Contagious yawning is a well-documented
phenomenon in humans and has recently attracted
much attention from developmental and
comparative sciences. The function, development
and underlying mechanisms of the phenomenon,
however, remain largely unclear.
Contagious yawning has been demonstrated in
dogs and several non-human primate species, and
theoretically and empirically associated with
empathy in humans and non-human primates.
Evidence of emotional closeness modulating
contagious yawning in dogs has, nonetheless,
been contradictory. Humans show a developmental
increase in susceptibility to yawn contagion,
with typically developing children displaying a
substantial increase at the age of four, when a
number of cognitive abilities (e.g. accurate
identification of others' emotions) begin to
clearly manifest.
Explicit tests of yawn contagion in
non-human animals have, however, thus far only
involved adult individuals. Here, we report a
study of the ontogeny of domestic dogs' (Canis
lupus familiaris) susceptibility to yawn
contagion, and whether emotional closeness to
the yawning model affects this. Thirty-five
dogs, aged 4-14 months, observed a familiar and
unfamiliar human repeatedly yawn or gape. The
dogs yawned contagiously, but emotional
closeness with the model did not affect the
strength of contagion, raising questions as to
recent evidence of emotionally modulated
auditory contagious yawning in dogs. The dogs
showed a developmental effect, with only dogs
above 7 months evidencing contagion.
The results support the notion of a
developmental increase in dogs' attention to
others and identification of others' emotional
states and suggest that yawn contagion is
underpinned by developmental processes shared by
humans and other animals.
-Harr AL,
Gilbert VR Do dogs show contagious yawning ?
Anim Cogn. 2009;12(6):833-837
-Lindsay
SR Coping with fear and stress: licking and
yawning. Handbook of applied dog behavior and
training 2000
-Madsen EA,
Persson T. Contagious yawning in domestic
dog puppies (Canis lupus familiaris): the effect
of ontogeny and emotional closeness on low-level
imitation in dogs. Anim Cogn. 2012
-O'Hara SJ, Reeve
AV A test of the yawning contagion and
emotional connectedness hypothesis in dogs,
Canis familiaris. Animal Behaviour
2011;81:335-340
-Perkins
JR Teaching Dogs to Yawn, Sneeze, and
Implications for Preparedness Theory and
Observational Learning.
In: Kusonose, Ryo and Sato, Shusuke 39th
Congress of the International Society for
Applied Ethology, Kanagawa, Japan. 20-24 August,
2005
-Silva K, Bessa J, de
Sousa L. Auditory contagious yawning in
domestic dogs (Canis familiaris): first evidence
for social modulation. Anim Cogn. 2012.
Introduction
Contagious yawning (yawning in response to
perceiving others' yawn, henceforth CY) is a
well-documented phenomenon in humans (Provine
1986; Anderson and Meno 2003; Norscia and Palagi
2011). Viewing others yawn elicits yawning in
45-55 % of adults (under experimental
conditions: Provine 1986; Platek et al. 2003),
and the mere thought (Provine 1986), and sound
(Arnott et al. 2009) of yawning is sufficient to
evoke contagion. While the phe- nomenon has
recently attracted much attention from
developmental and comparative sciences (see
Walusinski 2010), the function, development and
underlying mechanisms remain largely
unclear.
Recent research has emphasised that CY may
be linked to and modulated by empathy (Platek et
al. 2003; Anderson and Matsuzawa 2006; Palagi et
al. 2009; Campbell and de Waal 2011; Norscia and
Palagi 2011). The term empathy encompasses a
spectrum of resonant emotional responses,
ranging from basic 'affective empathy' to
'cognitive empathy'. 'Affective empathy'
represents vicarious, visceral emotional
responses to others' experiences. It has been
argued that this affective element results from
an automatic motor mimicry response to others'
emotional expressions and is mediated by
emotional closeness (Lipps 1903; Preston and de
Waal 2002; Baron-Cohen and Wheelwright 2004).
Affective empathy is therefore, effectively, a
case of emotional contagion, which relies on two
inter-linked processes: non-conscious mimicry
and afferent feedback. Non-conscious mimicry
(the so-called 'chameleon effect': Chartrand and
Bargh 1999) is the tendency to minor the
behaviours of social partners, without awareness
or intent. Since facial expressions tend to
influence emotional experiences (e.g. smiling
provides an afferent feedback: Niedenthal et al.
2005), the 'chameleon effect' can give rise to
emotional contagion, whereby the mimicking and
synchronisation of others' movements and facial
expressions lead individuals to converge
emotionally (Hatfield et al. 2009).
