or in one word list (i.e. instance) the word categories changed from the typical ‘clothing’ words to ‘fruit’ words (i.e. Rubínová, Blank, Ost, et al., 2020). Two studies also examined memory when the deviation details changed the order of one instance in the series (Rubínová, Blank, Koppel, et al., 2020; Rubínová, Blank, Ost, et al., 2020). That is, an order deviation occurs when the typical content of an instance (i.e. what happens in the event) is presented in a different temporal order. Across these six studies, the researchers assessed memory accuracy (defined narrowly) for all instances in the repeated-event series between 10 min and one month after the final instance was experienced.

When there was a change to what happened during one instance, four out of the six studies showed that content deviations had a recall-enhancing effect on memory (Kontogianni, 2018; MacLean et al., 2018; Rubínová, Blank, Ost, et al., 2020). For example, three studies found that compared to the absence of a deviation, the presence of a content deviation during one instance of a repeated event had a general effect on memory (i.e. the enhancing effect of the deviation instance generalised to all other instances; MacLean et al., 2018, Experiment 1; Rubínová, Blank, Ost, et al., 2020). One study also found that content deviations had a targeted effect on memory; participants were more accurate in their recall of the deviation instance (i.e. Instance 3) and Instance 1 than of all the other instances (MacLean et al., 2018, Experiment 2).³ In contrast, two of the six studies found that deviations did not have a recall-enhancing effect on memory. In fact, Rubínová, Blank, Koppel, et al. (2020) found that the presence of a content deviation had a recall-disrupting impact on memory compared to the no content deviation group.⁴ In another study, Kontogianni (2018) found no difference in memory between the content deviation and no deviation groups. This absence of a content deviation effect on memory might possibly be because the deviation details implemented were too subtle to be noticed as a deviation to what typically occurs.⁵

In contrast to content deviations, three studies showed that order deviations had a recall-disruptive impact on memory recall. Rubínová, Blank, Ost, et al. (2020) found that participants who experienced a deviation to the order of one instance recalled less correct information (narrowly defined) about all instances than participants who experienced no deviation. Similarly, Rubínová, Blank, Koppel, et al. (2020) found that the order deviation group recalled less correct information (defined narrowly) about the deviation instance (i.e. Instance 4) than the no deviation group.

Together, the mixed findings outlined above make it difficult to draw clear conclusions about the effects of deviations on memory for repeated events. However, it does appear that deviations to one instance has implications for how all other instances in the series are remembered. In particular, it appears that content deviations have a recall-enhancing effect on memory. This notion is also supported by memory theory that suggests that the presence of a content deviation enhances memory because it initiates a ‘comparison process’ among all instances, which then causes all details (i.e. options) associated with a script to be further consolidated, thus better remembered (Connolly et al., 2016; Rubínová, Blank, Ost, et al., 2020). On the other hand, order deviations appear to have a recall-disrupting impact on memory. Given that recall was disrupted by order deviations, it suggests that changes to temporal order might impair initial encoding of the instance and/or hinder effective reconstruction during retrieval (Rubínová, Blank, Ost, et al., 2020).

A final factor that influences repeated-event memory is the position of an instance within the repeated-event series. For example, in this research, participants (i.e. university students) experienced four to five instances, after which they were instructed to recall information about all the instances in the repeated-event series (MacLean et al., 2018; Rubínová, Blank, Koppel, et al., 2020; Rubínová, Blank, Ost, et al., 2020). Collectively, the research found that participants often recalled the most correct information about the first instance and last instance compared to the middle instances, demonstrating primacy and recency effects, respectively. Thus, these findings suggest that some instances might be better remembered than others depending on when they occurred in the repeated-event series.

Forensic implications of the findings

Although limited by a handful of studies, the findings in this review might have implications for memory evidence requirements in cases of repeated abuse. Here, we refer to domestic violence as an example. In such cases, the requirements of a complainant’s statement might depend on which model of domestic violence a jurisdiction prescribes to. In many jurisdictions, the pervasive criminal justice response to domestic violence is based on a ‘violent-incident model’ (Stark, 2012; Wangmann, 2020). According to this model, domestic violence is understood as comprising discrete assaults or threats by a perpetrator. The seriousness of each assault is determined by the degree of harm inflicted or intended. This means that the types of incident that warrant a justice intervention are ‘high-level’ assaults such as physical and/or sexual violence. In accordance with this model, a complainant might be expected to particularise discrete acts of abuse (i.e. report details specific to discrete events). Thus, a narrow ability of memory accuracy would be most useful in such cases, as this assesses an individual’s ability to attribute details to instances (i.e. remember what happened during specific abusive incidents). The findings in the review showed that adults struggle to particularise instances of a repeated event when compared to a group that experienced a single event. Therefore, these findings do not necessarily support the memory evidence requirements used in jurisdictions that follow the violent-incident model (Stark, 2012; Wangmann, 2020).

