psyg-11-00425 Ma ch 13, 2020 Time: 17:38 # 1
ORIGINAL RESEARCH
published: 13 Ma ch 2020
doi: 10.3389/ psyg.2020.00425
Edi ed by:
Mon se a Comesaña,
Uni e si y o Minho, Po ugal
Re iewed by:
Glo ia Gaglia di,
Uni e si y o Naples “L’O ien ale”, I aly
Ru h Elaine Ma k,
Tilbu g Uni e si y, Ne he lands
*Co espondence:
Da id Facal
[email p o ec ed]
Special y sec ion:
This a icle was submi ed o
Language Sciences,
a sec ion o he jou nal
F on ie s in Psychology
Recei ed: 25 Oc obe 2019
Accep ed: 24 Feb ua y 2020
Published: 13 Ma ch 2020
Ci a ion:
Campos-Magdaleno M, Lei a D,
Pe ei o AX, Lojo-Seoane C, Mallo SC,
Nie o-Viei es A, Juncos-Rabadán O
and Facal D (2020) Longi udinal
Pa e ns o he Tip-o - he-Tongue
Phenomenon in People Wi h
Subjec i e Cogni i e Complain s
and Mild Cogni i e Impai men .
F on . Psychol. 11:425.
doi: 10.3389/ psyg.2020.00425
Longi udinal Pa e ns o he
Tip-o - he-Tongue Phenomenon in
People Wi h Subjec i e Cogni i e
Complain s and Mild Cogni i e
Impai men
Ma ía Campos-Magdaleno1, Da id Lei a2, A u o X. Pe ei o1, C is ina Lojo-Seoane1,
Sabela C. Mallo1, Ana Nie o-Viei es1, Onésimo Juncos-Rabadán1and Da id Facal1*
1Depa men o De elopmen al Psychology, Uni e si y o San iago de Compos ela, Galicia, Spain, 2Depa men o Social
Psychology and Quan i a i e Psychology, Ins i u e o Neu osciences, Uni e si y o Ba celona, Ba celona, Spain
Backg ound: The Tip-o - he-Tongue (ToTs) s a e is conside ed a uni e sal phenomenon
and is a equen cogni i e complain in old age. P e ious c oss-sec ional s udies ha e
ound ha ToT measu es success ully disc imina e be ween cogni i ely unimpai ed
adul s and adul s wi h Mild Cogni i e Impai men (MCI). The aim o his s udy was o
iden i y longi udinal pa e ns o ToTs in indi iduals wi h subjec i e complain s and wi h
MCI ega ding p og ess o hei cogni i e s a us.
Me hod: The s udy included 193 pa icipan s wi h subjec i e cogni i e complain s
(SCC) and 56 pa icipan s wi h MCI who comple ed a baseline and wo ollow-
up assessmen s, wi h an in e al o abou 18 mon hs be ween each assessmen .
Pa icipan s we e classi ied in o h ee g oups by conside ing cogni i e s abili y o
de e io a ion om he baseline diagnosis: SCC-s able, MCI-s able and MCI-wo sened.
Pa icipan s pe o med a ToT ask in ol ing ecogni ion and naming o amous people
depic ed in 50 pho og aphs. Gene alized Linea Mixed Models (GLMM) we e used
o model longi udinal changes in amilia i y, eeling o knowing, seman ic access,
phonological access and e bal luency.
Resul s: Phonological access di e en ia ed MCI pa ien s, s able and wo sened, om
adul s wi h SCCs a all e alua ion imes. Phonological access declined o e ime in he
h ee g oups, wi hou signi ican in e ac ions be ween g oups and ime.
Discussion: This s udy p o ides he i s longi udinal e idence o di e ences in ToT
measu es o adul s wi h MCI. The indings indica e ha phonological access measu es
success ully di e en ia ed be ween he diagnos ic g oups. Howe e , slopes emain
i espec i e o he diagnos ic g oup and p og ession owa d mo e ad ance s ages o
cogni i e impai men .
Keywo ds: ip-o - he- ongue, lexical access, mild cogni i e impai men , linea mixed models, longi udinal s udy,
Compos ela Aging S udy
F on ie s in Psychology | www. on ie sin.o g 1Ma ch 2020 | Volume 11 | A icle 425
psyg-11-00425 Ma ch 13, 2020 Time: 17:38 # 2
Campos-Magdaleno e al. Longi udinal Pa e n o ToTs
INTRODUCTION
Cogni i e impai men in old adul s has been conside ed
a con inuum including di e en s ages (Jack e al., 2018):
a cogni i ely unimpai ed phase (CU), wi h pe o mance
wi hin he expec ed ange o age and educa ion; p esence
o subjec i e cogni i e complain s (SCC), wi hou objec i e
cogni i e impai men (Jessen e al., 2014;Molinue o e al., 2017);
Mild Cogni i e Impai men (MCI), cha ac e ized by he p esence
o cogni i e complain s, objec i e mild cogni i e de e io a ion
and ela i e p ese a ion o ins umen al ac i i ies o daily li ing
(Pe e sen, 2004;Pe e sen e al., 2018); and demen ia o majo
neu ocogni i e diso de , cha ac e ized by cogni i e a ec a ion
and psychological symp oms ha cause dependency (Ame ican
Psychia ic Associa ion, 2013). In MCI, single and mul iple
domain sub ypes (de e io a ion in only one o in mo e cogni i e
domains) ha e been used o desc ibe di e en deg ees o se e i y,
wi h he mul iple domain sub ype being he mos se ious
condi ion (B amba i e al., 2009;Han e al., 2012;Campos-
Magdaleno e al., 2016). P og ession in o his con inuum is a
complex p ocess cha ac e ized by cogni i e changes, ansi ions
and diagnos ic ins abili y, wi h an inc eased isk o con e sion
o demen ia bu also he possibili y o eg ession o CU (Facal
e al., 2015;Pe e sen e al., 2018). MCI en i y is he e ogeneous,
and di e en sub ypes acco ding e olu iona y ajec o ies and
se e i y need o be add essed (Díaz-Ma domingo e al., 2017).
