Obey or Not Obey? Dogs (Canis familiaris) Behave Differently in
Response to Attentional States of Their Owners
Christine Schwab and Ludwig Huber
University of Vienna
Sixteen domestic dogs (Canis familiaris) were tested in a familiar context in a series of 1-min trials on
how well they obeyed after being told by their owner to lie down. Food was used in 1/3 of all trials, and
during the trial the owner engaged in 1 of 5 activities. The dogs behaved differently depending on the
owner’s attention to them. When being watched by the owner, the dogs stayed lying down most often
and/or for the longest time compared with when the owner read a book, watched TV, turned his or her
back on them, or left the room. These results indicate that the dogs sensed the attentional state of their
owners by judging observable behavioral cues such as eye contact and eye, head, and body orientation.
Keywords: dog– human relationship, Canis familiaris, attention, social cognition, communication
Communication by visual signals is a crucial feature in the
human– dog relationship. Working dogs such as hunting or shep-
herd dogs have probably been selected for their understanding of
human communicative signals, including visual ones. It is yet
undecided whether dogs’ understanding of human communicative
cues is a genetic trait or develops during ontogeny following close
contact with humans. Studies have shown that dogs understand
visual cues given by humans very well (Hare, Brown, Williamson,
& Tomasello, 2002; Hare, Call, & Tomasello, 1998; McKinley &
Sambrook, 2000; Miklo´si, Polga´rdi, Topa´l, & Csa´nyi, 1998; So-
proni, Miklo´si, Topa´l, & Csa´nyi, 2001, 2002) and are also able to
recognize and use visual attention of humans (Hare et al., 1998;
Hare & Tomasello, 1999; Miklo´si, Polga´rdi, Topa´l, & Csa´nyi,
2000; Vira´ny, Topa´l, Ga´csi, Miklo´si, & Csa´nyi, 2004). For com-
munication to fulfill the purpose of reliably transferring visual
signals from a sender to a receiver, attention plays a crucial role.
Either there is first attention contact between the individuals in-
volved followed by the signal to be transmitted or the signal can be
followed by looking toward the receiver to check whether he or
she was attending (Miklo´si et al., 2000). In either case, attention is
necessary to actively and usefully transfer signals and is mainly
described by two components.
The first component is the so-called shared attention mechanism
(SAM), a neurocognitive mechanism for identifying if one person
and another organism are both attending to the same object or
event (Baron-Cohen & Swettenham, 1996), which is expressed
either by following the gaze of others or by directing someone’s
attention to an object or event and is necessarily accompanied by
observable behavioral cues. Following the gaze of others cannot
occur without the model turning his or her eyes and often also the
head or body to the object or event of interest, and directing
someone’s attention to something is often realized through ges-
tures like showing or pointing; these are therefore main features of
communication (Baron-Cohen & Swettenham, 1996).
Eye contact is a second essential feature of attention. It has been
considered as an index of attention contact and is fundamental in
communicative situations (Go´mez, 1991). The general picture is
that attentional states of communicative partners are characterized
by overt signals (Sce´rif, Go´mez, & Byrne, 2004), perceived from
behavioral cues, like body posture, or from facial configuration,
like head orientation and gaze direction (Baron-Cohen, 1991).
The use of human bodily cues has been found in object-choice
tasks with a number of different species. Researchers have inves-
tigated an animal’s understanding of communicative cues by ac-
tively directing the animal’s attention to an object with gestures
like touching, pointing (with arm, hand, and finger), gazing (turn-
ing head and eyes), and glancing (eyes only; Baron-Cohen &
Swettenham, 1996). Horses (Equus caballus) show the ability to
use touching and pointing gestures (McKinley & Sambrook,
2000), as do domestic goats (Capra hircus; Kaminski, Riedel,
Call, & Tomasello, 2005) and one gray seal (Halichoerus grypus;
Shapiro, Janik, & Slater, 2003), but they all failed to understand
referential cues like gazing and glancing (if the latter was tested at
all). Capuchin monkeys (Cebus capella) show similar abilities in
comprehending human pointing gestures but fail to use head and
eye direction (Anderson, Sallaberry, & Barbier, 1995; Itakura &
Anderson, 1996). There is some ambiguity in studies of apes.