An alternative model, the ‘coercive control model’ (Stark, 2007, 2012), conceptualises domestic violence as a course of ongoing coercive and controlling conduct that can be composed of both physical forms of abuse and ‘low-level’, non-physical forms of abuse (e.g. psychological or emotional, financial and social abuse). This conceptualisation of abuse might be particularly important in situations where an instance of abuse, when considered alone, appears seemingly innocuous unless placed within the context of a pattern of abuse. For instance, on one occasion a victim might be ordered to ‘vacuum the carpet until you can see the lines’, with the victim knowing full well there would be consequences (e.g. physical punishment) unless the act was carried out accordingly. Alone, this incident might seem inconsequential, possibly not even criminal, but when placed within a context of a pattern of abuse it could be classified as coercive and controlling behaviour. Therefore, this conceptualisation of domestic violence has led to the introduction of a ‘coercive control’ law in several jurisdictions (e.g. Scotland, Wales), which lays charges of abuse based on a pattern of repeated coercive and controlling behaviour (Bettinson, 2020; McMahon & McGorrery, 2020). To meet the requirements of this law, a complainant might be expected to recount a series of interrelated abusive events that together establish a pattern of coercive and controlling behaviour. In such cases, a broad ability of accuracy would be most useful as it accounts for what is remembered across multiple abusive incidents (i.e. experienced details). The data from this review indicate that memory is not ‘impaired’ for repeated-event participants when they are able to remember details across multiple experiences, rather than being constrained to refer to specific instances. Thus, these data support, and are more consistent with, the memory evidence requirements as per the coercive control model, rather than the violent-incident model of domestic abuse (Bettinson, 2020; McMahon & McGorrery, 2020; Stark, 2007, 2012). However, one possible limitation of the coercive control model and the reliance on a broader definition of accuracy is that it can be more difficult for the accused to argue their innocence.

Limitations of the research and future research directions

Given that very few studies have investigated adult repeated-event memory, it is important to regard the findings of this review with caution. We encourage further replication of the findings in order to support the generalisations made to memory for repeated abuse noted above. In particular, one fruitful area of research would be to further examine the effect of deviations on repeated-event memory. As evidenced by the coercive control model, there is often a great deal of unpredictable variability and disparity between instances of abuse, ranging from physical to non-physical abuse (e.g. emotional, financial, social, spiritual, technology-facilitated, etc.). Thus, in order to understand how such disparity between instances impacts memory, future research should consider implementing greater unpredictability variability between instances, possibly by examining the effect of multiple deviations instances on memory.

One limitation of the literature is that 83% of the included studies relied on convenience samples (i.e. university students). This overreliance of university samples might limit the generalisability of the findings to samples beyond young adulthood. Further to that, there was one study in this review that found that age negatively predicted the number of specific details recalled for instances, suggesting possible age-differences in memory (Willén et al., 2016). Thus, one task of future research might be to investigate adult repeated-event memory across the adult life span, by comparing younger, middle-aged and older adults. This research would have important applied implications, as repeated abuse is not necessarily restricted to certain age groups (Australian Institute of Health & Welfare, 2018).

Future research might also consider examining whether memory for an instance of a repeated event is impacted by factors such as time delay, different types of emotional responses and susceptibility to misinformation acceptance. These factors are important to consider as they are frequently a consideration during legal proceedings.

