Introduction

Living in isolation has widespread impacts on cognition and emotional well-being, which disproportionately affect older adults. Approximately 24% of community-dwelling older adults are isolated1 as a result of a range of factors such as living alone, physical impairments, declines in cognitive and mental health, and diminished social networks2. Alarmingly, isolation increases the risk of dementia, depression, and myriad health issues in older adults3,4,5. However, many of the risk factors that tend to emerge from isolation are modifiable, including low social contact, low cognitive engagement, depression, and physical inactivity6,7. Thus, lifestyle interventions that target the impoverished experiential diversity and reduced cognitive engagement characteristic of isolation may enhance cognitive and emotional well-being for older adults experiencing isolation. Indeed, diversity in daily activities has been linked to enhancements in cognitive functions such as working memory8. Here, we examined the impact of engagement in unique events on episodic richness, mood, and perception of time, in older adults during a period of isolation.

While isolated, days tend to be filled with monotonous housebound activities in place of a diverse range of activities that occur in a variety of contexts. The relatively mundane, routine activities of isolation tend to be highly familiar and repetitive, leaving only schematic memories that are difficult to differentiate as specific, unique episodes9,10,11. Without salient context changes from one event to another event boundaries become blurry, both hindering the ability to segment and remember distinct events, as well as distorting the perception of time12,13. Specifically, when we are bored or disengaged, time feels as though it drags on, but when we look back on such a period of disengagement, time appears to have passed quickly14,15. Such distortions to our sense of time are more likely to occur in older adults, with increased stress, boredom, and sadness, as well as with reduced task load and decreased satisfaction with one’s social interaction16,17. Collectively, this highlights the tight links between memory, time distortion, and well-being.

The close association between isolation and negative mood state is well documented3,18,19,20. Elevated symptoms of depression are characteristic of those experiencing both unfulfilled social needs3,20 and environments that lack sufficient stimulation21,22. The aversive subjective state of boredom that arises from the monotony and environmental constraint imposed by under-stimulating or impoverished environments is associated with reduced mindfulness and increased anxiety, depression, and substance use21,22. As such, an enriched diversity of experiences may help to combat the reduced well-being that results from isolation. Indeed, greater variability in physical location over the course of 3–4 months has been associated with increased positive affect23. Taken together, the evidence suggests that participation in unique experiences should enhance not only memory and perception of time, but also well-being.

The effects of isolation on cognition and emotional well-being became more widely evident during the COVID-19 global pandemic when stay at home and social distancing orders were implemented to reduce disease transmission. From the onset of the pandemic, studies reported an increase in depression and anxiety19, stress24, subjective memory impairments25, and distortions to sense of time16,17. The unpleasant feeling of the days ‘blurring’ together became so pervasive that the term ‘Blursday’ was highlighted by Oxford Languages as a 2020 Word of the Year26. Critically, older adults were disproportionately vulnerable to more severe effects of the COVID-19 lockdowns, as advancing cognitive decline is compounded by social isolation27,28. Within the first year of the pandemic alone, significant increases in depression and declines in cognitive function were observed in those with mild cognitive impairment or dementia28,29. Beyond throwing into sharp relief the negative effects of living in isolation, the COVID-19 lockdowns also provided a unique experimental opportunity to examine how experiential diversity impacts memory, well-being, and time perception.

Current study

In the current study, older adults used a digital memory application called “HippoCamera”30 to record and replay unique and routine experiences over an 8-week period during the summer 2020 lockdowns when stringent social distancing guidelines were in place for all participants. HippoCamera is a smartphone-based tool that has been shown to enhance episodic richness and positive valence of memories for daily life events30. HippoCamera has two main functions that enable participants to record and replay life events using principles known to promote memory. Through a guided multistep process, participants deeply engage with an ongoing event as they create a distinctive audiovisual cue. Later, these cues are replayed in curated study (“replay”) sessions designed to optimize learning and long-term retention. Our research has shown that older adults were able to recall over 50% more event-specific details for events that were replayed with HippoCamera compared to baseline events that were recorded but not replayed30.

We instructed older adults to engage in unique, non-routine events, which they recorded and reviewed in replay sessions using HippoCamera. Specifically, over a period of 8 consecutive weeks, participants recorded at least one daily event and engaged in at least one replay session per day. We interleaved one week of Baseline Control (the recorded event was not included in replay sessions) with one week of Replay (the recorded event was included in replay sessions) (Fig. 1C). This Baseline Control vs. Replay design has proven successful in our past research wherein we show that replaying memory cues boosts recall of event details30. As such, we expected to replicate the finding that episodic richness of memories would be enhanced for events recorded and replayed using HippoCamera, relative to baseline events that were recorded but never replayed. In addition, we expected that the benefit from replay would be larger for routine events because these events are typically not well remembered and have more to gain from repeated reactivation with high-fidelity memory cues.

