Is choice bias in delayed ma ching a consequence o in e - ial
in e al e ec s?
Ca a ina Soa es
1
, Mai iele San os, Ca los Pin o
*,2
Uni e si y o Minho, Po ugal
ARTICLE INFO
Da ase link: Choice biases in delayed
ma ching: is he in e - ial in e al always o
blame?
Keywo ds:
Tempo al disc imina ion
Ma ching o sample
Delay
In e - ial in e al
Pigeons
ABSTRACT
Despi e hei in ended goal o simply sepa a ing ials, in e - ial in e als ha e been ound o a ec choice
beha io in delayed ma ching- o-sample p ocedu es, leading o choice biases. In he p esen s udy, we assessed
he e ec s o delay on choice wi hou he po en ial in luence o in e - ial in e als. Six een pigeons lea ned a
symbolic ma ching- o-sample ask wi h samples di e ing in du a ion (3 s s. 9 s) and no in e - ial in e als. In
es ing, a delay las ing 2.5, 5, 10 o 20 s was in oduced be ween sample and compa isons. The e we e wo
di e ing delay condi ions: o G oup Ligh O , he delay was spen in da kness and, o G oup Ligh On, he
delay was illumina ed by a houseligh . Fo G oup Ligh O , as he delay inc eased, ma ching accu acy ollowing
bo h samples dec eased owa ds indi e ence. Fo G oup Ligh On, ma ching accu acy emained ela i ely high in
sho -sample ials bu dec eased ab up ly in long-sample ials (choose-sho e ec ). Hence, we ound ha a
choice bias may occu e en wi hou in e - ial in e als.
1. In oduc ion
In a symbolic ma ching- o-sample ask, a sample s imulus is p e-
sen ed ollowed by a choice be ween wo compa ison s imuli. Fo
ins ance, in an ope an chambe , a key migh be illumina ed wi h a
whi e hue o a sho e (3 s) o a longe du a ion (9 s), ollowed by he
p esen a ion o one ed key and one g een key. A esponse on he ed key
is ollowed by a ein o ce only a e a 3-s s imulus, and a esponse on
he g een key is ollowed by a ein o ce only ollowing a 9-s s imulus.
Gi en i s associa ion wi h he sho sample, he ed key migh be
designa ed as he “sho ” compa ison, while he g een key migh be
designa ed as he “long” compa ison. Such a p ocedu e, some imes wi h
a delay be ween sample o se and compa ison onse (delayed symbolic
ma ching) has been widely used o s udy memo y, choice, s imulus
con ol, o coding s a egies ( o a e iew, see Zen all and Smi h, 2016).
Typically, ials a e sepa a ed by an in e - ial in e al (ITI) o p e-
en p oac i e in e e ence. Indeed, pigeons’ accu acy in delayed
ma ching- o-sample asks has been ound o imp o e wi h ITI du a ion
(e.g., Edhouse andWhi e, 1988; G an , 1975; Maki e al., 1977; Robe s,
1980; Robe s and K aeme , 1982). Howe e , despi e hei appa en
bene i s, when empo al samples a e used (i.e., samples ha di e in
du a ion, as in he opening example), ITIs may also b ing abou o he ,
a guably undesi able, e ec s.
Fo ins ance, when ITI and delay in e als a e signaled simila ly (e.
g., bo h spen in da kness), pigeons ha e shown a p e e ence o he
“sho ” compa ison, i espec i e o he sample p esen ed (choose-sho
e ec ; e.g., Kelly and Spe ch, 2000; K aeme e al., 1985; Pin o and
Machado, 2011; Spe ch, 1987; Spe ch and G an , 1993; Spe ch and
Wilkie, 1982). Whe eas his choose-sho e ec could be amed as a
memo y- ela ed bias, some au ho s sugges ed i esul s om con usion
be ween delay and ITI (e.g., Fe e man and MacEwen, 1989; She bu ne
e al., 1998; Zen all, 1997; 2006): he delay may be in e p e ed as a
ial-cancelling ITI and hus, when he compa isons become a ailable
ollowing a delay, he animals espond as i no sample had been p e-
sen ed. I , by s imulus gene aliza ion, he absence o a sample is hen
conside ed o be close o he sho han o he long sample, a p e e ence
o he “sho ” compa ison is expec ed. In ac , in some cases, signaling
ITIs and delays di e en ly has been ound o elimina e choice biases
(Kelly and Spe ch, 2000; She bu ne e al., 1998; Spe ch and Rusak,
1992); howe e , in o he cases a p e e ence o “sho ” subsis ed (Kelly
and Spe ch, 2000; Pin o and Machado, 2011; Pin o and Sousa, 2021).
Thus, while simila ITIs and delays consis en ly lead o a p e e ence o
* Co espondence o: Uni e si y o Minho, School o Psychology, Gual a Campus, B aga 4710-057, Po ugal.
E-mail add ess: [email p o ec ed] (C. Pin o).
1
h ps://o cid.o g/0000-0001-5680-7362
2
h ps://o cid.o g/0000-0002-4435-6880
Con en s lis s a ailable a ScienceDi ec
Beha iou al P ocesses
jou nal homepage: www.else ie .com/loca e/beha p oc
h ps://doi.o g/10.1016/j.bep oc.2025.105153
Recei ed 24 Augus 2024; Recei ed in e ised o m 28 Janua y 2025; Accep ed 30 Janua y 2025
Beha iou al P ocesses 226 (2025) 105153
A ailable online 1 Feb ua y 2025
0376-6357/© 2025 The Au ho (s). Published by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license ( h p://c ea i ecommons.o g/licenses/by-
nc-nd/4.0/ ).
he “sho ” compa ison, di e en ia ing hese in e als appea s o
weaken his e ec – simila i y wi h delays is an example o how ITIs may
a ec ma ching pe o mance.
