How do guide dogs and pet dogs (Canis familiaris) ask their owners

Anim Cogn (2010) 13:311–323
DOI 10.1007/s10071-009-0279-z
ORIGINAL PAPER
How do guide dogs and pet dogs (Canis familiaris) ask their
owners for their toy and for playing?
Florence Gaunet
Received: 28 January 2009 / Revised: 7 September 2009 / Accepted: 10 September 2009 / Published online: 1 October 2009
Springer-Verlag 2009
Abstract When apes are not fully understood by humans,
they persist with attempts to communicate, elaborating
their behaviours to better convey their meaning. Such
abilities have never been investigated in dogs. The present
study aimed to clarify any effect of the visual attentional
state of the owner on dogs’ (Canis familiaris) socialcommunicative signals for interacting with humans, and to
determine whether dogs persist and elaborate their behaviour in the face of failure to communicate a request. Gaze
at a hidden target or at the owner, gaze alternation between
a hidden target and the owner, vocalisations and contacts in
12 guide and 12 pet dogs were analysed (i) when the dogs
were asked by their owners (blind or sighted) to fetch their
inaccessible toy and (ii) when the dogs were subsequently
given an unfamiliar object (apparent unsuccessful communication) or their toy (apparent successful communication). No group differences were found, indicating no
effect of the visual status of the owner on the dogs’ sociocommunicative modes (i.e. no sensitivity to human visual
attention). Results, however, suggest that the dogs exhibited persistence (but not elaboration) in their ‘‘showing’’
behaviours in each condition, except that in which the toy
was returned. Thus, their communication was about a
specific item in space (the toy). The results suggest that
dogs possess partially intentional non-verbal deictic abilities: (i) to get their inaccessible toy, the dogs gazed at their
Electronic supplementary material The online version of this
article (doi:10.1007/s10071-009-0279-z) contains supplementary
material, which is available to authorized users.
F. Gaunet (&)
Laboratoire ‘‘Eco-Anthropologie et Ethnobiologie’’,
UMR 5145, Muse´um National d’Histoire Naturelle,
CP 135, 57 rue Cuvier, 75231 Paris Cedex 05, France
e-mail: gaunet@mnhn.fr
owners as if to trigger their attention; gaze alternation
between the owner and the target direction, and two
behaviours directed at the target were performed, apparently to indicate the location of the hidden toy; (ii) after the
delivery of the toy, the dogs behaved as if they returned to
the play routine, gazing at their owner whilst holding their
toy. In conclusion, this study shows that dogs possess
partially intentional non-verbal deictic abilities: they
exhibit successive visual orienting between a partner and
objects, apparent attention-getting behaviours, no sensitivity to the visual status of humans for communication,
and persistence in (but no elaboration of) communicative
behaviours when apparent attempts to ‘‘manipulate’’ the
human partner fail.
Keywords Dog Deictic behaviour Intentional communication Guide dogs Socialisation Play Social cognition
Introduction
Manual gestures (pointing) in chimpanzees (and other great
apes) have often been interpreted as a form of functionally
referential and intentional communicative action (e.g. Call
and Tomasello 1994; Leavens et al. 1996, 2004, 2005a;
Tomasello et al. 1994; Leavens and Hopkins 1998;
Hostetter et al. 2001; Leavens et al. 2005b). Leavens et al.
(2005b) characterise the apes’ use of their outstretched
arms and fingers to indicate distant objects/events as
referring to specific entities in space. Pointing combined
with gaze at others (and also oriented vocalisations, and
gaze alternation between a target and an observer) are
considered as triadic referential communicative behaviours. They are used to attract the attention of others to an
123
312
event/object and to ‘‘manipulate’’ them to act on their
environment (Go´mez 2007). The dependency of dogs on
humans for food and other needs provides incentive for the
study of similar behaviours in dogs and how they are used
in different contexts.
Because it is impossible to study intentional communication with reference to the exercise of will or to the
motivational state of the signaller (Bruner 1981), Leavens
et al. (Leavens 2004; Leavens et al. 2004, 2005b) have
highlighted several operational criteria as evidence for
intentional and referential communication. Agents communicate intentionally when (1) there is successive visual
orienting between a partner and distant objects or events
(i.e. gaze alternation or visual checking), (2) apparent
attention-getting behaviours are deployed (e.g. vocalisations), (3) an audience is required to exhibit the behaviours,
(4) there is an influence of the attentional status of an
observer on the propensity to exhibit behaviours, and (5)
there is persistence in and (6) elaboration of communicative behaviour when apparent attempts to manipulate the
partner fail (e.g. when the social partner does not attend/
respond). Criteria (1–4) are well established in the great
apes (Leavens et al. 2005b) and these authors have recently
found the last two criteria in chimpanzees (for criteria 5–6
in orangutans, see Cartmill and Byrne 2007).
Dogs also perform triadic communicative behaviours,
sometimes called ‘‘showing’’ or deictic behaviours on
account of the lack of studies investigating their referential
character per se (i.e. the ability to emit signals to inform a
recipient about an external object or event, Evans 1997).
Only Miklo´si et al. (2000) have shown a greater frequency
of behaviours directed at the dog’s owner and at a hidden
target when both are present, in contrast to when only the
owner is present. This suggests the referential specificity of
these behaviours (see also Seyfarth and Cheney 2003 for
the referential properties of animal calls) and thus functionally referential communication in dogs (Miklo´si et al.
2000). These behaviours combine an apparently attentiongetting component that directs the attention of the perceiver
to the informer, and an apparently directional component
towards an external target (Miklo´si et al. 2000). This takes
the form of gaze or looks directed at the human partner and
at the hidden target, gaze alternation between the target and
the human, and vocalisations accompanying these behaviours (Hare et al. 1998; Miklo´si et al. 2000; Gaunet 2008).
These results provide support for criteria (1) and (2) in
dogs. These signals are successfully used by the owner to
locate the hidden target, suggesting that they act as intentional attention-getting and directional communicative
signals (Hare et al. 1998; Miklo´si et al. 2000).
In Miklo´si et al. (2000), when both the target and the
owner were present in the room, the dogs gazed at their
owner and at the target, sniffed at the target, licked their
123
Anim Cogn (2010) 13:311–323
mouths and their gaze alternated between owner and target
more markedly than when only the target was present
(Miklo´si et al. 2000). These results are in line with the
requirements to validate criterion (3).
Hare et al. (1998) tested criterion (4). In the first condition of Study 3, the experimenter stood in front of three
buckets; in the second condition, he stood facing the
buckets covering his eyes with his hands, and then
uncovered them; in the third, covering the eyes was
replaced by turning his back, after which the experimenter
turned back to face the buckets. The communicative signals displayed by the dogs did not differ among the
conditions. However, only two dogs were tested. Additional studies are thus required to test criterion (4). Nevertheless, other studies show that dogs are sensitive to the
direction of visual attention of the observer (head ? eyes)
(criterion (4)). This applies in situations where the dog
receives verbal commands (Study 1 in Vira´nyi et al.
2004), performs forbidden actions (Call et al. 2003) or
begs (e.g. Experiment 2 in Ga´csi et al. 2004; Study 2 in
Vira´nyi et al. 2004). Moreover, in dogs engaged in dyadic
rough-and-tumble play, Horowitz (2009) has also shown
sensitivity to the visual attention of other dogs (head
direction). Further to this, it is interesting to note that
Ga´csi et al. (2004) found indications that dogs were also
sensitive to the visibility of a human’s eyes: hesitant
behaviour in dogs offering their toy to their owner facing
them, but with eyes blindfolded, increased in comparison
to when the owner was not blindfolded. The dogs also
preferred to beg from the person with visible eyes.
