Characterization of Music and Photograph Evoked Autobiographical Memories in People with Alzheimer’s Disease

Abstract

Background:

Music evoked autobiographical memories (MEAMs) have been documented in people with Alzheimer’s disease (AD), but it is unclear whether music is more effective than other familiar stimuli at evoking memories.

Objective:

To explore the frequency and specificity of memories in response to famous songs compared with photographs of famous events (photograph evoked autobiographical memories, PEAMs), and whether stimuli from the period of the reminiscence bump (10–30 years of age) were more likely to elicit memories.

Methods:

10 participants with AD and 10 aged-matched healthy elderly people reported memories following exposure to 2 songs (longest time at number one in Australian music charts) and 2 photographs (of prominent famous events) from each decade from 1930 to 2010.

Results:

PEAMs were more frequent than MEAMs in healthy elderly (p < 0.05), but no such differences were observed among people with AD. There was no difference in the frequency of MEAMs between groups, but people with AD showed a significant decline in the frequency of PEAMs. In both groups, MEAMs were typically less specific than PEAMs and comprised semantic knowledge or repeated/extended events. Stimuli from when participants were aged 10–30 years triggered more frequent memories compared with stimuli from later decades, but this was only statistically significant for MEAMs.

Conclusion:

Our findings indicate a preserved mnemonic effect of music relative to pictures in this patient population, corroborating suggestions that MEAMs represent an island of preservation during the progression of AD.

Keywords

Alzheimer’s disease autobiographical memory music reminiscence bump

INTRODUCTION

The most common cause of dementia is Alzheimer’s disease (AD), with the hallmark symptom of impaired memory function. AD related memory impairment includes a decline in the ability to retrieve, and the specificity of ‘autobiographical’ or personal memories. Autobiographical memory comprises personal information that is both semantic (general knowledge about oneself) and episodic (relating to specific personal events), and it is considered to be crucial for our sense of self and identity [1]. The complex relationship between autobiographical memory and the self in the context of AD has been addressed by several authors, and it has been argued that autobiographical memory decline in AD corresponds with a diminished sense of self [2], or a failure to update self-knowledge resulting in a static, outdated self concept [3] or a ‘petrified’ self [4]. In light of this, it is crucial to explore methods of facilitating autobiographical memory function which may then enhance the sense of self in this patient population.

The majority of studies of autobiographical memory in people with AD have used verbal tasks, but recent literature has explored how other stimuli, such as music, can cue personal memories. Early research on the effects of music on autobiographical memory function in AD explored how background music impacted on performance of verbal autobiographical memory tasks such as the Autobiographical Memory Interview [5]. Some studies documented enhanced performance in a music condition compared with a silence condition [6 –8] whereas other research has failed to show any difference in autobiographical memories in the presence of music and cafeteria noise [9]. Taken together, these results could suggest that auditory stimulation, rather than music specifically, was responsible for enhancing memory.

More recent research on people with AD has revealed that spontaneous episodic memories are commonly elicited after or during music listening, a phenomenon referred to as ‘music evoked autobiographical memories’ or MEAMs [10]. The music stimuli used in studies of people with AD has typically been self-chosen personally preferred music [6 , 11] or researcher selected music, with or without lyrics [11, 12]. These studies have typically compared memories evoked after listening to the music excerpts or a period of silence and have demonstrated a memory enhancing effect for music. For individuals with memory decline associated with AD, this enhancement effect may be considered a form of preserved memory function. For example, Cuddy and colleagues [12, 13] reported that the frequency of MEAMs (in response to 12 famous instrumental tunes) in people with mild to moderate AD was no different from that of younger or older healthy adults, suggesting a striking maintenance of function in the presence of music. Further, various characteristics of MEAMs of people with AD were consistent with characteristics of MEAMs among healthy elderly individuals, including the length (word count) of the memories and vividness ratings [12]. Participants also dated their memories in the latter study, and these dates were consistent with the well-known “reminiscence bump”, in which memories from the period of life from 10–30 years of age are especially salient [14]. A ‘positivity effect’ of the topic and valence of MEAMs has also been described in people with AD [12].

In a series of studies in people with mild AD, El Haj and colleagues [6 , 11] used self or researcher chosen music. They found that memories evoked while listening to self-chosen music (compared with silence or researcher chosen music) were more specific, recalled more quickly and had higher emotional content. Verbal descriptions of MEAMs also had higher grammatical complexity [7] and contained more ‘self-defining’ references (defined by the authors as how the person saw themselves or events related to personality construction, concerns or unresolved conflicts) compared with memories evoked after a period of silence [6, 7].

Although research has established that music can enhance autobiographical memory, it is less clear whether music is more efficient than other types of stimuli at eliciting such memories. Insights can be gained from studies that compare MEAMs with memories evoked by other types of stimuli in healthy populations. Belfi, Karlan, and Tranel [15] compared memories evoked by music and famous faces. They found that faces evoked more memories than songs, but MEAMs were more vivid than face evoked memories. They suggested that these results may reflect the more rewarding and emotional nature of music relative to faces. They also found a gender difference: Women reported more frequent and vivid memories than men, regardless of the condition. Zator and Katz [16] compared memories evoked by music and visually presented word cues (‘lifetime period’ such as ‘10 years of age’ or a specific event such as ‘Hurricane Katrina’) from when participants were aged 5, 10, and 16 years old. Their focus was on the linguistic differences between the memories evoked by these different cues. They found that MEAMs were evoked more quickly, had fewer total words, and contained more words related to motion and spatial terms (relating to ‘embodiment’) than the word evoked memories. There was no difference in the frequency of memories evoked by music or words.

