Content uploaded by Rahel Noser
Author content
All content in this area was uploaded by Rahel Noser on Jan 20, 2019
Content may be subject to copyright.
ORIGINAL PAPER
Wild chacma baboons (Papio ursinus) remember single foraging
episodes
Rahel Noser
1
•Richard W. Byrne
2
Received: 21 November 2014 / Revised: 5 March 2015 / Accepted: 16 March 2015
ÓSpringer-Verlag Berlin Heidelberg 2015
Abstract Understanding animal episodic-like memory is
important for tracing the evolution of the human mind.
However, our knowledge about the existence and nature of
episodic-like memory in non-human primates is minimal.
We observed the behaviour of a wild male chacma baboon
faced with a trade-off between protecting his stationary
group from aggressive extra-group males and foraging
among five out-of-sight platforms. These contained high-
priority food at a time of natural food shortage. In 10
morning and eight evening trials, the male spontaneously
visited the platforms in five and four different sequences,
respectively. In addition, he interrupted foraging sequences
at virtually any point on eight occasions, returning to the
group for up to 2 h. He then visited some or all of the
remaining platforms and prevented revisits to already de-
pleted ones, apparently based on his memory for the pre-
vious foraging episode about food value, location, and
time. Efficient use of memory allowed him to keep mini-
mal time absent from his group while keeping food intake
high. These findings support the idea that episodic-like
memory offers an all-purpose solution to a wide variety of
problems that require flexible, quick, yet precise decisions
in situations arising from competition for food and mates in
wild primates.
Keywords Foraging Episodic-like memory Baboons
Feeding platforms Flexibility
Introduction
Declarative memory systems offer potential solutions to
two distinct kinds of challenge for a foraging animal, both
of which are fitness-relevant. On the one hand, a semantic
memory system would provide information about envi-
ronmental regularities (Tulving 1983), such as the above-
ground aspect of plants with palatable bulbs, the spatial
location of a water hole, and the animal’s social status
relative to others. On the other hand, an episodic memory
system would provide detail about entire events that are
likely to occur in the same form only once: human
episodic (Tulving 1983) and animal episodic-like (Clayton
and Dickinson 1998; see Beran 2014 for alternative ter-
minology) memories have been viewed as integrated rep-
resentations of several co-occurring percepts that allow
remembering the broader context in which an important
event was embedded (e.g. an entire scenario). For example,
in addition to remembering a leopard attack with infor-
mation about its location, a baboon may remember that all
other group members had been out of sight, that the sun
had just disappeared behind the horizon, and that a flock of
crested guinea fowls had been foraging in a nearby thicket.
Although none of these percepts alone are directly linked to
leopard attacks, they may together indicate an increased
likelihood of an attack and, in the baboon, evoke the cor-
responding episodic memory. Episodic(-like) memory
shares neural substrates in humans and animals (Allen and
Fortin 2013), and therefore is likely to also share an evo-
lutionary past (Raby and Clayton 2012; Allen and Fortin
2013). Understanding the nature and circumstances in
&Rahel Noser
rnoser@dplanet.ch
1
Cognitive Ethology Lab, German Primate Center,
Kellnerweg 4, 37077 Go
¨ttingen, Germany
2
Centre for Social Learning and Cognitive Evolution, School
of Psychology, University of St Andrews, Fife KY16 9JP,
Scotland, UK
123
Anim Cogn
DOI 10.1007/s10071-015-0862-4
which episodic memories are formed by animal minds is
thus crucial for understanding the human mind.
Animal episodic-like memories were initially demon-
strated in scrub jays by Clayton and Dickinson (1998).
They were viewed as integrated memories for content
(‘what’), location (‘where’), and time (‘when’), leaving
aside the much-disputed role of a sense of subjective ex-
perience (e.g. resulting from autonoetic consciousness,
Tulving 1983,2005; mental time travel, Schwartz et al.
2005; or autobiographical memory, Fivush 2011)asan
additional, compulsory component. At the same time, a
language-trained chimpanzee was reported to sponta-
neously recall the nature (‘what’) and locations (‘where’)
of objects several hours or days after they had been hidden
outside her cage, even though she witnessed the corre-
sponding hiding event only once (Menzel 1999,2005).
Clearly, these behaviours could not be explained by the
traditional view that animal memory is confined to envi-
ronmental regularities. This sparked a hefty controversy
about the relation between animal and human episodic
(-like) memory (e.g. Tulving 1983,2005; Suddendorf and
Corballis 2007; Suddendorf and Busby 2003; Roberts and
Feeney 2009; Raby and Clayton 2012 for reviews).
Although the role of self-consciousness in human episodic
memory still remains unclear to date, it was the ‘when’
component that has been particularly questioned as a valid
indicator of episodic-like memory in animal studies
(Roberts 2002; Crystal 2010; Raby and Clayton 2012): It
was suggested that animals can distinguish two events that
occurred in the past with relatively simple cognitive
mechanisms, for example based on relative familiarity
(Crystal 2010) or memory trace strength estimation (e.g.
