See instructions and articles below Culture and Selective

Culture and Selective Social Learning
in Wild and Captive Primates

Stuart K. Watson, Jennifer Botting, Andrew Whiten, and Erica van de Waal

Abstract Once thought to be a unique human trait, the presence of culture in
non-human primates has been confirmed and studied by researchers for several
decades. What has been discovered is evidence for between-group traditions in a
wide range of primate taxa, including all of the great apes, macaques, capuchins and
spider monkeys, as well as many non-primate species. The capacity to learn from
others is a powerful means by which animals can acquire adaptive ways of
interacting with their environment and each other without engaging in time-
consuming and potentially risky trial-and-error learning. However, much remains
to be understood about the exact mechanisms and processes that underpin social
learning and how these lead to the cultures identified in wild populations of primates,
including humans. In the current chapter, we review what is known about

The authors “Stuart K. Watson” and “Jennifer Botting” contributed equally. Order of authorship
was determined by coin-toss.

S. K. Watson
Centre for Social Learning and Cognitive Evolution, and Scottish Primate Research Group,
School of Psychology and Neuroscience, University of St Andrews, St Andrews, UK

J. Botting
Centre for Social Learning and Cognitive Evolution, and Scottish Primate Research Group,
School of Psychology and Neuroscience, University of St Andrews, St Andrews, UK

Inkawu Vervet Project, Mawana Game Reserve, Swart Mfolozi, KwaZulu Natal, South Africa

Smithsonian National Zoological Park, Washington, DC, USA

A. Whiten
Centre for Social Learning and Cognitive Evolution, and Scottish Primate Research Group,
School of Psychology and Neuroscience, University of St Andrews, St Andrews, UK

Inkawu Vervet Project, Mawana Game Reserve, Swart Mfolozi, KwaZulu Natal, South Africa
e-mail: [email protected]

E. van de Waal (*)
Inkawu Vervet Project, Mawana Game Reserve, Swart Mfolozi, KwaZulu Natal, South Africa

Anthropological Institute and Museum, University of Zurich, Zurich, Switzerland
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018
L. D. Di Paolo et al. (eds.), Evolution of Primate Social Cognition, Interdisciplinary
Evolution Research 5, https://doi.org/10.1007/978-3-319-93776-2_14

211

non-human primate culture with a particular emphasis on the emerging field of social
learning biases. Theoreticians and field researchers alike have suggested that animals
may exhibit biases in whom they obtain information from, either as by-products of
social dynamics or as adaptive strategies that allow animals to selectively acquire the
most useful information. Here, we review the theoretical arguments and current
empirical evidence for proposed biases in social learning, including majority-based
biases and model-based biases. We draw from field observations and experiments in
both captive and wild populations to examine how information may be transferred
between individuals and how this may affect the emergence of cultural behaviours
across primate species.

Keywords Culture · Primates · Social learning biases · Social transmission
mechanisms · Conformity · Model-based biases

1 Introduction

Culture pervades every aspect of human life, from the way we communicate and the
values we hold to the way we think and shape our environment. The extent and
complexity of human culture has created such a palpable gulf between us and other
members of the animal kingdom that the capacity for culture has historically been
argued to be unique to humans (e.g. Galef 1992). However, this depends on how
culture is defined, operationalised and tested for. Here we adopt the definition
offered by Reader and Laland (2003) that cultures are ‘group-typical behaviour
patterns shared by members of a community that rely on socially learned and
transmitted information’ (p. 151). This seems sufficiently broad to account for all
instances which one might consider ‘cultural’ in humans, but it means that theoret-
ically, culture might emerge in any species with sufficient propensity for social
learning. We humans are pre-eminent (Dean et al. 2014) in our capacity for ‘cumu-
lative culture’—the ability to modify learned behaviours to become more complex
and/or efficient, which can be transmitted and further improved by others—but an
increasingly large body of research shows that our fellow animals are also capable of
a functionally significant degree of cultural inheritance and diversity.

The adaptive benefits of social learning appear clear; the alternative, individual
innovation, is potentially costly as it may require a significant time or energy
investment, produce only minor or non-existent rewards or be physically dangerous.
Instead, the capacity to learn from others allows individuals to reap the benefits of
others’ useful innovations whilst minimising the costs. Furthermore, this may benefit
not only the learner but also any offspring or other kin to whom the behaviour
subsequently spreads. In this sense, culture acts as nature’s ‘second inheritance
system’ (Whiten 2005). However, social learning may not always be adaptive. An
indiscriminate social learner runs the risk of copying costly or other suboptimal
behaviours (Laland and Williams 1998; Kendal et al. 2005). To avoid these, the
evolution of social learning ‘biases’ (or ‘strategies’ in some literature) has been

212 S. K. Watson et al.

predicted, to guide when to use social information and whom best to acquire it from
(Giraldeau et al. 2002; Laland 2004).

This chapter reviews what is known about social learning in non-human primates,
with particular emphasis on the cognitive biases that underpin it, and the important
light it sheds on our understanding of the emergence, propagation and maintenance
of culture. Studies of social learning in non-primate species will also be noted when
pertinent, in order to situate the primate literature within a broader biological
context.

2 Observations of Culture in the Wild

Early whispers of non-human primate culture were heard when, in 1953, a young
Japanese macaque, Imo, carried pieces of sand-covered sweet potato to a stream and
washed off the sand before eating them (Kawai 1965). In the months and years that
followed, other macaques began using this technique, and it eventually became a
common behaviour in Imo’s group, leading the researchers studying the macaques to
label this behaviour as ‘pre-culture’. Whilst later scrutiny of the spread of sweet
potato washing behaviour suggested it did not provide strong evidence for cultural
transmission (e.g. due to factors such as human provisioning; Galef 1992), the case
of the Japanese macaques established the study of social transmission and culture in
wild primates, which led to a plethora of exciting discoveries in the animal kingdom.

A further influential finding came when, following observations of group differ-
ences in chimpanzee behaviours across different sites (Goodall 1986; Boesch et al.
1994), Whiten et al. (1999, 2001) systematically collated data from seven long-term
chimpanzee field sites across Africa and found evidence of multiple variations in
behaviours between communities, inferred to be cultural. The researchers listed
39 behavioural traditions which were common (either customary or habitual) in
some groups, yet absent in others, without obvious ecological explanation, ranging
from handclasp grooming to different termite fishing and nut-cracking behaviours
(see Fig. 1). The discovery of this large number of putative cultural variations in our
closest relatives challenged the assumption of complex culture being uniquely
human.