Recent work has compared MEAMs with memories evoked by other familiar stimuli including words [17] or odors [18] in people with AD. Chevreau [17] found that both people with AD and healthy elderly had more frequent memories cued by single words than by music (familiar French songs). They found no difference between AD and healthy groups in the ‘quality’ of memories evoked by music (ratings of specificity), but higher quality of memories cued by words in the healthy compared with AD group. El Haj et al. [18] compared memories evoked after exposure to music (Spring movement from Vivaldi’s Four Seasons, and La Boheme performed by Charles Aznavour), odors (coffee and vanilla), or silence in 28 people with mild AD. Both odors and music were more effective than silence at eliciting autobiographical memories, and retrieval time was significantly shorter when triggered by odors compared with music, possibly because recognition of music requires encoding of sequential events. They found similar beneficial effects of music and odors for memory characteristics including specificity, emotion, and mental time travel. These studies demonstrate that music may not be any more efficient than other familiar stimuli at evoking memories in people with AD.

The aim of this study was to characterize the frequency and specificity of memories evoked by music (famous songs) and photographs (of famous events) in people with AD. We also explored the relationship between the frequency of music and photograph evoked memories and demographic variables such as age, gender, and level of cognitive function. In view of previous findings, we predicted no difference in the frequency of MEAMs between AD and healthy elderly groups, but a decline in specificity of evoked memories in the AD group. We also predicted that music and photograph stimuli from the time period of the reminiscence bump (when participants were aged 10–30 years) would trigger autobiographical memories more frequently than stimuli from later decades.

METHODS

Materials

Songs and photograph stimuli

Sixteen famous songs and sixteen photographs depicting world famous events across eight decades, from 1930–2010, were chosen (see Supplementary Material 1). Two songs and two photographs were selected from each decade. Song selection was based on those that had spent the longest duration at Number One in the Australian music charts and had the highest music sales according to two sources: The Kent Music Report (for songs from 1930–1989) and Australian Recording Industry Association (for songs from 1990–2010). Songs were purchased via iTunes and played on a laptop through a Bluetooth speaker. Photographs of prominent world events were based on famous events for each decade listed in Wikipedia and verified using online news resources. We aimed to choose specific ‘one off ‘events (e.g., “victory over Japan day”) rather than events that spanned a long period of time (e.g., WWII). Each photograph was individually printed on A5 paper and laminated.

MEAM and PEAM task

This task was based on the MEAMs questionnaire developed by Baird and Samson [19] (see Supplementary Material 2), and was administered orally.

Standardized cognitive tasks

The Mini-Addenbrooke’s Cognitive Examination (M-ACE) [20] was used to assess cognitive functioning. It comprises 5 items which assess attention (/4), memory (/7), animal fluency (/7), clock drawing (/5), and memory recall (/7), with a maximum score of 30. There are two recommended cut-offs; 25/30 (a score which is 5 times more likely to have come from a person with dementia than without) and 21/30 (almost certainly a diagnosis of dementia). The M-ACE has been deemed more sensitive than the Mini Mental State Examination (MMSE) and less likely to have ceiling effects [20]. Verbal fluency in the letter form was also assessed using the letters FAS (one minute per letter).

Music experience questionnaire (MusEQ) [21]

The MusEQ is a 35-item questionnaire that was specifically designed for people with dementia to assess music listening and engagement across six domains: 1) ‘Daily’, which addresses the role of music in routine daily life, 2) ‘Emotion’, which concerns the emotion and mood regulatory aspects of musical experience, 3) ‘Perform’, which involves music in social contexts such as musical performance and identity, 4) ‘Consume’, or consumer musical choices, 5) ‘Response’, or responses made in synchrony to music, and 6) ‘Prefer’, or the extent a person shows preferences or dislikes to certain music styles. The self- rated version was used for the healthy elderly and the informant version (close family member, spouse/partner, or a residential care facility staff member who had known the participant for at least 12 months) was used for the participants with AD.

Participants

Ten individuals with a clinical diagnosis of dementia (as determined by a geriatrician and clinical neuropsychologist) participated in the study. Nine participants (4 males, 5 females) were diagnosed with probable AD and one female was diagnosed with vascular dementia/AD. Participants were recruited from the aged care service at Rankin Park Unit, Hunter New England Health (n = 8) in Newcastle and Macquarie Park Baptist Care aged care residential facility in Sydney (n = 2). Ten age matched healthy elderly (6 males, 4 females) were recruited via the Macquarie University Australian Research Council Centre of Excellence in Cognition and its Disorders research participant registry (n = 9) and one family member of the AD participants (n = 1).

Inclusion criteria were native English speaking and residing in Australia since aged 10 years (to ensure familiarity with stimuli). During the assessment session, it was discovered that one AD participant had moved to Australia from Scotland at the age of 29 years, and one control participant had moved to Australia from England at the age of 30 years. A comparison of their results with the others did not reveal any differences, therefore their results were included. Exclusion criteria were a comorbid neurological condition (e.g., Parkinson’s disease) or severe psychiatric disorder (e.g., schizophrenia), and any visual, hearing, or language impairments that would prevent communication or ability to hear music and engage with the stimuli. Table 1 displays the demographic characteristics of the participants. Two participants with AD played a musical instrument while the remaining had no music training. No healthy elderly participants had any formal music training.

Table 1

Demographic information of the AD and healthy elderly groups

	AD	Healthy elderly
Number of participants	10	10
Gender (M/F)	4/6	6/4
Age (yr)	77.7 (12.7)	76.0 (9.3)
M-ACE^*	18 (6.0)	29 (0.8)
FAS^*	22 (12.3)	42 (7.3)
MusEQ	2.87 (0.8)	3.03 (0.5)

^*p < 0.005. M-ACE, The Mini-Addenbrooke’s Cognitive Examination; FAS, Verbal fluency task (Letters F, A and S); MusEQ, Music Experience Questionnaire.

Ethical approval

The study was approved by the Hunter New England Health and Macquarie University ethics committees.