Language measu es such as e bal luency, naming and wo d
lea ning ha e been success ully used as p edic o s o MCI
and i s p og ession o demen ia (Mu phy e al., 2006;Clague
e al., 2011;Campos-Magdaleno e al., 2017). Tip-o - he-Tongue
(ToT) cons i u es one o he mos equen age- ela ed language
complain s and is cha ac e ized as a s ong eeling o knowing
in pa allel wi h an inabili y o ecall a lexical i em which is
known and ha migh e en ually be ecalled i enough a en ion
and encoding eedback is p o ided (B own, 2012;Bloom e al.,
2018). Age- ela ed inc eases in ToT expe iences (he ea e ToTs)
a e no ela ed o inc eased ocabula y knowledge h oughou
adul hood (Facal e al., 2012;Sal house and Mendell, 2013;
Sha o e al., 2017). Consis en e idence suppo s he hypo hesis
ha he highe equency o ToTs in olde adul s is caused
by a decline in ansmission o he ac i a ion om seman ic
o phonological ep esen a ions (Bu ke e al., 1991;James
and Bu ke, 2000;Sha o e al., 2007;Juncos-Rabadán e al.,
2010;Whi e e al., 2013). Acco ding o his hypo hesis, ToTs
occu when he ac i a ed seman ic ep esen a ion o a wo d
ails o sp ead he necessa y ac i a ion o i s co esponding
phonological ep esen a ion, making lexical access impossible.
The inc ease in he equency o ToTs in olde adul s is
consis en wi h an age- ela ed decline in ac i a ion ansmission,
and p ope names seem o be mo e ulne able o his decline
han common nouns, as p ope names a e ep esen ed by he
indi idual cha ac e is ics o a pe son a he han by mo e gene al
in o ma ion connec ed o mul iple seman ic nodes (Bu ke e al.,
1991). O he ele an hypo hesis on cogni i e aging, such as
he inhibi ion de ici , explained ToT as a de icien inhibi ion o
di e en phonological ep esen a ions (compe i o s) ha a ise
when seman ic ep esen a ion o he a ge wo d is success ully
ac i a ed (Woodwo h, 1938). Howe e , ew expe imen al s udies
(Jones and Lang o d, 1987;Jones, 1989) ha e suppo ed ha
hypo hesis, and o he s udies ha e no been able o eplica e hem
(Meye and Bock, 1992;Pe ec and Hanley, 1992).
Acco ding o he cogni i e con inuum be ween unimpai ed
cogni ion and demen ia, MCI ep esen s an in e media e s age
in he abili y o e ie e p ope names and is cha ac e ized by
g ea e di icul y in phonological access, ela i e o cogni i ely
unimpai ed old adul s, and only mild di icul ies in seman ic
access mo e commonly associa ed wi h he onse o Alzheime ’s
disease (Juncos-Rabadán e al., 2014). Se e al ToT measu es,
including seman ic access (calcula ed as a p opo ional measu e
ha ep esen s success ul access in he o al numbe o
a ge names) and phonological access (calcula ed as he
p opo ion o success ul seman ic e ie als in which success
in phonological access is also achie ed) (Gollan and B own,
2006;Juncos-Rabadán e al., 2010), ha e been success ully used
as language p edic o s o MCI (Juncos-Rabadán e al., 2013).
A mul i a ia e logis ic eg ession model including eeling o
knowing, seman ic knowledge, seman ic access and phonological
access was used o assess he p edic i e alue o ToT measu es o
disc imina ing be ween no mal con ols and MCI pa ien s wi hin
he Compos ela Aging S udy (CompAS). In a c oss-sec ional
s udy, Juncos-Rabadán e al. (2013) ound ha a model including
hese ou ToT measu es oge he co ec ly classi ied 70% o
con ols (speci ici y) and 71.6 o MCI pa ien s (sensi i i y), wi h
an A ea Unde Cu e Roc (AUC) alue o 0.74, and accoun ed
o 23.5% o he a iance. Al hough he model comp ised all
ToT a iables, only he phonological access measu e emained
signi ican ly associa ed wi h amnes ic MCI. The au ho s also
ound ha speci ici y, sensi i i y and AUC alues we e highe
han hose ob ained using seman ic luency as a language measu e
o disc imina e MCI ( o al classi ica ion alue, AUC = 0.66 and
accoun ed a iance = 15.4%). We ha e o men ion wo s udies
(Poppe e al., 2006;Oh and Ha, 2015) ha did no ind di e ences
be ween no mal oldes people and MCI pa ien s, bu hey used
he pe cen o he o al numbe o p oduced ToTs ha ha e
been c i icized as no app op ia e measu es because hey do
no explain he seman ic and phonological ep esen a ion and
p ocesses in ol ed in ToT (Gollan and B own, 2006).