Friedman 1993; Roberts 2002; but see Schwartz et al. 2005
for an example of gorilla palinscopy). It was pointed out
that these time estimation mechanisms can be performed in
the present, but fail to reflect a ‘sense of pastness’. For
example, scrub jays may use the state of their memory
trace of the location where they have previously cached
worms to semantically predict whether or not the worms
are decayed, whereby, for example, a weaker memory trace
meant decayed worms and a stronger memory trace meant
fresh worms (Roberts 2002; Schwartz 2005).
However, it has also been pointed out that the time
component of human episodic memories can be rudimen-
tary (Zentall et al. 2001) or entirely absent (Friedman
1993). For example, humans may distinguish two past
dinners out with the same friend by binding each episode to
the different contexts provided by the two restaurants
(Clayton et al. 2007). Thus, time has been proposed to
consist of just one of several possible ‘occasion specifiers’
that are used by humans and animals alike to distinguish
similar episodes (Eacott and Norman 2004; Eacott et al.
2005). As a consequence, episodic-like memory has been
redefined as integrated long-term memories for events (one
or several ‘what’ components) in context (‘which’, such as
‘where’, ‘when’, ‘who’, or other details, see also Menzel
2005), to support adaptive behaviour in novel situations
(Allen and Fortin 2013; Clayton et al. 2003,2007).
Episodic-like memory has been claimed for several
species (e.g. scrub jays: Clayton and Dickinson 1998; rats:
Eacott and Norman 2004; Eacott et al. 2005; Babb and
Crystal 2005,2006; chickadees: Feeney et al. 2009; mag-
pies: Zinkivskay et al. 2009; cuttlefish: Jozet-Alves et al.
2013; meadow voles: Ferkin et al. 2008; chimpanzees:
Menzel 1999, Menzel 2005; orangutans and bonobos:
Martin-Ordas et al. 2010; minipigs: Kouwenberg et al.
2009; adult humans: Holland and Smulders 2011; human
children: Scarf et al. 2013; and hummingbirds: Gonzalez-
Gomez et al. 2011). Surprisingly, monkeys have to date not
shown convincing evidence of episodic-like memory
(Hampton et al. 2005; Hoffman et al. 2009; Basil and
Hampton 2011), alongside of pigeons (Zentall et al. 2001).
The experimental paradigms used by Hoffman et al. (2009)
and Basil and Hampton (2011) used delays of seconds and
thus examined working memory. This raises doubt about
whether a link of the results to the concept of episodic
memory can be made with confidence. In contrast, the
monkeys (Macaca mulatta) tested by Hampton et al.
(2005) failed to show that they differentiated between short
and long delays (‘when’). The authors concluded that their
subjects were not sensitive to the time of learning where
preferred food was hidden in a room, and that their
memory was restricted to the ‘what’ and the ‘where’
components. Thus, convincing evidence that monkeys use
episodic-like memory is still lacking, and our understand-
ing of the factors under which episodic-like memory may
be beneficial in natural settings is at best vague.
The daily lives of wild savannah baboons, as those of
many other animal species, contain a large number of re-
occurring event sequences. Baboons feed on a wide variety
of stationary food sources, and their revisit rate to par-
ticular food patches and water holes is high. They possess
excellent memory skills for the locations of different food
types. They anticipate the value of out-of-sight food they
will obtain up to 2 h later (Noser and Byrne 2007,2010).
Also, there is evidence that they infer food value of out-of-
sight resource sites from the movement directions of group
members alone (Stolba 1979; King et al. 2008, Figure S4).
This impressive knowledge about resource value and lo-
cation allows them to take important, but re-occurring
foraging decisions during social and ecological stability,
allowing them to survive and reproduce in areas of low
food density.
However, baboons also regularly face a wide variety of
unpredictable or novel events, such as the sudden ap-
pearance of an intra-group competitor, a competing group,
Anim Cogn
123
or a predator, causing a deviation of their usual foraging
routine. Do they link such an event with additional details
surrounding the scenario and use this information in the
future? To answer this question, we examined whether a
wild alpha male chacma baboon under naturally occurring
but unusually high time pressure spontaneously formed a
memory for single foraging episodes. Stressor was the
highly unusual presence of close-by extra-group males
that for him generated a trade-off between two fitness-
relevant tasks: foraging for highly valuable fruit (orange)
among five out-of-sight feeding platforms during a time of
food shortage, when the most valuable naturally occurring
fruit type consisted of tiny, dry, and at best mildly sweet
berries (Grewia spp) located at much larger distances
(Noser and Byrne 2007) on the one hand, and protecting
his females and offspring from aggression and takeover by
these males (e.g. Palombit et al. 2000) on the other. We
hypothesised that knowledge about the spatial locations of
platforms (i.e. a cognitive map of some form, ‘where’)
and their contents (‘what’, resource value) would allow
him to forage for some or all orange sites at a time, and
perhaps to keep short overall foraging distance. In con-
trast, only by integrating a cognitive map and contents
with an episodic-like memory of temporal or categorical
information of single platforms could he tell recent for-
aging episodes apart from older ones, and as a result,
predict which platforms were still baited after foraging
had been unexpectedly disrupted. If available, this form of
memory would allow him to flexibly switch between the
two tasks given by the trade-off, and ultimately to coor-
dinate his foraging with the group’s behaviour, adjusting
the timing and extent of foraging to the current social
requirements of a particular situation while maximising
foraging benefit.