Procedure

Participants were seen individually and completed the standardized cognitive tasks first, followed by the MEAM/PEAM task, with the order of the block of songs or photographs counterbalanced across all participants. The 16 songs and 16 photographs were individually presented in random order. This was done by pressing “shuffle” on iTunes for the songs, and manually shuffling the photographs. Each song was played from the beginning as would typically occur in a real-life music listening scenario. The participant was asked the MEAMs task questions (see Supplementary Material 2) after approximately 10–15 seconds, while the song excerpt continued to play for approximately 30 seconds, or until the chorus was heard (typically 30–60 seconds). Some participants reported a memory within seconds of the song commencing, before they were formally asked the MEAM questions. If this occurred then the order of the questions was rearranged to start with those related to the memory. For the PEAM task, each photograph was handed to the participant to look at for as long as required to respond to the PEAM task questions (see Supplementary Material 2). Participants were asked at the initial presentation of the first song and photograph as to whether or not they were able to hear/see the stimulus. These questions were asked in the same manner by the researcher as those in the MEAM task. Participants were filmed using a small digital camera on a tripod and all participants were asked for their consent before filming proceeded.

Familiarity ratings

Participants were asked to rate their familiarity verbally with each stimulus on a 3-point scale (1 = not familiar, 2 = somewhat familiar, and 3 = extremely familiar (“I know it almost perfectly”).

Specificity ratings

Audio transcripts of the MEAM/PEAM task were coded by three raters to determine the specificity of the memories reported. This was based on coding used in previous studies by El Haj and colleagues [6 , 11] adapted from the TEMPau test [22], and a recent method of coding specificity of semantic memories as ‘general’ or ‘personal’ [23]. Scores ranged from 0 –7 with a higher score indicting greater specificity of memory; 0 = no memory, 1 = Semantic General (knowledge with no reference to a personal event or episode), 2 = Semantic Personal (knowledge with reference to a personal event or episode), 3 = Lifetime Period (with no reference to time or place), 4 = Repeated Event (with no reference to time or place), 5 = Repeated/Extended Event with reference to time or place, 6 = Specific Event in time and place, and 7 = Specific Event with internal details. Two raters scored half of the transcripts (10 per judge), and a third independent rater scored half of each judges’ transcripts (five from each rater or 10 in total). There was a high interrater agreement (95.3%). Any discrepancies were discussed by all three raters and the majority decision was taken.

Topic of memories

The topic content of MEAMs and PEAMs were coded according to the following categories: Person/people, period of life (e.g., high-school days), place, or specific event. Memories were coded for their primary topic content. The proportion of memories for each topic was calculated (expressed as a percentage) by dividing the number of memories per topic by the total number of memories.

RESULTS

The frequency of MEAMs and PEAMs was calculated for each participant as the proportion of stimuli that triggered a reported autobiographical memory, expressed as a percentage (out of a total of 16). For example, if 6/16 songs elicited a reported memory, then that would result in a score of 37.5% for MEAMs.

Statistical analyses were performed using an Analysis of Variance (ANOVA) to examine group differences (AD versus healthy elderly) between MEAMs and PEAMs, including frequency and specificity. Group (AD patients versus healthy elderly) was the between-participants factor and stimuli (songs versus photographs) was the within-participants factor. Follow up t-tests when run, were corrected for multiple comparisons using False Discovery Rate.

Pearson correlations were performed to examine the relationship between frequency of MEAMs and PEAMs and demographic variables including cognitive function and age, and chi square analysis was used to examine any gender differences. A p value of less than 0.05 after correction, was used as the statistical cut off for significant results. Descriptive statistics for MEAMs and PEAMs frequency, topic and specificity are shown in Table 2.

Table 2

Means (and standard deviations) of the percentage of stimuli evoking memories

	AD (n = 10)		Healthy elderly (n = 10)
	Songs	Photographs	Songs	Photographs
Proportion of stimuli with memories:	33.75 (13.56)	25.62 (18.74)	36.25 (14.96)	53.12 (16.73)
Min-max	12.50 –56.25	0 –62.50	18.75 –62.50	18.75 –81.25
Memory Content:
Person	46.94 (18.45)	28.2 (25.17)	25.22 (15.33)	43.83 (31.85)
Period of life (POL)	51.30 (18.24)	20.00 (32.91)	51.79 (27.84)	17.21 (25.99)
Event	2.86 (6.02)	33.50 (32.06)	20.76 (30.72)	60.12 (39.62)
Place	2.85 (9.03)	15.75 (26.25)	5.00 (11.25)	3.00 (4.83)
Memories evoked from stimuli during specific life-time periods (age):
10–30 yr	58.50 (26.36)	39.60 (28.33)	53.50 (23.33)	62.30 (19.73)
31–50 yr	20.00 (21.47)	19.00 (30.62)	31.50 (28.48)	62.50 (35.84)
51+ yr	4.40 (7.15)	22.78 (20.78)	15.11 (19.43)	48.22 (26.28)

Frequency of MEAMs and PEAMs

A 2×2 mixed ANOVA was conducted to explore group differences (AD versus healthy elderly) in the frequency of MEAMs and PEAMs. A significant main effect was found for Group, with healthy adults having more memories evoked overall than the AD Group (45% versus 30% ; F(1,36) = 8.705, p = 0.006, $η_{p}^{2}$ = 0.195). A significant interaction effect was also evident between Group and Stimuli (F(1,36) = 5.456, p = 0.025, $η_{p}^{2}$ = 0.132). Follow up t-tests confirmed this interaction was due to higher frequency of PEAMs in the healthy compared with the AD group, (53.75% versus 26.87% respectively; t(18) = –3.37, p = 0.014 FDR corrected, 95% CI [–43.6%, –10.1% ], d = 1.51), but no difference in frequency of MEAMs between these groups (36.25% versus 33.13% respectively; t(18) = –0.495, p = 0.626, d = 0.22), see Table 2 and Fig. 1). The healthy elderly group also reported significantly more PEAMs than MEAMs (53.75% versus 36.25% respectively; t(9) = 2.905, p = 0.035 FDR corrected, 95% CI [3.8%, 31.1%], d = 1.14). In contrast, there was no significant difference between the frequency of MEAMs and PEAMs in the AD group (see Table 2, and Fig. 1). Post-hoc power analysis confirmed that the power to detect these effects was above 0.88 on these tests.

Fig.1

Percentage of memories evoked by songs and photographs (mean±standard error of the mean) in AD (participants with Alzheimer’s disease) and healthy elderly participants. ^*p < 0.05.