Longi udinal s udies on ToTs in MCI a e e y sca ce. As a as
we know, apa o he a o emen ioned by Poppe e al. (2006) ha
used he o al numbe o epo ed ToTs, only one ollow-up s udy
o changes in ToT in MCI has been ca ied ou o da e (Facal e al.,
2016a). In he a o emen ioned s udy, p opo ional measu es o
change be ween baseline and one ollow-up assessmen (a ound
18 mon hs) we e calcula ed o amilia i y, seman ic access,
phonological access and seman ic luency in a sample o 15
indi iduals wi h mul iple domain amnes ic MCI, 41 indi iduals
wi h single domain amnes ic MCI and 41 cogni i ely unimpai ed
con ols. Compa isons e ealed signi ican di e ences be ween
baseline and ollow-up only in seman ic and phonological access,
wi h imp o emen s in seman ic access in he con ol g oup and
decline in phonological access in he wo g oups wi h amnes ic
MCI. Ne e heless, ull longi udinal models ha e been used o
s udy change in seman ic and phonological access and hei
po en ial ole in explaining diagnos ic change in MCI.
F on ie s in Psychology | www. on ie sin.o g 2Ma ch 2020 | Volume 11 | A icle 425
psyg-11-00425 Ma ch 13, 2020 Time: 17:38 # 3
Campos-Magdaleno e al. Longi udinal Pa e n o ToTs
FIGURE 1 | Rep esen a ion o he asks included in he ToT p ocedu e. The pho og aph shown in he example is o Ra ael Nadal, a amous Spanish ennis playe .
Pho og aphy by Valen ina Alemanno (CC), modi ied acco ding o igu e size equi emen s om h ps://www. lick .com/pho os/ he hale/14309864633.
Facal e al. (2016a) also conside s “ amilia i y” as a ToT
measu e o me a-cogni i e p ocesses in ol ed in ToTs ha
indica e ha he name knowledge is p esen (Schwa z and
Me cal e, 2011). Al hough some c oss-sec ional s udies sugges
ha amilia i y-based memo y measu es may be sensi i e
ma ke s o p eclinical and p od omal Alzheime ’s Disease (AD,
Wolk e al., 2013;Pi a que e al., 2016), he longi udinal
app oach did no show any e idence o hei p edic i e alue
(Facal e al., 2016a).
The aim o he p esen s udy was o de e mine longi udinal
pa e ns o se e al ToT measu es (mainly seman ic and
phonological access) by using linea mixed models and da a om
longi udinally assessed indi iduals wi h SCC and MCI classi ied
on he basis o diagnos ic s abili y o de e io a ion. Wi h his
objec i e we expec ed o ob ain new e idence ega ding he
use ulness o hese measu es as linguis ic ma ke s o cha ac e ize
he cogni i e p o ile o adul s wi h MCI.
MATERIALS AND METHODS
Pa icipan s
Two hund ed o y-nine adul s in he ange o 50–87 yea s old
al eady pa icipa ing in he Compos ela Aging S udy (CompAS)
and who comple ed 3 ex ensi e clinical and neu opsychological
assessmen s (Baseline, Time 1, and Time 2) we e included in his
s udy. A baseline he e we e 407 pa icipan s who pe o med
he ToTs asks, bu only 249 comple ed he 3 assessmen s, being
he o al a e o a i ion a ound 38% (158 pa icipan s) due
o mo i a ion, mobili y o mo bidi y. CompAS is an ongoing
longi udinal p ojec (Juncos-Rabadán e al., 2012) in which
pa icipan s a e ec ui ed a e e e al by gene al p ac i ione s
om p ima y ca e cen e s in Galicia (an au onomous egion in
no h-wes e n Spain) subjec i e cogni i e complain s. A s udy
on he a i ion in he gene al CompAS p ojec and hei
aisons may be see in Facal e al. (2016b). Exclusion c i e ia
included p e ious diagnosis o any neu ological o psychia ic
disease, demen ia, MCI, clinical s oke, mo o -senso y de ec s,
alcohol o d ug abuse/dependency and auma ic b ain inju y
a baseline. All pa icipan s unde wen he same ex ensi e
assessmen , and we e classi ied in o SCC o MCI g oups a a
special mee ing o he esea ch eam. MCI subjec s we e classi ied
in o ou sub ypes ollowing s anda d c i e ia (Pe e sen, 2004;
Dubois e al., 2007;Albe e al., 2011): single-domain amnes ic
MCI (sda-MCI); mul iple-domain amnes ic MCI (mda-MCI);
single-domain non-amnes ic MCI (sdna-MCI); and mul iple-
domain non-amnes ic MCI (mdna-MCI). All MCI pa icipan s
ul illed he gene al c i e ia ou lined by he Na ional Ins i u e
F on ie s in Psychology | www. on ie sin.o g 3Ma ch 2020 | Volume 11 | A icle 425
psyg-11-00425 Ma ch 13, 2020 Time: 17:38 # 4
Campos-Magdaleno e al. Longi udinal Pa e n o ToTs
TABLE 1 | Mean alues and s anda d de ia ions (in pa en heses) o he demog aphic, cogni i e and TOT measu es by he h ee cogni i ely no mal g oups: subjec i e cogni i e complain s (SCC) ha emained S able
(SCC S able), Mild Cogni i e Impai men ha emained S able (MCI S able), and MCI ha wo sened (MCI Wo sened).
