This is a sel -a chi ed e sion o an o iginal a icle. This e sion
may di e om he o iginal in pagina ion and ypog aphic de ails.
Au ho (s):
Ti le:
Yea :
Ve sion:
Copy igh :
Righ s:
Righ s u l:
Please ci e he o iginal e sion:
CC BY 4.0
h ps://c ea i ecommons.o g/licenses/by/4.0/
Emo ional and musical ac o s combined wi h song-speci ic age p edic he subjec i e
au obiog aphical saliency o music in olde adul s
© The Au ho (s) 2023
Published e sion
Salakka, Ilja; Pi käniemi, Anni; Pen ikäinen, Emmi; Saa i, Pasi; Toi iainen, Pe i;
Sä kämö, Teppo
Salakka, I., Pi käniemi, A., Pen ikäinen, E., Saa i, P., Toi iainen, P., & Sä kämö, T. (2023).
Emo ional and musical ac o s combined wi h song-speci ic age p edic he subjec i e
au obiog aphical saliency o music in olde adul s. Psychology o Music, Online Fi s .
h ps://doi.o g/10.1177/03057356231186961
2023
h ps://doi.o g/10.1177/03057356231186961
Psychology o Music
1 –17
© The Au ho (s) 2023
A icle euse guidelines:
sagepub.com/jou nals-pe missions
DOI: 10.1177/03057356231186961
jou nals.sagepub.com/home/pom
Emo ional and musical ac o s
combined wi h song-speci ic
age p edic he subjec i e
au obiog aphical saliency o
music in olde adul s
Ilja Salakka1,2,3 , Anni Pi käniemi1,2,
Emmi Pen ikäinen1,2, Pasi Saa i4,
Pe i Toi iainen2,4 and Teppo Sä kämö1,2
Abs ac
Music ha e okes s ong emo ional esponses is o en expe ienced as au obiog aphically salien .
Th ough emo ional expe ience, he musical ea u es o songs could also con ibu e o hei subjec i e
au obiog aphical saliency. Songs which ha e been popula du ing adolescence o young adul hood
(ages 10–30) a e mo e likely o e oke s onge memo ies, a phenomenon known as a eminiscence
bump. In he p esen s udy, we sough o de e mine how song-speci ic age, emo ional esponsi eness
o music, musical ea u es, and subjec i e memo y unc ioning con ibu e o he subjec i e
au obiog aphical saliency o music in olde adul s. In a music lis ening s udy, 112 pa icipan s a ed
exce p s o popula songs om he 1950s o he 1980s o au obiog aphical saliency. Addi ionally, hey
illed ou ques ionnai es abou emo ional esponsi eness o music and subjec i e memo y unc ioning.
The song exce p s’ musical ea u es we e ex ac ed compu a ionally using MIR oolbox. Resul s showed
ha au obiog aphical saliency was bes p edic ed by song-speci ic age and emo ional esponsi eness o
music and musical ea u es. Newe songs ha we e mo e simila in hy hm o olde songs we e also
a ed highe in au obiog aphical saliency. O e all, his s udy con ibu es o au obiog aphical memo y
esea ch by unco e ing a se o ac o s a ec ing he subjec i e au obiog aphical saliency o music.
Keywo ds
emo ion, memo y, aging, hy hm, imb e, onali y, modeling
1 Music, Ageing and Rehabili a ion Team, Cogni i e B ain Resea ch Uni , Depa men o Psychology and Logopedics,
Facul y o Medicine, Uni e si y o Helsinki, Helsinki, Finland
2
Cen e o Excellence in Music, Mind, Body and B ain, Uni e si y o Jy äskylä and Uni e si y o Helsinki, Helsinki,
Finland
3Rehabili a ion Founda ion, Helsinki, Finland
4Depa men o Music, A and Cul u e S udies, Uni e si y o Jy äskylä, Jy äskylä, Finland
Co esponding au ho :
Ilja Salakka, Depa men o Psychology and Logopedics, oom F212, P.O. Box 21, FI-00014, Uni e si y o Helsinki,
Finland.
Email: [email p o ec ed]
1186961POM0010.1177/03057356231186961Psychology o MusicSalakka e al.
esea ch-a icle2023
O iginal Empi ical In es iga ions
2 Psychology o Music 00(0)
Memo ies a e impo an o he sense o sel , and music is an excep ionally powe ul s imulus
o e oking hem, e en in people wi h demen ia (Cuddy e al., 2015). Because music ends o be
pe sonally salien o mos people, i is o en used o emo ional sel - egula ion h oughou
adul hood (Saa ikallio, 2010). Tha being said, s ong emo ional esponses o music a e o en
connec ed o s ong musical memo ies (Jana a e al., 2007; Salakka e al., 2021; Schulkind e
al., 1999; Schulkind & Woldo , 2005) and hey can se e as a s eng hening ac o in episodic
memo y encoding (Mado P o e bio e al., 2015). Musical memo y o ma ion has been ound o
be in ima ely linked o he expe ience o ewa d om music, modula ed by he dopamine gic
mesolimbic sys em (Fe e i e al., 2021, 2019), which can, a leas pa ially, explain his s ong
connec ion be ween musical emo ions and memo ies. Con e sely, when a epea edly hea d
song has become amilia , i ends o be expe ienced as highe in emo ional in ensi y and
alence (Ali & Peyni ciogˇgˇlu, 2010; Pe e z e al., 1998). In gene al, hese indings sugges ha
bo h he consolida ion o new musical memo ies and he ecall o old musical memo ies a e
s ongly d i en by emo ions. S ill, indi idual di e ences in emo ional esponsi eness o music
a ec wha kind o emo ional eac ion music is likely o e oke (Juslin e al., 2022); he e o e, hese
di e ences migh also play a pa in he subjec i e au obiog aphical saliency o music.