Methods
This experiment was part of a larger study that we carried
out in the woodland savannah at Blouberg Nature Reserve
(BNR; 22°580S, 29°090E), Limpopo Province, South
Africa, between August 2000 and August 2002 with a
group of chacma baboons (Papio ursinus; see Noser and
Byrne 2007 for details). The group consisted of the adult
alpha male, an adolescent male, and of eight adult females
with offspring, totalling 24 individuals. At the time of the
test phase, they had been followed on foot daily by a team
of two human observers for 23 months and thus were well
habituated to our presence. The alpha male had been
dominant over an additional male who had emigrated from
the group at an earlier stage of the study, and thus was
likely to have fathered a large number of offspring present
in the group. The group used a single sleeping site
throughout the study and, as a consequence, could be
regularly met in the adjacent area in the early mornings and
late afternoons.
Between August 2001 and May 2002, we habituated the
group to the presence of novel feeding platforms, com-
mercial oil barrels (approximately 88 cm in height, 60 cm
in diameter), with the outer steel walls painted in brown.
We filled the barrels with heavy rocks to ensure stability
and occasionally baited them with two slices of peeled
orange each, so that the oranges could only be seen by a
baboon when sitting on a platform, or when standing on the
hind legs in front of it, but not when walking on the
ground. The relatively small platform diameter enabled
single baboons to monopolise a platform. All group
members could reach the oranges.
In May 2002, we placed a total of five platforms (P1–
P5) in the vicinity of the sleeping site (Fig. 1), so that two
criteria were met: (1) the platforms were out of sight of
each other, and (2) all the platforms as well as the sleeping
site area were well visible from an elevated observation
point (large star in Fig. 1). We excluded inter-platform
visibility: two persons stood each next to a platform, one
waving a white plastic bag on a stick at the height of the
upper platform rim, and the other one determining with
binoculars from which point along the line connecting the
two platforms the flag became visible through the bushes.
60 Meters
N
P1
P2
P5
P4
P3
Fig. 1 Map of experimental setting. Five monopolisable feeding
platforms P1–P5 (grey circles), situated next to the sleeping site
(large grey area), were baited with two slices of peeled orange each.
The foraging behaviour of subjects was monitored from an elevated
observation point (large star). Thick grey lines are car tracks, and thin
grey line is line of most pronounced change in elevation. Small
symbols (suns and crossed rectangles) represent landmarks
Anim Cogn
123
This revealed that platforms became visible from a distance
of 40 m or less. The sleeping site area was out of sight
from P3, P4, and P5, and partly visible from P1 and P2. We
took GPS readings of the sleeping site area, the platforms,
and of prominent landmarks in the experimental area (car
tracks, a fire break, a dry river bed, tall trees, rock outcrops)
and plotted maps of this layout.
The learning phase took place between May and July
2002, for a total of 21 trials, and ended when at least one
individual had discovered all platforms. Learning trials
were carried out opportunistically, either in the mornings
before the group left the sleeping site, or in the evenings
when they re-approached it for the night. To indicate that
the platforms had been baited, we placed two slices of
orange on the ground at a distance of approximately 5 m
from P1, so that they were visible from the sleeping site.
Note that, as a result, the value of P1 (with two orange
slices on the platform and two next to it) was larger than
the value of P2–P5 (each containing two slices). For
morning trials, platforms were baited before dawn, at
around 5 a.m., when the group was still asleep. For evening
trials, baiting took place at around 3 p.m., when the group
still foraged far from the experimental area.
The sleeping site had been used exclusively and at a daily
basis by the focal group for at least 21 months (August
2000–May 2002). This, together with our observation that
our focal group’s day journey lengths were large in com-
parison with those observed at other baboon study sites
(Noser and Byrne, unpublished), suggested that the Blou-
berg baboon groups competed intensely for suitable sleep-
ing sites. Indeed, an additional unknown baboon group
consisting of approximately 35 individuals (two adult males,
11 adult females) suddenly started to co-use our focal
group’s sleeping site during the learning phase (June 2002).
From this point onward, the two groups shared the site every
night, using a small vertical elevation in the cliff as a natural
visual barrier to separate the groups’ actual sleeping ledges.
At dawn (approximately 6 a.m.), both groups climbed the
cliff. This was the point when aggressive inter-group in-
teractions started, including emission of a large number of
male aggressive inter-group calls (‘roar-grunts’), alarm calls
(‘wahoos’, ‘barks’), many herding bouts, and screaming.
We did not observe any friendly interactions between any of
the two groups’ members. Aggressive displays persisted in
the mornings until one of the groups left the area at around 8
a.m. to forage separately and out of sight of the other during
the day. In the evenings, the two groups returned from their
foraging trips at around 5 p.m. Aggressive displays were
taken up immediately. They lasted until around 6:30 p.m.,
when the two groups entered the actual sleeping ledges. The
novel group was unhabituated to observers and platforms,
and never participated in trials.