To examine the impact of unique experiences on memory, mood, and time perception, we scored each individual event according to the frequency of its occurrence, the frequency its location was visited, and the typicality of the overall day in which the individual event was situated (see “Uniqueness composite score” section). We hypothesized that incorporating unique events into daily life would disrupt the monotony of lockdown, providing a distinctive experience that should not only enhance recent memory, but also improve well-being overall. Additionally, we predicted that days with more unique events would be associated with higher positive affect, reduced boredom, increased mindfulness, and the perception that time passed more quickly, as assessed on a series of daily measures. As predicted, we found independent effects of both an event’s uniqueness and whether or not it had been replayed on the number of episodic details subsequently recalled about that event. We also found that unique events were related to increased positive affect, decreased boredom, and the perception of time moving faster. Collectively, our results highlight the importance of promoting active engagement in unique events as a simple, yet very effective way for future interventions to improve episodic recollection and emotional well-being for older adults.

Results

Overall, participants collected 670 memory cues of events from their daily lives using HippoCamera (See Fig. 1 for overview of procedure). Each participant recorded an average of 39.4 memory cues (SD = 9.84) and watched each memory cue in the Replay condition an average of 11.2 times (SD = 2.05). At both the 4-week and 8-week time points, we conducted extensive memory testing (an average of 6.5 h per participant) using an adapted version of the Autobiographical Interview31 (see Supplementary Information, Memory Test Methods). We prompted retrieval by presenting participants with their HippoCamera cues over a videoconference call and asked that they verbally recall their memory for the corresponding event (Fig. 1D).

Fig. 1
figure 1

(A) Cues are recorded in four steps: Users (1) select a real-world event and initiate recording, (2) record a 24-second video, (3) record an 8-second spoken verbal description, and (4) rate the event’s significance. (B) Replay sessions consist of up to 5 cues played sequentially. During replay, the 8-second audio is played concurrently with a speeded (3x) version of the video. Each cue is preceded by a text display that indicates approximately how much time has passed since the event occurred as well as the event’s exact date and time. (C) The experiment alternated between weeks where cues were assigned to either the Replay or Baseline Control conditions (8 weeks in total). Each day, participants captured a cue and completed a replay session. Additionally, at the end of each day, participants completed a series of daily measures to assess their well-being and typicality of their day. (D) After 4 weeks and again after 8 weeks participants were tested on their memory for the events from the prior month. After recalling the event, participants rated the frequency of each event and its location. (E) Event and Location Frequency ratings were combined with the Typicality of Day measure to create the Uniqueness Composite Score corresponding to each event. We related this uniqueness score to both their measures of well-being provided on the day the event occurred and their subsequent memory for the event.

Full size image

Uniqueness composite score

To quantify the uniqueness of each event, we created a composite score using participants’ ratings of Location Frequency and Event Frequency (provided after recalling each event during the Autobiographical Interview and specific to each individual event), as well as their Typicality of Day daily measure (provided at the end of each day and specific to the events of the day overall) (Fig. 1E). Each of these elements captures a potentially different aspect of an event’s uniqueness. Location Frequency probed the uniqueness of the spatial location in which the event occurred, which is known to be integral to episodic memory32,33. Event Frequency provided a rating of how often the participant engaged in that specific type of event, with more unique events being those that occurred less frequently. Finally, Typicality of Day provided an assessment of the overall uniqueness of the day in which each event was situated. A set of Pearson correlations revealed significant but weak positive correlations between these three elements of the Uniqueness Composite Score (Table 1), suggesting that although the various components of the Uniqueness Composite Score are similar, they do not perfectly overlap and capture distinct aspects of each event.

Responses for each of these three elements were normalized within each participant, added together, and then again normalized within each participant to create the Uniqueness Composite Score. This Uniqueness Composite Score was used as our primary assessment of uniqueness in all following analyses as a metric for the uniqueness of each event for each participant. In addition, we also conducted these same analyses for each of the three elements of the Uniqueness Composite Score individually. The pattern of results did not meaningfully change compared to those obtained with the Uniqueness Composite Score (Supplementary Information, Table S2–S4).