Musical memo ies a e no s a ic h oughou he li e cou se. In younge people, hey mo e
ypically esemble speci ic li e e en s, whe eas in olde people, hey end o be mo e a ec i ely
ocused and engage do somedial p e on al egions (Fo d e al., 2016), which a e known o
bo h ack he onal s uc u e o music and encode i s emo ional and au obiog aphical saliency
(Jana a, 2009; Jana a e al., 2002). Impo an ly, musical memo ies can o en be e ec i ely
e ie ed e en in neu ological diso de s in which memo y e ie al is ypically s ongly
impai ed, such as in Alzheime ’s disease (El Haj, Faso i, & Allain, 2012; El Haj, Pos al, & Allain,
2012). Gene ally, olde adul s end o emembe memo ies be e om hei you h when com-
pa ed o newe memo ies ( e y ecen memo ies excluded) o memo ies om hei i s 10 yea s.
Known as a eminiscence bump, his is a obus cogni i e phenomenon ound in adul s ac oss
di e en ages (Janssen e al., 2005) and also c oss-cul u ally (Conway e al., 2005). I is cha -
ac e ized by an inc eased ease o e ie al o memo ies om age 10 o 30, usually peaking
somewhe e be ween 13 and 18 yea s on a e age (Janssen e al., 2005). Howe e , he es ima ed
age ange o he eminiscence bump a ies o some ex en depending on he me hod used o
ac i a e he memo ies (Munawa e al., 2018). The eminiscence bump has also been obse ed
in he con ex o musical memo ies, ypically occu ing be ween 5 and 24 yea s and peaking
somewhe e a ound 15 yea s (Jakubowski e al., 2020; Pla z e al., 2015; Schulkind e al., 1999;
Schulkind & Woldo , 2005). In e es ingly, K umhansl and Zupnick (2013) ound e idence o
he cascading eminiscence bump, meaning ha people show a bump in au obiog aphical
memo y ecall o he songs hei pa en s lis ened o as young adul s, because as child en hey
we e exposed o hose songs in hei home en i onmen .
Al hough musical memo ies can usually be ecalled ela i ely well e en unde neu odegen-
e a i e condi ions (Cuddy & Du in, 2005; Cuddy e al., 2012; El Haj, Pos al, & Allain, 2012;
Vans one & Cuddy, 2009b), gene al subjec i e memo y changes also become mo e common
du ing heal hy aging. While subjec i e memo y di icul ies do no necessa ily e lec a concu -
en decline in objec i e memo y pe o mance, hey a e some imes ela ed o de e io a ion in
o he cogni i e domains, such as slowe p ocessing speed (Vaski uo e al., 2018), and may p e-
cede objec i e memo y p oblems (Reid & MacLullich, 2006), and e en demen ia (Schmand e
al., 1996). Regula music lis ening has been obse ed o imp o e subjec i e memo y in olde
adul s expe iencing ea ly signs o cogni i e decline (Innes e al., 2017), sugges ing a ela ion-
ship be ween music and gene al memo y. Howe e , i is s ill unknown i subjec i e memo y
di icul ies a e ela ed o he au obiog aphical saliency o musical memo ies.
Salakka e al. 3
While he emo ional expe ience o music plays a cen al pa in he o ma ion and
e ie al o musical memo ies, musical ea u es migh also independen ly con ibu e o he
subjec i e au obiog aphical saliency o a song. B oadly comp ised o onal, hy hmic, and
imb al aspec s, music can be iewed as s uc u ed sound. I can be b oken down in o
smalle sou ces o in o ma ion, such as spec al p ope ies o musical signals, he ea e
e e ed o as musical ea u es. Music In o ma ion Re ie al (MIR) me hods ha e made i
possible o ex ac musical ea u es ela ed o basic musical elemen s such as imb e,
hy hm, and onali y om he audio signal, which in u n can be used in s a is ical analy-
ses explo ing hei ela ionship o he subjec i e expe ience o music. These compu a ion-
ally ex ac ed musical ea u es ha e p e iously been success ully used o p edic bo h he
subjec i e emo ional expe ience (Ging as e al., 2014; Schube , 2004; Singe e al., 2016)
and he au obiog aphical memo ies e oked by a song (Salakka e al., 2021). Using MIR,
Salakka e al. (2021) epo ed ha he salience o he musical pulse, he ela i e s eng h
o high ha monics, and luc ua ion in equencies be ween 200 and 800 Hz p edic ed bo h
music-e oked emo ions and memo ies in olde adul s. The esul s o hese p e ious s udies
lea e us wi h an open ques ion o whe he musical ea u es may indi idually con ibu e o
he au obiog aphical saliency o a song a e accoun ing o age and emo ional esponsi e-
ness o music.