The test phase started in July 2002, during the dry
season. Due to poor quantity and quality of naturally oc-
curring fruit during this time of year, the study group
regularly travelled for up to 2 h each morning at high speed
to breakfast on small, dry Grewia berry (Grewia spp) si-
tuated at distances of 5–6 km from the sleeping site. This
was the turning point of their journey after which they
returned to the sleeping site while slowly travel-feeding on
abundant dry matter such as pods and kernels (Noser and
Byrne 2007). Both the natural fruit scarcity and the asso-
ciated natural foraging behaviour suggested that our baited
platforms constituted extremely valuable food resources to
these baboons. The test phase comprised a total of 18 trials
(10 morning, eight evening trials). During this time, none
of the focal group’s females were in oestrus.
Two observers watched the experimental area from the
observation point (Fig. 1): one continuously followed the
alpha male’s behaviour, and the other observer scanned the
experimental area for additional baboons using the plat-
forms. The following variables were spoken onto a voice
recorder: identity of individuals that foraged for oranges
(thereafter ‘subjects’), time when subjects climbed indi-
vidual platforms P1–P5, and whether or not they found and
ingested oranges. We reconstructed the routes that subjects
had taken by additionally recording the time when they
passed by the prominent landmarks in the experimental
area (Fig. 1). Maps of these routes were drawn from the
recordings after returning to the research camp the same
day. Since travel among platforms was essentially linear,
we used the order of platform visits to assess spatial flex-
ibility (Fig. 2).
We viewed a trial as ‘split’ into two parts when a subject
interrupted and then resumed orange foraging. This re-
sulted in a ‘first part’ and a ‘second part’ of orange for-
aging, with either of two types of behaviour occurring
intermittently: returning to the sleeping site to socialise or
engage in group defence, and climbing trees to inspect the
area. During the second part of orange foraging, one ore
more platforms were visited. Foraging time for oranges
was the time (accurate to 1 min) when a subject crossed the
imaginary outer lines of the experimental area connecting
the platforms.
A baboon was viewed as absent from the group when he
was engaged in orange foraging, while the majority of the
group stayed in the sleeping site area (mornings; see Fig. 1
for distances) or foraged for natural foods far from the
experimental area (evenings; unknown distances between
forager and other group members), thereby temporally
losing sight of them. In contrast, subjects were viewed
as present when they coordinated their behaviour with
their own group, so that orange foraging occurred during
the time when their group crossed or passed by the
Anim Cogn
123
experimental area, and spatial proximity to their group was
retained.
We used R software (R Core Team 2013) for statistics.
We performed a hypothesis test with a linear mixed model
fit by maximum likelihood (lme4 package, Bates et al.
2014), comparing a null model with an extended model
using BIC and Anova. After depletion of a given orange
site, subjects needed to decide which platform to visit next.
We viewed long-term memory to be involved in this choice
when a ‘split’ occurred between depletion of platform n
and choice for platform n?1. For these cases, we devised
a random choice model each for the third, the fourth, and
the fifth platform simulating 1000 decisions from random
numbers. We determined the likelihood that the observed
number of choices to unvisited platforms was due to
chance.
Results
The alpha male participated in 18 test trials (100 %) and
obtained 0.68 ±0.48 orange slices per min spent in the
experimental area (mean ±SD). In comparison, the ado-
lescent male participated only in three trials and obtained
only 0.14 ±0.27 slices per min. Two additional group
members, an adolescent male and an adult female, spent 1
and 12 min in the experimental area, but did not obtain any
oranges at all. For the rest of this study, we focus on the
performance of the alpha male only.
Within-trial revisits to platforms occurred three times:
twice in the learning phase, and once in the test phase (Fig.
2), with a delay between the two visits of 14, 28, and
14 min, respectively, and intermittent visits to additional
platforms. Inter-trial intervals ranged between 8.3 and
37.7 h (median 18.7 h), and some or all platforms that had
been visited in trial xwere revisited in trial x?1.
The alpha male’s spatial and temporal decisions during
orange foraging exhibited a high degree of variation
(Fig. 2). It is reasonable to assume that trials in which he
left one or several platforms unvisited (indicated by 0 in the
platform symbols; N=8) as well as split trials (a split in a
sequence of visits is indicated by II;N=8; Fig. 2) reflect
socially induced interruptions of a given trip. Thus, they
can be interpreted as parts of intended, but unfinished full
trips. Taking this view, the alpha male exhibited five spatial
variants of orange foraging in the mornings and four in the
evenings. Abandoning orange foraging after only two
platform visits without later resumption occurred in three
trials; it remains open whether these trips, if finished,
would have resulted in additional variants.