SCC S able
G oup 1 N= 193
MCI S able G oup
2N= 33
MCI Wo sened
G oup 3 N= 23
G oup di e ences
-K uskal–Wallis
χ2(gl)
G oups
compa ison –
Mann–Whi ney
es s
G oups
di e ences-
ANOVAs
F(2,146)
G oups
compa ison
Bon e oni
es s
Age 64.94 (8.84) Range:
50–87
69.27 (7.35) Range:
54–82
73.91 (7.15) Range:
61–87
24.15 (2)** G oup3 >G oup2
>G oup1
41.66** G oup3 >G oup2
>G oup1
Yea s Educa ion 10.18 (4.66) Range:
2–22
9.06 (4.58) Range:
3–21
8.35 (3.70) Range:
2–18
4.51 (2) 6.64*G oup3 <G oup1
CCI 0.87 (0.89) Range:
0–3
0.75 (−93) Range:
0-4
1.04 (0.97) Range:
0–3
1.56 (2) 0.68
Law on 7.60 (0.88) Range:
4–8
6.84 (1.40) Range:
3–8
6.55 (1.93) Range:
3–8
16.33 (2)** G oup3,
G oup2 <G oup1
11.80** G oup3 <G oup2,
G oup1
SCC- Pa ien 18.66 (4.34) Range:
7–32
20.22 (4.21) Range:
10–29
19.47 (4.45) Range:
13–33
7.09 (2) *G oup2 >G oup1 2.02
SCC-In o man 15.59 (4.49) Range:
2–29
16.53 (3.72) Range:
9–22
18.25 (4.31) Range:
12–26
6.83 (2)*G oup3 >G oup1 3.78*G oup3 >G oup1
MMSE 28.16 (1.64) Range:
21–30
25.75 (2.63) Range:
21–30
25.17 (2.53) Range:
19–29
50.82 (2)** G oup3,
G oup2 <G oup1
42.62** G oup3,
G oup2 <G oup1
CVLT-SDFR 10.64 (2.71) Range:
3–16
4.96 (3.15) Range:
0–11
4.43 (3.08) Range:
0–11
87.23 (2)** G oup3,
G oup2 <G oup1
96.77** G oup3,
G oup2 <G oup1
CVLT-LDFR 11.44 (2.91) Range:
3–16
6.27 (3.91) Range:
0–14
4.82 (2.99) Range:
0–11
79.95 (2)** G oup3,
G oup2 <G oup1
78.25** G oup3,
G oup2 <G oup1
WAIS ocabula y
subscale
51.18 (13.07)
Range: 19–75
41.27 (13.00)
Range: 20–71
41.48 (14.47)
Range: 15–64
20.89 (2)** G oup3,
G oup2 <G oup1
36.43** G oup3,
G oup2 <G oup1
Seman ic access
p opo ion
0.92 (0.09)
Range:0.40–1
0.87 (0.12)
Range:0.40–1
0.74 (0.20) Range:
0.22–1
15.97 (2)** G oup3,
G oup2 <G oup1
70.54** G oup3 <G oup2
<G oup1
Phonological
access p opo ion
0.82 (0.13)
Range:0.09 – 1
0.66 (0.20)
Range:0.08–1
0.62 (0.61)
Range:0.18–0.98
13.95 (2)** G oup3,
G oup2 <G oup1
70.37** G oup3,
G oup2 <G oup1
Familia i y 222.87 (28–28)
Range: 89–250
223.44 (27.96)
Range: 127–250
204.14 (36.42)
Range: 116–250
16.98 (2)** G oup3 <G oup2,
G oup1
10.78** G oup3 <G oup2,
G oup1
Feeling o Knowing 48.05 (3.57) Range:
23–50
46.75 (4.64) Range:
24–50
41.40 (9.53) Range:
17–50
45.29 (2)** G oup3 <G oup2
<G oup1
51.79** G oup3 <G oup2,
G oup1
Seman ic luency 17.75 (5.79) Range:
6–35
13.66 (4.11) Range:
7–25
11.17 (4.13) Range:
5–20
26.32 (2)** G oup3,
G oup2 <G oup1
54.67** G oup3 <G oup2
<G oup1
CCI, Cha lson Como bidi y Index. SCC, Subjec i e Cogni i e Complain s. MMSE, MiniMen al S a e Examina ion. CVLT-SDFR, Cali o nia Ve bal Lea ning Tes , Sho Delay F ee Recall. CVLT- LDFR, Cali o nia Ve bal
Lea ning Tes , Long Delay F ee Recall. *p<0.05, **p<0.01.
F on ie s in Psychology | www. on ie sin.o g 4Ma ch 2020 | Volume 11 | A icle 425
psyg-11-00425 Ma ch 13, 2020 Time: 17:38 # 5
Campos-Magdaleno e al. Longi udinal Pa e n o ToTs
TABLE 2 | Summa y o models compa ed o Familia i y.
Dependen a iable: Familia i y
Model 1 Model 2 Model 3
Age a baseline 1.283 (1.529) 2.558 (1.556) 2.552 (1.555)
Vocabula y-
WAIS
2.453 (1.527) 1.953 (1.549) 1.938 (1.549)
E alua ion Time 4.764*** (1.196) 5.050*** (1.332)
MCI-S able 1.103 (4.529) 6.311 (8.679)
MCI-Wo sened −18.798*** (5.526) −19.531 (0.134)
E alua ion
Time ×MCI-
S able
−2.596 (3.685)
E alua ion
Time ×MCI-
Wo sened
0.429 (4.815)
In e cep 221.769*** (1.468) 213.637*** (2.904) 213.064*** (3.134)
Obse a ions 676 676 676
Log Likelihood −3,186.721 −3,166.436 −3.161.473.160
Akaike In . C i . 6,391.442 6,356.871 6,350.947
Bayesian In .
C i .
6,432.048 6,410.959 6,414.007
Bayes Fac o – 37966.384 0.2178
All models include andom e ec s o in e cep s and slopes, he e oskedas ici y due
o he g oup, and Age and Vocabula y a baseline as co a ia es. Model 1 is he null
mixed model (i.e., in e cep s and co a ia es only); Model 2 is he mixed model wi h
main e ec s; and Model 3 is he mixed model wi h main e ec s and in e ac ions.