Taken oge he , many di e en ac o s ela ed o he lis ene o o he music i sel po en ially
unde lie he au obiog aphical saliency o a song. In he p esen s udy, ou ocus was o iden i y
i some o hese, namely song-speci ic age (SSA, he age o he lis ene when he song was
popula ), emo ional esponsi eness o music, subjec i e memo y unc ioning, and musical ea-
u es, ha e an e ec on he o e all au obiog aphical saliency o music. Conce ning musical
ea u es, we ook in o accoun some o he undamen al elemen s o music: onali y, hy hm,
and imb e. Salakka e al. (2021) p oposed ha because olde music, compa ed o mo e con-
empo a y music, ends o ha e a di e en kind o musical “p o ile” (i.e. di e ences in musical
ea u es), people who g ew up lis ening o ha kind o music may ind simila ea u es mo e
appealing also in newe music. The e o e, we expec ed o see ha in he g oup o newe songs,
hose ha ing simila musical p o iles wi h olde songs (measu ed wi h MIR sco es) would be
expe ienced as mo e au obiog aphically salien .
Me hods
Pa icipan s
The pa icipan s o he lis ening expe imen we e 112 heal hy olde adul s, 86 women and 26
men, aged 60–86 yea s (M = 70.7, SD = 5.4), wi h a mean educa ion le el o 4.8 (SD = 2.0,
ange = 1–8) on he In e na ional S anda d Classi ica ion o Educa ion (ISCED) 8-poin scale.
All we e om he Helsinki me opoli an a ea and we e pa icipa ing in an ongoing s udy on he
neu ocogni i e e ec s o senio choi singing (78 pa icipan s we e ama eu choi singe s, and
35 we e nonsinge s). The possible esponses when asking abou gende we e “Man,” “Woman,”
and “O he .” The pa icipan s we e ec ui ed om he Adul Educa ion Cen e s o he Ci ies o
Helsinki, Espoo, and Van aa and om di e en senio ci izens’ associa ions and independen
choi s h ough p esen a ions, lye s, and email ad e isemen s. All pa icipan s we e Finnish-
speaking and had no his o y o neu ological (e.g. demen ia, s oke) o se e e psychia ic (e.g.
schizoph enia, bipola diso de ) diso de s. The s udy was app o ed by he E hical Re iew Boa d
in he Humani ies and Social and Beha io al Science o he Uni e si y o Helsinki, and all pa -
icipan s ga e w i en in o med consen .
4 Psychology o Music 00(0)
Music lis ening expe imen s and ques ionnai es
The pa icipan s pe o med an old- ime music a ing ask (OMRT) consis ing o 70 song
exce p s (see “S imuli and musical ea u e ex ac ion” sec ion). A web applica ion o OMRT
was de eloped speci ically acco ding o he needs o his s udy. The Finnish company Sen ina
L d was esponsible o he echnical implemen a ion o he applica ion and da a epo ing sys-
em, which an on Sen ina’s cloud se ice. The pa icipan s a ed each o he song exce p s by
answe ing i e ques ions assessing expe ienced alence, emo ional in ensi y, a ousal, amilia -
i y, and he au obiog aphical saliency o he song (5-poin Like scale). In he p esen s udy, we
used only au obiog aphical saliency as a dependen a iable on p edic ion models. I was meas-
u ed wi h a ques ion, “Kuinka paljon omakoh aisia muis oja kappale sinussa he ä i?” which
ansla es di ec ly as “How much did he song e oke pe sonal memo ies?” The a ing was om
“ei lainkaan (none a all)” (1) o “e i äin paljon ( e y much)” (5). E en hough he ansla ion
sounds somewha quan i a i e in na u e, he Finnish ph asing has a conno a ion no jus o
he numbe o memo ies bu also o hei s eng h o in ensi y, indica ing how “pe sonally
memo y-e oking” he song was. The answe ing o ma was a o ced choice, and ques ions we e
p esen ed one a a ime. When answe ing ques ions, he pa icipan s we e able o lis en o he
song exce p again as many imes as hey wan ed. Fo each song, he in e ace also allowed he
pa icipan s o sha e any memo ies ela ed o he song by w i ing ex o eco ding audio ( his
quali a i e da a is no pa o his s udy). The pa icipan s we e able o do he whole lis ening
expe imen in hei p e e ed place (e.g. a home), using hei own compu e o a able com-
pu e which was len and se up by he esea che s. The use o headphones o ex e nal speake s
was ecommended, and he pa icipan s we e asked o se he olume o a com o able bu
clea ly audible le el. The expe imen s a ed wi h a sho p ac ice session o amilia ize he
pa icipan s wi h he basic unc ionali y o he in e ace. The whole expe imen ook 2.5 h o
comple e on a e age.
The pa icipan s sel -assessed hei subjec i e memo y by illing ou he P ospec i e-
Re ospec i e Memo y Ques ionnai e (PRMQ; Smi h e al., 2000). In his s udy, we used a
PRMQ o al sco e as a measu e o subjec i e memo y (n = 112, M = 31.08, SD = 7.50,
ange = 16–52). The pa icipan s also illed ou he Musical Engagemen Ques ionnai e
(MusEQ; Vans one e al., 2016) o sel -assess hei engagemen wi h music in daily li e. We
decided a p io i o use only he Emo ion subscale o he MusEQ (n = 111, M = 3.60, SD = 0.65,
ange = 1–5) in his s udy o assess he ela ionship be ween pa icipan s’ emo ional espon-
si eness o music and hei a ings o he au obiog aphical saliency o music.