Familiarity ratings of songs and photographs

A 2×2 ANOVA was conducted to explore group differences (AD versus healthy elderly) in the familiarity ratings of the stimuli (songs versus photographs). One participant from each group did not provide familiarity ratings for both stimuli, and so were removed from this familiarity analysis. A significant main effect was found for stimuli (F(1,32) = 25.533, p < 0.001, $η_{p}^{2}$ = 0.444), with photographs being rated as more familiar overall than songs (2.48 versus 2.09 respectively). A significant main effect was also found for group (F(1,32) = 11.90, p = 0.002, $η_{p}^{2}$ = 0.271), with healthy adults rating the stimuli as more familiar than AD (2.42 versus 2.15 respectively). Follow up t-tests revealed no differences in the familiarity ratings of songs between AD and the healthy group (p = 0.054) but did show healthy elderly rated the photograph stimuli as more familiar compared to the AD participants (mean familiarity ratings of 2.65 versus 2.32 respectively; t(16) = –2.76, p = 0.018, FDR corrected, 95% CI [–0.58, –0.08], d = 1.31). Consistent with the main effect finding, photographs were rated as more familiar than songs within both the AD Group (2.32 versus 1.98) (t(8) = 3.63, p = 0.0132 FDR corrected, 95% CI [–0.57, –0.13], d = 1.25) and the healthy adults (2.65 versus 2.19; (t(8) = 4.19, p = 0.012 FDR corrected, 95% CI [–0.710, –0.206], d = 2.36).

To determine whether music-evoked memories were more likely for music that was given high ratings of familiarity, a 2×2×2 ANOVA of Group (AD, healthy adults)×Stimuli (song, photograph) × Memory evoked (yes, no) was conducted. In addition to a main effect of Stimulus and Group as seen previously, there was also a main effect of Memory evoked, with stimuli that evoked memories being rated as more familiar than stimuli that did not evoke a memory (2.64 versus 1.93; F(1,64) = 114.41, p < 0.001, $η_{p}^{2}$ = 0.641). In addition, a significant two way interaction effect between stimuli type and whether a memory was evoked was also found (F(1,64) = 12.03, p < 0.001, $η_{p}^{2}$ = 0.158). This suggests in both AD and healthy elderly groups, songs and photographs that triggered memories were rated as more familiar than those that did not trigger memories. Post-hoc power analysis confirmed that the power to detect these effects was above 0.79 on these tests.

Relationship between frequency of MEAMs and PEAMs and demographic variables

Correlations were calculated for all participants to compare frequency of MEAMs and PEAMs with demographic variables of cognitive function (total score on M-ACE), verbal fluency (FAS) performance and age. There was a significant positive relationship between frequency of PEAMs and cognitive function as determined by M-ACE score (r = 0.65, p = 0.012 FDR corrected) as well as between frequency of PEAMs and verbal fluency (r = 0.64, p = 0.018 FDR corrected). In contrast, there was no relationship between frequency of MEAMs and cognitive function (either M-ACE or verbal fluency scores) and no significant relationship between frequency of MEAMs or PEAMs and age.

Topic of memories

To compare the prevalence of different types of memories, a 4×2×4 ANOVA of Topic Content (person, period of life, place, event) × Group (AD, healthy elderly) × Stimuli (song, photograph) was conducted. There was a significant main effect for Topic (F(3,144) = 12.604, p < 0.001, $η_{p}^{2}$ = 0.208), as well as a significant interaction between Stimuli (song and photograph) and Topic Content (F (3,144) = 12.936, p < 0.001, $η_{p}^{2}$ = 0.212; see Figs. 2 and 3). Follow up t-tests comparing the stimuli for each of the memory topics across groups revealed significant differences for Period of Life (t(19) = 4.244, p = 0.005 FDR corrected, 95% CI [16.7%, 49.2%], d = 1.263) and Event (t(19) = –3.181, p = 0.007 FDR corrected, 95% CI [–54.2%, –15.8% ], d = 1.116). In both AD and healthy groups, MEAMs were more frequently about a period of life than PEAMs (51.54% versus 18.60%, respectively), whereas PEAMs were more often about a specific event compared with MEAMs (46.81% versus 11.81% respectively, see Table 2). To determine whether this pattern was similar in the two groups, these t-tests were also run at the individual group level. The results are displayed in Figs. 2 and 3. For the AD group, after FDR correction, only the topic Event was significantly more prevalent for PEAMs than MEAMs (t(9) = –3.05, p = 0.041 FDR corrected, 95% CI [–53.3%, –8.0% ], d = 1.33). In the Healthy group, after FDR correction, period of life memories were significantly more prevalent for MEAMs than PEAMs (t(9) = –3.64, p = 0.022 FDR corrected, 95% CI [13.09, 56.1], d = 1.28).

Fig.2

The content of autobiographical memories evoked by songs and photographs (mean±standard error of the mean) in participants with AD, **p < 0.005.

Fig.3

The content of autobiographical memories evoked by songs and photographs (mean±standard error of the mean) in healthy elderly participants. ^*p < 0.05.

In addition, a significant interaction was also found between Group and Memory Topic (F(3,144) = 2.747, p < 0.0452, $η_{p}^{2}$ = 0.054). However, after FDR correction, none of the follow up t-tests revealed a significant difference between the groups. There was no significant interaction between stimuli, memory topic, and group (Figs. 2 and 3). Post-hoc power analysis confirmed that the power to detect these effects was above 0.76 on these tests.

Specificity of MEAMs and PEAMs

The specificity of MEAMs and PEAMs was coded according to the categories described above. For each memory type (MEAMs or PEAMs), the specificity of the memories was calculated as a proportion of all reported memories and expressed as a percentage. For example, if a person had three MEAMs and two were coded as a score of “1” (semantic general) and the other was a score of “2” (semantic personal), then they had a proportion of 66.7% Semantic General and 33.3% Semantic Personal memories. Examples of the specificity coding of MEAMs and PEAMs are shown in Supplementary Material 3. Figure 4A displays the mean specificity of MEAMs and PEAMs in the AD group, and Fig. 4B shows the mean specificity of MEAMs and PEAMs in the healthy group. These figures demonstrate that MEAMs typically comprised semantic associations (personal and general), lifetime periods or repeated/extended events, while PEAMs tended to comprise mostly specific events and events with internal details.