Spe ch and Rusak (1989) ained pigeons o disc imina e be ween 2
and 10-s samples in a delayed symbolic ma ching- o-sample ask wi h
45-s ITIs and 10-s delays. In es ing, hey a ied he ITI du a ion: In some
sessions, ITIs we e 5 o 90-s long and, in o he sessions, 15 o 75-s long.
Rega dless o he sample du a ion, when he ITI was sho e han in
aining (i.e., 5 o 15 s), pigeons showed a p e e ence o he “long”
compa ison (choose-long e ec ). When he ITI was longe han in
aining (i.e., 75 o 90 s), he e was a p e e ence o he “sho ” com-
pa ison (choose-sho e ec ). Thus, pigeons’ pe o mance was a ec ed
by ITI du a ion. Spe ch and Rusak (1989, 1992) sugges ed ha he an-
imals judged he sample du a ion ela i e o a empo al backg ound
composed by he ITI and he delay – a sample would seem sho e in he
con ex o a leng hy ITI and longe in he con ex o a sho ITI ( ela i e
du a ion hypo hesis). Once mo e, he ITI e eals i sel o be mo e han a
simple ial sepa a o .
Besides in luencing pe o mance unde delayed choice, he ITI i sel
may be used o lea n a ask, a he expense o o he s imuli. Fo ins ance,
Pin o and Machado (2015, 2017) ained pigeons in a
ma ching- o-sample ask wi h h ee sample du a ions and wo compa -
isons; one compa ison was co ec ollowing one sample and he o he
compa ison was co ec ollowing he o he wo samples. A 30-s ITI
illumina ed by a houseligh sepa a ed ials. The houseligh o se
signaled he end o he ITI and coincided wi h sample onse . Thus, pi-
geons could ha e igno ed he sample and used he houseligh as a ime
ma ke o lea n he ask. When delays we e in oduced (which leng h-
ened he in e al be ween ITI o se and compa ison onse ), pigeons
showed a p e e ence o he compa ison associa ed wi h he long sam-
ple. In es s whe e no sample was p esen ed, p e e ence o he com-
pa ison associa ed wi h he long sample inc eased wi h ime elapsed
since he ITI. Taken oge he , hese esul s sugges ha pigeons we e
using ITI o se (signaled by he houseligh ) as a ime ma ke .
All a o emen ioned ins ances o choice biases occu ed in asks wi h
ITIs. No only ha , bu ITIs ha e been equen ly in oked o explain
hese biases. In he p esen s udy we ask whe he choice biases occu
only because o ITIs. So, pigeons lea ned a empo al ma ching- o-sample
ask wi hou ITIs. Ou goal was o assess he e ec o delays on ma ching
pe o mance wi hou he po en ial in e e ence o ITI e ec s. Addi-
ionally, in asks wi h ITIs, delay illumina ion appea s o a ec pigeons’
choices (e.g., Do ance e al., 2000; Spe ch and Rusak, 1989; Kelly and
Spe ch, 2000). To explo e he po en ial e ec o delay illumina ion in a
ask wi hou ITIs, we included wo delay condi ions: Fo hal o he pi-
geons, he delays we e spen in da kness, and o he o he hal he
delays we e houseligh -illumina ed.
2. Me hod
2.1. Subjec s
Six een pigeons (Columba li ia), main ained a 85 % o hei ee-
eeding body weigh , pa icipa ed in he expe imen . The bi ds we e
housed indi idually in a colony oom wi h a ligh /da k cycle o 13:11 h
(ligh s on a 8 AM) and con olled empe a u e (be ween 20 and 22ºC).
All pigeons had wa e and g i always a ailable. The expe imen was
conduc ed once a day, a app oxima ely he same ime o each pigeon,
six days a week. All pigeons had expe ience wi h ma ching- o-sample,
choice, and empo al disc imina ion asks.
2.2. Appa a us
Fou ope an chambe s we e used: h ee Lehigh Valley Elec onics
(LVE) chambe s and a homemade chambe . The LVE chambe s
measu ed 340 mm high by 350 mm long by 310 mm wide and we e
equipped wi h an exhaus an ha ci cula ed ai and masked ou side
noise. The on panel con ained h ee ci cula esponse keys (25 mm in
diame e ), ho izon ally aligned, 90 mm apa om each o he (cen e -
o-cen e ), loca ed 225 mm abo e he wi e mesh loo . The keys we e
equipped wi h an Indus ial Elec onics Enginee s (IEE) 12-s imulus in-
line p ojec o and each s imulus was illumina ed wi h a 28-V, 0.1-A ligh
bulb. On he back panel, 300 mm abo e he loo , a houseligh (28 V,
0.1 A) p o ided gene al illumina ion o he chambe . Mixed g ain se ed
as ein o cemen and was deli e ed h ough a LVE eede h ough an
opening measu ing 60 mm wide by 50 mm high, cen e ed ho izon ally
on he esponse panel, 85 mm abo e he loo . When he eede was
aised o allow access o ein o cemen , a 28-V, 0.04-A ligh illumina ed
he opening.