Table 1 Pearson correlation matrix for factors included in the uniqueness composite score, including location frequency, event frequency, and typicality of day.
Full size table

Qualitatively, events that were rated as more ‘unique’ on the Uniqueness Composite Score captured moments where participants engaged in activities that diverged from their typical routines or found ways to transform their ordinary everyday life experiences into something special. Common themes of unique events included trying a new hobby, learning new skills, reconnecting with friends, and experimenting with small ways to add more variety into typical life (e.g., exploring a new section of one’s neighborhood, cooking a dish for the first time, planting a flower bed after renovating the garden).

Relationship between Uniqueness Composite Score and Replay Condition. A Point-Biserial Correlation revealed that Uniqueness Composite Score and Replay condition had an overall moderate correlation (r = 0.33, p < 0.001), indicating that although unique events were more likely to be replayed, events in both the Replay and Baseline Control conditions had varying levels of uniqueness (Fig. 2). That is, there were Baseline events that were unique and Replayed events that were typical. Our original instructions to participants emphasized engaging in unique events on replay weeks with the intention to compare unique replayed events to routine baseline events, however, the variability in naturally occurring unique events in the real-world resulted in a range of uniqueness scores for events in both the replayed and baseline conditions. As a result, we were able to disentangle the relative influence of uniqueness and replay, confirmed by analyses to check for multicollinearity. Specifically, we assessed whether there were issues with collinearity between Uniqueness and Replay that would prevent us from assessing the independent contributions of Uniqueness and Replay to our variables of interest. Using the performance package in R34,35 we found that all tolerance and Variance Inflation Factor (VIF) values in our models showed no evidence of multicollinearity between these two variables, with values within the recommended range (tolerance values above 0.2 and VIF values less than 536): Internal details Uniqueness Tolerance = 0.46, VIF = 2.19; Internal details Replay Tolerance = 0.74, VIF = 1.35; External details Uniqueness Tolerance = 0.92, VIF = 1.08; External details Replay Tolerance = 0.97, VIF = 1.03; Proportion of Internal details Uniqueness Tolerance = 0.86, VIF = 1.08; Proportion of Internal details Replay Tolerance = 0.88, VIF = 1.07. Thus, multicollinearity between an event’s Replay status and its uniqueness was not a concern for any of our models, thereby allowing us to disentangle the relative importance of each variable to memory for event details.

Fig. 2
figure 2

The distribution of uniqueness composite scores for events in the baseline control and replayed conditions. Higher density of events signifies a greater number of events with a given uniqueness composite score.

Full size image

Event recollection with the Autobiographical Interview

Memory recall was scored using the Autobiographical Interview scoring protocol31 to classify individual details as being either Internal or External (Fig. 1). Internal details are event-specific and reflect episodic reexperiencing (e.g., recollecting the conversation you had when hiking Doe Lake Trail with your son), whereas External details reflect retrieval of general semantic information (e.g., knowing that Doe Lake Trail is in Frontenac National Park), personal semantics (e.g., knowing that your son loves to hike), or details from a nontarget event (e.g., recollecting a different time you went hiking in a provincial park). Details evident in either the verbal description or video components of the memory cue (In-cue details) were excluded from the External and Internal detail counts. Additionally, to compare Internal and External detail recall directly, we also calculated the Proportion of Internal details (number of Internal details divided by total number of Internal and External details).

Detail counts were analyzed using multilevel modeling in R 4.1.037 with the lme4 package38, which was an appropriate analytic approach given that multiple data points (event memories) were analyzed for each individual participant. The performance package was used to obtain both conditional and marginal coefficients of determination (R2C and R2M, respectively) to determine model fits34,35. Significant interactions were probed using simple slopes tests with the emmeans package39. To test hypotheses related to autobiographical memory test detail counts, we used multilevel generalized Poisson models with individual events nested within participants (for fit statistics and fixed-effects parameters, see Supplementary Information, Table S5). Poisson models were used to account for the count-based nature of the number of details recollected40. Uniqueness (Uniqueness Composite Score), Condition (Replay vs. Baseline Control), and the Uniqueness x Condition interaction were included as fixed effects for all models (The same analyses were conducted individually for each of the three elements of the Uniqueness Composite Score (i.e., Location Frequency, Event Frequency, and Typicality of Day). The pattern of results did not meaningfully change compared to those obtained above with the Uniqueness Composite Score (Supplementary Information, Tables S2–S4).). Consistent with recommended best practices, the models were fit with maximal random effects structure41, with a random slope fit for each fixed effect and a random intercept fit for each participant. Uniqueness Composite Scores were standardized within each participant and Replay was effect-coded (Baseline Control = − 1, Replay = 1). Events assigned to the Replay condition were included in replay sessions whereas events assigned to the Baseline Control condition were not included in replay sessions. Individual models were specified for both Internal and External details to obtain separate estimates for each detail type. Additionally, to compare Internal and External detail recall directly, the above process was repeated using a Gaussian mixed model to assess the impact of Uniqueness and Condition on the Proportion of Internal details (for fit statistics and fixed-effects parameters, see Supplementary Information, Table S6). All models were estimated with an unstructured covariance matrix.