However, the study concluded overall that the visibility of
the eyes might be of lesser significance than the other
bodily cues for direction of attention. In Call et al. (2003),
dogs were quicker in taking the forbidden food when the
experimenter had her eyes closed. In addition, in a twoway object choice test, dogs appropriately ignored a
human’s gaze when the person stared into space (eye
direction only) above the correct hiding location. It was as
if this posture was used by the dog to infer human inattention; their ability to use eye direction as a cue developed across sessions (Soproni et al. 2001). Further to this,
dogs do not follow eye direction at all if there is no
attractive object (Agnetta et al. 2000). Finally, in the
presence of effective and ineffective barriers, dogs make
the decision not to approach forbidden food when they can
see a human or when a human can see them (Bra¨uer et al.
2004), showing that dogs have knowledge about the visual
perspective of others.
Altogether, these studies show that dogs have a propensity to request performance of actions by humans, using
various types of behaviours, when faced with an unsolvable
problem (Cooper et al. 2003; Hare 2004). In addition, they
show that the use of human gaze (the direction of the head,
Anim Cogn (2010) 13:311–323
and less markedly eye direction) as a social cue for
determining the direction of a human’s attention is a key to
their interaction with humans. However, additional studies
are necessary to clarify how this is used; more generally,
the degree of intentionality of referential communication in
dogs needs to be investigated.
One way to study whether dogs use their owners’ eyes
as a cue for inferring visual attention of an observer is to
compare the communicative behaviour modes of pet dogs
and guide dogs in interaction with their sighted and nonsighted (blind) owners, respectively (criterion (4)). Indeed,
because dogs learn to communicate and collaborate with
humans throughout their lives, they have ample opportunity to learn to use human behaviours, both intentional and
unintentional, to predict future events from simple learnt
contingencies (Cooper et al. 2003). Guide dogs and pet
dogs are similarly socialised: they are both reared in
sighted families in the earlier stages of their lives. However, they differ in terms of their adult living conditions:
the ownership of guide dogs is transferred to blind people
generally after the age of about 1.5–2 years. They also
have the ability to interactively exchange roles with
humans as the initiators of actions when they synchronise
their actions with that of their owners during navigation
(Naderi et al. 2001). Blind owners and guide dogs also
have extensive interactions during free time periods. Furthermore, the different visual status of sighted and blind
owners means that they attend differently to their dogs and
respond differently to their behaviours: sighted owners can
see their dogs, whereas the blind cannot. For instance,
when guide dogs look at their blind owner or at a target, the
owners cannot respond appropriately because they cannot
perceive these behaviours. In addition, blind owners are not
likely to gaze at the guide dog’s eyes when talking to it
because they cannot perceive the exact location and orientation of the dog’s head. The daily interactions between
guide dogs and their blind owners could thus result in an
altered inter-species interaction pattern in guide dogs
compared to pet dogs. Guide dogs present a unique
opportunity to test whether incidental socio-communicative apprenticeship in dogs occurs earlier during their cohistory with their owner (for apprenticeship/enculturation
in apes see Cooper et al. 2003; Bering 2004; Tomasello and
Call 2004). In Gaunet (2008), guide and pet dogs were
prevented from accessing food that they had previously
learnt to access. Gazing at the container where the food
was and gazing at their owner (accompanied or not by
silent mouth licking), gaze alternation between the container and the owner, vocalisation and contact with the
owner did not differ among the groups. However, the guide
dogs licked their mouths audibly for a longer time and
more frequently than the pet dogs. These results overall
suggest that dogs are not sensitive to their owner’s visual
313
attentional state. However, the guide dogs showed incidental learning, suggesting that they supplemented their
means of triggering their owners’ attention with a new
distal cue. This behaviour cannot however be considered as
intentional: the task situation did not include, amongst
other possible criteria, a condition with the owner absent.
Although there are further indications that dogs use the
eyes of humans as a cue for referential communication
(Miklo´si et al. 1998, 1 dog; McKinley and Sambrook 2000,
2 dogs out of 11; Soproni et al. 2001; Reid 2009), the
literature indicates that the exact conditions under which
sensitivity to human eyes holds true remain unclear
(Gaunet 2008).
The purpose of the present study is to clarify any effect
of the visual status of the owner on dogs’ social-communicative signals for interacting with humans (criterion (4))
and to study whether dogs persist in and elaborate their
communicative behaviours in the face of failure to communicate a request (criteria (5) and (6)). More precisely, it
aimed to compare the interactive modes in guide dogs with
the modes in pet dogs (criterion (4)). This was enacted in a
play session based on a fetch task: the gazing behaviours,
vocalisations and contacts were analysed for the two
groups in the following conditions: (i) when the dogs were
asked by their owner to fetch their inaccessible toy; (ii)
when the dogs were subsequently given an unfamiliar
object (apparent unsuccessful communication by the dog)
or their toy (apparent successful communication by the
dog). These experimental situations model play situations
commonly experienced by humans and dogs and were
designed to test criteria (5) and (6).
If dogs meet criterion (4), in guide dogs sound and
contact should to some extent replace engaging the owner’s
gaze to trigger attention: more emissions of sound and
more contacts in the guide dog group than in the pet dogs
should be observed if a dog’s social cognition is attuned to
the owner’s visual status. Guide dogs could also be
expected to gaze less towards the box where the unreachable toy is hidden, and there is likely to be less gaze
alternation between their owners and the box than amongst
pet dogs. Finally, when an unfamiliar object or a toy is
returned to the dogs, if the dogs communicate about the
toy, guide dogs could be expected to present it to the
owner’s hand, whereas pet dogs would be expected to gaze
at the owner, holding the toy, or their gaze would alternate
between the owner and the toy.
If dogs meet criterion (5), they should display (i) cessation of communicative behaviours in a successful postdelivery period compared with an unsuccessful postdelivery
period, (ii) complete cessation of communication after
successful communication, and no cessation or facilitation
after unsuccessful communication in comparison with the
predelivery phase. To validate criterion (6), dogs should
123
314
exhibit new communicative behaviours and/or an increase
in the number of different kinds of communicative behaviours in case of return of an unfamiliar object compared to
the predelivery phase. In addition, the dogs should engage
in a behaviour apparently intended to communicate with
their owners in case of return of their toy more markedly
than for an unfamiliar object.
Methods
Participants
One group of 12 pet dogs belonging to sighted owners and
one group of 12 guide dogs belonging to blind owners,
matched for breed, age (t22 = -0.12, P = 0.90) and life
span spent together (t22 = -1.42, P = 0.16), were included in the experiment (see Table 1 for the characteristics of
the dogs). The guide dogs were raised in a sighted family
before receiving their training between the ages of 6 and
18–24 months. They were transferred to the ownership of
blind persons when they were adult (1.82 years old on
average). On average, the pet dogs spent 2 h per day
interacting with their owner (play, educational games,
obedience training…) in addition to morning, evening and
week-end walks and play. The selected dogs were known
to have a favourite toy, but were also used to playing
occasionally with their owner with other objects such as
pieces of wood, plastic bottles, etc. The owners received
€12 payment for their participation.