S imuli and musical ea u e ex ac ion
The song pool consis ed o 140 songs in o al di ided in o wo lis s (A and B, 70 songs in each),
ma ched o song gen e and e a. Each lis comp ised 10 olk songs plus 15 songs om each o
he ou decades om he 1950s o he 1980s, ep esen ing di e en musical gen es (popula ,
ock, and jazz) and languages (Finnish o English). This ime pe iod o s imulus songs was
selec ed because i was op imal o e oking au obiog aphical memo ies in he pa icipan s o
he p esen s udy (Jakubowski e al., 2020; Janssen e al., 2005; Pla z e al., 2015; Schulkind e
al., 1999; Schulkind & Woldo , 2005). The pa icipan s we e alloca ed in o wo g oups (co -
esponding o A and B song lis s) ma ching o age, gende , and choi singing backg ound.
The e was no no able di e ence in he au obiog aphical saliency a ings be ween he g oups,
Akaike in o ma ion c i e ion (AIC) 24,756 -> 24,755; χ2(1) = 3.506, p = .061. The ull song
lis wi h a mo e de ailed desc ip ion o o ming i is p esen ed in Supplemen al Ma e ial 1. Fo

Salakka e al. 5
each song, an exce p o he mos cha ac e is ic and well-known pa (e.g. he cho us) was
selec ed. The a e age leng h o he exce p s was 30 s ( ange = 18–37 s), and hal sine wa e
ade-ins (1 s) and ade-ou s (3 s) we e added o each exce p o make he lis ening expe ience
smoo h. The MP3 o ma was used o all audio iles.
Conce ning he selec ion o musical ea u es, many s udies ha e used a p incipal compo-
nen analysis (PCA) app oach o combine many ea u es in o ewe componen s. Fo example,
Salakka e al. (2021) used a se o 24 musical ea u es and applied he PCA, esul ing in a six-
componen solu ion. Howe e , none o he componen s had a clea onal in e p e a ion despi e
some onal ea u es in he o iginal se . In some o he s udies u ilizing he PCA o simila p oce-
du es, pulse cla i y and key cla i y ha e had s ong loadings in hei own dis inc componen s,
being also he only componen s ha do no ep esen imb e in any way (Allu i e al., 2012;
Bu una e al., 2016; Toi iainen e al., 2020). In he p esen s udy, we wan ed o emphasize he
in e p e abili y o he ea u es, so we decided o use pulse cla i y and key cla i y o ep esen
hy hm and onali y. Tempo al and spec al cla i y has also led o be e ecogni ion o song
ma e ial (Albouy e al., 2020), which u he suppo s using cla i y-based ea u es o hese
domains. Fo imb e, we chose spec al cen oid o be he ep esen ing ea u e, which has p e i-
ously been ound o cha ac e ize undamen al aspec s o imb e, such as b igh ness (Allu i e
al., 2012; McAdams e al., 1995; Salakka e al., 2021). In his way, we o med a se o h ee
musical ea u es, ep esen ing onali y (key cla i y), hy hm (pulse cla i y), and imb e (spec-
al cen oid), wi h clea in e p e a ions o use in he p esen s udy. Supplemen al Ma e ial 3
shows all h ee musical ea u es plo ed agains he yea in which he song was popula . All
musical ea u es we e ex ac ed wi h MIR oolbox 1.7 (La illo & Toi iainen, 2007) on MATLAB
R2018a using a de aul sampling a e (44,100 Hz, ypical o MP3 iles). F ame leng hs o
0.025 sec ( o spec al cen oid) and 3 sec ( o pulse cla i y and key cla i y) we e used wi h a
ame o e lap o 50%. The mean o each musical ea u e (compu ed ac oss he ames) was
used in he s a is ical analyses. Because o he la ge skewness in pulse cla i y, loga i hmic ans-
o ma ion was applied o a oid biasing e ec on he esul s (skewness = 0.94–> 0.37).
S a is ical analyses
S a is ical analyses we e conduc ed wi h R-language e sion 4.0.4 (R Co e Team, 2021) in
RS udio en i onmen e sion 1.4.1717. Linea mixed-e ec s models (LMM) we e applied o i
models p edic ing he au obiog aphical saliency o he songs using he R package lme4 (Ba es
e al., 2015). LMM was used in his s udy o accoun o he s uc u e o he da a ( wo di e en
song lis s wi h di e en pa icipan s) and o indi idual a iance in he au obiog aphical sali-
ency a ings. The i o he models was assessed using ma ginal (R2m) and condi ional (R2c) R2
alues o LMM (Nakagawa & Schielze h, 2013) compu ed wi h he R package MuMIn (Ba on,
2020), and AIC and χ2 alues om he co e R packages (R Co e Team, 2021). Mul icollinea i ies
in LMM p edic o s we e assessed by inspec ing VIF alues using he R package ca (Fox &
Weisbe g, 2019). Residuals o he models we e examined isually using he co e R packages (R
Co e Team, 2021). Di e ences and in e ac ions be ween SSA (low ⩽ 20 yea s/high > 20 yea s)
and au obiog aphical saliency (low sco e ⩽ 3/high sco e > 3) g oups in he domains o he
musical ea u es we e assessed wi h LMM. All g aphical plo s we e c ea ed wi h ggplo 2-pack-
age (Wickham, 2016). Whe e compa ison be ween di e en p edic o s was needed, he s and-
a diza ion o he p edic o a iables was done by sub ac ing he mean and di iding by he
s anda d de ia ion o each a iable. This compu a ion esul ed in scaled a iables ha ing a
mean o 0 and a s anda d de ia ion o 1. The SSA was calcula ed as he di e ence be ween he
yea when he song was popula in he Finnish music cha s and he bi h yea o he
6 Psychology o Music 00(0)
pa icipan . Fo he pu poses o his pape , olk songs we e emo ed om he da a because com-
pu ing hei SSA is ambiguous.