The homemade chambe measu ed 319 mm high by 330 mm long by
330 mm wide and was placed inside a PVC sound a enua ion cubicle
(Med Associa es, ENV-018 V) equipped wi h an exhaus an. In he on
panel, 210 mm abo e he wi e mesh loo , he e we e h ee esponse
keys, 25 mm in diame e , ho izon ally aligned, 90 mm apa (cen e - o-
cen e ). S imuli we e p esen ed h ough a 12-s imulus IEE in-line p o-
jec o and illumina ed by a 28-V, 0.1-A lamp. Cen e ed ho izon ally on
he esponse panel, 65 mm abo e he loo , an opening (60-mm wide x
45-mm high) p o ided access o ein o cemen . When ood was a ail-
able, he eede opening was illumina ed by a 28-V, 0.04-A ligh . On he
opposi e wall, 275 mm abo e he loo , he e was a 28-V, 0.1-A ligh
ix u e. A compu e wi h he ABET II so wa e (La aye e Ins umen
Company) con olled he expe imen and eco ded he da a.
2.3. P ocedu e
2.3.1. T aining
All pigeons we e ained in a symbolic ma ching- o-sample ask ha
ea u ed wo samples and wo compa isons. A ial s a ed wi h he
illumina ion o he cen e key wi h a whi e hue o ei he 3 s (sho
sample) o 9 s (long sample). A e he sample du a ion elapsed, he
cen al key was u ned o and he wo side keys we e illumina ed, one
wi h a ed hue and he o he wi h a g een hue (compa isons). One
compa ison was co ec ollowing he 3-s sample and he o he com-
pa ison was co ec ollowing he 9-s sample. A peck on ei he key,
u ned bo h compa isons o . When a esponse was co ec , a ein o ce
(mixed g ain) was deli e ed. A co ec ion p ocedu e was in e ec : When
a esponse was inco ec , he ial epea ed un il a esponse was co ec ,
up o a maximum o h ee epe i ions. A e h ee consecu i e inco ec
esponses, only he co ec compa ison was p esen ed. Following a
co ec esponse, a ein o ce was deli e ed and he session p og essed
o a new ial.
Excluding co ec ion ials, each session consis ed o 60 ials di ided
in h ee 20- ial blocks. Each block con ained en ials o each sample
du a ion, andomly dis ibu ed. The loca ion o he compa isons was
coun e balanced wi hin each block o ensu e ha each compa ison was
p esen ed he same numbe o imes in each side key. The di ision o he
session in blocks ensu ed a mo e balanced dis ibu ion o ials
h oughou he session. The colo associa ed wi h each compa ison was
coun e balanced ac oss pigeons: o hal o he pigeons, ed was co ec
ollowing a sho sample and g een ollowing a long sample, whe eas o
he o he hal i was he o he way a ound.
T aining las ed a minimum o 15 sessions and con inued un il he
pe cen age o co ec esponses ollowing each sample was a leas 80 %
wi hin a session (excluding co ec ion ials), o h ee consecu i e
sessions. To minimize eeding ou side he expe imen al sessions, he
du a ion o he ein o cemen was adjus ed indi idually, anging om
1.5 o 4 s.
2.3.2. Delay es
Each session consis ed o 64 ials di ided in wo 32- ial blocks.
Each block comp ised 24 aining ials (12 sho -sample ials and 12
long-sample ials) and eigh delay- es ials ( ou o each sample).
T aining ials we e simila o hose o he aining phase and he
C. Soa es e al.
Beha iou al P ocesses 226 (2025) 105153
2
co ec ion p ocedu e was also main ained o hese ials. On delay- es
ials, a delay was in oduced be ween sample o se and compa isons
onse . The delay could las 2.5, 5, 10 o 20 s. Fo each sample, he e we e
eigh delay- es ials – wo ials pe delay du a ion – and ein o cemen
was deli e ed on hal o hose, non-di e en ially. The e was no
co ec ion p ocedu e on es ials. Rein o cemen du a ion was indi-
idually adjus ed and a ied be ween 1.5 and 4 s. Tes ing las ed o i e
sessions.
Pigeons we e di ided in wo g oups: Fo hal o he pigeons (G oup
Ligh O ), he delays we e spen in he da kness, and o he o he hal
(G oup Ligh On), he delays we e illumina ed by he houseligh .
2.4. Da a analysis
We analyzed ma ching accu acy o each sample in he las h ee
sessions o aining and in he i e sessions o delay es ing. S a is ical
analyses we e conduc ed using Mic oso Excel and jamo i o Windows
(Ve sion 2.3). Type-1 e o a e was se a 0.05. Fo he delay es s, we
conduc ed F iedman es s o assess he e ec o delays on accu acy in
each condi ion. Condi ions we e compa ed ia an adjus ed ank ans-
o m es (Leys and Schumann, 2010). A K uskal-Wallis es compa ed
speed o acquisi ion be ween he p esen s udy and an expe imen ha
employed in e - ial in e als. Fo each pigeon, o compa e choices
ollowing delays wi h chance le els, we calcula ed 95 % con idence
in e als (No mal app oxima ion o he binomial) o choices o he
“sho ” compa ison.
3. Resul s
3.1. T aining
On a e age, pigeons om G oup Ligh O ook 27 ( ange: 15–42)
sessions o comple e aining. On he las h ee sessions o aining,
ma ching accu acy was 89 % o 3-s samples ( ange: 80–100 %) and
90 % o 9-s samples ( ange: 80–100 %). Pigeons om G oup Ligh On
ook, on a e age, 23 ( ange: 15–39) sessions o comple e aining. One
pigeon ailed o each he lea ning c i e ia a e 60 sessions and did no
p og ess o es ing. Fo he se en pigeons ha did ad ance o es ing,
ma ching accu acy on he las h ee sessions o aining was 92 % o 3-s
samples ( ange: 83–100 %) and 90 % o 9-s samples ( ange: 80–100 %).