Anim Cogn (2010) 13:311–323
Table 1 Characteristics of the dogs (ages are in years)
Dog
Owner and dog
Breed
Sex
Age
No of years of
living together
Labrador
M
7
5.5
Labrador
Golden retriever
F
F
7
6.5
5.5
3.5
Labrador
F
6
4.5
Golden retriever
F
4
2
Golden retriever
F
7.5
5.5
Labrador
M
7
4
Guide dog
Golden retriever
M
3
2
Hovawart
F
5
3
Golden retriever
F
5
3.5
Golden retriever
M
5.5
4
Shepherd
F
6
4.5
5.79/1.35
3.95/1.23
Mean/SD
Pet dogs
Labrador
M
6.5
6.33
Labrador
F
8
6.17
Golden retriever
Labrador
F
F
6
6
5.75
5.67
Golden retriever
F
2.5
2
Golden retriever
F
7.5
4.5
Labrador
M
8
7.75
Golden retriever
F
3
2.75
Labrador
M
5
4.75
Golden retriever
F
5
2.5
Experimental settings
Golden retriever
M
8
5
Shepherd
M
5
4.75
A carpeted room (8 9 5 m) located in a laboratory was
dedicated to the experiment. The room was novel to the
participants. The whole experiment was videotaped by two
cameras for future analysis. A heavy box (1.2-m
wide 9 1.2-m high 9 0.5-m deep) was placed in a corner
between a wall and a cupboard. The dog’s toy (usually its
ball) and three small objects unfamiliar to the dog that
could be held in the mouth were used: a plastic milk bottle,
a plastic shampoo bottle and a brick-shaped piece of rubber, 15–20-cm long. The three unfamiliar objects were
placed behind the box before the dog entered the room. The
location of the owner was 1.5 m away from the box along
the wall.
Mean/SD
5.87/1.87
4.82/1.7
conditions (each including two different types of trial),
each repeated three times:
A. Unfamiliar object condition (failed communication):
it tested the behaviours of the dogs when they were given
back an unfamiliar object instead of their toy after the toy
had been made inaccessible. The experimenter remained
hidden behind another cupboard. The owner first called the
dog by its name, showed the dog its toy and then threw it
away for play. It was repeated three times, in the manner
usual to their fetch-game situations.
1.
Procedure
Before starting the experiment, the dogs were allowed to
explore the experimental room (20–25 min). During this
period, the owners were briefed (in the absence of their
dogs) about the procedure. This consisted of two different
123
Fetch followed by Unfamiliar Object trial (Fetch f
Unf.Obj. trial): The owner then called the dog by its
name to ensure that it was attending, showed, and
threw the toy behind the box. The experimenter
checked that the dog was attentive during these three
actions by glancing unobtrusively towards the experimental area and then returning behind the cupboard.
Anim Cogn (2010) 13:311–323
2.
Once the toy was inaccessible, the owner asked the
dog to fetch it, once or twice. The owner then waited
40 s at his/her place, without interacting with the
dog.
Unfamiliar Object trial (Unf.Obj. trial): The owner was
then required to retrieve an unfamiliar object from
behind the box (this was verbally signalled by the
experimenter), and to place it on the ground on a
marked spot (between the owner’s standing position
and the box, and 1 m away from the wall). The owner
then waited 40 s at his/her place without interacting
with the dog.
B. Toy condition (successful communication): this tested
the behaviours of the dogs when their toy was returned
after it had been inaccessible. The same free play session as
for Condition A was first performed and the experimenter
remained hidden behind a cupboard.
1.
2.
Fetch followed by Toy trials (Fetch f Toy trial): same
as for the Fetch f Unf.Obj. trial.
Toy trial (Toy trial): same as for the Unf.Obj. trial,
except that it was the dog’s toy that was returned from
behind the box.
Owners were asked to remain standing during the
experiment. The order of presentation of the conditions was
counterbalanced across dyads (using a Latin squares
design). A tactile code for the blind owners or a written
sheet for the sighted owners, giving the order of the conditions, was placed on the wall next to the owners’ standing
position. The blind owners practised throwing the toy
behind the box before the experiment started. The duration
of the trials was determined during pre-tests (40 s, measured with a handheld stopwatch by the experimenter). The
experiment overall lasted 20–30 min. Breaks were allowed
between conditions.
The motivation of the dogs for playing with their
owner with three unfamiliar objects of the same type as
those used during the experiment was checked before
starting the procedure. They were placed on the ground.
The dogs spontaneously took them in their mouths or
played with each of them with their owner, or engaged
in a fetching game with their owner when asked to.
The criteria for inclusion of the dogs in the experimental
group were (1) that the dog had a favourite toy and occasionally played with the owner with objects found on the
ground, (2) the involvement of the dog in a play session
with its owner with its toy and the three unfamiliar objects
in the laboratory, and (3) the display of interest towards the
box during and after the hiding of the toy. Of 27 dogs
initially recruited, 3 were unwilling to play with their
owner in the laboratory and were not included in the
experiment.
315
Data collection and analyses
Data on the number of occurrences and the duration of the
following behaviours were collected during the 40 s-periods of the Fetch f Unf.Obj. trials, Fetch f Toy trials,
Unf.Obj. trials and Toy trials (see Hare et al. 1998; Miklo´si
et al. 2000; Gaunet 2008).
Fetch trials:
•
•
•
•
•
•
•
GazeOwner: the dog’s head/nose was oriented towards
the owner’s head.
GazeBox: the dog’s head/nose was oriented towards the
box.
GazeAlternation: only the number of looks at the owner
that were followed directly within 2 s by a look at the
box, or vice versa, was collected.
Vocalisation: the dog barked and/or whined.
Contact: the dog pawed the owner, touched him/her
with its nose or head, or performed another/other
form(s) of bodily contact.
MouthLickingNS: silent mouth-licking behaviour was
displayed by the dog.
MouthLickingS: sonorous (audible) mouth-licking
behaviour was displayed by the dog.
Certain combinations of the interactive behaviours were
also observed and collected: GazeOwner ? Vocalisation,
GazeBox ? Vocalisation, GazeOwner ? MouthLickingS
and GazeBox ? MouthLickingS.
Unfamiliar Object and Toy trials:
The same 11 and 6 behaviours involving the toy or the
unfamiliar object (termed object below) were collected:
•
•
•
•
•
•
SniffObject: the dog sniffed the object.
GazeObject: the dog gazed at the object.
GazeObject ? MouthLickingS: the dog gazed at the
object and performed sonorous mouth licks.
HoldObject: the dog held the object in its mouth.
HoldObject ? GazeOwner: the dog held the object in
its mouth and gazed at the owner’s head.
HoldObject ? Contact: the dog held the object in its
mouth and touched its owner.
Two trained observers recorded the occurrences of
behaviours independently (one of them was naı¨ve with
respect to the aim of the experiment). Before data analyses,
inter-observer agreement was assessed for the most frequent and relevant behaviours by means of parallel coding
of 100% of the sample for duration of the behaviours.
Kendall’s concordance coefficient was calculated. For the
three Fetch f Toy trials, this yielded W = 0.89 with
P = 0.0001 for GazeBox, W = 0.94 with P = 0.0001 for
GazeOwner, W = 0.09 with P = 0.0005 for Vocalisation,
W = 0.38 with P = 0.0001 for Contact, and W = 0.77
123
316
with P = 0.0001 for MouthLickingS. For the three
Unf.Obj. trials pooled and the three Toy trials pooled, W
for HoldObject ? GazeOwner were, respectively, 0.31 and
0.85, with P = 0.0001. The results indicate good agreement between raters.
For each dog and each trial, the relative number and
duration of behaviours (expressed in percentages) were
computed. For each dog, the relative number and duration
of the three Fetch f Unf.Obj. trials, the three Fetch f Toy
trials, the three Unf.Obj. trials and the three Toy trials were
averaged separately to enable comparisons, for each individual behaviour. These relative numbers and durations of
behaviours were first submitted to between-group analyses
for each of the four types of trial, to study the effect of the
visual status of the owner (criterion (4)). Both variables
(number and duration) were then submitted to betweentrial analyses to evidence whether the dogs’ behaviours
showed persistence and elaboration (criteria (5) and (6)).
