The present short note aimed at further
exploring data from a recent study showing
socially modulated auditory contagious yawning
in domestic dogs (Canis familiaris).
Two independent observers further extended
the analysis of all video-recordings made in
that previous study and coded both the number of
yawns performed by the dogs and the frequencies
or durations of stress-related behaviors
exhibited throughout the presentation of
familiar and unfamiliar yawns. By showing no
significant different between conditions in the
frequencies or durations of the coded behaviors,
nor any association between the number of yawns
and the frequencies or durations of
stress-related behaviors, results raised doubt
on the stress-induced yawn hypothesis, thus
supporting social modulation. The exact
mechanism underlying contagious yawning,
however, needs further research.
-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
The recent interest in the phenomenon of
dogs yawning contagiously at humans has raised
several challenges as conflicting results from
different studies have left a number of
unresolved issues (for detailed discussions on
the topic see Campbell & de Waal 2010 and
Silva et al. 2012). In particular, the question
of whether contagious yawning in dogs might have
a basis in empathy has stimulated debate among
researchers (e.g., Joly Mascheroni et al. 2008
and Silva et al. 2012 vs. O'Hara & Reeve
2010 and Madsen & Persson 2013; see also
Yoon & Tennie 2009).
By showing that dogs are subject to auditory
contagious yawning and that, in addition, this
is modulated by familiarity with the model
yawner, Silva et al. (2012) provided the first
direct test suggesting empathy-based emotionally
connected yawn contagion in dogs (but see the
complete paper for alternative interpretations).
Such evidence, however, strongly contrasts with
observations from a previous test on social
modulation in dogs (O'Hara & Reeve 2010) and
also with more recent data on puppies (Madsen
& Persson 2013). In neither of these two
studies, was the strength of yawn contagion
related to dogs' emotional closeness with the
yawning models. Based on this, Madsen &
Persson (2013) raised questions about the
interpretation of the results presented in Silva
et al (2012) and brought up the possibility of a
stress-artifact underlying the apparent social
modulation reported in Silva et al. (2012):
"While the study showed that the sound of
familiar yawns elicited more contagious yawning
than unfamiliar yawns, the dogs' increased
yawning to familiar yawns may have been induced
by mildly heightened tension" (Madsen &
Persson 2013). That is, according to Madsen
& Persson (2013), (at least) some of the
yawns observed in response to the sound of
familiar yawns as 'natural' contagious yawns
might have been, in fact, 'tension' yawns evoked
by that particular experimental condition: "The
sound of the 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"
(Madsen & Persson 2013).
The possibility of a stress-related effect
underlying Silva et al.s' (2012) findings is a
valid concern and deserves serious
consideration, particularly given lack of
evidence of empathy based-emotionally connected
yawning contagion in dogs exposed to familiar
and unfamiliar human yawns presented by a live
model (O'Hara & Reeve 2010). In light of
this, the purpose of the present short study was
to further explore recordings from Silva et al.
(2012) so as to obtain new data on the potential
levels of stress in the tested dogs during the
presentation of familiar and unfamiliar
yawns.
Assuming that 'tension' yawns, as opposed to
'natural' contagious yawns, tend to co-occur
with a number of associated signs of anxiety,
then, under Madsen & Perssons' (2013)
argument that the increase in dog yawns in
response to the sounds of familiar yawns might
have been accounted for by elicitation of
tension yawns, one should be able to i) notice
more frequent (or longer) stress-related
behaviors in that particular experimental
condition and ii) find a correlation between the
number of yawns performed by each dog and the
total frequency, or duration, of stress-related
behaviors.
METHODS
Two independent researchers, naive to the
purpose of the analysis and blind to the
treatment conditions, re-coded the video
recordings of all experimental sessions
conducted by Silva et al. (2012). Both coded i)
the total number of yawns performed by each dog
during the presentation of familiar and
unfamiliar yawns and ii) the total frequency, or
total duration, of the behaviors exhibited by
the dogs that could potentially be related to
some level of tension. Regarding point ii), a
shorter version of the ethogram presented in
Beerda et al. (1998) was considered for
analysis. Behaviours scored only in terms of
frequency of occurrence included: body shaking,
circling, open mouth, oral behaviours, sighing,
stretching and vocalizing. Behaviours scored as
state and event included: autogrooming, panting
and trembling.