Chimpanzees (Pan troglodytes) seem to need some training to
understand pointing cues (Povinelli, Bierschwale, & Cech, 1999),
whereas the ability to use gazing (Call, Hare, & Tomasello, 1998;
Povinelli et al., 1999) or glancing as a referential cue in apes is
debated (Call, Agnetta, & Tomasello, 2000).
In contrast, the majority of dogs effectively use many different
visual cues given by humans in object-choice tasks (Hare et al.,
2002). They understand gestures of pointing very well (Hare et al.,
1998; McKinley & Sambrook, 2000; Miklo´si et al., 1998; Soproni
Christine Schwab and Ludwig Huber, Department for Behavior, Neu-
robiology and Cognition, University of Vienna, Vienna, Austria.
We thank all dogs and their owners for spending time and energy to
participate in this study and Viennese dog schools for allowing us to recruit
participants. Special thanks goes to Elisabeth Sowka for discussion and
suggestions and Bernhard Voelkl and Hermann Prossinger for statistical
Correspondence concerning this article should be addressed to Chris-
tine Schwab, who is now at the Konrad-Lorenz-Forschungsstelle,
Gruenau/Almtal, Fischerau 11, Gruenau A-4645, Austria. E-mail:
Journal of Comparative Psychology Copyright 2006 by the American Psychological Association
2006, Vol. 120, No. 3, 169 –175 0735-7036/06/$12.00 DOI: 10.1037/0735-7036.120.3.169
et al., 2002) and are also able to use gazing to find the hidden food
(Hare et al., 1998; Soproni et al., 2001), whereas glancing seems
to be the most difficult cue for dogs to understand (Hare et al.,
1998; McKinley & Sambrook, 2000; Miklo´si et al., 1998; Soproni
et al., 2001).
But still, direct eye contact not only seems to be a crucial feature
in communicative situations but also plays an important role in
predator–prey relations. When disturbed, hognose snakes (Heter-
odon platirhinos) often puff, hiss, coil, and strike, followed by
energetic writhing behavior that ends in a quiescent inverted
posture with the mouth open, the tongue extruded, and no overt
signs of breathing (Burghardt & Greene, 1988). Recovery time
from this death-feigning behavior has been investigated with the
presence of a human gazing at the snake and with the presence of
a human with his eyes averted from the snake. Results showed that
recovery times from the death feigning in the human-gazing con-
dition were significantly higher than in a control condition and
suggest that hognose snakes possess good visual acuity and can
use this ability in a rather sophisticated way to adaptively modify
their behavior (Burghardt, 1991). Furthermore, snakes might well
be more sensitive to visual cues from a predator’s eyes than just its
presence (Burghardt & Greene, 1988).
Similar results concerning gazing humans in a predator–prey
context were obtained from Burger and colleagues, who investi-
gated fleeing responses to humans in black iguanas (Ctenosaura
similis). Iguanas moved earlier, ran earlier, and ran further when
the face of an approaching human was clearly visible than when an
approaching human’s face was hidden by hair (Burger & Goch-
feld, 1993). Likewise the iguanas responded differently when a
human approached directly or tangentially with direct or averted
gaze: Their escape distances and distances the iguanas ran were
greatest when the human looked and walked toward the iguana and
were least when the human approached tangentially and looked
away (Burger, Gochfeld, & Murray, 1992). The same results in
escape behavior were found when the approaching human wore a
mask with large eyes than when he wore a mask with small eyes
(Burger, Gochfeld, & Murray, 1991). Burger et al. (1992) con-
cluded that the iguanas’ ability to perceive differences in predator
behavior could be attributed to many factors, including body
orientation and direction of gaze.