Internal details

A significant main effect of Uniqueness (b = 0.499, SE = 0.055, z = 9.127, P < 0.001) was found, indicating that participants recalled significantly more Internal details for events with higher Uniqueness ratings (Fig. 3A). In addition to a main effect of Uniqueness, we found a main effect of Condition (b = 0.272, SE = 0.059, z = 4.592, P < 0.001), such that memories for Replayed events were recalled with significantly more Internal details than were events in the Baseline Control condition (Fig. 3D). Additionally, we found a Uniqueness x Condition interaction (b = − 0.157, SE = 0.044, z = − 3.575, P < 0.001), indicating that the effect of Uniqueness was larger for events in the Baseline Control condition than in the Replay condition (Fig. 3G). A simple slopes analysis indicated an effect of Uniqueness in both the Baseline Control (b = 0.657, SE = 0.043, z = 7.99, P < 0.001) and Replay conditions (b = 0.342, SE = 0.090, z = 7.34, P < 0.001).

Fig. 3
figure 3

The effect of Uniqueness and Replay on the number of details recalled in the autobiographical memory test. There was a significant effect of Uniqueness on recall of (A) Internal details, (B) External details, and (C) Proportion of Internal details, indicating that while memory for events with higher Uniqueness ratings contained both more Internal and External details, they ultimately contained a larger Proportion of Internal details. There was a significant effect of Replay condition for (D) Internal details, but not for (E) External details, and a significant effect of Replay condition for (F) Proportion of Internal details. There was a significant interaction between Uniqueness and Replay condition for recall of (G) Internal details, but not recall of (H) External details or (I) Proportion of Internal details. For the panels above, individual participants are represented by grey lines, group average data is represented by bold lines, and standard error is represented by shaded bands.

Full size image

External details

A significant main effect of Uniqueness (b = 0.098, SE = 0.032, z = 3.022, P = 0.003) was found, indicating that participants recalled significantly more External details for events with higher Uniqueness ratings (Fig. 3B). There was no main effect of Condition (b = 0.037, SE = 0.031, z = 1.185, P = 0.236) (Fig. 3E), or Uniqueness x Condition interaction (b = − 0.039, SE = 0.023, z = − 1.722, P = 0.085) (Fig. 3H).

Proportion of internal details

A significant main effect of Uniqueness on the Proportion of Internal details recalled (b = 0.072, SE = 0.008, t(766) = 9.261, P < 0.001) was found (Fig. 3C). That is, memories for events with higher Uniqueness ratings contained a larger Proportion of Internal details. We also found a main effect of Condition (b = 0.041, SE = 0.008, t(766) = 5.357, P < 0.001), such that Replayed memories contained a larger Proportion of Internal details (Fig. 3F). There was no Uniqueness x Condition interaction (b = − 0.008, SE = 0.008, t(766) = − 0.977, P = 0.329) (Fig. 3I).

Impact of uniqueness on daily well-being measures

In addition to recording at least one event per day and watching at least one replay session per day, participants also answered daily short questions at the end of each day using Qualtrics42 about valence of mood, how mindful they felt, their level of boredom, subjective passage of time, and how typical the events from that day were (see Materials, Daily Measures of Well-being and Typicality of Day). These measures probed the specific day in question, allowing us to track well-being on a daily basis. Completion of the daily measures had a high compliance rate of 89%.

Raw scores on each of the daily well-being measures were analyzed using multilevel modeling in R 4.1.037 with the lme4 package38. The performance package was used to obtain both conditional and marginal coefficients of determination (R2C and R2M, respectively) to determine model fits34,35. For each of the daily well-being measures we ran a linear mixed-effects model with Uniqueness (Uniqueness Composite Score) included as fixed effect, with a corresponding random slope and a random intercept fit for each participant, to determine whether each of our measures differed on a daily basis depending on Uniqueness (for fit statistics and fixed-effects parameters, see Supplementary Information, Table S7). Uniqueness Composite Scores were standardized within each participant. Given that participants replayed every day, and the same event cue was shown on multiple days, we could not associate a given event’s Replay status with a specific day’s well-being measures. Additionally, given that multiple cues with varying levels of uniqueness ratings were included in each replay session, it was not possible to examine the effect of replaying unique events on daily mood because we could not isolate the effect of a single event’s uniqueness. Therefore, Replay condition was left out of these models. The average scores on each of the daily measures across the duration of the 8-weeks of study participation are available in the Supplementary Information file, alongside available normative data (for descriptive statistics of the daily measures, see Supplementary Information, Table S8).