While no empirical research has yet
demonstrated that CY provides afferent feedback
(increases sleepiness in the subject), CY may be
considered an emotional contagion, arising
nonconscious mimicry (Yoon and Tennie 2010). In
contrast to 'affective empathy', 'cognitive
empathy' is considered to emerge
phylogenetically and ontogenetically with other
'indicators of mind' and requires a capacity for
self-other differentiation and
perspective-taking and that the subject
represents the state of another' s feelings (in
terms of theory-of-mind), without this
necessarily involving emotional matching
(Preston and de Waal 2002).
Multiple lines of psychological, clinical,
developmental and neuroscientific evidence
suggest that empathy, linked with social
bonding, influences yawn contagion. For example,
in human adults, CY is positively correlated
with visual self-recognition and performance on
theory-of-mind tasks (Platek et al. 2003), two
abilities that contribute to 'cognitive
empathy'. Moreover, CY is diminished in
individuals with empathy-related disorders
(schizotypy: Platek et al. 2003; autism: Senju
et al. 2007, 2009; Giganti and Ziello 2009; Helt
et al. 2010).
Typically, developing children begin to yawn
contagiously at 4-5 years. (Anderson and Meno
2003; HeIt et al. 2010; Millen and Anderson
2011), an age, at which a host of cognitive
abilities (such as false-belief theory-of-mind
understanding) begin to clearly manifest, and
children begin to correctly identify the
emotions of others (Wiggers and van Lieshout
1985; Wellman et al. 2001; Saxe et al. 2004;
Singer 2006). While this is consistent with the
suggestion that CY shares a developmental basis
with 'cognitive empathy' and theory-of-mind
capacities, CY has been recorded in a number of
non-human species, of which none have passed the
paradigmatic false-belief test of theory-of-mind
understanding, suggesting that the phenomenon
relies on processes other than those involved in
'cognitive empathy'.
Aside from humans, CY has been recorded in
chimpanzees (Anderson et al. 2004; Campbell et
al. 2009; Campbell and de Waal 2011), gelada
baboons, Theropithecus gelada (Palagi et al.
2009) and domestic dogs, Canis lupus familaris
(Joly-Mascheroni et al. 2008), with some
suggestion of CY in stumptailed macaques, Macaca
arctoides (Paukner and Anderson 2006) and
budgerigars, Melopsittacus undulatus (Miller et
al. 2011). In contrast, no CY has been
demonstrated in tortoises, Geochelone carbonaria
(Wilkinson et al. 2011), a solitary species,
lending some empirical support to the notion
that CY serves to coordinate and synchronise
group behaviour (e.g. Deputte 1994; Palagi et
al. 2009).
For humans, the strength of the contagion is
proportional to degree of familiarity (i.e. it
is greatest in response to kin, then friends,
then acquaintances and lastly strangers: Norscia
and Palagi 2011). Moreover, adult chimpanzees
are more likely to yawn contagiously in response
to watching videos of familiar than unfamiliar
conspecifics (Campbell and de Waal 2011), and
among gelada baboons, social bonding predicts CY
(Palagi et al. 2009). Overall, these findings
are consistent with the notion that familiarity
and emotional closeness increase empathy
(Preston and de Waal 2002; de Waal 2008;
Langford et al. 2006) and, consequently,
CY.
Dogs are an interesting species for
addressing questions regarding social cognition
and the domestication thereof. By virtue of
social domestication over the past 15,000 years
(Miklósi 2008), dogs have been
selectively bred for their function in human
society (Cooper et al. 2003) and attachment to
humans (as if they were conspecifics:
Tópal et al. 1998, 2005). Dogs, for
example, are sensitive to human emotions and
respond to their owners' distress with enhanced
negative emotional arousal (Jones and Josephs
2006) and behave seemingly to comfort owners as
well as strangers, who pretend to be distressed
(Custance and Mayer 2012).
Joly-Mascheroni and colleagues showed that
72 % of dogs tested yawned in response to
viewing an unfamiliar human yawn
(Joly-Mascheroni et al. 2008). Subsequent
studies have, however, failed to replicate this
result and found little evidence for
hetero-specific as well as conspecific CY in
dogs, and none suggests that dogs respond
differently when viewing familiar and unfamiliar
humans yawn (Harr et al. 2009; O'Hara and Reeve
2010). Consequently, doubts have been raised
regarding the cross-species CY reported by
Joly-Mascheroni and colleagues and have led to
the suggestion that the results were explained
by a high degree of 'tension yawns' (O'Hara and
Reeve 2010), which may be evoked in situations
of uncertainty (as yawning indexes social stress
in dogs: Beerda et al. 1998).