Furthermore, this differential sensitivity to external stimuli has
been shown in chickens (Gallus gallus) in shock situations. They
showed significantly differential tonic immobility (freezing) be-
havior with regard to the presence of conspecifics compared with
the presence or absence of artificial eyes (Rodd, Rosellini, Stock,
& Gallup, 1997). The latter are considered as a fear stimulus for
chickens in the context of predation (Gallup, Nash, & Ellison,
In our study we investigated to what extent dogs are able to
discriminate attentional states of humans expressed in different
body postures and modify their behavior accordingly. The exper-
imental design is similar to the one reported by Call and colleagues
(Call, Bra¨uer, Kaminski, & Tomasello, 2003), in which the dog
“competes” with a more or less attentive human over food. In that
study, the experimenter placed a piece of dry dog food on the floor
of an experimental room and the dogs were forbidden to take the
food. Then the experimenter either sat in a chair watching the dog
(control condition), sat in a chair facing the dog but with eyes
closed, sat in a chair facing the dog but engaged in a distracting
activity, or sat in a chair facing the wall with the back turned to
the dog. In comparison with the eyes-open condition, dogs
approached the food in a more direct or quicker way and
retrieved more food in the test conditions (Call et al., 2003).
These results provided first evidence that dogs are able to
discriminate at least some attentional states of humans and
Our study deviates from Call et al.’s (2003) study in some
aspects that we consider important for finding the true abilities of
dogs’ understanding of attentional states shown by their human
partner. Rather than testing the dogs’ reaction to a completely
unknown person, to whom the dog could not have established any
closer relationship, we tested the dog’s responses to its owner. We
thought that familiarity would enable the dog to use experiences on
how to interpret attention-dependent behavioral cues in its owner
in a quite familiar situation. In contrast, in the “punishment avoid-
ance” game used by Call et al. (2003), the dog could not know
which consequences the verbal command of a strange person
would have. This state of uncertainty was perhaps further strength-
ened by the fact that until the end of the first session and thereafter
in four of five trials, the experimenter did not react contingently to
the dog’s behavior either during the trial or after the trial was
Our decision to exploit the dog– owner relationship for the dog’s
reaction to the different attentional states of humans rests on the
assumption, shared with others (e.g., Topa´l, Miklo´si, & Csa´nyi,
1997), that dogs show their best cognitive performance and their
strongest sensitivity to human attention cues in a highly familiar
situation and with their most familiar human partner. As noted by
Ga´csi, Miklo´si, Varga, Topa´l, & Csa´nyi (2004, p. 152), “situations
used for testing are often ‘caricatures’ of natural situations” and
“frozen gestures are not the best candidates for tackling the exis-
tence of understanding of attention.” Therefore, we created as
much as possible natural testing conditions for the dogs. In contrast
to Call et al. (2003), who tested the dogs in an unfamiliar, quiet,
and sterile testing room, we tested them in their owner’s living
room. And in contrast to Call et al. (2003) seeking to tempt the
dogs by presenting all subjects the same piece of dry dog food,
which is unlikely to be very attractive for all the dogs, we used the
dogs’ favorite food.
Additionally, dogs were only given the command to lie down
without any explicit prohibition of eating the presented food. This
allowed a comparison between food and no-food conditions, with
the latter aimed at testing dogs without any additional stimulus
(food), a situation they often encounter in their everyday lives. To
test for the effects of behavioral cues reflecting the degree of
attention toward the dogs, we asked owners to engage in everyday
activities like reading or watching TV.
Dogs in our study were told to lie down on the floor while their
owner looked at them, read a book, watched TV, turned his or
her back on them, or left the room. We predicted that the dogs
would stand up (or stand up and eat the food in those trials in
which food was used) more often or more quickly the less the
owner was attentive to them. If the dog would discriminate
between all of the different attentional states of the owner, the
dog’s responses should be related to the above sequence of
situations in a stepwise function.