On the heels of these findings, researchers collated data from other great ape
species. Van Schaik et al. (2003) conducted a similar analysis of behavioural
traditions in orangutans across six different field sites and identified 24 cultural
variants that were habitual or customary at some sites and absent at others. The
authors concluded that orangutans also possess multiple-tradition cultures. A more
recent analysis supports these earlier conclusions, finding that ecological and genetic
differences accounted for only a small proportion of the variation seen between
groups (Krützen et al. 2011). Emerging evidence has also been found for a number
of traditions in wild gorillas (Robbins et al. 2016, but see Neadle et al. (2017) for a
discussion on the ontogeny of food cleaning behaviours) and bonobos (Hohmann

Culture and Selective Social Learning in Wild and Captive Primates 213

and Fruth 2003), although we have less information on bonobos than for chimpan-
zees or orangutans.

Whilst the ‘exclusion’ method of inferring social learning where ecological or
genetic explanations appear untenable can never be conclusively watertight (Laland
and Janik 2006; Langergraber et al. 2011: Robbins et al. 2016, explicitly refer to the
traditions they identify as ‘putative’), additional lines of evidence have more recently
supported the role of social learning in the maintenance of the above behaviours.
First, a series of diffusion experiments in which different foraging techniques are
‘seeded’ in different groups has shown that these will spread in ways that confirm a
capacity to transmit and sustain multiple traditions (Whiten 2011; Whiten et al.
2016). Second, recent advances in statistical techniques have allowed researchers to
assess the role of social learning in the spontaneous spread of a novel behaviour in
the wild. For example, Hobaiter et al. (2014) used a dynamic form of network-based
diffusion analysis (NBDA, Franz and Nunn 2009) to systematically chart the spread
of a novel leaf-sponging behaviour in a group of wild chimpanzees via social
transmission, providing the first direct evidence that at least some of the observed
behavioural variation in wild chimpanzees is likely due to cultural learning. This
conclusion is further supported by Lonsdorf (2006), who found a correlation
between the amount of time infant chimpanzees spend with their mother and the
rate at which they become proficient at termite fishing, as well as Luncz and Boesch
(2014, see below) comparing neighbouring communities where neither genetic nor
ecological appear able to explain behavioural variations. It is also worth noting that
the exclusion method of identifying animal cultures has been criticised for being too
conservative, as it precludes examination of ecologically driven cultural differences
(Koops et al. 2014; Sanz and Morgan 2013).

Whilst the great apes, in particular chimpanzees, display the most diverse reper-
toire of cultural behaviours documented thus far, the first candidates for cultural

Fig. 1 The putative cultures of wild chimpanzees (after Whiten 2005). ‘Customary’ acts are those
typical in a community; ‘habitual’ are less frequent yet consistent with social learning. Each
community displays its own profile of such local behavioural variants, providing evidence of a
unique culture for each locality. Numbers identify behaviour patterns in the catalogue attached to
Whiten et al. (1999). For a more recent and detailed version focused on nut-cracking variations, see
Carvalho and McGrew (2012)

214 S. K. Watson et al.

behaviours came from monkeys. Evidence has since accrued that several monkey
species exhibit their own forms of culture. Notably, Perry et al. (2003) described a
number of ‘social customs’ found in some, but not all, groups of capuchins studied in
Costa Rica. These included a number of social ‘games’ and other, seemingly bizarre,
conventions such as poking fingers in each other’s nostrils and eyes. Whilst the
functions of these conventions are difficult to pin down, the authors suggest that they
may serve to enhance social bonds between participants and perhaps highlight them
to others. Whatever the function, the distributions of conventions across groups are
strongly suggestive of cultural transmission. Additionally, Santorelli et al. (2011)
identified a number of behavioural traditions, many in the social domain, in wild
spider monkeys, and peculiar stone-handling behaviours seen in numerous provi-
sioned groups of Japanese macaques also appear to display the pattern of multiple, if
narrowly constrained, cultural traditions (Leca et al. 2007).

Whilst it might be tempting for us to think of primates as distinctive in their
capacity for cultural transmission, researchers have uncovered convincing evidence
for culture in other animal taxa, notably in cetaceans and in birds. Whitehead and
Rendell (2014) discuss evidence for a range of putative traditions in cetaceans, most
notably in the domain of song transmission in the baleen whales (also well
documented in birds; Slater 1986), but also including a number of foraging traditions
(Krützen et al. 2005; Allen et al. 2013). Implementation of the aforementioned
technique, NBDA, has provided some of the strongest evidence for social learning
in the wild in any species, documenting the gradual spread of a particular foraging
method, lob-tail feeding, in a population of over 600 humpback whales (Allen et al.
2013).

In addition to vocal culture, there may also be tentative evidence for tool-making
traditions in birds. Hunt and Gray (2003) described evidence for sophisticated tool-
making in wild New Caledonian crows and, finding that the distribution could not be
linked to ecological correlates, suggested a role for social learning. They argued that
the distribution of tool types across New Caledonia indicates cumulative technolog-
ical culture. However, the extent to which the tool manufacture is necessarily
socially learned remains in doubt, as hand-reared crows in captivity also display
tool-making skills (Kenward et al. 2005, but see also Holzhaider et al. 2010), and
experiments have failed to support social learning of alternative techniques (Logan
et al. 2016). Recently, researchers also found that the experimental removal and
replacement of individuals from a group of homing pigeons improved the efficiency
of homing routes over successive generations through social learning and refine-
ment, satisfying the main criteria for cumulative culture (Sasaki and Biro 2017). This
finding is important not only for identifying a putative case of cumulative culture in a
non-human species, a capacity many argue to be uniquely human, but also for
emphasising that culture extends beyond foraging behaviour which is the primary
focus of a majority of social learning studies.