The Positive and Negative Affect Schedule (PANAS)

Positive Affect. Positive Affect on the PANAS was higher on days with more unique events (as assessed by the Uniqueness Composite Score), (b = 1.021, SE = 0.254, t = 4.013, P < 0.001) (Fig. 4A).

Fig. 4
figure 4

(A–E) The effect of Uniqueness, as measured by the Event Uniqueness Composite Score, on each of the daily measures. A significant effect of Uniqueness was found on Positive Affect, Boredom, and Perception of Time. Individual participant data is represented by grey lines, group average data is represented by orange lines, and standard error is represented by shaded bands.

Full size image

Negative Affect. Negative Affect on the PANAS was not affected by event uniqueness, (b = 0.035, SE = 0.144, t = 0.246, P = 0.809) (Fig. 4B).

Multidimensional State Boredom Scale (MSBS-8)

Boredom ratings on the MSBS-8 were lower on days with more unique events, (b = − 0.783, SE = 0.350, t = − 2.238, P = 0.038) (Fig. 4C).

The state Mindful Attention Awareness Scale (MAAS-State)

Mindfulness ratings on the MAAS-State was not affected by event uniqueness, (b = 0.041, SE = 0.031, t = 1.327, P = 0.201) (Fig. 4D).

Perception of the passage of time

Responses to the question “How quickly do you feel the day today passed for you?” were higher (i.e., faster perception of time) on days with more unique events, (b = 0.091, SE = 0.036, t = 2.538, P = 0.021) (Fig. 4E).

Discussion

The limited experiential diversity that accompanies living in isolation has extensive negative impacts on both memory and mental health. Critically, older adults are a population that has always been much more likely to experience living in isolation1,2, putting these individuals at risk of advancing cognitive decline27,28,43. Here, we used a novel smartphone application with which older adults captured and replayed cues of unique and routine events during the lockdowns of the summer of 2020. Our goal was to characterize the links between the impoverished experiential diversity characteristic of isolation with memory, mood, and the perception of time. During the COVID-19 lockdowns, the near-universal baseline was a stream of monotonous activities that lacked distinctiveness and diversity, providing the novel opportunity to assess the impact of individual unique events on memory and well-being. We showed that unique events were recalled in richer episodic detail, and that increased uniqueness was associated with greater positive affect, decreased boredom, and the perception of time passing faster for the entire day. While it is typically assumed that experiential diversity is generally beneficial, no previous study has demonstrated the impact of individual unique real-world events on memory and well-being.

Our findings suggest an inextricable relationship between unique experiences and well-being. This is the first evidence to suggest that engaging in non-routine experiences enhances positive affect, reduces boredom, and alters perception of time. Previous work suggests that diversity in one’s physical location in everyday life enhances positive affect23. Here, we show that Location Frequency alongside the perceived uniqueness of an individual event directly impacts both memory and well-being. A potential mechanism for the observed effect of novelty on well-being is the disruption of dull, wearisome monotony. Boredom is typically described as an unpleasant emotional state driven by an inability to engage with the environment, with prolonged boredom linked to increases in anxiety and depression22. While we demonstrate here that engagement in unique events reduces boredom and increases positive affect on a daily basis, there may also be beneficial longer-term outcomes to well-being, especially in the context of isolation with a limited diversity of experiences day to day. It is possible that experiential diversity only provides enhancements to well-being in the context of isolation, however, evidence of a link between diversity in geographic location and positive affect prior to the pandemic23 suggests that our effects are likely generalizable beyond the pandemic lockdowns.

Experiences that elicit positive affect tend to be more accessible in memory44, as are relatively distinctive experiences9,45. Conversely, routine activities are highly familiar and repetitive, making it difficult to differentiate specific episodes from one another9,10. We suggest that memory for unique experiences in everyday life may additionally be enhanced through increased positive affect spurred on by engaging in a novel event, whereby both uniqueness and emotional valence contribute to greater episodic richness. As we demonstrate here, however, real-world events do not need to be exceedingly rare or exciting to produce these benefits; experiences that simply deviate from our normal routines (such as trying a new recipe or visiting a new park) are distinctive enough to benefit memory and well-being.