SCHWAB AND HUBER
Because the response of the dog in the experiment is likely dependent on
the dogs’ preexperimental obedience-training history, the preexperiences
made in similar situations, their personality (bold vs. shy), and possibly
many more factors, we selected dogs that showed an intermediate level of
obedience in a preexperimental test. To fit the purpose of the experiment,
they should have already learned to obey commands but must not be too
strictly trained, too lazy, or too anxious in the test situation. Therefore, we
selected the experimental subjects from a big sample depending on their
behavior in the first session that we used as a preexperimental test. Only
those dogs that fulfilled two criteria were used in the main experiment: (a)
They obeyed their owner’s command to lie down on the floor and kept
lying there for 1 min in more than half of the trials without food, and (b)
they moved within 1 min and took the food that was placed in front of them
when left alone in the room. In this first session there were four trials
conducted without food (Conditions 1, 2, 3, and 4) and two trials with food
(Conditions 1 and 5) with one of them representing Condition 5 when the
owner left the room and the dog was left alone with the food. Of 41 dogs
tested, 5 dogs were excluded because they failed to meet the first criterion
and 13 dogs because they failed to meet the second criterion. Three other
dogs could not be used any further because their owners did not follow the
instruction not to talk to or feed the dogs during or in between those
selection trials. Thus there were 20 dogs selected to participate in the main
experiment. After finishing the experiment, the first criterion was tested
again to make sure that all dogs were in fact well enough trained to obey
their owner’s commands. Four dogs did not meet the first criterion to obey
the command in more than half of the trials without food and were
subsequently excluded as experimental subjects.
The experimental subjects included in the video and statistical analyses
were 16 domestic dogs (Canis familiaris; 8 female, 8 male; mean age ⫽
4.25 years; range ⫽ 1.25– 8 years). Seven dogs were mongrels, and 9 dogs
were members of seven different breeds. Nine dogs lived together with
their owner since they were puppies, and the others spent most of their lives
with their owners. The age of the 14 female owners and the 1 male owner
(2 dogs had the same owner) ranged from 25 to 64 years. We ascertained
that the owners were the most familiar persons to their dogs, and both dogs
and owners were recruited from dog training schools in Vienna, Austria.
Owners were asked not to feed their dog 2 hr prior to testing.
The experiment was conducted in the living room of the owners’ house
or apartment, where the dog was usually not fed but allowed to take food.
Food that was used in the trials was always the favorite food of each dog.
So it differed between dogs but stayed the same throughout the sessions for
each dog. Each dog was tested by its owner without the experimenter being
present in the room during the experimental tests. We used a video camera
with a wide-angle lens to record the behavior of the dog and its owner
simultaneously. The camera was directed at the face of the dog, with the
food and the owner in line with it. Dogs were allowed to inspect the video
camera and tripod prior to every session to habituate them to the technical
equipment. The experiment was carried out from November 2002 to
The procedure for the experimental tests was as follows. Prior to every
session, which always consisted of six trials, the owner was instructed what
to do or not do using a written instruction. Owners were not allowed to
move, speak, give signs, or react in any other way to the dog’s action
during the trial. They were told to sit on the chair in a relaxed manner and
keep engaged in their activity (reading, watching TV, etc.) no matter what
the dog did. Before the trials started, the owner gave the dog the one and
only verbal command to lie down on the floor. As soon as the dog reliably
obeyed the command, the owner took his or her position and behaved in the
predetermined manner depending on the experimental condition. In those
trials in which food was used, the owner placed the favorite food item at
a distance of 1.5 m from the dog after it obeyed the command. Then the
owner took his or her predetermined position again at a distance of 1.5 m
from the dog. So the distance between dog and owner and dog and food
was always 1.5 m, and it was 3 m between owner and food. While the
owner placed the food on the floor, he or she looked at the dog and took
care that the dog was always attentive and looked in his or her direction.
As soon as the owner had taken his or her position, the trial started.
Between trials there was a short break of 2 min in which the owner was
encouraged to talk to the dog and get the dog up again. If in a food trial the
dog did not eat the food item, the owner removed it in the intertrial interval.
Conditions were chosen to reflect everyday situations, with which the
dog is assumed to be familiar, and with the assumption that attention in
humans is behaviorally expressed especially through three bodily cues: eye
gaze direction, head direction, and body posture. Thus differences in these
three cues should express different and (in the following order) decreasing
degrees of attention (see Table 1).
Condition 1 (look at dog). The owner sat straight on a chair and looked
at the dog without moving his or her body. Eyes, head, and body of the
owner were turned to the dog. The owner watched the dog during the whole
trial and tracked the dog with his or her head and gaze if the dog moved,
thereby signaling high attention. This condition resembles the eyes-open
condition of Call et al. (2003).