Resul s
He e, we p esen h ee se s o LMM analyses o p edic au obiog aphical saliency. Fi s , we
inspec he ela ionship be ween SSA and he au obiog aphical saliency accoun ing o he
eminiscence bump. Second, we p esen i e models wi h mul iple p edic o s (subjec i e emo-
ional esponse o music, subjec i e memo y unc ioning, musical ea u es, SSA, and age and
gende as demog aphic co a ia es) o inspec hei indi idual p edic i e abili y a e accoun ing
o o he s. Las , we p esen h ee models in which we examine he in e ac ions be ween SSA
and au obiog aphical saliency in di e en musical domains.
The ela ionship be ween SSA and he au obiog aphical saliency
Along wi h he pa icipan ’s age and he yea he song was popula on Finnish adio channels,
Figu e 1 p o ides dis ibu ions o SSA and au obiog aphical saliency. To examine he eminis-
cence bump, we plo ed he au obiog aphical saliency agains SSA. Visual inspec ion o he
LOESS cu e (locally es ima ed sca e plo smoo hing) shows a clea bump a ound he eenage
yea s, peaking a he age o 11 (Figu e 2, dashed line). Because o his nonlinea ela ionship
be ween he SSA and he au obiog aphical saliency, i was possible ha a simple linea unc ion
was no he mos app op ia e model o use he e. Thus, we decided o i highe o de SSA e ms
o he da a o see i hey do be e in accoun ing o he eminiscence bump phenomenon han
a simple linea unc ion. As can be seen om Figu e 1 and Table 1, he cubic unc ion i s qui e
well o he LOESS cu e, p o iding a mo e pa simonious i o e he quad a ic and linea unc-
ions. The qua ic unc ion does no inc ease he i o he model enough compa ed o he cubic
unc ion bu makes he model mo e complex and uns in o he dange o o e i ing. Based on
hese examina ions, we used he cubic SSA model as a p edic o in he nex se o LMM analyses
by including all ele an SSA e ms o he models.
S ep-by-s ep models p edic ing he au obiog aphical saliency
We buil i e LMMs s ep-by-s ep by adding new a iable(s) on op o he o he s o see how much
hey imp o ed he model (in e ms o R2m/R2c and AIC) and wha con ibu ion each o he
p edic o s had beside o he s (Table 2; VIF alues p esen ed in Supplemen al Ma e ial 2). Age
and gende (Model 1) and subjec i e memo y unc ioning (PRMQ; Model 2) had no explana-
o y powe o au obiog aphical saliency. Subjec i e emo ional esponsi eness o music (Model
3) signi ican ly p edic ed he au obiog aphical saliency a e adjus ing o Model 2 e ec s. This
is seen as a dec ease in he be ween-pa icipan s a iance, a ising om he ac ha emo ional
esponsi eness e lec s indi idual di e ences be ween pa icipan s. Model 4 added he cubic
SSA e ec o he model. Unde se e e, bu ine i able, mul icollinea i y be ween di e en SSA
e ms, hei be a coe icien s canno be eliably compa ed o o he e ms in he model. This,
howe e , does no a ec hei added p edic i e alue in he model. Compa ed o Model 3, Model
4 inc eased R2 alues by o e 6% and dec eased esidual a iance by ~10%, indica ing ha
SSA is he bes single p edic o o au obiog aphical saliency among he a iables used he e. In
he ull model (Model 5), musical ea u es we e added, inc easing he ma ginal R2 o almos
11%. All h ee musical ea u es (pulse cla i y, key cla i y, and spec al cen oid) had a signi i-
can s a is ical e ec . O he musical ea u es, pulse cla i y was he bes p edic o o
Salakka e al. 7
au obiog aphical saliency, while he e ec s o key cla i y and spec al cen oid we e clea ly
smalle . Fo bo h models (4 and 5), he e ec s we e seen only in he esidual a iance, e lec ing
a e age consis ency ac oss he pa icipan s in hese ma e s.
Figu e 1. Dis ibu ions o (a) Pa icipan Age (M = 70.7, SD = 5.4), (b) The Yea he Song was Popula in
Finnish Radio Channels (M = 1,969, SD = 10.7), (c) SSA (M = 22.6, SD = 12.0), and (d) The Au obiog aphical
Saliency (M = 2.5, SD = 1.3).
Figu e 2. The Cubic Func ion (Smoo h Line) Ba ely Misses he Peak and he Ve y Ends o he LOESS
Cu e (Dashed Line) bu O he wise Fi s Well Wi h he Reminiscence Bump Phenomenon in he P esen
Da a.