Splits occurred after visits to two, three, or four plat-
forms. One, two, or three platforms were visited after re-
sumption of orange foraging in the second part of split
trials. Splits never occurred during the learning phase, but
appeared for the first time during the third test trial. Re-
visits to platforms did not occur in split trials. Median time
elapsed between the end of the first and the beginning of
the second part of orange foraging in split trials was
92.5 min (range 8–113 min). The onset of the second part
was triggered by the spatial displacement of the focal group
in seven of eight split trials: the alpha male re-entered the
experimental area when the group either finished socialis-
ing/engaging in aggressive displays (mornings) or feeding
(evenings) and started moving by either directly crossing
the experimental area or passing by at a short distance. In
the remaining case, the adolescent male entered the ex-
perimental area shortly after the alpha male had interrupted
orange foraging. This caused the alpha male to re-enter the
area after only 8 min and to resume orange foraging.
5
4
0
2
1
0
3
2
0
0
4
0
0
3
0
3
0
0
3
5
4
0
2
0
2
4
5
15
a
( )
2
2
5
143
b4
0
3
20
5
5
4
1
3
0
5
0
( )
1)
Fig. 2 Flexibility in number, order, and timing of visits to platforms
in 18 trials. In the mornings (a;N=10), when the sleeping site was
the starting point of orange foraging, the alpha male first visited P1 on
nine occasions and then visited either P2, P4, or P5. He then either left
the experimental area to entirely abandon orange foraging (indicated
by three zeros for the remaining platforms), or to resume orange for-
aging, sometimes after returning to the sleeping site and spending
8–113 min with the group (II, occurrence of a ‘split’, see text). Note
that a split applies to upper of two path legs when the sign is located
above arrow, and to lower path leg when the sign is below. Stars
indicate identical number and order of visits in two trials. (II)
indicates that only one of two otherwise identical sequences of orange
foraging was split. In the evening trials (b;N=8), P2 and P3 were
the first platforms visited, corresponding to the overall direction from
where the sleeping site was approached after natural foraging. (1) A
single within-trial revisit to a platform was recorded in an evening
trial, with the order of visits P2–P1–P5–P4–P3–P1-sleeping site
Anim Cogn
123
Overall, the alpha male stayed in the experimental area
for 14.4 ±5.8 min per trial (mean ±SD). However, he
stayed absent from the group for only 10.1 ±5.1 min, due
to his ability to split trials. Statistically, splitting orange
foraging into two parts significantly reduced the duration of
absence from the group [linear mixed null model: fixed
effect =number of barrels visited, random effect =date
(with two or only one experimental trials per date),
BIC =-8.3; full model: fixed effects =Number of plat-
forms visited, occurrence of split (yes/no), random ef-
fect =date, BIC =-18.7, v
2
=12.4, P\0.001, Fig. 3].
Were the platform choices during the second parts of
split trials informed by memory? Splits after depletion of
two platforms occurred on five occasions (Fig. 2). The
random choice model revealed a likelihood of subsequently
targeting an unvisited platform on five occasions (e.g.
without targeting an already depleted platform in one or
several trials) of P=0.093. In contrast, a split after three
visited platforms occurred in a single trial, and the corre-
sponding likelihood of targeting an unvisited platform after
the split was P=0.386. A split after visits to four plat-
forms occurred twice, and the likelihood of targeting the
single unvisited platform after the splits was P=0.033.
Thus, when splitting the orange foraging into two parts, the
likelihood of never targeting an already visited platform
after the splits in a total of eight trials by pure chance,
without the involvement of long-term memory, was highly
significant (P=0.001). This suggests that the alpha male
used a memory of earlier platform visits when resuming
orange foraging after a split.
Discussion
During 10 morning and eight evening experimental trials, a
wild male chacma baboon faced two mutually exclusive
tasks: the presence of extra-group males next to his group
required him to engage in group defence and to stay next to
the group, while obtaining rare high-priority food at distant
out-of-sight feeding platforms early (see Noser and Byrne
2007; Janmaat et al. 2014) required him to leave. In this
situation, the male applied a flexible spatial and temporal
strategy. He visited five baited platforms in several dif-
ferent sequences, with the visit order possibly depending
on the behaviour and spatial location of extra-group males
in a given trial. In addition, he interrupted some of these
sequences at virtually any point to rejoin the social group
after having depleted only a fraction of the platforms.
Later, he returned to visit some or all remaining platforms,
apparently by choosing a time when he could coordinate
his own foraging behaviour with the general movement of
his group. This strategy allowed him to significantly reduce
the duration of absence from the group, but required long-
term memory for previous specific foraging episodes to
avoid redundant revisits to platforms.
The flexibility of the order of platform visits as well as
the ability to resume a randomly interrupted trip after a
delay without revisits to depleted platforms implies that he
did not primarily pursue a distance minimising strategy, but
rather reflects a disposition to adapt to an unpredictable and
unstable social environment, perhaps while keeping over-
all-distance short. Indeed, a short-term maximisation for-
aging strategy, predicting choice of the next best option at
each decision point, is expected under such circumstances
(Sayers and Menzel 2012 and references therein). On a
cognitive level, semantic knowledge about resource value
and location cannot entirely explain the high degree of
behavioural flexibility required in such situations. When
interrupting and then resuming orange foraging, the alpha
male baboon never revisited an already depleted platform.