Non-parametric analyses were used. In ‘‘Results’’, relative
number and duration of behaviours are, respectively,
referred to as number and duration to simplify the text. In
the figures, non-parametric data are presented as medians,
and the box indicates the interquartile range of 50% of the
data. Whiskers extend to the smallest and largest values
and exclude outliers. To test for criterion (6), the dogs were
classified as either having exhibited multiple behaviours or
not having done so, and non-parametric and parametric
statistics were used.
Results
Between-group analyses
The Mann–Whitney test applied to the number and duration of the behaviours did not show differences between the
guide dogs and the pet dogs for the Fetch f Unf.Obj. trials,
the Fetch f Toy trials, the Unf.Obj. trials or the Toy trials
(44.5 B U B 72; 0.11 B P B 0.99; see S1 for the detailed
results). Thus, the two groups did not differ for any of the
behaviours and trials, indicating that criterion (4) is not
verified in the present experimental paradigm.
Between-trial analyses
Given the absence of differences between the two groups,
the data from the two groups were pooled for each individual behaviour and for each type of trial. The Wilcoxon
test was used to compare Fetch f Unf.Obj. trials versus
Fetch f Toy trials, Fetch f Unf.Obj. trials versus Unf.Obj.
trials, Fetch f Toy trials versus Toy trials, and Unf.Obj.
trials versus Toy trials. Cochran’s Q test (a non-parametric
test appropriate for dichotomous variables in repeated
123
Anim Cogn (2010) 13:311–323
measure designs, Siegel and Castellan 1988) was also used
in a few instances to disprove or confirm the Wilcoxon
results, and to study the elaboration of behaviours.
Predelivery
None of the 11 behaviours differed between the two Fetch
trials (Wilcoxon tests: 2 B T B 141; 0.09 B P B 0.99; see
S2 for the detailed results): the dogs apparently did not
discriminate between the Fetch trials followed by unfamiliar object delivery and followed by toy delivery on the
basis of any uncontrolled differences in the two predelivery
phases. These results validate the experimental design,
allowing comparison of the predelivery phases with postdelivery phases, and comparison of the two postdelivery
phases.
Postdelivery
For criterion (5) to be validated, behaviours should show
no change in their number and duration between the Fetch f
Unf.Obj. trials and the Unf.Obj. trials, and a decrease in
number and duration during the Toy trials compared with
the Fetch f Toy trials and with the Unf.Obj. trials. This is
not the case for most of the behaviours. Indeed, we did not
evidence this pattern of results for any of the following: the
three vocal and the two mouth-licking behaviours; gazing
at the object; gazing at the owner or at the object accompanied with sonorous mouth licks; contacts and holding the
object accompanied by touching the owner (Wilcoxon
tests: 0 B T B 129; 0.01 B P B 0.99; with all P ns with
the BL FDR adjustment; see S3 for detailed results).
In contrast, two behaviours fulfilled criterion (5). Firstly,
the numbers and durations for gaze at the box (Fig. 1) did
not differ between the predelivery phase and the delivery of
the unfamiliar object, but the values for both variables were
greater during the predelivery phase than after delivery of
the toy. Finally, the number and duration were greater
when the unfamiliar object was given back than when the
toy was returned (Table 2). Secondly, the number of gaze
alternations followed the same directional pattern of
change, except for predelivery versus unfamiliar object
delivery (Fig. 2; Table 2): they were greater during the
predelivery phase than following the return of an unfamiliar object. Thus, delivery of the unfamiliar object had
no influence on gaze at the box and some influence on
gaze-alternating behaviours, but delivery of the toy resulted
in a near-complete cessation of these activities. Hence,
communication in this experimental context is about the
toy. A plausible explanation is that the decrease in this
communicative behaviour following the return of an
unfamiliar object is attributable to the time spent sniffing
the unfamiliar object (see in Table 2 the greater number
Anim Cogn (2010) 13:311–323
GazeBox
**
45
Relative duration (%)
35
30
25
20
15
10
Unf. Obj.
and longer duration for sniffing the unfamiliar object, a
novelty-detection behaviour, Berlyne 1950, than for sniffing the toy). Non-cessation was not, however, attributable
to an increase in gazing at the object with or without mouth
licks following delivery of an unfamiliar object: the number and duration of these behaviours during the trials do not
differ from that observed during the Toy trials (Table 2).
Additionally, for the number and the duration of gaze at
the box accompanied by sonorous mouth licks, the comparisons between predelivery and delivery phases verify
criterion (5): no difference was evidenced between the
predelivery phase and the delivery of an unfamiliar object,
whilst this behaviour was displayed with greater frequency
and for a longer time when the toy was hidden than when it
was given back. However, no significant differences were
found for either variable when the unfamiliar object was
delivered compared to when the toy was given back (Fig. 3;
Table 2). However, the number of dogs that performed this
behaviour during the Unf.Obj. trials (7) was significantly
greater than during the Toy trials (1), and this difference
was significant using the Cochran test (Q = 6, P \ 0.014),
suggesting that this behaviour fulfils criterion (5).
GazeAlternation
**
**
Number
8
6
4
2
0
Unf. Obj.
30
25
20
15
10
0
Fetch f
Unf.Obj.
Fetch f
Unf.Obj.
35
5
0
**
**
40
5
10
**
45
**
40
Relative number (%)
Fig. 1 Relative number (left)
and duration (right) of gaze at
the box for the guide dogs and
the pet dogs pooled
317
Fetch f
Toy
Toy
Fig. 2 Number of gaze alternations for the guide dogs and the pet
dogs pooled
Fetch f
by Toy
Toy
Fetch f
Unf.Obj.
Unf. Obj.
Fetch f
Toy
Toy
Gaze at the owner only partially fulfilled criterion (5).
Indeed, the number and duration of gaze at the owner did
not differ between the predelivery phase and the delivery of
an unfamiliar object. But when it was the toy that was
returned, the numbers were greater during the predelivery
phase than after the return; however, the difference in
duration for the latter comparison was not statistically
significant. Finally, neither the number nor the duration of
the behaviour differed between return of the unfamiliar
object and return of the toy (see Table 2). This result was
confirmed by the Cochran test (Q = 0.33, P = 0.5, ns,
with no. of dogs for Unf.Obj. trials = 23 and no. of dogs
for Toy trials = 22).
The number and duration of holding the object (Fig. 4a)
and only the number for holding the object accompanied by
gaze at the owner (Fig. 4b) were significantly greater when
the toy was returned than when the unfamiliar object was
returned (Table 2). These results were confirmed by the
Cochran test (for HoldObject Q = 15, P \ 0.0001, with
no. of dogs for Unf.Obj. trials = 9 and no. of dogs for Toy
trials = 24; for HoldObject ? GazeOwner Q = 8.8,
P \ 0.0028, with no. of dogs for Unf.Obj. trials = 4 and
no. of dogs for Toy trials = 17).
Finally, the fact that dogs sniffed at the unfamiliar
objects more than at the toy is fully congruent with the
results obtained by Kaulfuss and Mills (2008). However,
the present results differ from those obtained by these
authors for holding unfamiliar objects versus the toy, where
neophilia was evidenced: their dogs chose to hold the
unfamiliar objects. The difference might be explained by
the weaker co-history between the dog, the familiar object
and the experimenter in Kaulfuss and Mills’s experiment
(10 min) compared to the long co-history between the
dogs, the favourite toy and their owners in the present
study. The differences in task contexts may also contribute:
in the present study, the dogs played with their owners and
the toy was rendered inaccessible, and the owner asked the
dogs to fetch it once or twice.