3.2. Delay es
The uppe panel o Fig. 1 depic s ma ching accu acy o each sample
as a unc ion o delay du a ion o G oup Ligh O . A F iedman es
e ealed no signi ican di e ence in ma ching accu acy ollowing 3-s
and 9-s samples,
χ
²(1) =0.11, p =.739. Tha is, he pe cen age o co -
ec esponses signi ican ly dec eased wi h delay o bo h samples, o-
wa ds chance le el.
To cha ac e ize he da a a he indi idual le el, a 95 % Con idence
In e al (CI) o choices o he “sho ” compa ison on delayed ials was
calcula ed o each pigeon. Hal o he pigeons a e well desc ibed by he
a e age unc ion: Ma ching accu acy dec eased wi h he delays, wi h no
clea p e e ence o any compa ison, and choice app oached chance
le els – he CIs included 50 %; [35.3 %, 57.2 %], [37.8 %, 59.7 %],
[40.3 %, 62.2 %], and [46.7 %, 68.3 %]. Fo he emaining pigeons, wo
showed a p e e ence o he “sho ” compa ison, signi ican ly abo e
chance, 95 % CI [61.3 %, 81.2 %] and [72.7 %, 89.8 %]. The las wo
pigeons showed he opposi e bias, and p e e ed he “long” compa ison,
95 % CI [28.1 %, 49.4 %] and [15.5 %, 34.5 %]. As a g oup, he e was
no clea p e e ence o one o he op ions, and he end owa ds
indi e ence shown by he a e age unc ion e lec s ha .
The bo om panel o Fig. 1 shows, o G oup Ligh On, he pe cen age
o co ec esponses ollowing each sample in he delay es . A F iedman
es e ealed a signi ican di e ence in ma ching accu acy ollowing 3-s
and 9-s samples,
χ
²(1) =8.00, p=.005, showing ha he pe cen age o
co ec esponses changed di e en ly o he wo samples – o 3-s
samples accu acy dec eased e y slowly, bu o 9-s samples accu acy
dec eased s eeply when a ial had a delay.
The asymme y be ween samples was con i med a he indi idual
le el, wi h ou ou o he se en pigeons showing a s a is ically signi i-
can p e e ence o he “sho ” compa ison, 95 % CI [53.2 %, 74.3 %],
[71.2 %, 88.8 %], [74.2 %, 90.8 %], and [86.7 %, 98.3 %]. The
emaining h ee pigeons did no clea ly p e e any compa ison, and
choices did no di e signi ican ly om chance le els, 95 % CI [34.1 %,
55.9 %], [35.3 %, 57.2 %], and [42.8 %, 64.7 %].
The di e ence be ween G oup Ligh O and G oup Ligh On was
con i med by an adjus ed ank ans o m es ha e ealed a signi ican
in e ac ion be ween delay illumina ion and accu acy ollowing each
sample, F(1, 146) =7.41, p=.007,
η
2=0.05. In o he wo ds, he
impac o delays on ma ching accu acy di e ed when delays we e illu-
mina ed o spen in da kness.
4. Discussion
In e - ial in e als (ITI) a e commonly employed o p e en in e -
ial in e e ence. Howe e , he impac o ITIs can go beyond ha sim-
ple pu pose, a ec ing pe o mance in unexpec ed ways. Fo ins ance,
how an ITI is signaled, o i s du a ion, can lead o choice biases. As such,
ITIs can complica e ou unde s anding o pe o mance in s anda d asks.
Gi en he innume ous ways an ITI may a ec delayed ma ching, we
in es iga ed whe he choice biases we e exclusi ely a by-p oduc o ITIs,
by assessing choices in delayed empo al ma ching wi hou he in e -
e ence o ITI e ec s. To do so, pigeons we e ained in a empo al
symbolic ma ching- o-sample ask (3 s s. 9 s) wi hou ITIs. Then, a
delay was in oduced be ween sample o se and compa isons onse .
This delay could be ei he spen in da kness o houseligh -illumina ed.
Gi en ha in e - ial in e e ence is mo e likely wi hou ITIs, he
p esen ask could be mo e di icul o lea n. To ha e a sense o how he
absence o ITIs migh ha e a ec ed acquisi ion – and gi en ha in his
s udy he e was no condi ion wi h ITIs – we compa ed he p esen esul s
Fig. 1. Mean pe cen age co ec esponses (wi h s anda d e o o he mean) as
a unc ion o delay du a ion, ollowing 3-s (emp y da a poin s) and 9-s ( illed
da a poin s) samples, o G oup Ligh O (uppe panel) and G oup Ligh On
(bo om panel).
C. Soa es e al.
Beha iou al P ocesses 226 (2025) 105153
3
wi h ano he expe imen om ou labo a o y (Pin o and Cas anhei a
Dinis, 2024). Tha expe imen used he same s imuli o signal he sample
and compa isons, and e en hough sample du a ions we e no he same
(2 and 6 s e sus 3 and 9 s in he p esen s udy), hey had he same 1:3
a io, so he disc imina ions should be equi alen (Webe ’s law; see, e.g.,
Gibbon, 1977). T ials we e sepa a ed by a 30-s, houseligh -illumina ed
ITI. Session leng h was simila be ween he wo s udies (64 ials
e sus 60 ials in he p esen s udy). In he p esen expe imen , aining
could be comple ed a e a minimum o 15 sessions, so Fig. 2 compa es
how o e all ma ching accu acy e ol ed in he i s 15 sessions o
aining in bo h expe imen s.