8 Psychology o Music 00(0)
O e all, a e SSA, he bes single p edic o was pulse cla i y, ollowed by a subjec i e emo-
ional esponse o music. The condi ional R2 o he ull model was ~30%, meaning ha a ound
a hi d o he a iance in he expe ienced au obiog aphical saliency was explained by a combi-
na ion o ou ixed p edic o s (SSA, emo ions, and musical ea u es) and he di e ences in he
o e all le el o au obiog aphical saliency o music be ween di e en pe sons ( andom ac o ).
Compa ing he musical ea u es o he songs wi h low/high SSA and he
au obiog aphical saliency
He e, we es ed ou hypo hesis ha he songs wi h low SSA/high au obiog aphical saliency
would ha e mo e simila i y o he songs wi h high SSA/high au obiog aphical saliency han o
he songs wi h high SSA/low au obiog aphical saliency. Based on he isual inspec ion o
Figu e 2 and he esul s o Jakubowski e al. (2020), we ca ego ized he SSA ange o he songs
indi idually o each pa icipan using a cu -o o 20 yea s, o low SSA (⩽ 20 yea s) o high
SSA (> 20 yea s). Simila ly, he au obiog aphical saliency o he songs was classi ied indi idu-
ally o each pa icipan as low (sco e ⩽ 3) o high (sco e > 3). Fi s , we es ed i an in e ac ion
be ween low/high SSA and low/high au obiog aphical saliency was p esen . Second, we
inspec ed he means and con idence in e als o low/high SSA/au obiog aphical saliency ca -
ego ies o see he ela i e di e ences be ween hem. In he domains o spec al cen oid and
pulse cla i y, he s a is ical in e ac ion was clea ly p esen , AIC 123,491 -> 123,367;
χ2(1) = 126.53, p < .001 and AIC 6,270.5 -> 6,178.7; χ2(1) = 93.84, p < .001, espec i ely.
Fo key cla i y, he in e ac ion was no s a is ically signi ican , AIC -17054 -> -17055;
χ2(1) = 2.83, p = .093.
When inspec ing Figu e 3 and Table 3, we see ha o he songs om pa icipan s’ you h
(low SSA), all musical ea u es we e ela i ely independen o he song’s a e age au obiog aph-
ical saliency. In con as , o he la e songs (high SSA), he e we e clea di e ences be ween
high and low au obiog aphical saliency in spec al cen oid and pulse cla i y. Fo pulse cla i y,
in line wi h ou hypo hesis, i is clea ha he songs om he pa icipan s’ you h we e mo e
simila o he la e songs wi h high au obiog aphical saliency han o he songs wi h low au o-
biog aphical saliency. This means ha la e and less au obiog aphically salien songs ha e a
clea e pulse on a e age han he es o he songs in ou sample. As an example, one can imag-
ine ha songs om he 50s o he 60s ypically had less clea bea , and people who we e young
Table 1. Compa ing Di e en Models P edic ing he Au obiog aphical Saliency by Akaike In o ma ion
C i e ion and χ2- es .
AIC χ2p alue
Linea a20,430 11.859 .0006
Quad a icb20,420
Quad a icb20,420 81.48 <.0001
Cubicc20,341
Cubicc20,341 0.6807 .4094
Qua icd20,342
No e. AIC: Akaike in o ma ion c i e ion.
aSSA + ID andom, R2m = 0.0537, R2c = 0.2595.
bSSA + SSA2 + ID andom, R2m = 0.0558, R2c = 0.2608.
cSSA + SSA2 + SSA3 + ID andom, R2m = 0.0604, R2c = 0.2678.
dSSA + SSA2 + SSA3 + SSA4 + ID andom, R2m = 0.0605, R2c = 0.2678.
Salakka e al. 15
Ee ola, T. (2011). A e he emo ions exp essed in music gen e-speci ic? An audio-based e alua ion o da a-
se s spanning classical, ilm, pop and mixed gen es. Jou nal o New Music Resea ch, 40(4), 349–366.
h ps://doi.o g/10.1080/09298215.2011.602195
El Haj, M., Faso i, L., & Allain, P. (2012). The in olun a y na u e o music-e oked au obiog aphi-
cal memo ies in Alzheime ’s disease. Consciousness and Cogni ion, 21(1), 238–246. h ps://doi.
o g/10.1016/j.concog.2011.12.005
El Haj, M., Pos al, V., & Allain, P. (2012). Music enhances au obiog aphical memo y in mild Alzheime ’s
disease. Educa ional Ge on ology, 38(1), 30–41. h ps://doi.o g/10.1080/03601277.2010.515897
Fe e i, L., Mas-He e o, E., Ca dona, G., Za o e, R. J., An onijoan, R. M., Valle, M., Riba, J., Ripollés,
P., & Rod iguez-Fo nells, A. (2021). Dopamine modula ions o ewa d-d i en music memo y con-
solida ion. Annals o he New Yo k Academy o Sciences, 1502(1), 85–98. h ps://doi.o g/10.1111/
nyas.14656
Fe e i, L., Mas-He e o, E., Za o e, R. J., Ripollés, P., Gomez-And es, A., Alica , H., Oli é, G., Ma co-
Palla és, J., An onijoan, R. M., Valle, M., Riba, J., & Rod iguez-Fo nells, A. (2019). Dopamine modu-
la es he ewa d expe iences elici ed by music. P oceedings o he Na ional Academy o Sciences, 116(9),
3793–3798. h ps://doi.o g/10.1073/pnas.1811878116
Fo d, J. H., Rubin, D. C., & Gio anello, K. S. (2016). The e ec s o song amilia i y and age on phenom-
enological cha ac e is ics and neu al ec ui men du ing au obiog aphical memo y e ie al.