To sum up, relative to the predelivery phase and to the
delivery of an unfamiliar object, delivery of the toy had a
123
318
Anim Cogn (2010) 13:311–323
Table 2 Wilcoxon tests results for the comparisons of Fetch followed
by Unfamiliar Object trials (Fetch f Unf.Obj) versus Unfamiliar
Object trials (Unf.Obj.), Fetch followed by Toy trials (Fetch f Toy)
versus Toy trials (Toy), and Unfamiliar Object trials (Unf.Obj.)
versus Toy trials (Toy) with n = 24 for all comparisons. Significant
Behaviours
differences are in bold. After the FDR BL adjustment (Benjamini
et al. 2001) only the P values shown in italics remain statistically
significant (threshold for number: 0.0013; threshold for duration:
0.0012). Medians are given on the second line
Number
Duration
Fetch followed by Unfamiliar Object versus Unfamiliar Object
GazeOwner
GazeBox
GazeAlternation
T = ? 73, P = 0.026
T = ? 141, P = 0.81
Fetch f Unf.Obj. = 23.80; Unf.Obj. = 20.09
Fetch f Unf.Obj. = 22.25; Unf.Obj. = 16.74
T = ? 36, P = 0.006
T = ? 85, P = 0.064
Fetch f Unf.Obj. = 19.45; Unf.Obj. = 9.83
Fetch f Unf.Obj. = 13.65; Unf.Obj. = 6.59
T = ? 1, P = 0.0001**
–
Fetch f Unf.Obj. = 1.78; Unf.Obj. = 0.054
GazeBox ? MouthLickingS
T = ? 32, P = 0.12
T = ? 23, P = 0.035
Fetch f Unf.Obj. = 0.85; Unf.Obj. = 0
Fetch f Unf.Obj. = 0.2; Unf.Obj. = 0
GazeOwner
T = ? 15, P = 0.0001**
Fetch f Toy = 23.55; Unf.Obj. = 11.76
T = ? 69, P = 0.019
Fetch f Toy = 19.95; Toy = 8.72
GazeBox
T = ? 0, P = 0.0001**
T = ? 0, P = 0.0001**
Fetch followed by Toy versus Toy
GazeAlternation
Fetch f Toy = 16.4; Toy = 0
Fetch f Toy = 12.55; Toy = 0
T = ? 0, P = 0.0001**
–
Fetch f Toy = 1.6; Toy = 0
GazeBox ? MouthLickingS
T = ? 0, P = 0.0001**
T = ? 0, P = 0.0001**
Fetch f Toy = 2.4; Toy = 0
Fetch f Toy = 1.15; Toy = 0
T = ? 71, P = 0.022
T = ? 83, P = 0.056
Unf.Obj. = 20.09; Toy = 11.76
Unf.Obj. = 16.74; Toy = 8.72
T = ? 1, P = 0.0001**
T = ? 1, P = 0.0001**
Unfamiliar Object versus Toy
GazeOwner
GazeBox
Unf.Obj. = 9.83; Toy = 0
Unf.Obj. = 6.59; Toy = 0
GazeAlternation
T = ? 0, P = 0.0005**
–
GazeBox ? MouthLickingS
T = ? 1, P = 0.031
Unf.Obj. = 0; Toy = 0
T = ? 0, P = 0.0156
Unf.Obj. = 0; Toy = 0
SniffObject
T = ? 10.5, P = 0.0001**
T = ? 12, P = 0.0001**
Unf.Obj. = 0.054; Toy = 0
Unf.Obj. = 8.19; Toy = 0
Unf.Obj. = 4.67; Toy = 0
GazeObject
T = ? 117, P = 0.35
T = ? 129, P = 0.54
Unf.Obj. = 11.02; Toy = 14.03
Unf.Obj. = 3.31; Toy = 3.71
GazeObject ? MouthLickingS
T = ? 1, P = 0.031
T = ? 23, P = 0.15
Unf.Obj. = 0; Toy = 0
Unf.Obj. = 0; Toy = 0
HoldObject
T = ? 24, P = 0.0001**
T = ? 42, P = 0.0012*
Unf.Obj. = 0; Toy = 23.43
Unf.Obj. = 0; Toy = 27.52
HoldObject ? GazeOwner
T = ? 15.5, P = 0.0003**
T = ? 32, P = 0.0049
Unf.Obj. = 0; Toy = 6.38
Unf.Obj. = 0; Toy = 0.99
suppressive or a reduction effect on the values of the two
variables (number and duration of behaviours) for GazeBox
(no. of dogs for Fetch f Toy = 24, no. of dogs for
Unf.Obj. = 22 and no. of dogs for Toy = 6), GazeAlternation (no. of dogs for Fetch f Toy = 19, no. of dogs for
Unf.Obj. = 12 and no. of dogs for Toy = 0) and GazeBox ? MouthlickingS (no. of dogs for Fetch f Toy = 14,
no. of dogs for Unf.Obj. = 7 and no. of dogs for Toy = 1).
123
This effect was clearly less marked for GazeOwner; but
given that it was the toy that was returned and that the dogs
might then be expected to solicit play from their owners
(and look for their gaze), the cessation of these gazing
behaviours was not to be expected. Moreover, these four
behaviours did not differ between the predelivery phase and
the delivery of an unfamiliar object. This shows that the
GazeBox, GazeAlternation and GazeBox ? MouthlickingS
Anim Cogn (2010) 13:311–323
GazeBox+MouthLickingS
13
12
11
10
9
8
7
6
5
4
3
2
1
0
**
Relative duration (%)
Fetch f
Unf.Obj.
Fetch f
Toy
**
Fetch f
Unf.Obj.
Toy
(a)
Unf. Obj.
HoldObject
Relative number (%)
Fig. 4 Relative number (left)
and duration (right) of holding
the object (unfamiliar or toy)
(a), and holding the object
accompanied by gazing at the
owner (b), for the guide dogs
and the pet dogs pooled
Unf. Obj.
13
12
11
10
9
8
7
6
5
4
3
2
1
0
**
100
90
80
70
60
50
40
30
20
10
0
Relative duration (%)
Relative number (%)
Fig. 3 Relative number (left)
and duration (right) of gaze at
the box accompanied by
sonorous mouth licking for the
guide dogs and the pet dogs
pooled
319
Unf.Obj.
(b)
Toy
*
100
90
80
70
60
50
40
30
20
10
0
Unf.Obj.
Toy
Fetch f
Toy
Toy
HoldObject+GazeOwner
**
20
Relative duration (%)
Relative number (%)
20
15
10
5
0
15
10
5
0
Unf.Obj.
fulfil criterion (5). In addition, HoldObject and HoldObject ? GazeOwner showed an increase in values when the
toy was given back compared to when the unfamiliar object
was returned, and conversely for SniffObject. For all the
other behaviours, the results do not meet the requirements to
validate criterion (5).
Elaboration
For the analysis of behavioural elaboration (criterion (6)),
the dogs were dichotomously classified either as having
exhibited multiple behaviours or as not having done so:
initially, the range was 0–2 behaviours per subject, with
behaviours being gaze at the target and gaze at the target along with sonorous mouth licks. In the two predelivery
phases, the proportion of dogs that exhibited multiple
behaviours was randomly distributed: the number of dogs
Toy
Unf.Obj.