Fig.4

A) Distribution of specificity of MEAMs and PEAMs in AD participants. B) Distribution of specificity of MEAMs and PEAMs in healthy elderly.

We conducted a Fishers Exact Test for count data with simulated p-value (based on 1,000,000 replications) using the eight specificity categories. Four tests were conducted and the distributions of specificities were significantly different across samples and conditions: healthy elderly group MEAMs versus healthy elderly group PEAMs (p < 0.001); AD group MEAMs versus AD group PEAMs, (p = 0.011); AD group MEAMs versus healthy elderly group MEAMs, (p = 0.047); and AD group PEAMs versus healthy elderly group PEAMs (p < 0.001).

To investigate where these differences occurred, 28 pairwise t-tests were calculated, and Bonferroni correction was used to conservatively control for multiple comparisons. The only significant t-test was between the PEAMs of AD and healthy elderly, in the Specific Event category (t(18) = –3.788, p = 0.0377, 95% CI [–25.3%, –7.2% ], d = 1.694), with healthy participants having more ‘Specific Event’ category memories from photographs than AD participants (18.75% versus 2.5%, respectively).

Frequency of MEAMs and PEAMs in response to stimuli from the reminiscence bump

In keeping with previous literature on this phenomenon [24, 25], for the purpose of our analyses the reminiscence bump was defined as 10–30 years of age. To determine if stimuli from the time of the reminiscence bump (10–30 years) were more likely to trigger memories we explored differences in frequency of MEAMs and PEAMs in response to stimuli from three lifetime periods, specifically, when each participant was aged: 10–30 years, 31–50 years and 51 years and over. For each participant we determined what year they were 10 years old and then considered their responses to the corresponding stimuli for the period of time they were aged 10–30 years, 31–50 years, and 51+ years. For example, if the participant was born in 1935, stimuli from the period of 1945–1965 were considered for their lifetime period 10–30 years (songs 4–7 in Supplementary Material 1). For their lifetime period of 31–50 years, we considered responses to stimuli from 1966–1985 (songs 8–11 in Supplementary Material 1), and for their lifetime period of 51+ years, we considered responses to stimuli from 1986 onwards (songs 12–16 in Supplementary Material 1). Of note, for two participants (one AD and one healthy elderly aged 55 and 57 years, respectively), their lifetime period of 51+ did not include any photograph stimuli or song stimuli.

A 3×2×2 ANOVA was calculated to compare Time period (10–30 years, 31–50 years, 51+ years)×Group (AD, healthy)×Stimuli (song, photograph). A significant main effect of Time period was found (F(2, 108) = 17.443, p < 0.001, $η_{p}^{2}$ = 0.244) as well as an interaction effect of Stimuli and Time period (F(2,108) = 3.279, p = 0.041, $η_{p}^{2}$ = 0.057). Overall, the stimuli from when participants were aged 10–30 years evoked the most frequent memories compared with stimuli from when they were aged 31–50 years (t(39) = 4.21, p < 0.001 FDR corrected, 95% CI [10.5%, 30%], d = 0.68) or 51+ years (t(39) = 7.064, p < 0.001 FDR corrected, 95% CI [23.6%, 42.5%], d = 1.32).

To determine if both groups showed a similar pattern of memories across time periods and both stimuli, planned pairwise comparisons were also run at the individual group level and results displayed in Fig. 5. In the AD participants, MEAMs were more frequently reported in response to songs from when the participants were aged 10–30 years (58.5%) compared with when they were aged 31–50 years (20%) (t(9) = 4.27, p = 0.006 FDR corrected, 95% CI [18.1%, 58.9%], d = 1.602) or 51 years and over (4.4%) (t(9) = 6.55, p < 0.001 FDR corrected, 95% CI [35.4%, 72.8%], d = 2.80]. However, this pattern was not observed for PEAMs, whereby the frequency of memories was more evenly distributed over the three different time periods and there were no significant differences (Table 2, Fig. 5). Healthy elderly also showed more memories evoked by songs during their lifetime period of 10–30 years (53.5%) compared to 31–50 years (31.5%) but after FDR correction this failed to reach significance (t(9) = 2.38, p = 0.077 FDR corrected, d = 0.844). There was, however, a significant difference in the frequency of memories evoked by songs from age 10–30 year compared with those from 51+ years (15.1%) (t(9) = 6.15, p < 0.001 FDR corrected, 95% CI [25.2%, 54.6%], d = 1.877). There was no difference in the frequency of PEAMs across these three periods of life for the Healthy group (Fig. 5). Post-hoc power analysis confirmed that the power to detect these effects was above 0.82 on these tests.

Fig.5

Mean percentage of songs and photographs that evoked memories across three lifetime periods (10-30, 31-50 and 51+ years) (mean ± standard error of the mean) in AD and healthy elderly groups. *p < 0.05, **p < 0.005.

DISCUSSION

The aim of this investigation was to provide the first comparison of autobiographical memories evoked by music (famous songs) and photographs (of famous events) in people with AD. Previous studies in people with AD have typically compared memories cued by music with those reported after a period of silence, precluding any evaluation of whether music is more effective than other stimuli at evoking memories. Our findings contribute to recent work that compared memories evoked by music with other stimuli (single words [17] or odors [18]) in this patient population. A discussion of the current findings relating to the frequency and specificity of MEAMs and PEAMs, the relationship between frequency and demographic variables, and our exploration of the frequency of memories in response to stimuli from the reminiscence bump, will each be addressed in turn below.