Moreover, the absence of context changes between events may make it more difficult to parse discrete events when one is largely housebound. According to Event Segmentation Theory, event boundaries (i.e., the beginning or end of an episode, often signaled by a shift in spatiotemporal context) are critical to shaping memory because they enable the segmentation of distinct events from the continuous stream of experience in our everyday lives12. Additionally, the presence of multiple defined events or ‘context changes’ results in the perception that more time has passed, and that time is passing faster in the moment14,15. As such, when staying home and not engaging in a variety of activities we have not only fewer events to segment, but also fewer salient event boundaries to aid in event segmentation. These factors will both negatively impact event memory and distort perception of the passage of time. Indeed, not only were unique events better remembered, but on days with more unique events time was perceived to pass faster, possibly due to better established event boundaries for these experiences.

It is important to highlight that the present work investigated both the effect of engaging in unique activities as well as replaying these events using HippoCamera. We have demonstrated in previous work that replaying events with HippoCamera benefited recall of Internal, but not External, details30. Here, we replicated this finding and additionally observed distinct effects of both uniqueness and replay, with each independently contributing to enhanced episodic richness of memories. We also observed an interaction between uniqueness and replay, which was driven by the fact that uniqueness had a greater impact on episodic recall of Baseline Control events relative to Replayed events, suggesting that the benefit from uniqueness was larger for memories that did not already benefit from replay. In addition, episodic recall of events that were less unique benefited more from replay, perhaps because memory for these more routine events was worse and had more gain from replay-based reactivation. Notably though, the interaction between uniqueness and replay was not significant when considering the Proportion of Internal details, indicating either that there was a similar pattern for External details which was not borne out by the External detail count analysis, or, more likely, that the Proportion analysis simply lacked the statistical power to reliably detect the interaction. Future research is required to disentangle these possibilities, however in the absence of a variety of unique experiences in daily life, HippoCamera may be particularly effective in enhancing episodic richness.

The present study utilized HippoCamera to investigate whether experiencing unique events was associated with better memory and self-reported well-being and was not designed to address directly whether HippoCamera in and of itself promotes increased engagement in unique events. However, feedback from participants after the study suggests that HippoCamera might indeed lead to increased experiential diversity: “It inspired me to have something to do that was useful” (Participant 117); “It made every day more interesting” (Participant 122); “I started finding things, so my life didn’t seem so dull and boring, like the least little thing. My whole day felt better.” (Participant 223). Future work will directly investigate the influence of recording and reminiscing of events on older adults’ tendency to engage in more unique experiences. Additionally, it may be that the mere act of capturing an event, regardless of whether that event is replayed, causes that event to be perceived as more unique in memory, either due to intentional direction of attention to the present moment during encoding or enhanced significance of the event one chooses to record. Indeed, we observed a relationship between the significance rating gathered at event capture (Fig. 1A and 4th screen) and the Uniqueness Composite Score (b = 0.29, SE = 0.06, t = 4.62, P < 0.001). The specific circumstances that result either in the perception of uniqueness or in the promotion of engagement in unique events should be examined in future research to develop interventions that target experiential diversity.

Collectively, the present work demonstrated that unique events were remembered in greater episodic detail and that engaging in these events boosted well-being for those living in isolation. We suggest that novel experiences have a powerful impact on well-being by enhancing positive affect and disrupting monotonous routine. Although we studied this at a singular moment in time, the experience of isolation is something that affects millions of older adults and those experiencing health and mobility issues. Ultimately, a life enriched with a variety of experiences gives rise to enhanced episodic richness and positive well-being in our day-to-day lives.

Methods

Participants

Eighteen older adults ranging in age between 60 and 81 years old (M = 70.78, SD = 4.54, 10 women) were recruited through the University of Toronto Adult Volunteer Pool or via direct contact with the researchers. We aimed to recruit as many participants as feasible given constraints of the lockdown period, obtaining a sample size which exceeded that in previously published work using the HippoCamera wherein 12 participants were assessed in a long-term intervention30. Scores on the Montreal Cognitive Assessment (MoCA) in our sample ranged between 24 and 29 (M = 27.29, SD = 1.72), with three participants scoring below the 26-point cut-off (e.g., 24, 24, 25) indicating potential mild impairments to cognitive function46. The pattern of results for all analyses remained unchanged when the three participants scoring below 26 on the MoCA were excluded. Additionally, to further characterize our participant sample, at intake all participants completed a set of questionnaires on Qualtrics to assess well-being on a variety of measures such as depression, stress, and loneliness (for a full list of measures see Supplementary Information, Participant Characterization Measures). Our participants scored comparably to normative data on all scales aside from the Perceived Stress Scale (PSS)47, which indicated heighted levels of perceived stress experienced over the previous month (see Supplementary Information, Table S1). As such, our sample was normal on measures of mental health and well-being, with elevated stress levels likely due to the onset of the pandemic.