Toy
was 23 for the predelivery of the unfamiliar object phase
and 22 for the predelivery of the toy phase, Cochran’s
Q = 1, P = 0.31, ns. In contrast, the proportion of individuals that exhibited multiple behaviours was non-randomly distributed in the postdelivery phase, Cochran’s
Q = 16, P \ 0.00001: the number of dogs was 22 following the delivery of an unfamiliar object, and 6 following the delivery of the toy.
A 2 (phase) 9 2 (trial) repeated measures ANOVA was
performed, in which the number for the two behaviours
served as the dependent variable. If a dog gazed at the
target and gazed at the target along with sonorous mouth
licks, it exhibited two different kinds of behaviour. This
revealed a main effect for phase, F1,23 = 50.54,
P \ 0.000001: there were more communicative behaviours
exhibited in the predelivery phase than in the postdelivery
phase. There was also a main effect for condition,
123
Mean (+/-SE) number of multiple behaviours
320
Anim Cogn (2010) 13:311–323
Unfamiliar Object
Toy
1,5
1,0
0,5
0,0
Predelivery phase
Postdelivery phase
Fig. 5 Interaction between phases and trials in the number of kinds
of communicative behaviours (gaze at the box with and without
sonorous mouth licks) exhibited by the 24 dogs
F1,23 = 21.67, P \ 0.0001: there were fewer multiple
communicative behaviours exhibited during the Toy trials
than during the Unfamiliar Object trials. Finally, there was
a phase 9 trial interaction, F1,23 = 23,99, P \ 0.00006,
which is depicted in Fig. 5. Paired t tests confirmed first
that there was no apparent suppression (but no elaboration)
of the numbers of two kinds of communicative behaviours
exhibited after delivery of the unfamiliar object compared
with the predelivery phase for the unfamiliar object,
t23 = 1.54, P = 0.13, ns. In contrast, after delivery of the
toy, there was a marked decrease in the numbers for the
different behaviours exhibited: t23 = 9.16, P \ 0.0000001.
Thus, by this aggregate measure, as for the above individual analyses, obtaining the unfamiliar object apparently
did not suppress communication in this study. This suggests that the quasi-cessation of communicative behaviours
observed in the Toy condition is not attributable to the
retrieval action of the owner or to the delivery of an object
to the dog, but rather to the dogs’ perceived success in
obtaining the desired communicative outcome.
It is important to note that, relative to the predelivery
phase, delivery of the unfamiliar object had no facilitating
effect on the behaviour, either quantitatively or qualitatively: for the latter, no new communicative behaviours
were found during the analysis of the 40 s sequences
overall. This indicates that criterion (6) is not met in the
present experimental paradigm.
Discussion
The first aim of this study was to determine if the modes of
interaction of dogs with their owners are attuned to the
visual status of the owners (sighted or blind). This could
123
indicate whether there is an influence of the visual attentional status of an observer on the propensity to exhibit
particular communicative behaviours (criterion (4)). The
behaviours of guide dogs and pet dogs towards their
respective owners were compared in tasks during which
dogs were first prevented from accessing their toy, and were
next given back either an unfamiliar object or their toy. In
none of the experimental situations did the two groups
differ in novelty detection, object-directed behaviours or in
modes of interaction (gazing at the box and at the owner,
gaze alternation, mouth licks, vocalisations, contacts with
the owner and any combination of these behaviours). The
results thus show that in commonly experienced interactive
situations of this type, dogs do not show sensitivity to their
owner’s visual attentional state. This result converges with
the findings by Ittyerah and Gaunet (2009), who showed
that the same two categories of dogs did not differ when
required to respond to pointing, pointing and gaze and gaze
cues for referential communication given by their blind and
sighted owners. The visual status of the owner was not a
factor in the use of referential communication cues. In
addition to this, and in contrast to the findings by Gaunet
(2008), the present two groups did not differ for sonorous
mouth licks. This could be explained by the fact that the
present situation does not involve food, whereas the study
by Gaunet (2008) did. However, in the present task, when
the two groups were pooled, sonorous mouth licks accompanying gaze at the box almost disappeared when the toy
was returned (only one dog displayed this behaviour) contrasting with the rates observed during the two Fetch trials
and the return of the unfamiliar object. Sonorous mouth
licks hence seem more likely to be displayed when dogs
apparently indicate the direction of the desired object
(or make requests), and it is also possible that sonorous
mouth licks alone are more likely (and more likely to be
reinforced) in food-related situations (Gaunet 2008).
Overall, the similar behaviours in the two groups could
be explained by the following: (i) the visual status of the
owner may be too subtle a behavioural cue to be used by
(all) dogs (see also Miklo´si et al. 1998; McKinley and
Sambrook 2000; Soproni et al. 2001; Ga´csi et al. 2004;
Ittyerah and Gaunet 2009); (ii) guide dogs are raised and
continue to live surrounded by sighted people; (iii)
domestication, which has led dogs to use human gaze (Hare
et al. 2002, 2005; Hare and Tomasello 2005; see also
Miklo´si and Soproni 2006 for a discussion), may be
stronger than any socio-communicative apprenticeship or
incidental learning that could apply to the other modes of
interaction in particular human–dog dyads that have spent
at least 2 years of life together.
The comparison of the dogs’ behaviours when the toy
was inaccessible, when an unfamiliar object was returned
and when the toy was returned casts light on whether there
Anim Cogn (2010) 13:311–323
is (5) persistence in and (6) elaboration of communicative
behaviours when attempts to manipulate the partner fail
(the second aim of the study). There is no indication that
the dogs elaborated their communicative behaviours (i.e.
no new communicative behaviours or facilitation were
evidenced with the present methodology). However, the
results indicate some persistence in their communicative
actions when the unfamiliar objects were returned (see the
combination of results obtained from the comparisons of
the predelivery trials with the Unf.Obj. trials, of the predelivery trials with the Toy trials and of the two postdelivery trials, respectively, for gaze at the box, gaze
alternation and gaze at the box accompanied by sonorous
mouth licks). The fact that there was no complete cessation
of gaze at the owner following toy-return trials can be
easily explained by the fact that there was still opportunity
for interacting with the owner in the room, i.e. the toy that
had been returned.
On the whole, the results suggest that one criterion (5)
for intentional communication by dogs has been demonstrated for three behaviours directed at the hidden target
(see Hare et al. 1998, Study 3; Miklo´si et al. 2000; for an
extensive review of gaze alternation in apes see Leavens
et al. 2005b and Anderson et al. 2007 for squirrel monkeys). This converges with the view of Miklo´si et al.
(2000), who argued that dogs might be able to engage in
functionally referential communication with their owner,
and that their behaviour could be described as a form of
intentional ‘‘showing’’. Furthermore, the way in which
gaze directed towards the owner was displayed across the
trials suggests that this behaviour is used as a form of
getting attention.
The results reveal a further communicative behaviour.
Interestingly, the dogs held the toy gazing at the owner more
often and longer than they did with the unfamiliar object.
This behaviour may have an attention-getting or deictic
component. Indeed, a similar behaviour develops at
9 months in infants, namely the give-and-take behaviour
(Iverson and Meadow 1997; Shwe and Markmen 1997;
Messinger and Fogel 1998), and these games reach their
maximum when children are between 15 to 24 months old
(Bruner 1977; Restoin et al. 1985). It is thought that infants
may use an offer to ask the partner to do something with the
offered object (Messinger and Fogel 1998). In dogs, the
fetch-game is of a cooperative nature, which means that if
dogs are ordered to fetch something, they ‘‘rightly’’ expect
appropriate coordinated behaviour on the part of the receiver
based on their previous experiences (Mitchell and Thompson 1991; Ga´csi et al. 2004; see also Kerepesi et al. 2005 for
human–dog synchronised activities). Here, it is possible that
the dogs gazed at their owners along with holding the toy in
the mouth as if to communicate about the returned toy
because of the ongoing cooperative context of play.