The ask wi h ITIs had consis en ly highe accu acy, and also seem-
ingly as e acquisi ion. To complemen he igu e, we compa ed he
numbe o sessions needed o a pigeon o i s each 80 % co ec o
bo h samples simul aneously. The ask wi h ITIs equi ed an a e age o
9.8 sessions ( ange: 5–18), app oxima ely hal he numbe o sessions
equi ed o he ask wi hou ITIs, 18.4 ( ange: 6–40). This di e ence
was s a is ically signi ican ,
χ
²(1) =5.30, p=.021. Despi e ha , by he
end o 15 sessions he wo asks showed simila ma ching accu acies. I
is in e es ing o no e ha in he wo expe imen s he e we e pigeons
lea ning he ask simila ly as , wi h he di e ence be ween expe imen s
s emming om how slow he slowe bi ds we e. In conclusion, pigeons
a e able o lea n he disc imina ion in he absence o ITIs, albei aking
longe han i he ask had included ITIs.
Rega ding he delay es s, he wo illumina ion condi ions yielded
di e en esul s. When he delays we e spen in da kness, ma ching
accu acy o bo h samples dec eased g adually wi h delay, app oaching
indi e ence. Tha is, choices ollowed a ypical e en ion cu e – he
longe he delay, he less likely an animal was o emembe he sample
and hus, andom choices became mo e likely. A simila pa e n is ound
in non- empo al disc imina ions, namely ea u ing isual s imuli, such
as colo s o shapes (e.g., Nelson and Wasse man, 1978; Robe s, 1972;
Robe s and G an , 1978). The e o e, wi hou he in e e ence o he ITI,
delay es ing o empo al disc imina ions appea s simila o o he
s imulus modali ies.
Howe e , when he delays we e illumina ed, esul s we e qui e
di e en . While accu acy emained ela i ely high o 3-s samples, i
dec eased ab up ly o 9-s samples: he e was an o e all p e e ence o
he compa ison associa ed wi h he sho sample (choose-sho e ec ).
In addi ion, he ac ha ma ching accu acy ollowing 9-s samples was
simila ly low o all delays (Fig. 1, bo om panel, illed da a poin s)
sugges s ha he in oduc ion o an illumina ed delay caused a gene -
alized dis up ion in pe o mance. Tha is, he choice bias is likely due o
a p ocess un ela ed o o ge ing (in case o o ge ing, he e ec o he
delay would be expec ed o become mo e p onounced as delay
inc eased).
The possibili y ha delay illumina ion is dis up i e has been
co obo a ed by s udies ea u ing isual disc imina ions, which ound
an ab up dec ease in ma ching accu acy when a delay was illumina ed
compa ed o a g adual and smalle dec ease when a delay was spen in
da kness (e.g., Calde and Whi e, 2014; Case e al., 2015; G an and
Robe , 1976; Ha pe and Whi e, 1997; Maki e al., 1977; Robe s and
G an , 1978; Whi e, 1985). To illus a e, Robe s and G an (1978, Exp.
1) ound ha , when delays we e spen in da kness, ma ching accu acy
p og essi ely dec eased wi h delay and emained abo e 75 % e en a
he longes delay (12 s). Ye , when delays we e illumina ed, accu acy
dec eased ab up ly ( o a ound 75 %) wi h a delay as sho as 0.5 s, and
app oached indi e ence a he longes delay. The impac o illumina ion
was clea ly con i med by G an and Robe (1976), who, by a ying he
in ensi y o he delay illumina ion, ound ha ma ching accu acy p o-
g essi ely dec eased as ligh in ensi y inc eased. A possible explana ion
o hese esul s is ha s imuli p esen ed du ing a delay (such as u ning
on a ligh o illumina e he chambe ) may compe e o he animals’
a en ion, inc easing he likelihood ha hey disengage om he ask,
hus educing accu acy.
Taking hese s udies in o accoun , ou indings sugges ha , when
he e a e no ITIs, he deg ee o dis up ion caused by he in oduc ion o a
delay may lead o quali a i ely di e en ou comes. A less dis up i e
delay (spen in da kness) yields a ypical o ge ing unc ion, while a
mo e dis up i e delay (illumina ed) esul s in a ma ked p e e ence o
one o he op ions. In e es ingly, illumina ed delays do no always lead
o a “sho ” bias. As desc ibed in he In oduc ion, simila i y be ween ITI
and delay may a ec delayed choices; when ha po en ial con ound was
con olled o by ha ing da k ITIs and illumina ed delays, a endency
owa ds indi e ence has been obse ed o bo h samples (Kelly and
Spe ch, 2000; She bu ne e al., 1998; Spe ch and Rusak, 1992). The
disc epancy be ween hose s udies and he p esen da a may sugges
ha , e en when ITI-delay con usion is con olled o , he ITI may s ill be
a ec ing choices.
Despi e o e all g oup di e ences, he e we e some simila i ies a he
indi idual le el be ween he wo illumina ion condi ions. In bo h da k
and illumina ed delay es s, some pigeons showed no bias and some
showed a bias o “sho ”. Di e ences in delay- es pe o mance may be
due o di e ences in s imulus con ol, o in he way he ask was lea ned.