All participants were initially contacted via telephone and given the study details and participation requirements prior to providing consent. Requirements for participation included being able to engage in daily app use, complete a set of daily measures for 8 consecutive weeks, and have a smartphone on which they could download the HippoCamera app or be able to use a borrowed smartphone from the lab. Compensation was $5 per day over the 8-week period plus an additional $18 per hour spent meeting with a researcher over phone or videoconference call for training on the HippoCamera app, memory testing, and online completion of questionnaires. The research protocol was reviewed and approved by the University of Toronto Research Ethics Board (Protocol 39014) and all experiments were performed in accordance with relevant guidelines and regulations. Informed consent was obtained electronically from all participants prior to study onset.

Materials

HippoCamera application

As described in Martin et al.30, the HippoCamera app guides users to create rich memory cues with a multistep process that takes less than a minute to complete (Fig. 1A, B). First, participants intentionally make the decision to record an ongoing event, which encourages increased attention to daily events. Second, the participant provides an 8-second audio recording in which they provide a spoken verbal description of the target event. Third, they record a 24-second high-definition video, capturing visual elements of the event. Fourth, they rate the personal significance of the event using a 1–5 scale. Meta-data are also automatically recorded, including time, date, and GPS co-ordinates, when available. Upon completing a recording, the application generates memory cues that combine the verbal description with a speeded version of the video. Specifically, audio is stripped from the 24-second video and the resulting file is accelerated by a factor of three, resulting in an 8-second speeded video. The speeded video is then coupled with the audio file containing the 8-second verbal description. When replayed, the participant hears the real-time verbal description while viewing the speeded video. Replay sessions consist of up to five sequentially presented cues that are automatically selected. The application is designed to select cues for replay in a manner that balances distributed learning and the prioritization of recent events with high significance ratings. Within a replay session, cues are separated by a text screen that includes date and time information about the upcoming cue.

The software used in the current manuscript was programmed by Tactica Interactive (Winnipeg, Manitoba). An updated public-facing version of the application can be obtained from Apple’s App Store and Google Play. A research-dedicated version of the application with access to participants’ cues and more control over the participant’s experience can be obtained from a corresponding author (MDB).

Daily measures of well-being and typicality of day

The Positive and Negative Affect Schedule (PANAS)

The PANAS assessed daily mood48. This measure comprises 20 adjectives, 10 of which reflect a positive affective state (interested, excited, strong, enthusiastic, proud, alert, inspired, determined, attentive, active) and 10 of which reflect a negative affective state (distressed, upset, guilty, scared, hostile, irritable, ashamed, nervous, jittery, afraid). Participants were asked to ‘Indicate to what extent you have felt this way today’ with responses made using a 5-point Likert scale (1 = very slightly or not at all; 5 = extremely). Responses to the set of 10 positive items and the 10 negative items were averaged separately, with higher scores on each set indicating greater positive affective state and negative affective state, respectively.

Multidimensional State Boredom Scale (MSBS-8)

The short form of the MSBS-8 assessed state boredom each day49. The MSBS-8 comprises 8 self-report statements, with responses made using a 7-point Likert scale (1 = strongly disagree; 7 = strongly agree). Participants were asked to consider their experience from that day. Sample items include: “I feel like I’m sitting around waiting for something to happen”; “I seem to be forced to do things that have no value to me”. Responses to the 8 items were summed, with higher scores reflecting a state of greater boredom.

The state Mindful Attention Awareness Scale (MAAS-State)

The MAAS-State assessed state mindfulness each day50. This test is a shortened form of the MAAS which measures state rather than trait mindfulness, it comprises 5 self-report statements, with responses made using a 7-point Likert scale (0 = not at all; 7 = very much). Participants were asked to consider their experience from that day. Sample items include: “I was finding it difficult to stay focused on what was happening”; “I was doing something automatically, without being aware of what I was doing”. Following reverse scoring, responses to the 5 items were averaged, with higher scores reflecting a more mindful state.