Frequency and specificity of MEAMs and PEAMs

We found that the frequency of MEAMs in people with AD was in keeping with the healthy group, which supports the findings of Cuddy et al. [12]. In contrast, there was a decline in the frequency of PEAMs in the AD compared with the healthy group. Within the AD group, there was no difference in the frequency of MEAMs and PEAMs, whereas in the healthy group, PEAMs were more frequent than MEAMs. In both groups and across both types of stimuli, higher familiarity ratings were reported for songs and photographs that triggered memories compared with those that did not. The photograph stimuli were rated as more highly familiar than songs by both groups. Further, healthy elderly rated the photographs as more highly familiar than the AD group. This may have contributed to the higher frequency of PEAMs than MEAMs in the healthy compared with AD group.

Recent research has compared the frequency of memories evoked by music and non-musical stimuli, specifically odors [18] or words [17], in people with AD. In keeping with our finding of a lack of differences in frequency of MEAMs and PEAMs in people with AD, El Haj and colleagues [18] found no difference in the frequency of odor or music evoked memoires. Chevreau et al. [17] found more memories in response to word cues compared with music in both AD and healthy groups. In summary, there are inconsistent findings regarding the efficiency (in regard to frequency) of music compared with other types of stimuli at evoking memories in the AD population. This inconsistency is also evident in research on healthy populations (as discussed above) [15, 16].

In regard to the specificity of MEAMs and PEAMs, we found that in both groups MEAMs were less specific than PEAMs and typically comprised semantic associations, lifetime periods or repeated/extended events. In contrast, PEAMs were more likely to be of specific events. The nature of the photograph stimuli is likely to have underpinned this difference and is a confound that should be addressed in future research. Overall, the AD group produced less specific memories than the healthy group. This is in keeping with previous research showing a decline in specificity of autobiographical memories in general, and more specifically in MEAMs in people with AD compared with healthy groups [2 , 12], despite methodological differences in how specificity has been rated between studies. Cuddy et al. [12] used two methods, namely self-report, and objective rating of memories as either episodic (specific or general) or semantic. They found that self-reported specificity of memories and ratings of ‘episodic specific’ was higher for healthy older and younger adults compared with people with AD. They concluded from results of analyses that collapsed across healthy older and AD groups compared with younger adults that “the effect of age is more evident than the effect of the disease”, in that “specific episodic memories are less available in older age”.

Chevreau et al. [17] found no difference in the ‘quality’ of MEAMs between people with AD and healthy elderly, but higher ‘quality’ of memories cued by words in the healthy compared with AD group. Overall, our findings add to the accumulating evidence for the preservation of MEAMs in the face of AD, but further research is required to clarify how music compares with other types of stimuli in regard to various features of memories, including specificity.

Relationship between frequency of MEAMs and PEAMs and demographics

We found no significant relationship between frequency of MEAMs and the severity of cognitive impairment (as assessed by the M-ACE and a verbal fluency task). This supports previous findings of Cuddy et al. [12] who used alternative measures of cognition, specifically the MoCA and MMSE. In contrast, we found a significant positive relationship between frequency of PEAMs and cognitive function. This suggests that the ability for photographs to trigger memories is more dependent on the integrity of cognitive abilities, whereas music is able to access memories despite severe cognitive impairment. These findings are in keeping with Baird and Samson [19] who found preserved MEAMs in a different neurological population, namely people with severe acquired brain injury. This provides further support for the notion that music abilities, including music memory, are relatively resistant to neurological injury compared with other non-music skills. This is likely to be due to the widespread brain regions that are activated by music, as demonstrated by accumulating functional neuroimaging research (e.g., [26]). Furthermore, several neuroimaging studies have identified a crucial role of the medial frontal regions in MEAM retrieval [27 –29]. In the case of preserved MEAMs in people with AD, it has been shown that the frontal brain regions are relatively less affected by AD pathology [30].

Cuddy et al. [12] found lower scores on the Dementia Severity Rating Scale (DSRS, informant version), in their group of ‘non-responders’ (those participants who did not report any MEAMs), suggesting greater functional impairment in these participants. Nevertheless, DSRS scores were not predictive of the number of MEAMs in ‘responders’ (those participants who did report MEAMs), suggesting that it cannot account for the differences between these groups. We did not have any non-responders to music in our AD or healthy groups, and this may be due to a difference in the music stimuli used (see further discussion of this issue below). We did, however, have one non-responder to photographs in the AD group.

Our prediction of a positive relationship between level of music engagement and frequency of MEAMs was not supported by the current results. We found no significant relationship between the frequency of MEAMs and level of music engagement, as measured by the total score or any subtest scores of the informant version of the MusEQ [21]. We also found no differences in the frequency of MEAMs between those people with AD who were musicians compared with those without any music training. These findings highlight that music can be a powerful mnemonic even in people with low levels of music engagement and no music training.

An unexpected finding was that of a gender differences in frequency of MEAMs (but not PEAMs), with males reporting more than females in both healthy and AD groups. No previous studies of MEAMs in AD have examined gender differences, but it is interesting to note that the two non-responders in Cuddy’s [12] study were both female. In their study of healthy people, Belfi et al. [15] found that compared to men, women produced more frequent and more vivid memories in response to music and photographs of famous faces. It is known that AD affects women more than men [31], and it may be that the memory advantage observed in healthy women reverses in the context of AD. Further studies exploring gender differences in MEAMs in both healthy and clinical populations are needed.

Topics of MEAMs and PEAMs

In both people with AD and healthy elderly, the most common topic of MEAMs was about a period of life (e.g. high school years), followed by a person. In people with AD, nearly half of the MEAMs were about a person/people, typically a spouse/partner. This finding is in keeping with other studies in healthy [11] and AD populations [12]. The observation that music provokes memories of significant people and times in our lives (often highly social times) is in keeping with the broader notion that music evolved to promote social cohesion (for a review, see [32]). In the specific case of people with AD, the current findings of the most common topics of MEAMs relates to our proposal that music is uniquely equipped to enhance various aspects of the self, particularly the ‘interpersonal’ self [33]. It is also consistent with a recent case study of a woman with severe AD, showing how familiar and significant songs can promote recognition of close relatives and correspondingly help to maintain interpersonal relationships [34].