This suggests an additional temporal memory component
that was specific to a particular foraging episode and al-
lowed him to differentiate between long and short delays of
previous visits to individual platforms: already visited
platforms were empty after short, but not after long delays.
How might he have achieved this? Using an external cue
such as time of day cannot account for our findings, due to
the irregular nature of our experiment. Neither is interval
timing (Friedman 1993; Roberts 1998) a likely explanation,
since this mechanism has only been documented for time
0.7 0.8 0.9 1.0 1.1 1.2 1.3
log duration (min)
no no no no yesyesyesyes
3425
split foraging trip
platforms visited / trial
3
22
4
2
5
Fig. 3 Duration of absence from group was shorter when the
alpha male split the foraging trip into two parts (white)and
longer when a split did not occur (grey), irrespective of the
number of platforms visited per trial. Numbers within boxes are
numbers of observations
Anim Cogn
123
estimates considerably shorter than those involved in the
present study. Relative familiarity and memory strength
estimation are two mechanisms often proposed as provid-
ing humans and animals with cues to distinguish more
recent from older memories when directly perceiving an
item. In the present experiment, recent memories were built
92.5 min ago on average, and older ones 18.7 h ago,
theoretically allowing a clear distinction. However, when
the alpha male started approaching the remaining platforms
after a split, he was unable to see them until he was close
by (i.e. 40 m or less apart). Thus, relative familiarity can-
not account for his performance unless we assume he
compared mental representations of familiarity or memory
trace strengths for several platforms.
A more parsimonious view is that upon visiting a plat-
form, the alpha male may have ‘tagged’ it as ‘depleted’,
internally checking it off a represented list or a cognitive
map containing all platforms, and holding a memory of the
platforms that remained baited. In each experimental trial,
he would then have used a represented map with different
locations still active after the delay. This assumption would
not necessarily assume a memory for the actual event of
depleting a platform, or a ‘sense of pastness’; rather, it
assumes an integrated memory that (semantically) encodes
food value and location of several resource places, and an
additional flexible component that regulates a categorical
‘tag’ for each location that temporally and individually
devalues platforms once they are visited. This ‘tag’ would
need to become inactive after a minimum of 2 h to allow
revisits of all platforms during a later trial.
How does do this idea match the natural foraging be-
haviour of baboons? During our two-year study with this
baboon group, most naturally occurring fruit resources
were visited only once a day, leaving the possibility of a
tagging mechanism intact. However, this was not true for
one naturally occurring food source, the Mountain Fig
(Ficus glumosa): our focal group did revisit particularly
productive Mountain Fig tree specimens after delays of
111 min minimally (median 221 min), with six within-day
revisits to two trees recorded on 12 days of observation
(Noser and Byrne, unpublished). Other trees, however,
were visited only once a day, and yet others after longer
time periods. In contrast to the present study, the decision
to revisit a fig tree was likely taken on the basis of dif-
ferential renewal rates of individual trees, with trees that
were ideally exposed to the summer sun possibly produc-
ing ripe figs at shorter time intervals. In the present study,
renewal rate was identical for all platforms, but orange
feeding (i.e. visiting and depleting one platform after the
other) was interrupted by another baboon group. As a
consequence, the baboon male needed to remember which
platforms he had already visited, and which ones remained
baited. In this situation, the longest delay between the two
parts of a ‘split’ trial was 116 min, yet no revisit occurred.
These observations combined are at odds with the idea that
a single, rigid ‘tagging’ mechanism governs revisits to
resource places in baboons, which would obviously lead to
a rigid revisit pattern. Instead, we need to assume that
different tags exist for different food types, and in the case
of fig foraging, also for different specimens within a spe-
cies. Given that our baboon group consumed at least 27
fruit species in addition to a wide variety of non-fruit food
(Noser and Byrne, unpublished), this would imply an ex-
tensive body of semantic knowledge about renewal pat-
terns (or a large number of different ‘tags’). Any advantage
of parsimony is thus lost.
Clearly, a more elegant way to solve the problem posed
in this study, and in foraging in large-scale space under
uncertain social conditions in general, is holding an inte-
grated memory containing information about food value
(or profitability) at multiple locations, about how to get
there (e.g. Noser and Byrne 2007; Janson 1998), and an
egocentric memory component that stores information
about own behaviour to reveal the locations already visited
and those that are still available. This cognitive toolbox
allows for choosing the next best option at each decision
point (‘short-term rate maximisation’, see Sayers and
Menzel 2012). This idea converges with the observations
of captive chimpanzees who recover hidden bags in the
order of profitability rather than in the order of the hiding
process (Sayers and Menzel 2012), and of wild capuchin
monkeys who visit feeding platforms in the order of
profitability (Janson 1998). In fact, there is growing neu-
roanatomical evidence that this ability may extend to
simian primates in general (Genovesio et al. 2014).
In summary, we have presented data on an adult male
baboon whose foraging journey among five out-of-sight
platforms containing high-priority food was repeatedly
disrupted due to the presence of extra-group males in
proximity to his females and offspring. The fact that he
continued any disrupted journey after several hours,
thereby never revisiting an already depleted platform,
suggested that he had formed an episodic-like memory for
the value and location of high-priority food and for the
locations that still contained food after a single foraging
episode. These findings highlight the adaptive value of
episodic-like memory in the wild when flexible, quick, yet
precise decisions must be taken in a complex foraging and
social environment.