Culture and Selective Social Learning in Wild and Captive Primates 215

3 Mechanisms of Information Transmission

The study of the social learning processes that underlie culture can broadly be split
into two main categories: mechanisms and biases. Mechanisms, or psychological
processes, refer to the how of information transmission. Whereas biases
(or strategies) refer to when and whom to copy (Laland 2004; Hoppitt and Laland
2013, but see Heyes (2016) for discussion of limitations in the identification of
underlying cognitive mechanisms). Whilst this chapter focuses primarily upon the
biases of social learning, here we provide a brief introduction to transmission
mechanisms. Some researchers argued early on that only the (supposedly) more
cognitively complex processes of imitation and teaching would allow for transmis-
sion of sufficient fidelity to create stable, between-group traditions (Galef 1992, but
see Claidiere and Sperber 2010). Thus, it was the study of mechanisms and partic-
ularly the exploration of the process of imitation which formed the primary research
focus in experiments on social learning that until recently were largely restricted to
captive primates and other animals, where rigorous experimental and contrasting
control conditions can be engineered.

Social learning mechanisms range from the simplest processes of local and
stimulus enhancement (increased attention respectively towards locations and
objects one observes others acting on) to more complex mechanisms such as
teaching and copying (Heyes 1994). Experimenters have distinguished between
two principal types of copying mechanisms: imitation, which refers to copying the
actions (often conceived of a bodily actions) of another individual (Whiten et al.
2004), and emulation, which refers to learning focused only on desirable environ-
mental results of other’s actions (Tomasello et al. 1987). Whilst initial tests with
chimpanzees suggested they are capable of only emulation (Tomasello et al. 1987),
later studies provided a diversity of evidence for imitation, including recognisable
successes in ‘do-as-I-do’ games and other tests that require matching of bodily
actions [Custance et al. (1995), Buttelmann et al. (2007), although it should be
noted that some of these studies were conducted with hand-reared chimpanzees; see
also Fuhrmann et al. (2014), but see Tomasello et al. (1997), Tennie et al. (2012) for
evidence of marked limitations in copying arbitrary or novel gestures in untrained
chimpanzees]. Other studies with captive chimpanzees (and children) found evi-
dence for flexible use of imitation and emulation in chimpanzees (Hopper et al.
2008), with replication of a whole sequence of actions being preferred when a
complex task was relatively opaque (“program-level imitation”, Byrne and Russon
1998) and a more emulative response made when it was sufficiently transparent that
some actions could be seen to be redundant and were not copied (Horner and Whiten
2005). Other studies revealed some contexts where emulation does not enable
chimpanzees to solve a complex task, whereas seeing another chimpanzee complete
it allows success by copying (Hopper et al. 2007, 2015) and other contexts where
chimpanzees are flexible enough to succeed by emulation when direct imitation is
made impossible (Tennie et al. 2010).

216 S. K. Watson et al.

Limited evidence for bodily imitation in monkeys has also emerged, from species
including marmosets (Voelkl and Huber 2000), vervet monkeys (van de Waal and
Whiten 2012) and capuchins (Fragaszy et al. 2011, although see Dean et al. 2012).
There is thus limited evidence in monkeys and apes for one of the two features that
Galef emphasised as crucial to human culture: imitation. However, perhaps more
important is that a range of cultural diffusion experiments have demonstrated that
species of monkeys and apes can transmit and sustain traditions, whether or not these
are driven by imitative or emulative copying processes (reviewed in Whiten et al.
2016, but see Bandini and Tennie 2017). By contrast, there is little evidence for
Galef’s other factor, teaching, in non-human primates (although see Musgrave et al.
2016 for a recent example of tool transfer interpreted as teaching in chimpanzees).
There is, however, evidence for teaching (defined in functional rather than inten-
tional terms) in non-primate animals such as meerkats (Thornton and McAuliffe
2006), ants (Franks and Richardson 2006) and pied babblers (Raihani and Ridley
2008). It should be noted, however, that whilst we include teaching here to illustrate
its importance in the study of cultural transmission, it is not a mechanism in the
learner, but rather in the teacher, coupled with complementary social learning
processes in the learner. For example, the teaching process of demonstration couples
well with a capacity for imitation in the learner (Hoppitt and Laland 2008). For
discussions on the evidence for teaching and its significance in non-human animals,
see Thornton and Raihani (2010) and Hoppitt and Laland (2008).

4 Social Learning Biases

More recently, research has begun to tackle the potential biases or ‘strategies’ which
influence when and from whom animals learn socially. Given that learning from
others indiscriminately may result in the transmission of maladaptive behaviours
(Laland and Williams 1998; Pongracz et al. 2003), Laland (2004) suggested that
individuals should be selective in when and from whom they engage in social
learning, highlighting a number of potential social learning strategies (or biases as
we refer to them here). Indeed Coussi-Korbel and Fragaszy (1995) had earlier
suggested that the social dynamics of groups would likely lead to biases in social
learning. An adaptive bias may allow individuals to select the most productive
behaviour between multiple options and overwrite pre-existing behaviours when
innovations or a changing environment renders them inefficient. Whilst a number of
such biases have been suggested by researchers, empirical investigation of many
remains lacking, and Heyes (2016) points out that the underlying mechanisms
remain largely unspecified. Here we limit our discussion to the social learning biases
that have received the most research attention to date (Table 1), ranging from
frequency-based copying (e.g. copying common behaviours) to model-based biases
(e.g. copying high-ranked individuals). One of the most studied—and contentious—
biases addressed in recent years, in both humans and animals, is that of conformity.

Culture and Selective Social Learning in Wild and Captive Primates 217

4.1 Conformity and Majority Bias

In the 1950s, Solomon Asch (Asch 1951) showed that when people were faced with
a unanimous majority giving incorrect answers on a line judgement task, roughly
one third of individuals conformed to the group decision even though it was visibly
wrong. These findings were replicated across cultures (Bond and Smith 1996) and
more recently in children (Haun and Tomasello 2011), suggesting a powerful
predisposition for humans to conform to the judgement of the majority.