A recent study, however, suggested that dogs
are subject to auditory contagious yawning (i.e.
as in humans, the mere sound of yawning is
sufficient to evoke contagion) and that this is
modulated by familiarity with the model yawner
(Silva et al. 2012). To examine whether CY in
dogs is modulated by familiarity (indexing
empathy), dogs were played recordings of,
respectively, yawns produced by their owner or
an unfamiliar experimenter, and two control
sounds consisting of a computer-reversed yawn
from either the owner or experimenter. While the
study showed that the sound of familiar yawns
elicited more CY than unfamiliar yawns, the
dogs' increased yawning to familiar yawns may
have been induced by mildly heightened tension.
Dogs recognise their owner's voice (Adachi et
al. 2007) and have acute sound localisation. The
dogs in Silva et al's study were tested in their
homes, positioned facing two speakers, but
without visual contact to the owner, from whom
the recorded familiar yawns derived.
Elainie Madsen during FICY
2010
The sound of their owner's yawns, emitted
from two speakers immediately in front of the
dogs, was from a direction where the owner was
not (the owner was outside the testing room).
Dogs sensing this discrepancy would likely
experience uncertainty and consequently mild
stress. The higher proportion of yawns to
familiar stimuli may therefore have represented
an increase in mild stress. Conversely, no such
stress effect should be evident in response to
the scrambled (reversed) familiar yawns, since
identification of the owner's voice was likely
impossible. By comparison, auditory CY in humans
is evoked by the sound of unfamiliar yawns
(Arnott et al. 2009), and CY in responses to
viewing unfamiliar yawns has been demonstrated
in a number of other studies (children: e.g.
Senju et al. 2007; chimpanzees: Madsen et al. in
prep.; dogs: Joly-Mascheroni et al. 2008),
making the finding of an auditory CY effect in
dogs, elicited only in response to familiar
stimuli, puzzling. The methodological issues,
detailed above, are sufficient to raise
questions about the interpretation of the
results, particularly given lack of evidence of
empathybased, familiarity biased CY in dogs
exposed to familiar and unfamiliar human yawns
presented by a live model (O'Hara and Reeve
2010). As all empirical results, evidence of a
potential familiarity effect on CY must be
evaluated in the light of the overall pattern of
results.
One intriguing, yet unstudied aspect of CY
in nonhuman animals is its developmental
trajectory. While humans show a developmental
increase in susceptibility to yawn contagion,
explicit tests of the contagion in nonhuman
animals have thus far only involved adults. One
study of chimpanzees, however, indicated a
potential age effect in susceptibility to
contagion, as videoed stimuli of conspecific
yawns failed to elicit yawning in any of three
infants accompanying their mothers to the test
(Anderson et al. 2004). There is, however, some
evidence that the medium may obscure the message
for younger and nonhuman subjects. For example,
while neither video stimuli nor stories, in
which a central character repeatedly yawned,
have evoked CY in children below 5 years
(Anderson and Meno 2003; Millen and Anderson
2011), live models have educed CY in younger
children (35 % of 4-year-olds tested: Helt et
al. 2010; for observation of CY in a
12-month-old, see Piaget 1951, p. 40). There is
some suggestion that dogs are sensitive to the
medium, as only live (Joly-Mascheroni et al.
2008) and not videoed (conspecific and human)
models (Han et al. 2009) have elicited CY in
this species.
In this study, we explored the ontogeny of
dogs' susceptibility to yawn contagion and
whether emotional closeness to the yawning model
affects this. We wished to increase ecological
validity and the probability of evoking CY in
younger subjects by using live rather than
videoed models. We used the same basic design as
the first study of CY in dogs (Joly-Mascheroni
et al. 2008), but strove to minimise the
possibility of evoking 'tension yawns' by
engaging the dogs in a bout of play and
cuddling, without requiring eye contact with the
experimenter prior to each yawn, thereby
explicitly making the dogs attend to the yawning
(as done by Joly-Mascheroni and colleagues).
Calling the dogs' name and attention to the
experimenter immediately prior to yawning may
have endowed the behaviour with a puzzling and
seemingly communicative function, and avoiding
this addresses the critique by O'Hara and Reeve
(2010) that 'tension yawns' might underlie the
high frequency of CY found by Joly-Mascheroni
and colleagues. Moreover, we included a 5-min
baseline phase and a familiar experimenter.