Psychomusicology: Music, Mind, and B ain, 26(3), 199–210. h ps://doi.o g/10.1037/pmu0000152
Fox, J., & Weisbe g, S. (2019). An R companion o applied eg ession (3 d ed.). h ps://socialsciences.mcmas-
e .ca/j ox/Books/Companion/
Gignac, G. E., & Szodo ai, E. T. (2016). E ec size guidelines o indi idual di e ences esea che s.
Pe sonali y and Indi idual Di e ences, 102, 74–78. h ps://doi.o g/10.1016/j.paid.2016.06.069
Ging as, B., Ma in, M. M., & Fi ch, W. T. (2014). Beyond in ensi y: Spec al ea u es e ec i ely p edic
music-induced subjec i e a ousal. Qua e ly Jou nal o Expe imen al Psychology, 67(7), 1428–1446.
h ps://doi.o g/10.1080/17470218.2013.863954
Innes, K. E., Sel e, T. K., Khalsa, D. S., & Kanda i, S. (2017). Medi a ion and music imp o e memo y and
cogni i e unc ion in adul s wi h subjec i e cogni i e decline: A pilo andomized con olled ial.
Jou nal o Alzheime ’s Disease, 56(3), 899–916. h ps://doi.o g/10.3233/JAD-160867
Jakubowski, K., Ee ola, T., Tillmann, B., Pe in, F., & Heine, L. (2020). A c oss-sec ional s udy o emi-
niscence bumps o music- ela ed memo ies in adul hood. Music & Science, 3, 205920432096505.
h ps://doi.o g/10.1177/2059204320965058
Jakubowski, K., & Ghosh, A. (2021). Music-e oked au obiog aphical memo ies in e e yday li e. Psychology
o Music, 49(3), 649–666. h ps://doi.o g/10.1177/0305735619888803
Jana a, P. (2009). The neu al a chi ec u e o music-e oked au obiog aphical memo ies. Ce eb al Co ex,
19(11), 2579–2594. h ps://doi.o g/10.1093/ce co /bhp008
Jana a, P., Bi k, J. L., Van Ho n, J. D., Leman, M., Tillmann, B., & Bha ucha, J. J. (2002). The co ical
opog aphy o onal s uc u es unde lying wes e n music. Science, 298(5601), 2167–2170. h ps://
doi.o g/10.1126/science.1076262
Jana a, P., Tomic, S. T., & Rakowski, S. K. (2007). Cha ac e isa ion o music-e oked au obiog aphical
memo ies. Memo y, 15(8), 845–860. h ps://doi.o g/10.1080/09658210701734593
Janssen, S., Chessa, A., & Mu e, J. (2005). The eminiscence bump in au obiog aphical mem-
o y: E ec s o age, gende , educa ion, and cul u e. Memo y, 13(6), 658–668. h ps://doi.
o g/10.1080/09658210444000322
Juslin, P. N., Sakka, L. S., Ba adas, G. T., & La illo , O. (2022). Emo ions, mechanisms, and indi-
idual di e ences in music lis ening. Music Pe cep ion, 40(1), 55–86. h ps://doi.o g/10.1525/
mp.2022.40.1.55
K umhansl, C. L., & Zupnick, J. A. (2013). Cascading eminiscence bumps in popula music. Psychological
Science, 24(10), 2057–2068. h ps://doi.o g/10.1177/0956797613486486
La illo , O., & Toi iainen, P. (2007). A Ma lab oolbox o musical ea u e ex ac ion om audio. h ps://da x.
lab i. /main/pape s/p237.pd

16 Psychology o Music 00(0)
Luck, G., Toi iainen, P., E kkilä, J., La illo , O., Riikkilä, K., Mäkelä, A., Pyhäluo o, K., Raine, H.,
Va kila, L., & Vä i, J. (2008). Modelling he ela ionships be ween emo ional esponses o, and
musical con en o , music he apy imp o isa ions. Psychology o Music, 36(1), 25–45. h ps://doi.
o g/10.1177/0305735607079714
Mado P o e bio, A., Lozano Nasi, V., Alessand a A ca i, L., De Benede o, F., Gua damagna, M., Gazzola,
M., & Zani, A. (2015). The e ec o backg ound music on episodic memo y and au onomic esponses:
Lis ening o emo ionally ouching music enhances acial memo y capaci y. Scien i ic Repo s, 5(1),
15219. h ps://doi.o g/10.1038/s ep15219
McAdams, S., Winsbe g, S., Donnadieu, S., De Soe e, G., & K impho , J. (1995). Pe cep ual scaling o syn-
hesized musical imb es: Common dimensions, speci ici ies, and la en subjec classes. Psychological
Resea ch, 58(3), 177–192. h ps://doi.o g/10.1007/BF00419633
Munawa , K., Kuhn, S. K., & Haque, S. (2018). Unde s anding he eminiscence bump: A sys ema ic
e iew. PLOS ONE, 13(12), e0208595. h ps://doi.o g/10.1371/jou nal.pone.0208595
Nakagawa, S., & Schielze h, H. (2013). A gene al and simple me hod o ob aining R2 om gene alized lin-
ea mixed-e ec s models. Me hods in Ecology and E olu ion, 4(2), 133–142. h ps://doi.o g/10.1111/
j.2041-210x.2012.00261.x
Pe e z, I., Gaud eau, D., & Bonnel, A.-M. (1998). Exposu e e ec s on music p e e ence and ecogni ion.