Circumstantial evidence for conformist behaviour in primates arose in open
diffusion studies such as that by Whiten et al. (2005) which utilised a puzzle box
(the ‘panpipes’) that could be opened using either of two tool-based techniques
(‘poke’ or ‘lift’). A high-ranking female from each of the two groups was trained in
one of these methods and demonstrated it in front of the rest of their group, who later
had opportunities to interact with the panpipes. After 2 months, it was found that
although some individuals did open with the alternative method, there was a
significant tendency for individuals to converge on the option most common in
their group. Whiten et al. (2005) suggested this to be indicative of a conformity
effect. A similar effect has since been recorded in capuchin monkeys (Dindo et al.
2008, 2009). Critics have pointed out that chimpanzees are often reluctant to give up
a first-learned behaviour (Hopper et al. 2011; Marshall-Pescini and Whiten 2008;
Hrubesch et al. 2009; Harrison and Whiten 2018), so it may be that individuals were
simply returning to their original method after a period of exploration (van Leeuwen
and Haun 2013). However, there are also several documented examples of flexible,

Table 1 Definitions and examples of selected social learning biases

Category Term Definition Selected examples

Frequency-
based
biases

Majority
bias

A tendency to copy the
behaviour of the majority when
learning a task

Chimpanzees, Haun et al. (2012)
Cf. van Leeuwen and Haun
(2013)

Conformity A tendency to forgo one’s own
behavioural preference in favour
of that used by the majority

Fish, Pike and Laland (2010)
Vervet monkeys, van de Waal
et al. (2013)
Chimpanzees, Luncz and Boesch
(2014)
Birds, Aplin et al. (2015a, b)

Model-
based
biases

Rank bias A tendency to copy individuals
of high social rank

Chimpanzees, Horner et al.
(2010); Kendal et al. (2015)

Sex bias A tendency to copy individuals
of one sex over the other

Vervet monkeys, van de Waal
et al. (2010)

Kin bias A tendency to copy one’s own
kin

Chimpanzees, Matsuzawa et al.
(2001)
Vervet monkeys, van de Waal
et al. (2012, 2014)

For a more exhaustive review, particularly of ‘when’ biases, see Laland (2004) and Rendell et al.
(2011)

218 S. K. Watson et al.

non-conservative problem-solving by chimpanzees (Hopper et al. 2015; Yamamoto
et al. 2013)

Subsequent evidence for conformity in chimpanzees has been mixed. No study
has yet found an example of conformity in which an individual gives up a first-
learned behaviour in favour of that used by the majority. However, Luncz and
Boesch (2014), having earlier identified consistent differences between
neighbouring groups in aspects of nut-cracking tool selection, found that females
who transfer between these groups are as likely as males to match their local group
preferences, suggesting they must have conformed to them since their arrival. A case
study of one wild female chimpanzee that had recently migrated to a new group
showed that, over time, her behavioural repertoire became progressively more
similar to that of the group. However, without systematic testing we cannot be
sure this was driven by majority influence, rather than other biases. Working with
captive chimpanzees on a token exchange task, van Leeuwen and Haun (2013)
found that individuals were motivated to switch methods by greater rewards, but not
by the behaviour of the majority. This suggests other factors, such as maximising
payoffs, may be more critical in motivating apparently conformist behaviour than
majority influence. Moreover, Vale et al. (2017) found that captive chimpanzees
who were trained to prefer one colour of food and then moved to a group with a
strong preference for the alternative colour did not conform to the foraging prefer-
ences of their new ‘host’ group, although they did feed from this food more
frequently than before. Watson et al. (2018) found that lone minority individuals
trained on a method of opening a puzzle box, whilst the rest of their group were
trained on an alternative method, rapidly converged on the behavioural preference of
the majority. However, this occurred after observing just one or two groupmates,
meaning that they did not have first-hand knowledge of the majority preference. In
contrast, dyads of chimpanzees trained on alternative methods never switched to
using their partner’s preferred method. The authors argue that it is possible that,
when in a group context, chimpanzees may make inferences about group-level
behaviour preferences based on a limited sample and act accordingly, in a potentially
conformist manner.

Arguably, the strongest evidence for conformist transmission in non-human
animals has been found in non-primate species. In Pike and Laland (2010), nine-
spined stickleback fish independently learned that one of the two feeders was richer
in food than the other. The positions of the feeders were then reversed, and the fish
could observe others feeding at the previously less rewarding location. When these
demonstrator fish were removed and the observer was once again allowed to choose
a feeding location, they preferentially used the one favoured by the majority. This
effect increased disproportionately according to the relative size of the majority,
thereby fulfilling the most stringent definitions of conformist bias. In another
convincing example of conformity in a non-primate species, Aplin and colleagues
(Aplin et al. 2015a, b) found that not only did experimentally induced innovations
(alternative methods of operating a foraging box) spread throughout groups of great
tits in a manner consistent with majority-biased transmission but they expressed an
exaggerated tendency to do so. In addition, when individuals migrated between

Culture and Selective Social Learning in Wild and Captive Primates 219

groups, if the local method was different from their own, then a majority of birds
adopted the behaviour most common in their new community [but see van Leeuwen
et al. (2015) and Aplin et al. (2015b) for further debate]. This echoes the results from
an experiment that seeded alternative options in wild vervet monkeys (van de Waal
et al. 2013). Four groups of wild vervet monkeys were repeatedly provided with
two boxes of coloured corn, one pink and one blue. For each group, one of these
colours was initially made to taste bitter and unpalatable. Then, after monkeys had
learned to avoid one colour, no more bitter material was added to either colour of
corn. When males later migrated between groups with different preferences, and
once they were not outranked by residents at the boxes, nine of ten switched to eat
the colour that was locally preferred, expressing an apparently striking degree of
conformity. In a follow-up study, the preferences of low-ranking females who had
permanently splintered from their natal group have been monitored. The low rank of
these females meant that in their original group, they ate more of the previously
bad-tasting food than others yet, after the split, they ate exclusively the colour
preferred in the group from which they split, even though both colours were
available and neither tasted bad, emphasising the potency and durability of socially
learned behaviours (van de Waal et al. 2017). However, whether the individuals
were influenced by the behaviour of the majority of individuals or by certain classes
of individual within that group (as discussed below) remains unclear. The lack of
clear experimental evidence for conformity in chimpanzees may suggest …

Is Overimitation a Uniquely Human Phenomenon? Insights From Human
Children as Compared to Bonobos

Zanna Clay
University of Birmingham and Durham University

Claudio Tennie
University of Birmingham and University of T€ubingen

Imitation is a key mechanism of human culture and underlies many of the intricacies of human social life,
including rituals and social norms. Compared to other animals, humans appear to be special in their readiness
to copy novel actions as well as those that are visibly causally irrelevant. This study directly compared the
imitative behavior of human children to that of bonobos, our understudied great ape relatives. During an
action-copying task involving visibly causally irrelevant actions, only 3- to 5-year-old children (N = 77) readily
copied, whereas no bonobo from a large sample did (N = 46). These results highlight the distinctive nature of
the human cultural capacity and contribute important insights into the development and evolution of human
cultural behaviors.