We hypothesised that if CY is related to the
development of empathy and emotional
understanding in humans, a similar developmental
effect might be found in dogs. We thus predicted
that (1) older dogs show stronger evidence of CY
than younger dogs. Since empathy and model
identity may play a significant role in
facilitating social behaviours, such as
imitation (Meltzoff and Moore 1994; see de Waal
2001, for the 'Bonding- and Identificationbased
Observational Learning' model), we presented
young dogs with a familiar yawning model, with
whom they had a strong and positive emotional
relationship (their owner), and an unfamiliar
model. We predicted that (2) the dogs would be
more likely to yawn contagiously to the familiar
model and that this model would elicit more CY
from the younger participants. We further
predicted (3) yawning frequency to increase in
response to viewing a human model performing
repeated yawns, but not gaping (mouth opening
and closing), or indeed when the experimenter
performed none of the behaviours (baseline
phase). While previous studies have used a
variety of control behaviours (smiles, coughs,
laughs and gapes, see Campbell and de Waal
2010), gaping has the advantage of mimicking
much of the motor pattern of a yawn, while being
a meaningless expression. If CY is an emotional
contagion (Hatfield et al. 2009) and reflects
perception and internalisation of the emotion
and/or physical state that another' s yawning
reflects, only yawn stimuli should evoke
yawning. A comparison of the rate of yawing in
response to yawning and gaping stimuli thus
excludes a more motoric stimulus-response
interpretation of CY.
Discussion
The current study confirms previous results
showing that dogs 'catch' human yawns. Viewing
an unfamiliar human yawn elicited yawning in 69
% of 4-14 months old dogs, compared to 72 % of
adult dogs, tested using the same basic paradigm
(Joly-Mascheroni et al. 2008). While the failure
to replicate the results of the original study
(by Harr et al. 2009; O'Hara and Reeve 2010) has
led to questions raised as to the veracity of
the findings and the suggestion that the results
stemmed from elicitation of 'tension yawns', the
present study reduced potential stress in the
dogs, yet confirmed the original findings.
The study demonstrates that, like humans
(Senju 2010; Helt et al. 2010), dogs are subject
to a developmental increase in susceptibility to
yawn contagion, as dogs above the age of 7
months yawned contagiously, while dogs between 4
and 7 months showed little evidence of
contagion. While studies of CY in dogs above the
age of one have shown no age effect
(Joly-Mascheroni et al. 2008; Harr et al. 2009;
O'Hara and Reeve 2010), it is likely the
developmental trend is more pronounced at
younger ages. Indeed, while explicit tests of CY
in other non-human species have thus far only
involved adults, a potential age effect on the
contagion has been suggested by the failure to
elicit CY in three infant chimpanzees (Anderson
et al. 2004). While an explicit test of a larger
sample of chimpanzees is needed to confirm this
suggestive trend, the pattern of results
suggests that the ontogenetic emergence of CY
reflects developmental processes shared by
humans and other animals.
The CY response in younger dogs is delayed
compared to that of adult dogs. In contrast to
the study by Joly-Mascheroni et al. (2008), in
which adult dogs yawned in only the 5-min yawn
phase (in which they viewed a human yawn), more
puppies yawned in the 5-min post-yawn phase (61
%) than the preceding yawn phase (39 %, when
tested under comparable conditions, i.e., with
an unfamiliar experimenter). While such a
'spill-over effect' may be accounted for by a
self-reinforcing and accumulating effect of
yawning, more than twice as many dogs yawned
uniquely in the post-yawn phase (37 %) compared
to the yawn phase (17 %) further supporting the
notion of a delayed effect. In contrast to adult
dogs, the puppies seemed to generalise the yawn
response to the gaping stimuli. While adult dogs
showed no yawning in response to the control
(gape) condition (Joly-Mascheroni et al. 2008),
46 % of puppies yawned during the
gaping/postgaping condition (in trials with an
unfamiliar experimenter-a pattern similar to
that exhibited in trials with a familiar
experimenter). Given the similarity of the
studies, it seems plausible that,
ontogenetically, dogs' CY response is initially
generalised and evokable by the gaping stimulus
which, while not reflecting the same
physical/emotional state as a yawn (e.g.
sleepiness), mimics much of the motor pattern.
This suggests that the CY response becomes
progressively more accurate with age, possibly
reflecting increased awareness of changes in the
environment, and an improvement in the
identification of others' emotions, which in
children co-emerges developmentally with
susceptibility to yawn contagion (i.e. at 4-5
years of age: e.g. Singer 2006; HeIt et al.