These di e ences could also be e lec ed in he speed o acquisi ion. Fo
ins ance, ins ead o es ablishing wo esponse ules – “i 3 s, choose ed”
and “i 9 s, choose g een” – pigeons may c ea e a speci ic esponse ule
o one sample and a gene al esponse o all o he cases – “i 3 s, choose
ed” and “i no 3 s, choose ed” (see single-code/de aul ; e.g., Pin o and
Machado, 2015; Singe , e al., 2006). The la e coding s a egy could
esul in as e lea ning and in a choice bias in delay es ing. Ac oss bo h
condi ions, pigeons wi h a “sho ” bias in es ing ook an a e age o 18
( ange: 10–31) sessions o each he lea ning c i e ion. Pigeons ha
showed no bias we e sligh ly slowe o lea n he ask, aking on a e age
26.7 ( ange: 16–41) sessions. This is a pos -hoc hypo hesis, ocused on a
subse o ou animals, so i should be aken conse a i ely. Howe e , i
may wa an u he explo a ion.
The momen o in oduc ion o he delays also appea s o a ec how
dis up i e hey can be. When delays a e in oduced only in es ing,
ab up dec eases in ma ching accu acy a e commonly ound, wi h
e en ion unc ions app oaching a s ep-like pa e n (e.g., G an and
Tala ico, 2004; Pin o and Machado, 2015; 2017; Pin o and Sousa, 2021;
She bu ne e al., 1998; Spe ch, 1987; Spe ch and Rusak, 1992).
Con e sely, when delays a e p esen since he beginning o aining (and
o he delay du a ions a e in oduced in es ing), e en ion unc ions
show a g adual dec ease as a unc ion o delay du a ion (e.g. Do ance
e al., 2000; Kelly and Spe ch, 2000; Spe ch and Rusak, 1989). This may
happen because in oducing delays only du ing es ing may induce
g ea e gene aliza ion dec emen and dis up ion o s imulus con ol
(ambigui y hypo hesis; Zen all, 1997, 2006). In ha ein, i would be
in e es ing o conduc a a ia ion o he p esen expe imen in which
pigeons a e ini ially ained wi h a delay and hen es ed wi h o he
Fig. 2. A e age ma ching accu acy o he i s 15 sessions o aining, in he
p esen expe imen (emp y da a poin s) and in a simila p ocedu e ha ea u ed
in e - ial in e als ( illed da a poin s) (Pin o and Cas anhei a Dinis, 2024).
C. Soa es e al.
Beha iou al P ocesses 226 (2025) 105153
4
delay du a ions. Tha expe imen would con ol o wo impo an as-
pec s ha seem o a ec pigeons’ pe o mance on delay asks, he in-
luence o ITIs and he no el y o he delays.
Ou esul s sugges ha choice biases occu when es ing induces
signi ican dis up ion o he disc imina ion lea ned. A simila esul has
been desc ibed by Wa d and Odum (2007), who ained pigeons in a
empo al bisec ion ask ha ea u ed eigh sample du a ions ( anging
om 2 o 8 s); one compa ison was co ec ollowing he ou sho e
du a ions, and he o he compa ison was co ec o he ou longe
du a ions. The e we e no delays be ween sample and compa isons. The
impac o se e al dis up o s was hen assessed; in one manipula ion,
ood was deli e ed du ing he ITI, and pigeons’ accu acy dec eased
mainly ollowing long samples. Thus, e en in he absence o delays,
choice biases can eme ge as a esul o ask dis up ions.
Some explana ions o he choose-sho e ec ha e sugges ed ha a
delay, ei he h ough a g adual (Spe ch and Wilkie, 1983) o sudden
(K aeme e al., 1985) p ocess, lea es he animal in a s a e simila o as i
no sample was p esen ed o begin wi h. The absence o a sample (a 0-s
sample) would lead, by s imulus gene aliza ion, o a p e e ence o he
compa ison associa ed wi h he sho es sample. O he accoun s (Gai an
and Wix ed, 2000; Wix ed and Gai an, 2004) sugges ed ha animals
lea n he disc imina ion on he basis o he p esence o absence o he
mos salien s imulus (in his case, he long sample). Then, he dis up-
ion b ough by a delay would mean ha long samples would be
pe cei ed as absen – leading o choices o he “sho ” compa ison.
These explana ions do no seem able o accommoda e he esul s o he
p esen s udy, whe e delays led o a choice bias in only one o ou wo
delay- es condi ions.
In closing, he ac ha we ound a choice bias in a ask wi h no ITIs
e eals ha o he mechanisms besides how ITI and delay a e signaled, o
hei ela i e du a ions, can lead animals o p e e one compa ison o e
ano he . Addi ionally, he p esen esul s sugges ha , e en hough
he e may be many ways ha s imulus con ol by he sample can be
weakened – such as due o memo y- ela ed p ocesses (such as ollowing
a long span o ime), o due o he in oduc ion o a dis up i e e en –
hey do no necessa ily lead o he same ou comes. Di e en ypes o
dis up ion may yield di e en esponses.
E en hough he absence o ITIs may make a ask mo e di icul o
lea n, i does no do so o an insu moun able deg ee. The e o e, gi en
he unin ended in luence ITIs ha e on pe o mance, u he explo a ion
o asks wi hou ITIs may p o e ui ul, as he p esen esul s sugges .
Compliance wi h e hical s anda ds
The esea ch was ca ied ou in ag eemen wi h he Eu opean
(Di ec i e 2010/63/EU) and Po uguese law (O dinance 1005/92 o
Oc obe 23), being app o ed by he Di ec o a e-Gene al o Food and
Ve e ina y, he Po uguese na ional au ho i y o animal heal h
(Au ho iza ion #024946).
CRediT au ho ship con ibu ion s a emen
Soa es Ca a ina: W i ing – e iew & edi ing, W i ing – o iginal
d a , Visualiza ion, So wa e, In es iga ion, Fo mal analysis, Da a
cu a ion. San os Mai iele: W i ing – o iginal d a , In es iga ion. Pin o
Ca los: W i ing – e iew & edi ing, Visualiza ion, Supe ision, So -
wa e, Resou ces, P ojec adminis a ion, Me hodology, In es iga ion,
Fo mal analysis, Da a cu a ion, Concep ualiza ion.