Debate over the uniqueness of human culture and
the role that imitation plays in its evolutionary and
ontogenetic development has become the focus of
increasing research attention (Caldwell & Millen,
2009; Meltzoff & Prinz, 2002; Tennie, Call, & Toma-
sello, 2009; Tomasello, 1999; Whiten, 2016). In par-
ticular, comparative research has attempted to
identify what makes human culture special as com-
pared to that of other great apes, and to identify

similarities and differences in the underlying social
learning mechanisms (Tomasello, 1996; Tomasello,
Savage-Rumbaugh, & Kruger, 1993; Vale et al.,
2016; Whiten, 2013; Whiten, 2016). Imitation, which
in this article we define as the faithful copying of
others’ body movements, has elicited particular
attention as it has been proposed to form a core
component of human culture, enabling the acquisi-
tion of causally opaque material culture and action-
based cultures (e.g., gestures and dance), as well as
contributing to their accumulation over time
(Acerbi & Tennie, 2016; Dean, Kendal, Schapiro,
Thierry, & Laland, 2012; Gergely & Csibra, 2006;
Tennie et al., 2009; but see Caldwell & Millen, 2009;
Caldwell, Schillinger, Evans, & Hopper, 2012;
Morin, 2015). Imitation is also involved in many of
the complexities of human social life, including
norms, rituals, and conventions (Legare & Nielsen,
2015; Legare & Watson-Jones, 2015; Meltzoff &
Prinz, 2002; Rakoczy, Warneken, & Tomasello,
2008).

A striking feature of human imitation is the
extent to which humans are prepared to imitate
actions that appear causally irrelevant (Horner &
Whiten, 2005; Lyons, Young, & Keil, 2007; McGui-
gan, Whiten, Flynn, & Horner, 2007). This phe-
nomenon, termed “overimitation,” emerges early
during childhood (Lyons et al., 2007; Over & Car-
penter, 2012). It occurs in both Western and non-

[Article updated on July 26, 2017, after first online publication
on July 24, 2017: University of T€ubingen was added to the affilia-
tions of Dr. Claudio Tennie.]

We thank Claudine Andr�e, Fanny Minesi-Andre, Raphael
Belais, Pierrot Mbonzo, Dominique Morel, and Valery Dhanani
for their collaboration at Lola ya Bonobo Sanctuary and the Min-
istry of Research and the Ministry of Environment in the Demo-
cratic Republic of Congo for supporting our research (MIN.RS/
SG/004/2016). We thank the staff members at the Lola ya
Bonobo Sanctuary for their support, particularly to Stany
Mokando and Jean-Claude Nzumbi. We thank Brian Hare for
support and Chris Krupenye for coordinating bonobo testing.
and Charles Clay for developing the stimuli prototype We thank
Lynsey Rutter, Lauren Deere, and the staff at ThinkTank Science
Museum for enabling our research. We are grateful to all the
children and families who participated in this research. We thank
Polly Cowdell for reliability coding and Harriet Over, Eva
Reindl, and Elisa Bandini for the comments on earlier drafts of
the manuscript. We thank three anonymous reviewers for their
valuable comments on this manuscript. This research was funded
by the People Programme (Marie Curie Actions) and the Euro-
pean Research Council under the European Union’s Seventh
Framework Programme for research, technological development,
and demonstration under REA grant agreement no. 628763
awarded to Zanna Clay. The research was also supported by a
grant awarded to Claudio Tennie from the Economic and Social
Research Council (ESRC; ES/K008625/1).

Correspondence concerning this article should be addressed to
Zanna Clay Department of Psychology, Durham University,
South Road, Durham, DH1 3LE, United Kingdom
Email: [email protected]

© 2017 The Authors
Child Development © 2017 Society for Research in Child Development, Inc.
All rights reserved. 0009-3920/2018/8905-0007
DOI: 10.1111/cdev.12857

Child Development, September/October 2018, Volume 89, Number 5, Pages 1535–1544

Western cultures (Berl & Hewlett, 2015; Nielsen &
Tomaselli, 2010) and gradually increases with age,
starting from around 3 years old (McGuigan, Glad-
stone, & Cook, 2012; McGuigan, Makinson, &
Whiten, 2011; McGuigan et al., 2007). Overimitation
is thought to underlie many human sociocultural
behaviors including rituals and other forms of nor-
mative behavior (Legare & Nielsen, 2015; Legare &
Watson-Jones, 2015; Nielsen, Kapit�any, & Elkins,
2015). It is also involved in cumulative technologi-
cal culture; thus, it was suggested that children’s
overimitation is driven by their need to learn about
causally opaque cultural artifacts (Lyons et al.,
2007). This may be especially important in cases
where cultural accumulation has led to artifacts
whose causal properties have become complex and
opaque; that is, copying is required to produce or
use them (Gergely & Csibra, 2006; Lyons et al.,
2007; Whiten, McGuigan, Marshall-Pescini, & Hop-
per, 2009). Nevertheless, recent research has shown
that overimitation is strongly motivated by social
factors, such as to affiliate with or “be like the
other” (Keupp, Behne, & Rakoczy, 2013; Nielsen,
2006; Nielsen & Blank, 2011) and to conform to per-
ceived conventions and norms (Herrmann, Legare,
Harris, & Whitehouse, 2013; Keupp et al., 2013;
Legare & Nielsen, 2015). For instance, children are
more likely to copy when the task is framed as
being normative (Keupp et al., 2013; Legare & Niel-
sen, 2015; Moraru, Gomez, & McGuigan, 2016) and
after being primed with third-party ostracism (Over
& Carpenter, 2009a, 2009b). They can infer friend-
ship and social status from watching others imitate
(Over & Carpenter, 2015) and trust individuals
more that have imitated them (Over, Carpenter,
Spears, & Gattis, 2013).