Coe icien s and s anda d e o s (in pa en heses). ***p<0.01.
on Aging-Alzheime ’s Associa ion (Albe e al., 2011): (a)
in o man -co obo a ed memo y complain s, assessed by a sho
e sion o he Subjec i e Memo y Complain s Ques ionnai e
(SMCQ; Benede and Seisdedos, 1996); (b) pe o mance o 1.5
s anda d de ia ions below age and educa ion no ms in a leas
one cogni i e domain, assessed by he subscales o he Spanish-
adap ed e sion o he Camb idge Cogni i e Examina ion
(CAMCOG-R, Huppe e al., 1996; Spanish e sion: López-
Pousa, 2003;Pe ei o e al., 2015), apa om he memo y
domain, which was assessed by he Sho and Long Delay
F ee Recall om he Spanish-adap ed e sion o he Cali o nia
Ve bal Lea ning Tes (CVLT, Delis e al., 1987; Spanish
e sion: Benede and Alejand e, 1998); (c) no signi ican
impac on ac i i ies o daily li ing, assessed by he Law on
and B ody Index (Law on and B ody, 1969); and (d) no
demen ia, acco ding he Na ional Ins i u e o Neu ological and
Communica i e Diso de s and S oke-Alzheime ’s Disease and
Rela ed Diso de s Associa ion (NINCDS-ADRDA), and he
Diagnos ic and S a is ical Manual o Men al Diso de s- Fou h
Edi ion (DSM-IV) c i e ia. Pa icipan s we e classi ied as SCC
when, p esen ing subjec i e cogni i e complains o hei gene al
p ac i ione s con i med by hei own esponses and ha om
hei ela i es o he SMCQ, hey pe o med as cogni i ely
unimpai ed adul s acco ding o no ms o age and yea s o
educa ion in gene al unc ioning and speci ic domain es s
assessed wi h CAMCOG-R and he CVLT.
All pa icipan s and hei p oxies we e in o med o he
longi udinal na u e o he p ojec and we e con ac ed wice
ega ding pa icipa ion in wo successi e ollow-up assessmen s
wi h an in e al o 18.67 ±2.73 mon hs be ween each assessmen .
This ime in e al maximizes pa icipa ion and mo i a ion, while
and educes a i ion due mo bidi y, mobili y and mo ali y
(Facal e al., 2016b). A e he second ollow-up assessmen ,
pa icipan s we e classi ied in o h ee g oups by conside ing
FIGURE 2 | Es ima ed ma ginal means and e o s ba s om Model 2 o Familia i y in he h ee g oups ac oss he h ee e alua ion imes. SE, S anda d E o ; BL,
Baseline assessmen ; T1, Time 1 assessmen ; T2, Time 2 assessmen .
F on ie s in Psychology | www. on ie sin.o g 5Ma ch 2020 | Volume 11 | A icle 425
psyg-11-00425 Ma ch 13, 2020 Time: 17:38 # 6
Campos-Magdaleno e al. Longi udinal Pa e n o ToTs
s abili y o p og ession om he diagnos ic es ablished a
baseline: SCC pa icipan s a Baseline assessmen who emained
s able a Time 2 ollow-up (SCC-S able g oup, n= 193, 77.52%,
136 women/56 men); MCI pa icipan s a Baseline assessmen
who emained s able a Time 2 ollow-up (MCI-S able g oup,
n= 33, 13.24%, 20 women/13 men); and sda-MCI o sdna-
MCI pa icipan s a Baseline assessmen who had p og essed o
mda-MCI, mdna-MCI o demen ia ei he a Time 1 o Time
2 ollow-up e alua ions (MCI-Wo sened g oup, n= 23, 9.24%,
16 women/7 men). Di e ences in he g oups size e lec he
incidence o MCI in people wi h subjec i e cogni i e complains
who a end p ima y ca e cen e s and he di e en a es o s abili y
o p og ession/wo sening (Facal e al., 2019).
Assessmen o pa icipan s who p og essed o p obable AD o
demen ia was conduc ed acco ding o he DMS-IV and NINCDS-
ADRDA c i e ia, by checking he medical his o y and eco ding
he da e o neu ological diagnosis.
All pa icipan s ga e hei w i en in o med consen p io o
pa icipa ion in he s udy. The esea ch p ojec was app o ed
by he Galician Clinical Resea ch E hics Commi ee (Xun a de
Galicia, Spain), and he s udy was pe o med in acco dance wi h
he e hical s anda ds es ablished in he Decla a ion o Helsinki,
upda ed in Seoul in 2008.
Ma e ials and P ocedu e
The a ge i ems we e 50 colo pho og aphs o amous people
o he las 50 yea s (ac o s, singe s, poli icians, spo smen,
a s pe sonali ies, e c. om Spain and o he coun ies, see
TABLE 3 | Summa y o models compa ed o Feeling o Knowing.
Dependen a iable: Feeling o Knowing
Model 1 Model 2 Model 3
Age a baseline −0.017*** (0.006) −0.008 (0.006) −0.008 (0.006)
Vocabula y-
WAIS
0.015** (0.006) 0.010 (0.006) 0.010 (0.006)
E alua ion Time 0.013 (0.007) 0.011 (0.008)
MCI-S able −0.017 (0.018) −0.056 (0.046)
MCI-Wo sened −0.132*** (0.024) −0.118** (0.060)
E alua ion
Time ×MCI-
S able
0.020 (0.022)
E alua ion
Time ×MCI-
Wo sened
−0.008 (0.031)
In e cep 3.857***(0.006) 3.845***(0.015) 3.849*** (0.017)
Obse a ions 751 751 751
Log Likelihood −1,999.809 −1,981.637 −1,981.159
Akaike In . C i . 4,007.618 3,977.273 3,980.318
Bayesian In .