The most common topic of PEAMs was an event, typically that depicted in the photograph (e.g., watching it on TV). This finding is not surprising given that our photograph stimuli depicted famous events that were viewed repeatedly on TV or in newspapers. It is likely that the significant world events depicted in some of our photograph stimuli (such as that of 9/11 and the assassination of John F. Kennedy) could have elicited ‘flashbulb’ memories [35], or autobiographical memories that involve the circumstances in which one learned of a public event. Future studies of PEAMs in people with AD could compare differences between memories elicited by such public events with other types of events, famous people (as in [36]) or landmarks.

Exploration of stimuli from the reminiscence bump

Our method of using stimuli from specific time periods allowed us to explore which stimuli were most effective at triggering memories. In keeping with our prediction, we found that music and photograph stimuli from when participants were aged 10–30 years (the time period of the reminiscence bump) triggered more frequent memories compared with stimuli from later decades (31–50 years of age or 51 years and over).

Although the typical method of assessing a reminiscence bump is to date the time of the memory, our results provide a unique means of assessing the mnemonic effect of stimuli from this time period and provide indirect support for the presence of a reminiscence bump for music and photograph stimuli. The time period of the reminiscence bump is crucial in the formation of our self identity, and one reason for the high accessibility of memories from this time of life is that they are ‘self defining’ memories that are important in the formation of self images [37]. Our finding that memories were more frequently elicited by music from this time period supports the notion that MEAMs represent an important link to the self [33, 38]. Future research could explore this issue by characterizing the ‘self defining’ nature of MEAMs from different time periods. Overall, our results support the accumulating evidence that music is a useful tool for improving autobiographical recall in people with AD and in doing so may provide access and enhance the sense of self.

In their study of healthy young adults, Krumhansl and Zupnick [24] found that in addition to the standard reminiscence bump for popular music, participants also displayed a second reminiscence bump for music that was popular when their parents were young adults. They termed this phenomenon “cascading reminiscence bumps”. The current study did not include song stimuli from the reminiscence bump of the parents of participants, so we were unable to assess whether the phenomenon of cascading reminiscence bumps was present in our sample. Examination of this phenomenon would be an interesting topic for future research of MEAMs in people with dementia.

Cuddy and colleagues [12] did not explicitly explore the reminiscence bump but did report results related to this phenomenon. They dated the reported memories according to the age of the participant when the memory occurred (given as an age range in ‘years from’ and ‘years to’). They found that for people with AD, “the age range that the memory was said to occur was negatively correlated with percent of excerpts yielding MEAMs. In other words ⋯ a larger number of MEAMs was associated with earlier memories” [12, p. 10]. Examining their table of results the age ‘from and to’ was 23–31 years, falling within the commonly regarded reminiscence bump time period of 10–30 years of age. This pattern, however, was not evident in their healthy older adult sample. Methodological differences in characterizing the reminiscence bump (using date of stimuli or date of memory) and differences in the time period of this phenomenon between types of cues has been reported [25]. Future studies of this phenomenon in people with dementia are needed.

Methodological differences between the current and previous studies of MEAMs in AD include the number of stimuli used (i.e., only 2 music stimuli in [18]), the instructions for recalling a memory (‘recall an event from your life’ in [18] versus our more explicit stimulus related cue of ‘does this song/photograph bring to mind a memory?’), and the type of musical stimuli used (researcher chosen classical or instrumental music or participant self-preferred music) could account for some of the variations in findings, such as the absence of any non-responders in the current study. We used popular songs from across the decades, for the purpose of matching the time of photographs and music stimuli and to allow us to explore the presence of any reminiscence bump for either stimuli. We chose famous events that occurred at a specific time (rather than over a period of time), but this method may have influenced our specificity findings, prompting memories of seeing the event on TV or in the newspaper. Future research could attempt to explore memories evoked by personal music and photographs with the help of carers to date the photograph stimuli. Furthermore, music and memory research in dementia has primarily been in people with AD, and there is a need to explore triggers of autobiographical memories in other dementia (non-AD) types, in particular fronto-temporal dementia types given the known autobiographical memory impairment in this population.

As it is often claimed that music is ‘special’ in regard to autobiographical memory, a call for caution was recently recommended when interpreting results from studies that do not actually compare music conditions with other conditions [39]. The current study directly addresses this concern in that we compared music with another familiar stimulus, namely photographs of famous events. Our findings suggest that music, particularly songs from their youth, is special for people with AD, in that it can trigger memories at the same frequency as healthy people, regardless of the severity of their cognitive impairment or level of engagement with music. Whereas these MEAMs may not be special in terms of the specificity of the elicited memory compared with PEAMs, they are often about special people or times of life, which are crucial memories for sustaining a sense of self. Understanding how preserved music related memory functions can be leveraged and optimized for therapeutic benefits remains an exciting challenge for future research.

Footnotes

ACKNOWLEDGMENTS

Thank you to all participants for their time. This work was supported by the National Health and Medical Council and Australian Research Council (NHMRC-ARC Dementia Research Development Fellowship to Amee Baird and ARC funding grants to William Thompson).

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Conway

(2005) Memory and the self. J Mem Lang 53, 594–628.

El Haj

, Antoine

, Nandrino

, Kapogiannis

(2015) Autobiographical memory decline in Alzheimer’s disease, a theoretical and clinical overview. Ageing Res Rev 23(Pt B), 183–192.

Klein

, Cosmides

, Costabile

(2003) Preserved knowledge of self in a case of Alzheimer’s dementia. Soc Cogn 21, 157–165.

Mograbi

, Brown

, Morris

(2009) Anosognosia in Alzheimer’s disease–the petrified self. Conscious Cogn 18, 989–1003.

Kopelman

, Wilson

, Baddeley

(2008) The autobiographical memory interview: A new assessment of autobiographical and personal semantic memory in amnesic patients. J Clin Exp Neuropsychol 11, 724–744.