The study that originally reported what came to
be known as overimitation (Horner & Whiten,
2005) contrasted children’s copying with an appar-
ent absence of this behavior in captive chim-
panzees, a finding that has since been replicated
for orangutans (Nielsen & Susianto, 2010). Children
were willing to insert a stick into both an opaque
and a clear box in order to retrieve a reward, even
though the insertion in the latter was visibly cau-
sally irrelevant. Although this influential study has
stimulated a plethora of studies, it is limited in its
ability to detect overimitation in the sense in which
we define it here (i.e., with a focus on action copy-
ing). This is because pure action copying could not
be distinguished from other forms of social learn-
ing due to the fact that the captive chimpanzees
were already competent stick users. In other
words, this stick-based task could detect copying

of the location of the stick insertion, rather than
copying the action itself. Thus, for both the apes
and the children, this task more accurately tested
“local overenhancement” and/or overemulation
learning (see Tennie, Call, & Tomasello, 2006 for
discussion). Note this experiment also involved a
“two-target task,” where objects could be moved
to one of two sides. Copying here was likewise
likened with imitation; however, although this task
controlled for local enhancement, it could not fully
pinpoint action-based imitation as it could not
exclude so-called “object-movement reenactment”
(Custance, Whiten, & Fredman, 1999; Heyes &
Ray, 2000). Later studies, which added actions that
neither changed nor moved objects, were generally
unable to find action copying in chimpanzees (Ten-
nie, Call, & Tomasello, 2012) but found it in chil-
dren (Legare, Wen, Herrmann, & Whitehouse,
2015).

Given these constraints and the fact that no
equivalent data are yet available for the capacities
of our other closest living relative, the bonobo (Pan
paniscus), the question of whether overimitation is
uniquely human among the great apes remains
unresolved. Nevertheless, it is acknowledged that
some animals will copy some actions under certain
conditions (Huber et al., 2009). This includes, for
example, the so-called “Do as I do” studies which
involve heavily trained animals (Call, 2001;
Custance, Whiten, & Bard, 1995; Miles, Mitchell, &
Harper, 1996). There is also evidence from “encul-
turated” great apes that have received extensive
experience in human-centered environments (Bjork-
lund, Bering, & Ragan, 2000; Buttelmann, Carpen-
ter, Call, & Tomasello, 2007; Byrne & Tanner, 2006;
Call, 2001; Carrasco, Posada, & Colell, 2009; Hayes
& Hayes, 1952; Miles et al., 1996). Importantly,
however, the extent to which ecologically relevant
animals—that is, those that are untrained and
also unenculturated—spontaneously copy actions
remains hotly debated (Whiten, 2016; Whiten, Cus-
tance, Gomez, Teixidor, & Bard, 1996; Whiten, Hor-
ner, Litchfield, & Marshall-Pescini, 2004; Zentall,
1996, 2006). The lack of resolution is partly due to
methodological constraints in distinguishing imita-
tion from other social learning processes (Heyes &
Ray, 2000; Tennie et al., 2006).

To date, most research on great ape social learn-
ing has focused on two-target tasks involving
experimental puzzle boxes that can be opened in
more than one way in order to retrieve a reward
(Horner & Whiten, 2005; Horner, Whiten, Flynn, &
de Waal, 2006; Whiten, Horner, & de Waal, 2005;
Whiten et al., 1996, 2009). Although two-target

1536 Clay and Tennie

tasks provide many key insights into the factors
shaping animal cultural transmission (Galef, 2015;
Whiten, 2016), they cannot clearly distinguish imita-
tion from other learning mechanism due to the fact
that the demonstrator movements are confounded
with the object movements (Custance et al., 1999;
Whiten et al., 2004, 2009). Thus, animals can plausi-
bly solve the tasks via emulation, which is the
copying of results of actions on the environment
(Heyes & Ray, 2000; Tennie et al., 2006). Moreover,
given that chimpanzees are able to copy the move-
ments of the apparatus in two-target tasks without
seeing actions leading to these results (Hopper,
Lambeth, Schapiro, & Whiten, 2008), emulation can-
not be ruled out. Successful performance on two-
target tasks (Custance et al., 1999) is also wide-
spread in the animal kingdom (Galef, 2015; even in
reptiles, Kis, Huber, & Wilkinson, 2015), thus limit-
ing its usefulness for determining what truly makes
human cultural learning special or why wild great
apes, especially chimpanzees and orangutans, are
such expert tool users (Meulman & van Schaik,
2013; Sanz, Call, & Boesch, 2013; Whiten et al.,
1999).

Although some great apes will spontaneously
copy familiar actions (Fuhrmann, Ravignani, Mar-
shall-Pescini, & Whiten, 2014; Tennie et al., 2012),
evidence of novel action copying— which is a core
component of human culture—has not been convinc-
ingly demonstrated using two-target tasks. This is
because the target actions generally fall within the
species-typical repertoire, such as pulling or poking
(Tennie et al., 2012). Given the importance of copy-
ing novel actions in human culture, it is essential to
determine whether great apes can copy novel
actions. So far, only two studies with captive chim-
panzees have addressed this question, accounting for
the various methodological confounds (Tennie et al.,
2012; Tomasello et al., 1997). Both tested imitation of
novel actions where no physical information about
the task was available, that is, removing the possibil-
ity of emulation. Although one of the studies found
some evidence of familiar action copying in a single
chimpanzee subject (Tennie et al., 2012), neither
detected novel action copying in any subject.

Here, we addressed the confounds of previous
studies by designing a paradigm that could test for
pure overimitation while excluding other social
learning mechanisms. We did this by using purely
manual gestures as the target actions where no
physical information was provided about the solu-
tion. In order to probe the potential for overimita-
tion, some of the target actions were visibly
causally irrelevant. We included target actions that

were, to our knowledge, novel or at least very unli-
kely to be part of a species-typical repertoire.

To promote the possibility of demonstrating imita-
tion by great apes, we focused our attention to bono-
bos, a species of great ape that is equally as related to
humans as chimpanzees yet comparatively less stud-
ied. For a number of reasons, bonobos may represent
a more promising candidate species to demonstrate
imitation than chimpanzees. This is because bonobos
outperform chimpanzees on sociocognitive tasks
(Herrmann, Hare, Call, & Tomasello, 2010) and show
enhanced social orientation (Kano, Hirata, & Call,
2015; Kret, Jaasma, Bionda, & Wijnen, 2016) and high
levels of social tolerance (Hare & Kwetuenda, 2010).
Given the inherently social nature of imitation, an
activity requiring both social attention and social tol-
erance, the enhanced social orientation of bonobos
may enhance their imitative capacity. The current
study explored evidence for pure, spontaneous
action imitation in a large sample of untrained and
nonenculturated sanctuary-living bonobos as com-
pared to 3- to 5-year-old children. This sample is the
largest of its kind ever used with a single great ape
species for a pure action imitation study. If lower
social tolerance and the methodological constraints
emerging from the nature of previous tasks impede
the performance of great apes, we should expect
bonobos to show evidence of overimitation. If
overimitation is a human unique behavior, we
should not expect bonobos to copy any of the visibly
causally irrelevant actions.