C i .
4,025.470 4,025.514 4,020.485
Bayes Fac o – 4806.453 0.0025
All models included andom e ec s o in e cep s and Age and Vocabula y a
baseline as co a ia es. Model 1 is he null mixed model (i.e., in e cep s and
co a ia es only); Model 2 is he mixed model wi h main e ec s; and Model 3 is he
mixed model wi h main e ec s and in e ac ions. Coe icien s and s anda d e o s
(in pa en heses). **p<0.05, ***p<0.01.
Supplemen a y Ma e ial 1) selec ed om a se o 70. They we e
p e iously p esen ed o a small con ol g oup o cogni i ely
unimpai ed use s (20 pe sons) o a li e-long lea ning associa ion
om San iago de Compos ela (ATEGAL) aged be ween 55 and
80 yea s. Final 50 pho og aphs co espond wi h hose ha
ob ained he highes punc ua ion in amilia i y and seman ic
in o ma ion (age, esidence, ma i al s a us. . . o he celeb i y), in
o de o maximize he p obabili y o ToT s a es.
The ToT p ocedu e included in he CompAS has been
desc ibed in de ail in a p e ious s udy (Juncos-Rabadán e al.,
2011). In b ie , he ToT p ocedu e consis ed o h ee asks: (i)
a naming ask; (ii) a ask o de e mine whe he he ToTs we e
posi i e (when he name on he ToT was indeed he co ec
name) o nega i e (when he name on he ToT was no he
a ge name); and (iii) a amilia i y ask, o assess he subjec i e
deg ee o knowledge ha each pa icipan decla ed ha ing abou
each celeb i y depic ed in he pho og aphs (see Figu e 1). In
he naming ask, 50 pho og aphs o celeb i ies we e p esen ed
sepa a ely on a sc een (wi h E-P ime o Windows). Pa icipan s
we e asked o p ess he g een key on a esponse box i hey knew
he name and he ed key i hey did no know he name. They
we e also asked o say he name ou loud o o say ei he “I don’
know he name” o “I can’ ecall he name a he momen ” a he
same ime as p essing he esponse key. The names and esponses
we e egis e ed as ollows: (a) co ec (CORs) o inco ec ,
acco ding o he accu acy o he name; (b) “Don’ know,” when
he pa icipan did no know he name; and (c) ToT s a e, when
he pa icipan said ha hey knew he name bu could no
ecall i a he momen . In he second phase, he pho og aphs
ha p oduced ToT esponses we e p esen ed in a second ask,
in which pa icipan s we e again asked o he celeb i y’s name.
I he pa icipan co ec ly p oduced he name du ing he ask,
he esponse was classi ied as a esol ed ToT. When he ToT
was main ained o an inco ec name was p oduced, pa icipan s
we e encou aged o answe se e al ques ions ha appea ed on he
sc een in o de o es hei knowledge abou he pe son and hei
name: ‘Wha is he pe son’s p o ession?’, ’Wha is he i s le e
o syllable o he name?’, ‘Does any name come o you mind?’.
A e hese ques ions we e sco ed, he a ge name was p esen ed
wi h wo non- a ge names. Fo each such iad, pa icipan s
we e asked o s a e which o he names p esen ed sepa a ely on
he sc een was he co ec name o he pe son in he p e iously
p esen ed pho og aph and i i was he name ha hey had been
ying o emembe when hey said “I know he name bu I can’
ecall i .” The ToT was hen classi ied as a posi i e ToT (pToT)
when pa icipan s co ec ly ecognized he a ge name and said
ha i was he name ha hey had been ying o emembe , and
nega i e ToT when hey ecognized i bu said ha i was no
name on hei mind. In he hi d phase, he 50 a ge pic u es
we e p esen ed o each pa icipan o de e mine how amilia he
amous people we e. Responses we e sco ed on a scale o 1–5
(whe e 5 ep esen s maximum amilia i y and 1, un amilia i y).
The ollowing measu es we e conside ed o he pu poses
o his s udy: (A) Familia i y, which ep esen s he subjec i e
knowledge ha pa icipan s had abou he people ep esen ed
in he a ge pic u es. This was calcula ed by summing he
amilia i y esponses o all 50 pho og aphs. (B) Feeling o
F on ie s in Psychology | www. on ie sin.o g 6Ma ch 2020 | Volume 11 | A icle 425
psyg-11-00425 Ma ch 13, 2020 Time: 17:38 # 7
Campos-Magdaleno e al. Longi udinal Pa e n o ToTs
FIGURE 3 | Es ima ed ma ginal means and e o s ba s om Model 2 o Feeling o Knowing in he h ee g oups ac oss he h ee e alua ion imes. SE, S anda d
E o ; BL, Baseline assessmen ; T1, Time 1 assessmen ; T2, Time 2 assessmen .
Knowing, which ep esen s he secu i y ha pa icipan s ha e
abou he knowing he name, independen ly o whe he he
name was ecalled o no (Schwa z, 2002). This was measu ed
as he numbe o imes ha pa icipan s p essed he g een key
in he naming ask. (C) Seman ic access was calcula ed by he
equa ion [(CORs +pToTs)/N] and ep esen s success ul access
o he seman ic ep esen a ions o he names. (D) Phonological
access was calcula ed by he equa ion [CORs/(CORs +pToTs)]
and ep esen s he p opo ion o bo h success ul seman ic and
phonological e ie als (Juncos-Rabadán e al., 2010).