Our findings raise the question whether all foraging
episodes, only those for high-priority foods, or only for-
aging episodes that are coupled with elevated stress levels
leave behind episodic-like memory traces in baboon brains.
Evidence for the latter comes from the finding that human
and rat declarative memory consolidation co-varies with
stress hormones levels: experienced stress at the time of
Anim Cogn
123
learning leads to increased long-term memory performance
(Sandi and Pinelo-Nava 2007; Wolf 2009), and humans
often form vivid episodic memories in stressful or emo-
tionally otherwise prominent situations (Joels et al. 2006;
Pause et al. 2013). For example, a need to bring an unac-
complished task to an end (‘ungestilltes Erledigungs-
bedu
¨rfnis’) has been suggested to trigger human episodic
memory formation (Zeigarnik 1927). Future research may
also examine whether the ability of non-human primates to
integrate an episodic component into existing semantic
memory extends to contexts other than foraging, for ex-
ample social or predatory, and thus, whether episodic
memory may be a multi-purpose tool suitable for solving a
wide variety of problems that arise daily in the lives of
primates.
Acknowledgments We thank the Parks Board of the Limpopo
Province, South Africa, for permission to conduct this study at
Blouberg Nature Reserve; Ralph Schwarz, Martina Bra
¨gger and Beat
Egger for assistance, and Janine and Peter Snyman for logistic sup-
port in the field; Ken Sayers and an anonymous reviewer for valuable
comments on an earlier draft of this manuscript. This study was
supported by grants from Zu
¨rcher Hochschulverein, Schweizerische
Akademie fu
¨r Naturwissenschaften, Stiftung Thyll-Du
¨rr, and Stiftung
Annemarie Schindler, to R.N. In memory of Peter Snyman.
References
Allen T, Fortin N (2013) The evolution of episodic memory. PNAS
110:10379–10386
Babb S, Crystal J (2005) Discrimination of what, when, and where:
implications for episodic-like memory in rats. Learn Mot
36:177–189
Babb S, Crystal J (2006) Episodic-like memory in the rat. Curr Biol
16:1317–1321
Basil B, Hampton R (2011) Monkeys recall and reproduce simple
shapes from memory. Curr Biol 21:774–778
Bates D, Maechler M, Bolker B, Walker S (2014) lme4: linear mixed-
effects models using Eigen and S4 classes. R package version
10–7
Beran MJ (2014) Animal memory: rats bind event details into
episodic memories. Curr Biol 24:R1159–R1160
Clayton N, Dickinson A (1998) Episodic-like memory during cache
recovery by scrub jays. Nature 395:272–274
Clayton N, Bussey T, Dickinson A (2003) Can animals recall the past
and plan for the future? Nat Rev Neurosci 4:685–691
Clayton N, Salwiczek L, Dickinson A (2007) Episodic memory. Curr
Biol 17:R189–R191
Crystal J (2010) Episodic-like memory in animals. Behav Brain Res
215:235–243
Eacott M, Norman G (2004) Integrated memory for object, place, and
context in rats: a possible model of episodic-like memory?
J Neurosci 24:1948–1953
Eacott M, Easton A, Zinkivskay A (2005) Recollection in an
episodic-like memory task in the rat. Learn Mem 12:221–223
Feeney M, Roberts W, Sherry D (2009) Memory for what, where, and
when in the black-capped chickadee (Poecile atricapillus). Anim
Cogn 12:767–777
Ferkin M, Combs A, delBarco-Trillo J, Pierce A, Franklin S (2008)
Meadow voles, Microtus pennsylvanicus, have the capacity to
recall the ‘‘what’’, ‘‘where’’, and ‘‘when’’ of a single past event.
Anim Cogn 11:147–159
Fivush R (2011) The development of autobiographical memory. Annu
Rev Psychol 62:559–582
Friedman W (1993) Memory for the time of past events. Psychol Bull
113:44–66
Genovesio A, Wise SP, Passingham RE (2014) Prefrontal-parietal
function: from foraging to foresight. TICS 18:72–81
Gonzalez-Gomez P, Bozinovic F, Vasquez R (2011) Elements of
episodic-like memory in free-living hummingbirds, energetic
consequences. Anim Behav 81:1257–1262
Hampton R, Hampstead B, Murray E (2005) Rhesus monkeys
(Macaca mulatta) demonstrate robust memory for what and
where, but not when, in an open-field test of memory. Learn Mot
36:245–259
Hoffman M, Beran M, Washburn D (2009) Memory for ‘‘what’’,
‘‘where’’, and ‘‘when’’ information in rhesus monkeys (Macaca
mulatta). J Exp Psychol 35:143–152
Holland S, Smulders T (2011) Do humans use episodic memory to
solve a what-where-when memory task? Anim Cogn 14:95–102
Janmaat KRL, Polansky L, Ban SD, Boesch C (2014) Wild
chimpanzees plan their breakfast time, type and location. PNAS
111:16343–16348
Janson CH (1998) Experimental evidence for spatial memory in
foraging wild capuchin monkeys. Anim Behav 55:1229–1243
Joels M, Pu Z, Wiegert O, Oitzl M, Krugers H (2006) Learning under
stress: how does it work? TICS 10:152–158
Jozet-Alves C, Bertin M, Clayton N (2013) Evidence of episodic-like
memory in cuttlefish. Curr Biol 23:R1033–R1035
King A, Douglas C, Huchard E, Isaac N, Cowlishaw G (2008)
Dominance and affiliation mediate despotism in a social primate.