Method

Participants

Seventy-seven typically developing children, aged
3–5 years, participated in this study
(Mage = 4.4 years, range = 3.1–5.9 years; N = 43
males). We selected this age range as children of this
age are already manually competent, show reliable
evidence of imitation behavior (e.g., Hopper et al.,
2008; Horner & Whiten, 2005; McGuigan et al., 2007;
Whiten et al., 1996), and are comfortable being tested
individually, enabling more cross-species compar-
isons. Children were recruited from ThinkTank
Science Museum in Birmingham, West Midlands,
UK, and randomly assigned to conditions. Child test-
ing took place between April and December 2015.
Using parental questionnaires, we determined that
all were typically developing, had normal or cor-
rected to normal vision, and spoke English as their
first language: 69 children were monolingual,
whereas eight were bilingual (English + Urdu/

Overimitation in Children as Compared to Bonobos 1537

Punjabi/Spanish/Sinhalese/French/Arabic/Polish).
The sample comes from an area of high ethnic diver-
sity consisting of approximately 58% Caucasian, 27%
Asian/British Asian, 9% Black/African/Caribbean,
6% mixed children; the participants came from
working middle-class backgrounds (estimated from
census data for each county, Office of National
Statistics, 2011). Five children refused to participate
in the task and were excluded from analyses. The
remaining children were randomly assigned to one
of three conditions (N = 27 in the rub and rotate
condition [uncommon actions]; N = 26 in the cross
and trace condition [typical actions] and N = 19 in
the control condition).

Forty-six nonenculturated and untrained bono-
bos also participated (Mage = 11.3 years, range = 3–
29 years; N = 25 male). Testing took place in June
2015. The bonobos were housed at Lola ya Bonobo
Sanctuary, a naturalistic forested sanctuary, in the
Mont Ngafula district, Kinshasa, DR Congo (see
Supporting Information). The majority of subjects
were orphans, having arrived at the sanctuary as
victims of the bushmeat and pet trades. Three were
born and mother reared at the Sanctuary. Following
several years of rehabilitation within a cohort
group, individuals are integrated into large, mixed
age groups. The majority of our subjects (N = 36)
were housed in large, outdoor enclosures. We addi-
tionally tested 10 juveniles housed in a nursery.
Nursery individuals were cared for by human sub-
stitute mothers within a naturalistic forested enclo-
sure with age-matched peers. For subjects from the
main enclosures, the experiments were conducted
in their sleeping dormitories and before their morn-
ing feed in order to maximize motivation. Testing
rooms (15 m2) had a meshed ceiling with wide bars
through which the experimenter could hand items
to the subject, which they could then manipulate
themselves inside their testing room. In the nursery,
the experiments were conducted face to face with
the experimenter within their enclosures and sleep-
ing dormitories.

Materials and Procedure

For all participants, the task involved the opening
of a small box (10 9 6 9 3 cm, Figure 1), made of
two halves of a single piece of wood. A small cham-
ber was carved out in the middle to place the reward,
held in place by a peg and hole mechanism.

For both test conditions, each participant first took
part in a demonstration phase followed by a test
phase. All participants were tested individually in a
quiet testing area. Children’s parents waited behind

an occluder so were not visible. All participants
observed a human demonstrator who, facing the
participant, looked at the box and then slowly per-
formed two consecutive actions onto it before open-
ing it to reveal the reward inside, which was
provided to the participant. Due to health and safety
reasons, children received stickers, whereas bonobos
received a food reward (apple piece)—as is typical in
such cross-species studies (e.g., Herrmann & Toma-
sello, 2015; Hopper et al., 2008). This procedure was
repeated three times. Between demonstrations, the
demonstrator refilled the box behind an occluder,
preventing the refilling and closing from being seen.

We tested imitation for actions that we considered
plausibly typical or uncommon based on our direct
observations of actions performed by bonobos and
children, and our knowledge of their typical manual
behaviors. In the “uncommon” action condition
(rub–rotate), the demonstrator placed the back of the
right hand on the top of the box and slowly rubbed
it in a clockwise circular motion four times. Next, the
demonstrator raised the right hand into the air next
to the box and slowly rotated the wrist four times.
Given the difficulty in ascertaining whether a
demonstrated behavior is truly novel for a long-lived
species (Zentall, 2001), we considered these two
actions to be “uncommon” on the basis that, to our
knowledge, they had not been previously observed
in the study population or any other observed by the
authors, and were also unlikely to occur within the
species-typical repertoire. We also included a “typi-
cal” action condition (cross–trace), which included
actions that were rare but nevertheless fell within
the ape species-typical repertoire and have also been
observed in this bonobo population (Z. Clay, per-
sonal observations). Here, the demonstrator held the
box (left hand) and with the index finger, slowly
traced a diagonal cross across the top of the box.
Next, the demonstrator used this finger to trace
around the groove of the box, around its full diame-
ter. There was also a control condition (children
only) in which everything remained the same except
that no target actions were demonstrated.

Following each demonstration, the demonstrator
pretended to refill the box behind the occluder but
swapped it with a replica box, which was identical in
dimensions and external appearance except that it
did not actually open (the groove resembled that of
the other box, but in reality was not deep enough to
open). The use of a replica maximized the chances of
observing imitation once species-typical solutions
were individually discovered to be ineffective.

During the test phase, each participant was pro-
vided with the replica box, without verbal

1538 Clay and Tennie

instruction. Participants were given up to 2 min to
interact with the box. Regardless of performance, all
participants were rewarded at the end of the trial. Tri-
als were videotaped using a digital Sony Handcam
CX330 camcorder (www.sony.co.uk) mounted on a
tripod.

Coding

The occurrence of accurate matches of any of the
four demonstrated actions was coded from video (yes
or no). A second coder, blind to the hypotheses and
conditions, recoded 25% of the videos. Interobserver
reliability across all conditions was excellent (Cohen’s
j = .94, SE = .05). Complete details of coding proto-
col are provided in the Supporting Information.

Ethical Statement

We received ethical clearance from the University
of Birmingham Ethical Review Committee (ERN_13-
1412) and the Marie Curie European Commission
Ethical Screening Program (no. 628763). This study
conformed to University of Birmingham’s Code of
Practice for Research. For children, we received full
approval and ethical clearance from ThinkTank
Museum and full informed consent from parents. For
the bonobos, we received full ethical approval to con-
duct this study from “Les Amis des Bonobos du
Congo” (ABC, Lola ya Bonobo Sanctuary). This study
complied with all legal requirements required for
conducting research in DR Congo (research permit:
MIN.RS/SG/180/011/2016).