To test whether stress could have accounted
for the results reported in Silva et al. (2012),
the above referred behaviors were scored and
compared between the two experimental
conditions. Potential associations between the
total frequencies, or durations, of the observed
behaviors and number of yawns performed during
the presentation of familiar and unfamiliar
yawns were also assessed.
All statistical analyses were performed
using STATISTICA 7.0 (StatSoft, Tulsa, Oklahoma,
USA), and a significance level of 0.05 was
used.
RESULTS
Results showed 100% agreement between the
number of yawns coded per dog for each of the
two experimental conditions (presentation of
familiar and unfamiliar yawns) and those
reported in Silva et al. 2012 for those same
conditions (i.e. both new coders fully agreed
with Silva et al.'s reported yawn
frequencies).
In respect to signs of stress, only four
behaviours were recorded throughout the entire
study: snout liking (coded as an oral
behaviour), scratching and licking-self (both
coded as autogrooming) and panting. No other
behaviour considered for analysis (see the
Methods section) was ever observed.
Since the total duration of the observed
tension-related behaviours was not normally
distributed (Shapiro-Wilk, goodness-of-fit,
tests, P>0.05) only non-parametric tests
(i.e., Wilcoxon matched pairs tests) were used
throughout to assess differences between
treatments with regard to apparent stress
levels.
Results showed no significant differences in
neither the total frequency of the observed
stress related behaviors (snout liking,
scratching, licking-self and panting; mean total
frequency ± S.D. during the presentation of
familiar yawns: 0.14 ± 0.44; mean total
frequency ± S.D. during the presentation of
unfamiliar yawns: 0.10 ± 0.41; N=29, Z=
0.26, P=0.79) nor in total duration (scratching,
licking-self and panting; mean total duration
± S.D. during the presentation of familiar
yawns: 0.52 ± 1.57; mean total duration
± S.D. recorded during the presentation of
unfamiliar yawns: 0.66 ± 1.88; N=29,
Z=0.63, P=0.53). Also, no correlations were
found between total frequency of stress-related
behaviors and number of yawns performed during
the presentation of familiar and unfamiliar
yawns (familiar yawns: Pearson's r =0.14, P
=0.46; unfamiliar yawns: Pearson's r =0.09, P
=0.65). Additionally, no correlations were found
between total duration of stress-related
behaviors and number of yawns (familiar yawns:
Pearson's r =0.11, P =0.58; unfamiliar yawns:
Pearson's r =0.19, P =0.33).
DISCUSSION
According to Madsen & Persson's (2013),
a stress-effect related to the particular
experimental procedure used by Silva et al.
(2012) - instead of an actual social effect -
might have accounted for the reported pattern of
yawning. The sounds of the familiar (i.e.,
owner's) yawns were not emitted from the exact
same place where the owners were hiding, which
according to Madsen & Persson (2013) could
have generated uncertainty, and consequently
mild stress, in the tested dogs. The behavioral
analysis here presented, however, does not to
support this hypothesis. Tested dogs only rarely
showed behavioral indicators of stress and,
crucially, there were no differences in stress
levels between conditions. Also, no association
was found between the number of yawns and the
level of observed stress, thus raising doubt on
the stress-induced yawn hypothesis.
Importantly, we only measured stress as
evidenced in stress-correlated behaviors,
following Beerda et al. (1998). That is, we did
not measure stress directly through
physiological measurements, and we therefore
cannot fully discard Madsen & Persson's
(2013) argument. It seems however unlikely that
the tested dogs might have been able to control
their outwards emotional responses throughout
the experimental sessions, thus showing
physiological markers of tension (such as heart
rate changes; Harr et al. 2010), but suppressing
behavioral indicators. Also, it is important to
note, that the speakers used in Silva et al.