In some cases of repeated abuse, there are many reasons why a victim does not immediately report an incident of abuse, including fear of revenge from the perpetrator, shame and the victim perceiving the incident as too trivial to report (Birdsey & Snowball, 2013). In such cases, factfinders and triers of facts might be interested in understanding how the delay between experiencing a criminal event and reporting the event affects memory (Read & Connolly, 2007). Indeed, it is well documented that memories can become increasingly less accessible over time (e.g. Ebbinghaus, 1913; Odinot & Wolters, 2006; Paz-Alonso & Goodman, 2008; Rubin & Wenzel, 1996; Tuckey & Brewer, 2003; Turtle & Yuille, 1994). However, in many of the reviewed studies on repeated events, the delay to test was relatively short, ranging from a few days to one week. Future research should examine whether different time delays to test (e.g. days, weeks, months) influence adults’ ability to particularise instances of repeated events.

Arguably, many repeated incidents of abuse are unpleasant and will likely elicit a range of negative emotional responses, such as negative affect, heightened levels of acute hormonal stress and/or trauma responses following an abusive incident (e.g. post-traumatic stress disorder; Kemp et al., 1991). Only two studies in this review employed stressful stimuli that elicited heightened levels of negative mood and anxiety (i.e. ‘imagined experiences’ of domestic abuse and traumatic car accident videos; Dilevski et al., 2020; Theunissen et al., 2017). Thus, future research should continue to examine the impact of various emotional responses on memory for repeated events. Further to that, research ought to examine the influence of such responses during different memory stages, including encoding and retrieval. This research would guide our understanding of how negative emotional responses impact the initial processing of abusive incidents, and, further, how such responses impact recalling abuse when required to give evidence in an investigative interview and testifying to a court (e.g. evidence-in-chief).

It is well known that encountering misleading post-event information (i.e. suggestive questions, co-witness discussion) following an event can lead to memory distortion, such that individuals incorporate the suggestive information into their memory for the event (Loftus, 2005). This phenomenon is commonly referred to as the misinformation effect (Loftus, 2005). No study has yet investigated the extent to which adult memory for instances of repeated events can be distorted by misleading post-event information. However, data from Dilevski et al. (2020) indicate that experiencing repeated events might make one more susceptible to incorporating misleading information into memory than an individual who encounters a single event. In this study, using a recognition test, repeated-event participants, compared to single-event participants, struggled to discern what event details were or were not experienced (i.e. suggested false details) in the target instance. Given that repeated-event participants showed more memory confusion, it is possible that they would be more likely to accept misleading information and incorporate it into their memories. However, future research should examine this empirically.

Given that particularisation of repeated events is difficult, and complainants are often required to describe instances with a certain degree of precision, future research should further aim to develop investigative interviewing techniques to support recall. It is noteworthy that during our search of the literature, we identified several studies that have examined interviewing techniques for repeated experiences (e.g. Bearman et al., 2019; Leins et al., 2014; Rivard et al., 2014; Willén et al., 2015). We did not review these studies here as the aim of this review was to understand basic memory processes in adults’ repeated-event memory. However, we refer interested readers to Brubacher and Earhart (2019) who reviewed many of these techniques in their article. In general, these studies examined the effect of various memory mnemonics in generating events to discuss and/or techniques to aid particularisation.

Lastly, as a general recommendation for any future research endeavour, we encourage researchers to replicate findings, preregister their studies and, where possible, make their materials, study protocols and data available on a public depository. Engaging in these open science practices should increase the transparency, reproducibility and, thus, credibility, of research examining memory for repeated events.

Conclusion

In conclusion, domestic violence typically involves repeated incidents of coercive and controlling behaviour. When a complainant of repeated abuse decides to lay charges of abuse, their statement, and therefore memory, can be important evidence during legal proceedings. So far, the research on repeated-event memory suggests that while adults might struggle to recall details about specific instances (narrow accuracy), they are still quite capable of remembering what they have experienced across the instances (broad accuracy). These findings favour legal responses to domestic violence based on the coercive control model rather than the violent-incident model. We also found that several factors might impact recalling repeated-event instances, including age, the number of experienced instances, rehearsing an event, repeated retrieval and event distinctiveness. Despite these findings, our understanding of adults’ memory for repeated events is still in its infancy. We therefore encourage researchers to devote further attention to this topic.

Ethical standards

Declaration of conflicts of interest

Natali Dilevski has declared no conflicts of interest

Helen M. Paterson has declared no conflicts of interest

Sarah A. Walker has declared no conflicts of interest

Celine van Golde has declared no conflicts of interest

Ethical approval

This article does not contain any studies with human participants or animals performed by any of the authors.