In addi ion o hese ToT measu es, wo lexical measu es
we e conside ed: (A) Ve bal luency-animals (Seman ic luency),
de ined as he abili y o p oduce wo ds wi hin a ixed ime in e al
(Lezak e al., 2004) and conside ed sui able o de ec ing MCI
(Tale and Phillips, 2009), was used as a gene al measu e o
lexical access; and (B) To al sco ing in he ocabula y es o he
Wechsle Adul In elligence Scale (WAIS; Wechsle , 1988), used
o measu e he gene al e bal knowledge o he pa icipan .
S a is ical Analysis
Conside ing he he e ogenei y in he sample size o he g oups,
non-pa ame ic es s (e.g., K uskal–Wallis and Mann–Whi ney
es s) we e used o analyze be ween-g oup di e ences in socio-
demog aphic and ToT measu es a baseline. Complemen a ily,
pa ame ic es s we e included also analyzing be ween-g oup
di e ences. We ini ially selec ed Gene alized Linea Mixed
Models (GLMM) o modeling longi udinal changes in he
language, including andom in e cep s and andom slopes. Thus,
pa e ns o pe o mance in lexical access can be ep esen ed by
di e en slopes and longi udinal ajec o ies can be de ined by he
in e cep s. Howe e , and due o con e gence p oblems, andom
slopes we e excluded om he analyses.
We c ea ed he s a is ical models including E alua ion Time
(Baseline, Time 1, and Time 2), G oup (SCC-S able, MCI-
S able and MCI-Wo sened), and he in e ac ions (E alua ion
Time ×G oup) as independen a iables o p edic o s as ixed
e ec s. Pai wise compa isons o he es ima ed ma ginal means
o he dependen a iables E alua ion Time and G oup was
ca ied ou a e hey we e speci ied as ac o s. We included
he e oskedas ici y due o g oup, andom e ec s o in e cep s,
and he co a ia es age a baseline and p e iously s anda dized
ocabula y sco e in all models (see Supplemen a y Ma e ial 2).
Sepa a e models we e ob ained o each dependen a iable,
wi h SCC-S able as he e e ence g oup and Baseline assessmen
as he e e ence e alua ion ime. LMMs assuming a Gaussian
esponse we e used o modeling changes in p opo ional
measu es. GLMMs assuming Poissonian esponse we e selec ed
o coun ing measu es. When s a is ical assump ions (e.g.,
o e dispe sion o da a) we e no me in he GLMMs, a nega i e
binomial dis ibu ion was used o modeling coun da a.
Log Likelihood, Akaike and Bayesian In o ma ion C i e ia
indices o goodness o i we e used o selec he bes models
o each esponse. Thus, in o de o selec he bes model
o p edic ing he in e cep s and slopes in each g oup by he
ToT measu es, we i s compa ed all possible models including
ixed e ec s (i.e., E alua ion ime, G oup and hei in e ac ion)
F on ie s in Psychology | www. on ie sin.o g 7Ma ch 2020 | Volume 11 | A icle 425
psyg-11-00425 Ma ch 13, 2020 Time: 17:38 # 8
Campos-Magdaleno e al. Longi udinal Pa e n o ToTs
TABLE 4 | Summa y o models compa ed o Seman ic Access.
Dependen a iable: Seman ic Access pe cen age
Model 1 Model 2 Model 3
Age a baseline −1.798*** (0.609) −1.357** (0.629) −1.335** (0.629)
Vocabula y-WAIS 2.418*** (0.611) 1.943*** (0.632) 1.961*** (0.632)
E alua ion Time 2.234*** (0.265) 2.162*** (0.280)
MCI-S able −3.594 (1.835) −5.578**(2.504)
MCI-Wo sened −15.690*** (2.623) −11.971** (4.855)
E alua ion
Time ×MCI-S able
1.065 (0.913)
E alua ion
Time ×MCI-
Wo sened
−2.199 (2.403)
In e cep 91.063***(0.584) 87.996***(0.826) 88.129*** (0.843)
Obse a ions 641 641 641
Log Likelihood −2,273.151 −2,221.715 −2,217.955
Akaike In . C i . 4,560.302 4,463.430 4,459.910
Bayesian In . C i . 4,591.511 4,507.967 4,513.315
Bayes Fac o – 1.38479 ·1018 0.069
All models included andom e ec s o in e cep s and Age and Vocabula y a
baseline as co a ia es. Model 1 is he null mixed model (i.e. in e cep s and
co a ia es only); Model 2 is he mixed model wi h main e ec s; and Model 3 is he
mixed model wi h main e ec s and in e ac ions. Coe icien s and s anda d e o s
(in pa en heses). **p<0.05, ***p<0.01.
and including andom e ec s o no . A e op imizing he
s uc u e o he ixed and andom e ec s, we hen added
he e oskedas ici y (be ween-g oup a iabili y) o he model and
chose he bes - i model. Finally, we included s anda dized
co a ia es in he model o allow in e cep in e p e a ion (see
Supplemen a y Ma e ial 2 o ep oduce he s eps o ge hese
in e media e models as well as he inal eg ession models
he eby de ailed).
C oss-sec ional s a is ical analysis was pe o med wi h SPSS
o Windows, e sion 21.0 (SPSS, Chicago, IL, Uni ed S a es);
(G)LMMs we e es ima ed in R en i onmen ( e sion 3.5.3;
R Co e Team, 2019) wi h he nlme ( e sion 3.1-1137;
Pinhei o e al., 2018) and lme4 packages ( e sion 1.1-21;
Ba es e al., 2015).