Curr Biol 18:1833–1838
Kouwenberg A, Walsh C, Morgan B, Martin G (2009) Episodic-like
memory in crossbred Yucatan minipigs (Sus scrofa). Appl Anim
Behav Sci 117:165–172
Martin-Ordas G, Haun D, Colmenares F, Call J (2010) Keeping track
of time: evidence for episodic-like memory in great apes. Anim
Cogn 13:331–340
Menzel CR (1999) Unprompted recall and reporting of hidden objects
by a chimpanzee (Pan troglodytes) after extended delays.
J Comp Psychol 113:426–434
Menzel CR (2005) Progress in the study of chimpanzee recall and
episodic memory. In: Terrace H, Metcalfe J (eds) The missing
link in cognition. Origins of self-reflective consciousness.
Oxford University Press, Oxford, pp 188–224
Noser R, Byrne RW (2007) Travel routes and planning of visits to
out-of-sight resources in wild chacma baboons (Papio ursinus).
Anim Behav 73:257–266
Noser R, Byrne RW (2010) How do wild chacma baboons (Papio
ursinus) plan their routes? Travel among multiple high-quality
food sources with inter-group competition. Anim Cogn
13:145–155
Palombit RA, Cheney D, Seyfarth R, Rendall D, Silk J, Johnson S,
Fischer J (2000) Male infanticide and defense of infants in
chacma baboons. In: van Schaik CP, Janson CH (eds) Male
infanticide and its implications. Cambridge University Press,
Cambridge, pp 123–151
Pause B, Zlomuzica A, Kinugawa K, Mariani J, Pietrowsky R, Dere E
(2013) Perspectives on episodic-like and episodic memory. Front
Behav Neurosci 7:1–12
R Core Team (2013) R: a language and environment for statistical
computing. R Foundation for Statistical Computing, Vienna
Raby C, Clayton N (2012) Episodic memory and planning. In: Vonk
J, Shackelford T (eds) The Oxford handbook of comparative
evolutionary psychology. Oxford University Press, Oxford,
pp 217–235
Anim Cogn
123
Roberts W (1998) Principles of animal cognition. McGraw-Hill,
Boston
Roberts W (2002) Are animals stuck in time? Psychol Bull
128:473–489
Roberts W, Feeney M (2009) The comparative study of mental time
travel. TICS 13:271–277
Sandi C, Pinelo-Nava M (2007) Stress and memory: behavioral
effects and neurobiological mechanisms. Neural Plast. doi:10.
1155/2007/78970
Sayers K, Menzel CR (2012) Memory and forging theory: chim-
panzee utilization of optimality heuristics in the rank-order
recovery of hidden foods. Anim Behav 84:795–803
Scarf D, Gross J, Colombo M, Hayne H (2013) To have and to hold:
episodic memory in 3- and 4-year-old children. Dev Psychobiol
55:125–132
Schwartz B (2005) Do nonhuman primates have episodic memory?
In: Terrace H, Metcalfe J (eds) The missing link in cognition.
Origins of self-reflective consciousness. Oxford University
Press, Oxford, pp 225–241
Schwartz B, Hoffman M, Evans S (2005) Episodic-like memory in a
gorilla: a review and new findings. Learn Mot 36:226–244
Stolba A (1979) Entscheidungsfindung in Verba
¨nden von Papio
hamadryas. Dissertation, University of Zurich
Suddendorf T, Busby J (2003) Mental time travel in animals. TICS
7:391–396
Suddendorf T, Corballis M (2007) The evolution of foresight: what is
mental time travel, and is it unique to humans? Behav Brain Sci
30:299–313
Tulving E (1983) Elements of episodic memory. Clarendon, Oxford
Tulving E (2005) Episodic memory and autonoesis: uniquely human?
In: Terrace H, Metcalfe J (eds) The missing link in cognition:
origins of self-reflective consciousness. Oxford University Press,
Oxford, pp 3–56
Wolf O (2009) Stress and memory in humans: twelve years of
progress? Brain Res 1293:142–154
Zeigarnik B (1927) Das Behalten erledigter und unerledigter Hand-
lungen. Psychol Forschung 9:1–85
Zentall T, Clement T, Bhatt R, Allen J (2001) Episodic-like memory
in pigeons. Psychon Bull Rev 8:685–690
Zinkivskay A, Nazir F, Smulders T (2009) What-where-when
memory in magpies (Pica pica). Anim Cogn 12:119–125
Anim Cogn
123