El Haj

, Fasotti

, Allain

(2012) The involuntary nature of music-evoked autobiographical memories in Alzheimer’s disease. Conscious Cogn 21, 238–246.

El Haj

, Clément

, Fasotti

, Allain

(2013) Effects of music on autobiographical verbal narration in Alzheimer’s disease. J Neurolinguistics 26, 691–700.

Irish

, Cunningham

, Walsh

, Coakley

, Lawlor

, Robertson

, Coen

(2006) Investigating the enhancing effect of music on autobiographical memory in mild Alzheimer’s disease. Dement Geriatr Cogn Disord 22, 108–120.

Foster

, Valentine

(2001) The effect of auditory stimulation on autobiographical recall in dementia. Exp Aging Res 27, 215–228.

10.

Janata

, Tomic

, Rakowski

(2007) Characterization of music-evoked autobiographical memories. Memory 15, 845–860.

11.

El Haj

, Postal

, Allain

(2012) Music enhances autobiographical memory in mild Alzheimer’s disease. Educ Gerontol 38, 30–41.

12.

Cuddy

, Sikka

, Silveira

, Bai

, Vanstone

(2017) Music evoked autobiographical memories in Alzheimer’s Disease: Evidence for a positivity effect. Cogent Psychol 4, 1277578.

13.

Cuddy

, Sikka

, Vanstone

(2015) Preservation of musical memory and engagement in healthy aging and Alzheimer’s disease. Ann N Y Acad Sci 1337, 223–231.

14.

Rubin

, Schulkind

(1997) The distribution of autobiographical memories across the lifespan. Memory Cogn 25, 859.

15.

Belfi

, Karlan

, Tranel

(2016) Music evokes vivid autobiographical memories. Memory 24, 979–989.

16.

Zator

, Katz

(2017) The language used in describing autobiographical memories prompted by life period visually presented verbal cues, event-specific visually presented verbal cues and short musical clips of popular music. Memory 25, 831–844.

17.

Chevreau

, Nizard

, Allain

(2017) Retrieval of memories with the help of music in Alzheimer’s disease. Geriatr Psychol Neuropsychiatr Vieil 15, 309–318.

18.

El Haj

, Gandolphe

, Gallouj

, Kapogiannis

, Antoine

(2017) From nose to memory: The involuntary nature of odor-evoked autobiographical memories in Alzheimer’s disease. Chem Senses 43, 27–34.

19.

Baird

, Samson

(2014) Music evoked autobiographical memory after severe acquired brain injury: Preliminary findings from a case series. Neuropsychol Rehabil 24, 125–43.

20.

Hsieh

, McGrory

, Leslie

, Dawson

, Ahmed

, Butler

, Rowe

, Mioshi

, Hodges

(2015) The Mini-Addenbrooke’s Cognitive Examination: A new assessment tool for dementia. Dement Geriatr Cogn Disord 39, 1–11.

21.

Vanstone

, Wolf

, Poon

, Cuddy

(2016) Measuring engagement with music: Development of an informant-report questionnaire. Aging Ment Health 20, 474–484.

22.

Piolino

, Desgranges

, Clarys

, Guillery-Girard

, Taconnat

, Isingrirni

(2006) Autobiographical memory, autonoetic consciousness, and self-perspective in aging. Psychol Aging 21, 510–525.

23.

Strikweda-Brown

, Mothakunnel

, Hodges

, Piguet

, Irish

(2018) External details revisited – a new toaxonmy for coding ‘non-episodic’ content during autobiographical memory retrieval. J Neuropsychol 10.1111/jnp.12160.

24.

Krumhansl

, Zupnick

(2013) Cascading reminiscence bumps in popular music. Psychol Sci 24, 2057–2068.

25.

Koppel

, Rubin

(2016) Recent advances in understanding the reminiscence bump: The importance of cues in guiding recall from autobiographical memory. Curr Dir Psychol Sci 25, 135–149.

26.

Koelsch

(2011) Toward a neural basis of music perception – a review and updated model. Front Psychol 2, 110.

27.

Janata

(2009) The neural architecture of music-evoked autobiographical memories. Cereb Cortex 19, 2579–2594.

28.

Ford

, Addis

, Giovanello

(2011) Differential neural activity during search of specific and general autobiographical memories elicited by musical cues. Neuropsychologia 49, 2514–2526.

29.

Ford

, Rubin

, Giovanello

(2016) The effects of song familiarity and age on phenomenological characteristics and neural recruitment during autobiographical memory retrieval. Psychomusicology 26, 199–210.

30.

Jacobsen

, Stelzer

, Fritz

, Chételat

, La Joie

, Turner

(2015) Why musical memory can be preserved in advanced Alzheimer’s disease. Brain 138, 2438–2450.

31.

Mazure

, Swendsen

(2016) Sex differences in Alzheimer’s disease and other dementias. Lancet Neurol 15, 451–452.

32.

Huron

(2001) Is music an evolutionary adaptation? Ann N Y Acad Sci 930, 43–61.

33.

Baird

, Thompson

(2018) The impact of music on the self in dementia. J Alzheimers Dis 61, 827–841.

34.

Baird

, Thompson

(2018) When music compensates language: A case study of severe aphasia in dementia and the use of music by a spousal caregiver. Aphasiology. doi: 10.1080/02687038.2018.1471657.

35.

Brown

, Kulik

(1977) Flashbulb memories. Cognition 5, 73–99.

36.

Belfi

, Tranel

(2014) Impaired naming of famous musical melodies is associated with left temporal polar damage. Neuropsychology 28, 429–435.

37.

Rathdowne

, Moulin

, Conway

(2008) Self-centered memories: The reminiscence bump and the self. Mem Cogn 36, 1403–1414.

38.

El Haj

, Antoine

, Nandrino

, Gely-Nargeot

, Raffard

(2015) Self-defining memories during exposure to music in Alzheimer’s disease. Int Psychogeriatr 27, 1719–1730.

39.

Halpern

, Talarico

, Gouda

, Williamson

(2018) Are musical autobiographical memories special? It ain’t necessarily so. Music Percept 35, 561–572.