Results

We observed high levels of spontaneous imitation
by children across both uncommon (rub–rotate)
and typical (cross–trace) action conditions. The
majority of children readily copied at least one of

the two observed actions in both conditions—rub–
rotate: 77.8% of children (21/27); cross–trace: 81%
of children (21/26). Of these children, approxi-
mately one-third spontaneously copied both actions
demonstrated to them—rub–rotate: 39% children
(8/27); cross–trace: 27% children (7/26), see Fig-
ure 2. For cases where children only copied one of
the two actions, in both conditions it was most
often the second demonstrated action that was cop-
ied, suggesting a working memory constraint and/
or a recency effect (for single action responses,
copying of the second demonstrated action

Figure 1. Image of the wooden box stimuli used in the imitation experiment (also showing a reward sticker provided to child partici-
pants). [Color figure can be viewed at wileyonlinelibrary.com]

A

B

0.00

0.20

0.40

0.60

0.80

1.00

0

0.2

0.4

0.6

0.8

1

Figure 2. Results showing proportion of child (N = 52) and
bonobo (N = 46) participants that spontaneously imitated the
observed actions in the (A) uncommon (“rub–rotate”) condition
and the (B) typical (“cross–trace”) condition. [Color figure can be
viewed at wileyonlinelibrary.com]

Overimitation in Children as Compared to Bonobos 1539

occurred in 10/13 cases for rub–rotate and 12/15
cases for cross–trace). During a control condition,
where everything remained the same except that no
demonstration was performed, no child (N = 19)
performed any of the target actions. In all cases of
copying, the children copied the demonstrated (cau-
sally irrelevant) actions first, before potentially per-
forming any causally relevant actions to open the
box (i.e., trying to pry open the box).

In contrast, no bonobo in our sample copied any
of the target actions in either condition. Instead,
they attempted to open the box using an array of
causally relevant, species-typical methods, which
included pounding, biting, kicking, and shaking. As
no bonobo demonstrated any of the actions, we did
not run a control condition for the bonobos.

Requests for assistance occurred in both species,
but more in children, which is not surprising given
their language skills. Of children, 48% (14/29) made
direct verbal requests (e.g., “It’s too hard for me,
can you do it?”) and/or gestural requests. Although
actively returning objects in one’s possession is not
typically observed in great apes, 21.8% of bonobos
(10/46) in our sample actively returned the box to
the experimenter after attempting to open it; thus,
outwardly resembling a request for assistance.

Discussion

Our study identified striking contrasts in young chil-
dren’s copying behavior as compared to that of bono-
bos, our closest living relatives. Children readily
copied the actions, which were visibly causally irrele-
vant, whereas not a single bonobo did. Whether or
not the bonobos were unable, unwilling, or both
unable and unwilling to copy the demonstrated
actions, the results highlight striking differences in
human children’s cultural behaviors as compared to
those of bonobos. Importantly, our study addressed
methodological constraints of previous studies, thus
providing a true test for overimitation which allowed
us to compare the performances of both children and
bonobos. Combining our results with earlier findings
for chimpanzees (Tennie et al., 2012; Tomasello et al.,
1997), our findings indicate that bodily overimitation
—at least in high frequencies—is a uniquely human
capacity, which likely plays a key role in explaining
why human culture can accumulate over time.

This study focused on bonobos, an understudied
species of great ape that might be expected to show
higher imitative potential than chimpanzees, given
their enhanced social orientation (Kano et al., 2015;
Kret et al., 2016) and high social tolerance (Hare &

Kwetuenda, 2010). The fact that the bonobos failed
to overimitate demonstrates that even enhanced
social orientation may not be enough to trigger
human-like cultural learning behaviors. These
results thus demonstrate an important qualitative
difference between humans and great apes with
regard to the capacity or motivation to copy visibly
causally irrelevant actions. Differences in the capac-
ity for action copying may relate to cognitive con-
straints in great apes’ abilities to understand goals
and intentions as humans do (Call & Tomasello,
2008). Differences in motivation are likely to relate
to the strong affiliative and normative drivers of
imitation in humans but not in great apes (Legare
& Nielsen, 2015; Over & Carpenter, 2012).

An alternative explanation to the lack of copying
by the apes is that it was due to methodological con-
straints. However, although small sample size is fre-
quently a critique of great ape studies, this was not
the case for our study. The combined results from the
two related studies also make this explanation unli-
kely for chimpanzees (Tennie et al., 2012; Tomasello
et al., 1997). Age is also unlikely to be an explanatory
factor, given that a full age range was tested, and no
subject showed evidence of copying. Another possi-
bility is that using a human demonstrator inhibited
the bonobos’ motivation to imitate. However, a con-
specific demonstrator was used in both chimpanzee
studies (Tennie et al., 2012; Tomasello et al., 1997),
yet no novel action copying occurred. Moreover, in a
review of 23 studies directly comparing chimpanzee
and human performance in experimental settings,
Boesch (2007) concluded that the use of human
demonstrators did not seem to influence observed
species differences. Lack of motivation also does not
appear to be a problem: The majority of apes per-
sisted in this task and employed many alternative
techniques while trying to open the box.

Although previous studies have shown that great
apes will sometimes copy in certain circumstances, it
appears to primarily occur after receiving extensive
training and/or enculturation (Bjorklund et al., 2000;
Byrne & Tanner, 2006; Call, 2001; Carrasco et al.,
2009; Custance et al., 1995; Hayes & Hayes, 1952;
Miles et al., 1996). Given that these factors are absent
in wild apes, ecologically relevant findings must
therefore come from untrained and unenculturated
apes. In our study, not a single untrained and nonen-
culturated bonobo copied any of the demonstrated
actions, thus providing qualitative and ecologically
valid evidence of the distinctive nature of the human
cultural capacity as compared to that great apes: The
copying of visibly causally irrelevant actions (espe-
cially novel actions) appears to be uniquely human.

1540 Clay and Tennie

One relevant question is why children were so
willing to copy these superfluous actions? It has been
suggested that children copy in a blanket fashion
due to the causal opaqueness of a task (Horner &
Whiten, 2005; Lyons et al., 2007). However, …