Perception of the passage of time

Each day participants answered the question “How quickly do you feel the day today passed for you?”. Responses were made using a 5-point scale (− 2 = very slow; 2 = very fast).

Typicality of day

Each day participants answered the question ‘How typical were the activities you engaged in today?’. Responses were made using a 4-point Likert scale (1 = very unique; 4 = very typical).

Procedure

HippoCamera Use

Over a period of 8 consecutive weeks, participants recorded and replayed events each day using HippoCamera, with interleaved weeks of Baseline Control (event not included in replay sessions) and Replay (event included in replay sessions). Participants were not explicitly told that events from the Baseline Control weeks would not be replayed, only that some events they recorded would never be replayed. If a participant missed a replay session, the app would send a prompt to replay. During the Baseline Control weeks, participants were instructed to record a routine event of their choosing that was reflective of their daily lockdown life. Participants still watched daily replay sessions during Baseline Control weeks, however, to serve as a baseline for typical pandemic life, the Baseline Control events did not appear in the replay sessions. For the Replay weeks, instruction was provided for participants to engage in at least one unique, non-routine activity per day to be recorded using HippoCamera (example suggestions included trying out a new recipe, playing a new game, or going for a walk in an area not frequently visited). Our goal was for the Replay week to emphasize both engagement in unique activities and reactivation of memory for these unique events through daily replay sessions. However, we found that this manipulation did not work as cleanly as intended: participants recorded very unique events during the Baseline Control week and very typical events during the Replay week (Fig. 2). Thus, our analyses to investigate uniqueness focused on the Uniqueness Composite Score, rather than condition assignment.

Prior to beginning HippoCamera use, participants met with a researcher over Zoom for training on how to use HippoCamera in the context of the study. This included instruction on how to record and replay memories using the app and guidance on the types of events they should record. Additionally, participants provided informed consent and completed a set of questionnaires and scales on Qualtrics (see Supplementary Information, Participant Characterization Measures). At the end of each week, participants received a reminder phone call from a researcher that they should switch to recording unique or baseline events for the coming week, depending on the type of events they engaged in during the previous week.

Autobiographical Interview memory test

At both the 4-week and 8-week time points, we conducted extensive memory testing on both baseline and replayed events (Fig. 1C). Memory for all events (N = 670) was assessed using an adapted version of the Autobiographical Interview31 (see Supplementary Information, Memory Test Methods). Instead of using standard Autobiographical Interview instructions that allow participants to choose which memories to recall, we prompted retrieval by presenting participants with their audio-visual HippoCamera cues, without the accompanying timestamp (Fig. 1D). They were encouraged to provide as many details as possible and were provided with instructions and an example response at the beginning of each test session (see Supplementary Information, Modified Autobiographical Memory Test, Administration). The memory tests were conducted over Zoom video calls, in which the interviewer shared their screen for memory cue presentations. The cues were presented in a pseudo-randomized order, with Replay and Baseline Control cues interleaved and no two cues from events occurring on adjacent days presented sequentially at test. Verbal responses to each cue were recorded, transcribed, and then scored using the Autobiographical Interview scoring protocol31. Specifically, individual details were identified and scored as being either Internal, External, or In-Cue. Internal details are episodic in nature and can include information such as happenings during the event, perceptual descriptions, thoughts or emotions experienced during the event, and details about when or where the event occurred. External details are largely semantic in nature and can include general semantic information, personal semantics, autobiographical facts, self-knowledge, or details about other events they have experienced throughout their lives. In-cue details referred to descriptions of information that were readily apparent in the cue (either the verbal description or video) and were excluded from External and Internal detail counts. In other words, we conservatively discounted details that a naïve observer could also describe based on having seen only the cue, but not having experienced the event. A total of three independent raters completed AI training provided by Brian Levine prior to data coding and achieved a high degree of reliability on all detail types (ICC = 0.92).

Following recall for each event, participants answered a series of questions including those about memory confidence, Event Frequency, Location Frequency, length of the event, and significance of the event. The Event Frequency and Location Frequency measures were used alongside Typicality of Day ratings to calculate the Uniqueness Composite Score, described below. Location Frequency was assessed following recall of each event by asking “How many times have you visited this location?”, with response options including ‘1–2 times’, ‘3–5 times’, ‘6–10 times’, ‘>10 times’, ‘>50 times’, or ‘Home’.​ Event Frequency was assessed by asking “How frequent is this type of event?” with response options including ‘Daily’, ‘Weekly’, ‘Monthly’, ‘Annually’, or ‘Unique’.