2010).
Anecdotally, there was some suggestion that
the dogs may have internalised the emotion that
yawning reflects. Although difficult to measure
consistently (in the absence of a physiological
measure), a considerable number of the dogs (43
%) responded to the yawning condition with a
clear reduction in arousal (to the extent that
the experimenter needed to prevent [e.g.
playfully shake] five dogs from falling
asleep during the yawning/post-yawning phase).
This effect was not observed in the
gaping/postgaping phase, providing some, albeit
anecdotal, suggestion that yawn contagion
provides an afferent feedback (Hatfield et al.
2009) and elicits a similar, synchronous state
in the observer.
CY in young dogs may help adjudicate between
cognitive (2) and affective (3) empathy as the
key mechanism underlying cross-species CY. In
contrast to affective empathy, cognitive empathy
considered to involve theoryof-mind attribution
and perspective-taking, of which there is no
evidence for the former in dogs and the latter
emerges developmentally late in children and
chimpanzees (in chimpanzees not until around 4.5
year: Povinelli et al. 1994). CY in young dogs
thus suggests that the phenomenon is underlain
by processes less cognitively complex than
cognitive empathy. There is ample evidence that
the affective and cognitive components of
empathy have different developmental
trajectories in humans (see Decety 2010) and
that affective empathy precedes cognitive
empathy, ontogenetically and phylogenetically
(Preston and de Waal 2002). Thus, it seems
possible that non-conscious mimicry and the
development of affective empathy (as well as an
improvement in the identification of others'
emotions) are sufficient to explain the
distribution of yawn contagion, ontogenetically
as well as phylogenetically.
Studies of CY in dogs have produced
conflicting results, yet have deployed a variety
of methodological designs, of which the effect
on CY is largely unknown, thus complicating
comparison of results across studies (Campbell
and de Waal 2010). To date, in visual paradigms,
only designs that have presented live human
models (Joly-Mascheroni et al. 2008; present
study), rather than videoed human (Harr et al.
2009) or conspecific (O'Hara and Reeve 2010)
models, have elicited CY in dogs. This
difference may be accounted for by videoed
stimuli having been displayed on laptops
optimised for human viewing (in terms of flicker
rates) and that many dogs attend little to 2D
screens (pers. obs.). Interestingly, O'Hara and
Reeve (2010) failed to show evidence of yawn
contagion in dogs viewing a live human model
yawn. While puzzling, it is possible that the
shorter duration of stimulus presentation (3 mm,
with no post-yawn observation phase) and the
smaller sample size (19 dogs), of which 37 %
resided at a dog rescue centre, and thus
possibly had disrupted attachment to humans, may
account for these differences in results.
Moreover, while the human models in O'Hara and
Reeve's study yawned "from the neck up only" (p.
337), models in the present study were asked to
yawn as naturally as possible, thus mostly also
engaging the arms and upper body (e.g. raising
the chest and shoulders during deep inhalation)
and tilting the body axis backwards. Since dogs
seem to pay much attention to body movements and
posture, the absence of these cues may
contribute to the failure to demonstrate CY in
dogs by means of paradigms using only facial
movements.
Model familiarity failed to predict yawning
in both younger and older puppies. Contrary to
prediction, a positive emotional bond with the
model was thus insufficient to elicit CY in
young dogs. While the strength of contagion was
similar when older puppies viewed a familiar and
unfamiliar model (i.e. they yawned with similar
frequency to yawns presented by the two models
types), familiar gapes elicited a higher (though
statistically non-significant) number of yawns
than unfamiliar gapes.
Consequently, the difference between yawns
elicited by familiar gapes and yawns failed to
reach significance. This slight increase in
yawns in response to familiar gapes may be
accounted for by elicitation of tension yawns,
when subjects viewed their owner repeatedly
gape-an unusual behaviour from a familiar
individual.
While recent research has emphasised that CY
may be linked to and modulated by empathy, a
social modulation of the contagion has thus far
only been demonstrated in adult individuals
(humans: Norscia and Palagi 2011; chimpanzees:
Campbell and de Waal 2011; gelada baboons:
Palagi et al. 2009; for potentially
controversial results with dogs, see Silva et
al. 2012) and not very young individuals
(humans: Millen and Anderson 2011; dogs: present
study; chimpanzees: Madsen et al. in prep.).
Jointly, this raises the possibility that in
species that exhibit an empathy-based social
modulatory effect on CY, the effect emerges only
at later stages of development ('developmental
hypothesis of empathy modulation of CY').