321
Finally, across trials, the following behaviours were not
differentially performed: vocalisations alone or associated
with gazing at the owner or at the box, sonorous or silent
mouth licks, gazing at the owner accompanied with sonorous mouth licks, contacts. These behaviours may thus
have less importance than gaze at the box, gaze alternation
and gaze at the box along with sonorous mouth licks for
performing/reiterating a request. The same remark applies
to gazing at the object, gazing at the object accompanied
with sonorous mouth licks and holding the object accompanied by touching the owner compared to holding the toy
along with gazing at the owner or not for communicating
about the returned toy.
Overall, the results show that the dogs display specific
patterns of communicative behaviours when communication fails (criterion (5) only), and that they behave as if they
are asking to be given back their toy in order to use it to
keep interacting with their owner (cf. when communication
has succeeded). The dogs did indeed hold their toys and
gaze at their owners after it was given back more than they
did with the unfamiliar objects. The study finally reveals
that, in the trials overall, the patterns of differences in the
behaviours were similar for the two variables analysed.
This suggests no dissociation between the frequency and
duration of these communicative behaviours in dogs.
In conclusion, dogs do not demonstrate any sensitivity to
the visual status of their owners in the present toy-fetching
context. This is in agreement with the very mild influence
of the human visual status found in guide dogs by Ittyerah
and Gaunet (2009) and Gaunet (2008). Another result is
that not only do dogs display apparent intentional ‘‘showing’’ behaviours to get the object they desire, but they also
display another communicative behaviour when the desired
target has been given back. More specifically, (i) dogs
behave as if they are asking for their inaccessible toy by
gazing at their owners, apparently to trigger their attention;
the values of gaze alternation from box to owner and for
two deictic behaviours directed at the box remain high until
the toy is returned, apparently performed to indicate the
location of the toy to the owner; (ii) dogs then hold their
toy and gaze at the owner as though trying to get their
owner’s attention with regard to the toy.
Are the intentional requesting behaviours evidenced
only behavioural manifestations (and hence probably based
on instrumental learning) or do they reflect a mental state
or a theory of mind about others? In vocalising wild animals, the motivational and referential properties of animal
calls coexist (Seyfarth and Cheney 2003). The way in
which these communicative behaviours appeared or
developed in animals could provide some insights into the
issue. In chimpanzees (and orangutans, gorillas and bonobos) pointing is variably expressed, which is rare in the
wild, but common in captive, animals not explicitly trained
123
322
in pointing, suggesting the involvement of a hitherto
unidentified learning process (Leavens et al. 2005a). For
Go´mez (2007), who reviewed the studies focusing on apes
performing intentional and referential signalling, the gestures of apes and monkeys can neither be dismissed as
simple conditioned responses, nor be uncritically accepted
as fully equivalent to human gestures (i.e. driven by an
understanding of the socio-cognitive causality underlying
their efficacy). In dogs, Miklo´si et al. (2000) have argued
that there are some circumstantial observations that make
apprenticeship less likely, although incidental learning
cannot be excluded. Further to this, in a task where dogs
were presented with desirable food out of their reach,
Bentosela et al. (2008) reported that three reinforcement
trials were sufficient to increase the pet dogs’ rates of
gazing towards an unfamiliar human face, and the dogs
were able to learn to ignore human cues and not to emit
communicative signals when these responses were no
longer successful. Bentosela et al. (2008) suggest that the
gaze response involves instrumental learning processes and
would not require complex cognitive explanations. However, they caution that the latter explanation cannot be
completely excluded and conclude that more research is
required. This converges with the review by Reid (2009),
on the use by dogs of human-given cues, who concluded
that present data do not make it possible to attribute
sophisticated abilities to dogs. To postulate mental states in
dogs during referential and partially intentional signalling
on the basis of the present or previous studies would thus
be premature.
Criteria (1) and (2) are now well established in dogs, and
there is evidence in studies on the dog’s socio-communicative skills that can support other criteria. This is further
backed up by the recent impressive results obtained by
Pongra´cz-Rossi and Ades (2008) with a female dog, Sofia.
She was submitted to a training schedule in which she
learnt to ask for objects or activities by selecting lexigrams
and pressing keys on a keyboard. The analysis of her
behaviour during a spontaneous interaction with an
experimenter showed that the lexigrams were used in an
appropriate, intentional way, in accordance with the
immediate motivational context. She utilised the keyboard
in the experimenter’s presence and gazed at him after key
pressing. We suggest that a full validation of Leavens
et al.’s criteria can be achieved with the following combination of experimental conditions: owner and target
present versus only target present for criterion (3) (cf.
Miklo´si et al. 2000); human being attentive versus not
attentive to the dog (eyes open versus closed; facing versus
back turned) for criterion (4); toy hidden versus toy given
back versus unfamiliar object given back versus less
familiar/preferred toy given back, with specific statistical
analyses for testing persistence and elaboration of
123
Anim Cogn (2010) 13:311–323
behaviours, for criteria (5) and (6), respectively. These
investigations could provide evidence that the cognitive
abilities (whatever the underlying processes) that are
implied by the successful ‘‘manipulation’’ of a social
partner through intentional referential communication by
human infants and chimpanzees are also present in dogs.
Acknowledgments This work was supported by the Centre
National de la Recherche Scientifique and conducted at the ‘‘Laboratoire Eco-Anthropologie et Ethnobiologie’’, Muse´um National
d’Histoire Naturelle, Paris, France. The experiments comply with the
current laws in France for animal and human research. The author
thanks the guide and pet dog owner dyads for their interest and
cooperation. The author is especially grateful to P. Piwowar for her
contribution to the analysis of the videos and to A. Le Jeannic for her
contribution to conducting the experiments and in analysing the tapes.
The author is also grateful to A. Miklo´si for helpful commentaries on
a previous version of the manuscript, and to the anonymous reviewers
for their helpful comments on the design and interpretation of the
study and for corrections to improve the manuscript.
References
Agnetta B, Hare B, Tomasello M (2000) Cues to food locations that
domestic dogs (Canis familiaris) of different ages do and do not
use. Anim Cogn 3:107–112
Anderson JR, Kuwahata H, Fujita K (2007) Gaze alternation during
‘‘pointing’’ by squirrel monkeys (Saimiri sciureus)? Anim Cogn
10(2):267–271
Benjamini Y, Drai D, Elmer G, Kafkafi N, Golani G (2001)
Controlling the false discovery rate in behavior genetics
research. Behav Brain Res 125:279–284
Bentosela M, Barrera G, Jakovcevic A, Elgier AM, Mustaca AE
(2008) Effect of reinforcement, reinforcer omission and extinction on a communicative response in domestic dogs (Canis
familiaris). Behav Proc 78(3):464–469
Bering JM (2004) A critical review of the ‘enculturation hypothesis’:
the effects of human rearing on great ape social cognition. Anim
Cogn 7:201–213
Berlyne DE (1950) Novelty and curiosity as determinants of
exploratory behavior. Brit J Psychol 41:68–80
Bra¨uer J, Call J, Tomasello M (2004) Visual perspective-taking in
dogs (Canis familiaris) in the presence of barriers. Appl Anim
Behav Sci 88:299–317
Bruner JS (1977) Early social interaction and language acquisition.
In: Schaffer HR (ed) Studies in mother–infant interaction.