Memo y & Cogni ion, 26(5), 884–902. h ps://doi.o g/10.3758/BF03201171
Pla z, F., Kopiez, R., Hasselho n, J., & Wol , A. (2015). The impac o song-speci ic age and a ec i e
quali ies o popula songs on music-e oked au obiog aphical memo ies (MEAMs). Musicae Scien iae,
19(4), 327–349. h ps://doi.o g/10.1177/1029864915597567
R Co e Team. (2021). R: A language and en i onmen o s a is ical compu ing. h ps://www.R-p ojec .o g/
Reid, L. M., & MacLullich, A. M. J. (2006). Subjec i e memo y complain s and cogni i e impai -
men in olde people. Demen ia and Ge ia ic Cogni i e Diso de s, 22(5–6), 471–485. h ps://doi.
o g/10.1159/000096295
Saa ikallio, S. (2010). Music as emo ional sel - egula ion h oughou adul hood. Psychology o Music,
39(3), 307–327. h ps://doi.o g/10.1177/0305735610374894
Salakka, I., Pi käniemi, A., Pen ikäinen, E., Mikkonen, K., Saa i, P., Toi iainen, P., & Sä kämö, T. (2021).
Wha makes music memo able? Rela ionships be ween acous ic musical ea u es and music-e oked
emo ions and memo ies in olde adul s. PLOS ONE, 16(5), e0251692. h ps://doi.o g/10.1371/
jou nal.pone.0251692
Schmand, B., Jonke , C., Hooije , C., & Lindeboom, J. (1996). Subjec i e memo y complain s may
announce demen ia. Neu ology, 46(1), 121–125. h ps://doi.o g/10.1212/WNL.46.1.121
Schube , E. (2004). Modeling pe cei ed emo ion wi h con inuous musical ea u es. Music Pe cep ion,
21(4), 561–585. h ps://doi.o g/10.1525/mp.2004.21.4.561
Schulkind, M. D., Hennis, L. K., & Rubin, D. C. (1999). Music, emo ion, and au obiog aphical mem-
o y: They’ e playing you song. Memo y & Cogni ion, 27(6), 948–955. h ps://doi.o g/10.3758/
BF03201225
Schulkind, M. D., & Woldo , G. M. (2005). Emo ional o ganiza ion o au obiog aphical memo y. Memo y
& Cogni ion, 33(6), 1025–1035. h ps://doi.o g/10.3758/BF03193210
Singe , N., Jacoby, N., Lin, T., Raz, G., Shpigelman, L., Gilam, G., G ano , R. Y., & Hendle , T. (2016).
Common modula ion o limbic ne wo k ac i a ion unde lies musical emo ions as hey un old.
Neu oimage, 141, 517–529. h ps://doi.o g/10.1016/j.neu oimage.2016.07.002
Smi h, G., Del Sala, S., Logie, R. H., & Maylo , E. A. (2000). P ospec i e and e ospec i e memo y
in no mal ageing and demen ia: A ques ionnai e s udy. Memo y, 8(5), 311–321. h ps://doi.
o g/10.1080/09658210050117735
Toi iainen, P., Bu una , I., B a ico, E., Vuus , P., & Allu i, V. (2020). The ch onnec ome o musical bea .
Neu oImage, 216, 116191. h ps://doi.o g/10.1016/j.neu oimage.2019.116191
Vans one, A. D., & Cuddy, L. L. (2009). Musical memo y in Alzheime disease. Aging, Neu opsychology,
and Cogni ion, 17(1), 108–128. h ps://doi.o g/10.1080/13825580903042676
Vans one, A. D., Wol , M., Poon, T., & Cuddy, L. L. (2016). Measu ing engagemen wi h music:
De elopmen o an in o man - epo ques ionnai e. Aging & Men al Heal h, 20(5), 474–484. h ps://
doi.o g/10.1080/13607863.2015.1021750
Salakka e al. 17
Vaski uo, L., Hokkanen, L., Hänninen, T., An ikainen, R., Bäckman, L., Laa ikainen, T., Paajanen,
T., S igsdo e -Neely, A., S andbe g, T., Tuomileh o, J., Soininen, H., Ki ipel o, M., & Ngandu,
T. (2018). Associa ions be ween p ospec i e and e ospec i e subjec i e memo y complain s
and neu opsychological pe o mance in olde adul s: The Finge s udy. Jou nal o he In e na ional
Neu opsychological Socie y, 24(10), 1099–1109. h ps://doi.o g/10.1017/S135561771800053X
Wickham, H. (2016). ggplo 2: Elegan g aphics o da a analysis. Sp inge . h ps://ggplo 2. idy e se.o g