(2012), and from which the sounds of yawns were
emitted, were positioned as close as possible
(< 1 meter) to the owner, and that previous
studies testing the ability of dogs to match an
auditory stimulus to a corresponding visual
stimulus were not compromised by dogs' acute
sound localization. In Taylor et al. (2012), for
example, the distance between the speakers from
which sounds were emitted and the associated
visual stimulus was 1.5 meters.
We conclude that stress differences per se
did not account for the results presented in
Silva et al. (2012). How, then, to interpret the
contradictory evidence of socially modulated
contagious yawning in dogs reported in O'Hara
& Reeve (2010) and Madsen & Persson
(2013) (see the introduction section)? With
respect to the latter, it seems to us, given the
age of the tested dogs (puppies ranging from 4
to 14 months of age), that the contradiction may
be related to the testing age. Indeed the
authors themselves refer to the possibility
that, in species that exhibit an empathy-based
social modulatory effect on contagious yawning,
the effect may emerge only at later stages of
development (Madsen & Persson 2013). With
respect to the lack of evidence of familiarity
biased contagious yawning reported in O'Hara
& Reeve (2010), we previously argued that
disparate findings could have been due to the
methodology followed by these authors, which
implied some dog-human interaction that could
have diverted the dogs' attention from the yawn
stimuli (for further details, see Silva et al.
2012). We, did, however, also highlight the
possibility that the social modulation of
contagious yawing evidenced by our previous
results - and supported by the present analysis
- could be related to differences in dogs'
capacity to form mental representations from
familiar and unfamiliar auditory input (see
Adachi et al. 2007).
Clearly, there is a need for some kind of
standardization in the methodologies used to
explore the phenomenon of dogs yawning
contagiously at humans, so that unequivocal
conclusions can be drawn about the particular
mechanism at its base - be it empathy related or
not. Also, particularly interesting insights
could be gained from longitudinal studies, in
which the same dogs are tested throughout their
developmental process (i.e., from puppies to
adults).
ACKNOWLEDGMENTS
We thank Elainie Alenkær Madsen and
Tomas Persson for calling attention on the
possibility of a stress related effect
underlying the results presented in our previous
study on contagious yawning in dogs. Also we are
grateful to Claudio Tennie for his valuable
comments.
REFERENCES
Adachi I, Kuwahata H, Fujita K (2007) Dogs
recall their owner's face upon hearing the
owner's voice. Anim Cogn 10: 17-21.
Beerda B, Schilder MBH, van Hooff JARAM, de
Vries HW, Mol JA
(1998) Behavioral, saliva, cortisol and
heart rate responses to different types of
stimuli
in dogs. Appl Anim Behav Sci 58:
365-381.
Campbell MW, de Waal FBM (2010)
Methodological problems in the study of
contagious yawning. Front Neurol Neurosci 28:
20-127.
Joly-Mascheroni RM, Senju A, Shepherd AJ
(2008) Dogs catch human yawns. Biol Lett 4:
446-448.
Madsen EA, Persson T (2013) Contagious
yawning in domestic dog puppies (Canis lupus
familiaris): the effect of ontogeny and
emotional closeness on low-level imitation in
dogs. Anim Cogn 6: 233-240.
O'Hara SJ, Reeve AV (2010) A test of the
yawning contagion and emotional connectedness
hypothesis in dogs, Canis familiaris. Anim Behav
81: 335-340.
Silva K, Bessa J, de Sousa L (2012) Auditory
contagious yawning in domestic dogs (Canis
familiaris): first evidence for social
modulation. Anim Cogn 15: 721-724.
Taylor AM, Reby D, McComb K (2011) Cross
Modal Perception of Body Size in Domestic Dogs
(Canis familiaris). PLos ONE 6: e17069.
Yoon JMD, Tennie C (2010) Contagious
yawning: a reflection of empathy, mimicry, or
contagion? Anim Behav 79: 1-3.