RESULTS
Socio-demog aphic, neu opsychological and ToT measu es o
he g oups a baseline a e summa ized in Table 1. The MCI-
Wo sened g oup was he oldes , ollowed by MCI-S able. The
SCC-S able g oup ob ained highe sco es han he wo MCI
g oups a baseline o he cogni i e measu es, MiniMen al S a e
Examina ion (MMSE; Fols ein e al., 1975; Spanish e sion:
Lobo e al., 1999) and CLVT Sho (CVLT-SDFR) and Long
Delay F ee Recall (CVLT-LDFR), as well as o he TOT
measu es, seman ic access and phonological access and seman ic
luency. Familia i y was highe in he SCC-S able and MCI-
S able g oups han in he MCI-Wo sened g oup. Vocabula y
le el was highes o he SCC-S able han he o he wo g oups.
The lowes eeling o knowing was ob ained in he MCI-
Wo sened g oup. No di e ences we e ound a baseline in
FIGURE 4 | Es ima ed ma ginal means and e o s ba s om Model 2 o Seman ic Access in he h ee g oups ac oss he h ee e alua ion imes. SE, S anda d E o ;
BL, Baseline assessmen ; T1, Time 1 assessmen ; T2, Time 2 assessmen .
F on ie s in Psychology | www. on ie sin.o g 8Ma ch 2020 | Volume 11 | A icle 425
psyg-11-00425 Ma ch 13, 2020 Time: 17:38 # 9
Campos-Magdaleno e al. Longi udinal Pa e n o ToTs
TABLE 5 | Summa y o models compa ed o Phonological Access.
Dependen a iable: Phonological Access pe cen age
Model 1 Model 2 Model 3
Age a baseline −4.946*** (0.854) −3.769*** (0.835) −3.742***(0.835)
Vocabula y-WAIS 5.350*** (0.854) 4.301*** (0.831) 4.332*** (0.830)
E alua ion Time −0.958** (0.449) −1.059** (0.485)
MCI-S able −11.044 ***(2.573) −15.663***(4.171)
MCI-Wo sened −12.160*** (2.898) −10.437** (4.215)
E alua ion
Time ×MCI-S able
2.461 (01.749)
E alua ion
Time ×MCI-
Wo sened
−0.987 (1.728)
In e cep 78.651***(0.830) 82.874***(1.219) 83.061*** (1.265)
Obse a ions 637 637 637
Log Likelihood −2,402.963 −2,382.531 −2,378.374
Akaike In . C i . 4,823.925 4,789.062 4,784.747
Bayesian In . C i . 4,863.994 4,842.429 4,846.965
Bayes Fac o – 4.65930445815727e +47 0
All models included andom e ec s o in e cep s and slopes, he e oskedas ici y
due o he g oup, and Age and Vocabula y a baseline as co a ia es. Model 1 is
he null mixed model (i.e., in e cep s and co a ia es only); Model 2 is he mixed
model wi h main e ec s; and Model 3 is he mixed model wi h main e ec s
and in e ac ions. Coe icien s and s anda d e o s (in pa en heses). **p<0.05,
***p<0.01.
como bidi y, and di e ences in yea s o educa ion we e only
ob ained be ween he SCC S able and MCI Wo sened g oup in
he pa ame ic compa isons.
Familia i y
GLMMs conside ing no mal esponse (Gaussian) showed ha
he bes i model o Familia i y sco e (see Table 2) was model
2, which includes andom e ec s o he in e cep s and slopes,
and ixed e ec s o E alua ion Time [χ2(1) = 15.87; p<0.001]
and G oup [χ2(2) = 12.33; p<0.001] bu no he e ec o he
G oup ×E alua ion ime in e ac ion. Nei he o he co a ia es
(Age and WAIS- ocabula y sco e a baseline) o he E alua ion
Time ×G oup in e ac ion we e signi ican .
Acco ding o his model, he es ima ed means showed a
signi ican inc ease in amilia i y ac oss he e alua ion imes
in all g oups (p<0.001). Familia i y was signi ican ly lowe
o he MCI-Wo sened g oup han o he SCC-S able o he
MCI-S able g oups a any E alua ion ime (p<0.001). The
G oup ×E alua ion ime in e ac ion did no each signi icance,
showing ha be ween-g oup di e ences in amilia i y we e
main ained o e ime (Figu e 2).
Feeling o Knowing
We used GLMMs selec ing a Poisson model because o he
p esence o equi-dispe sion. Model 2 p o ided he bes i
(Table 3) and included andom e ec s o he in e cep s and ixed
e ec s only o E alua ion ime and G oup. Model 2 showed
a signi ican e ec only o G oup [χ2(2) = 31.27; p<0.001],
bu no o E alua ion ime o o he G oup ×E alua ion ime
in e ac ion. The co a ia es Age and WAIS- ocabula y sco e a
baseline did no each signi icance.
The es ima ed means model indica ed ha Feeling
o Knowing sco ing was signi ican ly lowe o he
FIGURE 5 | Es ima ed ma ginal means and e o s ba s om Model 2 o Phonological Access in he h ee g oups ac oss he h ee e alua ion imes. SE, S anda d
E o ; BL, Baseline assessmen ; T1, Time 1 assessmen ; T2, Time 2 assessmen .
F on ie s in Psychology | www. on ie sin.o g 9Ma ch 2020 | Volume 11 | A icle 425