Decla a ion o Compe ing In e es
The au ho s decla e ha hey ha e no con lic o in e es .
Acknowledgmen s
Funding: The p esen wo k was conduc ed a he Psychology
Resea ch Cen e (PSI/01662), School o Psychology, Uni e si y o
Minho, suppo ed by he Founda ion o Science and Technology (FCT)
h ough he Po uguese S a e Budge (Re .: UIDB/PSI/01662/2020).
Da a a ailabili y
Da a and code a e a ailable in he Da aReposi o iUM eposi o y,
h ps://doi.o g/10.34622/da a eposi o ium/U01CBM
Choice biases in delayed ma ching: is he in e - ial in e al always
o blame? (Da aReposi o iUM)
Re e ences
Calde , A., Whi e, K.G., 2014. In sea ch o consolida ion o sho - e m memo y in
nonhuman animals. Lea n. Beha . 42, 83–92. h ps://doi.o g/10.3758/s13420-013-
0127-5.
Case, J.P., Laude, J.R., Zen all, T.R., 2015. Delayed ma ching o sample in pigeons:
e ec s o delay o ein o cemen and illumina ed delays. Lea n. Mo i . 49, 51–59.
h ps://doi.o g/10.1016/j.lmo .2015.01.001.
Do ance, B.R., Kaise , D.H., Zen all, T.R., 2000. E en -du a ion disc imina ion by
pigeons: he choose-sho e ec may esul om e en ion- es no el y. Anim. Lea n.
Beha . 28, 344–353. h ps://doi.o g/10.3758/BF03200268.
Edhouse, W.V., Whi e, K.G., 1988. Sou ces o p oac i e in e e ence in animal memo y.
J. Exp. Psychol. Anim. Beha . P ocess. 14, 56–70. h ps://doi.o g/10.1037/0097-
7403.14.1.56.
Fe e man, J.G., MacEwen, D., 1989. Sho - e m memo y o esponses: he “choose-
small” e ec . J. Exp. Anal. Beha . 52, 311–324. h ps://doi.o g/10.1901/
jeab.1989.52-311.
Gai an, S., Wix ed, J., 2000. The ole o “no hing” in memo y o e en du a ion in
pigeons. Anim. Lea n. Beha . 28, 147–161. h ps://doi.o g/10.3758/BF03200250.
Gibbon, J., 1977. Scala expec ancy heo y and Webe ’s law in animal iming. Psychol.
Re . 84, 279–325. h ps://doi.o g/10.1037/0033-295X.84.3.279.
G an , D.S., 1975. P oac i e in e e ence in pigeon sho - e m memo y. J. Exp. Psychol.
Anim. Beha . P ocess. 1, 207–220. h ps://doi.o g/10.1037/0097-7403.1.3.207.
G an , D.S., Robe , W.A., 1976. Sou ces o e oac i e inhibi ion in pigeon sho - e m
memo y. J. Exp. Psychol. Anim. Beha . P ocess. 2, 1–16. h ps://doi.o g/10.1037/
0097-7403.2.1.1.
G an , D.S., Tala ico, D.C., 2004. P ocessing o emp y and illed ime in e als in pigeons.
Anim. Lea n. Beha . 32, 477–490. h ps://doi.o g/10.3758/BF03196043.
Ha pe , D.N., Whi e, K.G., 1997. Re oac i e in e e ence and a e o o ge ing in
delayed ma ching- o-sample pe o mance. Anim. Lea n. Beha . 25, 158–164.
h ps://doi.o g/10.3758/BF03199053.
Kelly, R., Spe ch, M.L., 2000. Choice biases in delayed ma ching- o-sample du a ion wi h
pigeons: manipula ions o ITI and delay illumina ion. Q. J. Exp. Psychol. 53B,
309–323. h ps://doi.o g/10.1080/713932737.
K aeme , P.J., Mazmanian, D.S., Robe s, W.A., 1985. The choose-sho e ec in pigeon
memo y o s imulus du a ion: subjec i e sho ening e sus coding models. Anim.
Lea n. Beha . 13, 349–354. h ps://doi.o g/10.3758/BF03208009.
Leys, C., Schumann, S., 2010. A nonpa ame ic me hod o analyze in e ac ions: he
adjus ed ank ans o m es . J. Exp. Soc. Psychol. 46, 684–688. h ps://doi.o g/
10.1016/j.jesp.2010.02.007.
Maki, W.S., Moe, J.C., Bie ley, C.M., 1977. Sho - e m memo y o s imuli, esponses,
and ein o ce s. J. Exp. Psychol. Anim. Beha . P ocess. 3, 156–177. h ps://doi.o g/
10.1037/0097-7403.3.2.156.
Nelson, K.R., Wasse man, E.A., 1978. Tempo al ac o s in luencing he pigeon’s
successi e ma ching- o-sample pe o mance: sample du a ion, in e ial in e al,
and e en ion in e al. J. Exp. Anal. Beha . 30, 153–162. h ps://doi.o g/10.1901/
jeab.1978.30-153.
Pin o, C., Cas anhei a Dinis, J.M., 2024. Sample-compa ison mapping and join s imulus
con ol. Beha . P ocess. 196, 105006. h ps://doi.o g/10.1016/j.
bep oc.2024.105006.
Pin o, C., Machado, A., 2011. Sho - e m memo y o empo al in e als: con as ing
explana ions o he choose-sho e ec in pigeons. Lea n. Mo i . 42, 13–25. h ps://
doi.o g/10.1016/j.lmo .2010.05.001.