Academic Press, London, pp 271–289
Bruner JS (1981) Intention in the structure of action and interaction. In:
Lipsitt LP, Rovee-Collier CK (eds) Advances in infancy research
1. Ablex Publishing Corporation, Norwood, NJ, pp 41–56
Call J, Tomasello M (1994) Production and comprehension of
referential pointing by orangutans (Pongo pygmaeus). J Comp
Psychol 108:307–317
Call J, Bra¨uer J, Kaminski J, Tomasello M (2003) Domestic dogs are
sensitive to the attentional state of humans. J Comp Psychol
117:257–263
Cartmill EA, Byrne RW (2007) Orangutans modify their gestural
signaling according to their audience’s comprehension. Curr
Biol 17:1345–1348
Cooper JJ, Ashton C, Bishop S, West R, Mills DS, Young RJ (2003)
Clever hounds: social cognition in the domestic dog (Canis
familiaris). App Anim Behav Sci 81:229–244
Anim Cogn (2010) 13:311–323
Evans CS (1997) Referential signals. Persp Ethol 12:99–143
´ , Varga O, Topa´l J, Csa´nyi V (2004) Are readers
Ga´csi M, Miklo´si A
of our face readers of our minds? Dogs (Canis familiaris) show
situation-dependent recognition of human’s attention. Anim
Cogn 7:144–153
Gaunet F (2008) How do guide dogs of blind owners and pet dogs of
sighted owners (Canis familiaris) ask their owners for food?
Anim Cogn 11(3):475–483
Go´mez J-C (2007) Requesting gestures in captive monkeys and apes:
conditioned responses or referential behaviours? In: Liebal K,
Mu¨ller C, Pika S (eds) Gestural communication in nonhuman
and human primates. John Benjamins, Amsterdam, pp 81–94
Hare B (2004) Domestic dog use humans as tools. In: Bekoff M (ed)
Encyclopedia of animal behavior, vol 1. Greenwood Press,
Westport, pp 277–285
Hare B, Tomasello M (2005) Human-like social skills in dogs?
Trends Cogn Sci 9:439–444
Hare B, Call J, Tomasello M (1998) Communication and food
location between human and dogs (Canis familiaris). Evol
Comm 2:137–159
Hare B, Brown M, Williamson C, Tomasello M (2002) The
domestication of social cognition in dogs. Sci 298:1634–1636
Hare B, Plyusnina I, Ignacio N, Schepina O, Stepika A, Wrangham R,
Trut L (2005) Social cognitive evolution in captive foxes is a
correlated by-product of experimental domestication. Curr Biol
15:226–230
Horowitz A (2009) Attention to attention in domestic dog (Canis
familiaris) dyadic play. Anim Cogn 12:107–118
Hostetter A, Cantero M, Hopkins WD (2001) Differential use of vocal
and gestural communication in chimpanzees in response to the
attentional status of a human audience. J Comp Psychol
115:337–343
Ittyerah M, Gaunet F (2009) The response of guide dogs and pet dogs
(Canis Familiaris) to cues of human referential communication.
Anim Cog 12(2):257–265
Iverson JM, Meadow SG (1997) What’s communication got to do
with it? Gesture in children blind from birth. Dev Psychol
33(3):453–467
Kaulfuss P, Mills DS (2008) Neophilia in domestic dogs (Canis
familiaris) and its implications for studies of dog cognition.
Anim Cog 11:553–556
´ , Topa´l J, Csa´nyi V, Magnusson
Kerepesi A, Jonsson GK, Miklo´si A
MS (2005) Detection of long-term temporal patterns in dog–
human interaction. J Behav Proc 70:69–79
Leavens DA (2004) Manual deixis in apes and humans. Inter Stud
5:387–408
Leavens DA, Hopkins WD (1998) Intentional communication by
chimpanzees (Pan troglodytes): a cross-sectional study of the use
of referential gestures. Dev Psych 34:813–822
Leavens DA, Hopkins WD, Bard KA (1996) Indexical and referential
pointing in chimpanzees (Pan troglodytes). J Comp Psychol
110:346–353
Leavens DA, Hopkins WD, Thomas R (2004) Referential communication by chimpanzees (Pan troglodytes). J Comp Psychol
118:48–57
323
Leavens DA, Hopkins WD, Bard KA (2005a) Understanding the
point of chimpanzee pointing: epigenesis and ecological validity.
Curr DiR Psych Sci 14:185–189
Leavens DA, Russell JL, Hopkins WD (2005b) Intentionality as
measured in the persistence and elaboration of communication
by chimpanzees (Pan troglodytes). Child Dev 76:291–306
McKinley J, Sambrook TD (2000) Use of human given cues by
domestic dogs (Canis familiaris) and horses (Equus caballus).
Anim Cogn 3:13–22
Messinger DS, Fogel A (1998) Give and take: the development of
conventional infant gestures. Merrill-Palmer Quart 44:566–590
´ , Soproni K (2006) A comparative analysis of animals’
Miklo´si A
understanding of the human pointing gesture. Anim Cogn 9:81–
93
´ , Polga´rdi R, Topa´l J, Csa´nyi V (1998) Use of experimenter
Miklo´si A
given cues in dogs. Anim Cogn 1:113–121
´ , Polga´rdi R, Topa´l J, Csa´nyi V (2000) Intentional
Miklo´si A
behaviour in dog–human communication: An experimental
analysis of showing behaviour in the dog. Anim Cogn 3:159–166
Mitchell RW, Thompson NS (1991) Projects, routines, and enticements in dog–human play. In: Bateson PPG, Klopfer PH (eds)
Perspectives in ethology, vol 9. Plenum Press, New York, pp
189–219
´ , Do´ka A, Csa´nyi V (2001) Co-operative
Naderi SZ, Miklo´si A
interactions between blind persons and their dogs. App Anim
Behav Sci 74:59–80
Pongra´cz-Rossi A, Ades C (2008) A dog at the keyboard: using arbitrary
signs to communicate requests. Anim Cogn 11(2):329–338
Reid PJ (2009) Adapting to the human world: dogs’ responsiveness to
our social cues. Behav Proc 80:325–333
Restoin A, Montagner H, Rodriguez D, Giradot JJ, Laurent D, Kontar
F, Ullmann V, Casagrande C, Talpain B (1985) Chronologie des
comportements de communication et profils de comportement
chez le jeune enfant. In: Tremblay RE, Provost MA, Strayer FF
(eds) Ethologie et de´veloppement de l’enfant. Stock/Laurence
Pernoud, Paris, pp 93–130
Seyfarth RM, Cheney DL (2003) Meaning and emotion in animal
vocalizations. Ann NY Acad Scie 1000:32–55
Shwe HI, Markmen EM (1997) Young children’s appreciation of the
mental impact of their communicative signals. Dev Psychol
33(4):630–636
Siegel S, Castellan NJ Jr (1988) Nonparametric statistics for the
behavioral sciences, 2nd edn. McGraw-Hill, New York
´ , Topa´l J, Csa´nyi V (2001) Comprehension of
Soproni K, Miklo´si A
human communicative signs in pet dogs (Canis familiaris).
J Comp Psychol 115:122–126
Tomasello M, Call J (2004) The role of humans in the cognitive
development of apes revisited. Anim Cogn 7:213–215
Tomasello M, Call J, Nagell K, Olguin R, Carpenter M (1994) The
learning and use of gestural signals by young chimpanzees: a
trans-generational study. Primates 35:137–154
´ , Csa´nyi V (2004) Dogs
Vira´nyi Z, Topa´l J, Ga´csi M, Miklo´si A
respond appropriately to cues of humans’ attentional focus.
Behav Process 66:161–172
123