Pin o, C., Machado, A., 2015. Coding in pigeons: mul iple-coding e sus single-code/
de aul s a egies. J. Exp. Anal. Beha . 103, 472–483. h ps://doi.o g/10.1002/
jeab.153.
Pin o, C., Machado, A., 2017. Un a eling sou ces o s imulus con ol in a empo al
disc imina ion ask. Lea n. Beha . 45, 20–28. h ps://doi.o g/10.3758/s13420-016-
0233-2.
Pin o, C., Sousa, A., 2021. Choice biases in no-sample and delay es ing in pigeons
(Columba li ia). Anim. Cogn. 24, 593–603. h ps://doi.o g/10.1007/s10071-020-
01457-1.
Robe s, W.A., 1972. Sho - e m memo y in he pigeon: e ec s o epe i ion and spacing.
J. Exp. Psychol. 94, 74–83. h ps://doi.o g/10.1037/h0032796.
Robe s, W.A., 1980. Dis ibu ion o ials and in e ial e en ion in delayed ma ching o
sample wi h pigeons. J. Exp. Psychol. Anim. Beha . P ocess. 6, 217–237. h ps://doi.
o g/10.1037/0097-7403.6.3.217.
Robe s, W.A., G an , D.S., 1978. An analysis o ligh -induced e oac i e inhibi ion in
pigeon sho - e m memo y. J. Exp. Psychol. Anim. Beha . P ocess. 4, 219–236.
h ps://doi.o g/10.1037/0097-7403.4.3.219.
C. Soa es e al.
Beha iou al P ocesses 226 (2025) 105153
5
Robe s, W.A., K aeme , P.J., 1982. Some obse a ions o he e ec s o in e ial in e al
and delay on delayed ma ching o sample in pigeons. J. Exp. Psychol. Anim. Beha .
P ocess. 8, 342–353. h ps://doi.o g/10.1037/0097-7403.8.4.342.
She bu ne, L.M., Zen all, T.R., Kaise , D.H., 1998. Timing in pigeons: he choose-sho
e ec may esul om pigeons’ “con usion” be ween delay and in e ial in e als.
Psychon. Bull. Re . 5, 516–522. h ps://doi.o g/10.3758/BF03208831.
Singe , R.A., Klein, E.D., Zen all, T.R., 2006. Use o a single-code / de aul s a egy by
pigeons o acqui e du a ion sample disc imina ions. Lea n. Beha . 34, 340–347.
h ps://doi.o g/10.3758/BF03193197.
Spe ch, M., L., Rusak, B., 1989. Pigeons’ memo y o e en du a ion: in e ial in e al
and delay e ec s. Anim. Lea n. Beha . 17, 147–156. h ps://doi.o g/10.3758/
BF03207629.
Spe ch, M.L., 1987. Sys ema ic e o s in pigeons’ memo y o e en du a ion: in e ac ion
be ween aining and es delay. Anim. Lea n. Beha . 15, 1–5. h ps://doi.o g/
10.3758/BF03204897.
Spe ch, M.L., G an , D.S., 1993. Pigeons
′
memo y o e en du a ion in choice and
successi e ma ching- o-sample asks. Lea n. Mo i . 24, 156–174. h ps://doi.o g/
10.1006/lmo .1993.1010.
Spe ch, M.L., Rusak, B., 1992. Time p esen and ime pas . In: Honig, W.K., Fe e man, J.
G. (Eds.), Cogni i e Aspec s o S imulus Con ol. Law ence E lbaum Associa es Inc,
pp. 47–67.
Spe ch, M.L., Wilkie, D.M., 1982. A sys ema ic bias in pigeons’ memo y o ood and ligh
du a ions. Beha . Anal. Le . 2, 287–274.
Spe ch, M.L., Wilkie, D.M., 1983. Subjec i e sho ening: a model o pigeons’ memo y o
e en du a ion. J. Exp. Psychol. Anim. Beha . P ocess. 9, 14–30. h ps://doi.o g/
10.1037/0097-7403.9.1.14.
Wa d, R.D., Odum, A.L., 2007. Dis up ion o empo al disc imina ion and he choose-
sho e ec . Lea n. Beha . 35, 60–70. h ps://doi.o g/10.3758/BF03196075.
Whi e, K.G., 1985. Cha ac e is ics o o ge ing unc ions in delayed ma ching o sample.
J. Exp. Anal. Beha . 44, 15–34. h ps://doi.o g/10.1901/jeab.1985.44-15.
Wix ed, J., Gai an, S., 2004. S imulus salience and asymme ic o ge ing in he pigeon.
Lea n. Beha . 32, 173–182. h ps://doi.o g/10.3758/BF03196018.
Zen all, T.R., 1997. Animal memo y: he ole o “ins uc ions”. Lea n. Mo i . 28,
280–308. h ps://doi.o g/10.1006/lmo .1996.0968.
Zen all, T.R., 2006. Timing, memo y o in e als, and memo y o un imed s imuli: he
ole o ins uc ional ambigui y. Beha . P ocess. 71, 88–97. h ps://doi.o g/10.1016/
j.bep oc.2005.07.008.
Zen all, T.R., Smi h, A.P., 2016. Delayed ma ching- o-sample: a ool o assess memo y
and o he cogni i e p ocesses in pigeons. Beha . P ocess. 123, 26–42. h ps://doi.
o g/10.1016/j.bep oc.2015.07.002.
C. Soa es e al.
Beha iou al P ocesses 226 (2025) 105153
6