Condition 2 (read book). The owner sat on a chair with his or her head
and body turned toward the dog, but the eyes were focusing downward
because he or she was reading a book. The owner read the book during the
whole trial and did not look at the dog at all. This condition resembles the
distracted condition of Call et al. (2003), except that in their experiment the
experimenter was playing a handheld computer game.
Condition 3 (watch TV). The owner sat on a chair and his or her body
was turned toward the dog. Head and eyes were turned away from the dog
in a 90° angle toward a TV monitor. The owner watched TV during the
whole trial and did not look at the dog at all. There was no similar condition
in Call et al. (2003).
Condition 4 (back turned). The owner sat on a chair, and head and
body were facing away from the dog with the back turned to the dog and
the food. The owner kept reading a book during the whole trial and did not
look at the dog at all. This condition resembles the back-turned condition
of Call et al. (2003).
Condition 5 (leave room). The owner left the room immediately with-
out saying anything, closed the door behind him or her, and was out of
sight during the whole trial. This condition resembles the out-forbid
condition of Call et al. (2003).
The order of presentation of conditions was counterbalanced across
sessions. One condition per session was conducted twice, once with and
once without food. Dogs received five sessions to complete the experiment.
Four trials per session were carried out without food and 2 trials per session
were carried out with food, which were conducted on the third and sixth
position in every session. Each trial lasted for 1 min. Thus every dog
received 10 trials with and 20 trials without food. Intervals between
sessions were 2 to 3 weeks to reduce the probability of learning effects.
Summary of the Relationship Between Condition and Owner’s
Condition Presence of owner Body Head Eyes
Note. Crosses indicate that the owner was present in a certain condition
and that his or her body, head, or eyes were turned to the dog.
DOGS RESPOND TO HUMAN ATTENTIONAL STATES
One of us (Christine Schwab) coded the behavior of the dogs from the
videotapes in seconds for further analysis. Scoring of whether dogs stood
up and took the food did not demand more than one observer because it
could be determined without ambiguity. Analysis of latencies started as
soon as the owner had taken his or her predetermined position and referred
to lying-down behavior of the dogs because this was the only command
they got and were expected to obey. Data were not normally distributed,
and thus we used nonparametric tests (Friedman, Wilcoxon’s, or Mann–
Whitney U tests) to compare conditions. All statistical tests were
A comparison of food trials presented in the third trial with those
presented in the sixth trial in each session shows no significant difference
in the latencies of standing up (Wilcoxon’s test, n ⫽ 16, Z ⫽⫺1.758, p ⫽
.079). It implies that there was no effect of learning throughout the session,
thus we collapsed the data for further analysis. Furthermore, there were no
learning effects throughout the sessions concerning trials without food.
Comparisons of sessions with regard to different conditions did not reveal
any significant differences (Friedman test): Condition 1,
(3, N ⫽ 16) ⫽
6.077, p ⫽ .108; Condition 2,
(3, N ⫽ 16) ⫽ 0.259, p ⫽ .968; Condition
(3, N ⫽ 16) ⫽ 0.682, p ⫽ .877; Condition 4,
(3, N ⫽ 16) ⫽ 2.607,
p ⫽ .456; and Condition 5,
(3, N ⫽ 16) ⫽ 6.314, p ⫽ .097.
Figure 1 shows the proportion of trials in which the dogs obeyed
the command for the whole 60 s of the trials. In trials in which no
food was presented (Figure 1a), the proportion of trials in which
the dogs obeyed the command to lie down decreases from Condi-
tion 1 to 5. However, these differences were not significant across
conditions (Friedman test),
(4, N ⫽ 16) ⫽ 7.021, p ⫽ .135.
In trials with food (Figure 1b), there was a significant difference
across conditions (Friedman test),
(4, N ⫽ 16) ⫽ 19.652, p ⫽
.001, which was based on Condition 5 that differed significantly
from Condition 1 (n ⫽ 9, Z ⫽⫺2.373, p ⫽ .018) insofar as when
the owner left the room the dogs stood up more often than when
With regard to latencies, comparisons across conditions showed
significant differences in both conditions: trials without food,
(4, N ⫽ 16) ⫽ 9.832, p ⫽ .043; trials with food,
(4, N ⫽ 16) ⫽
15.529, p ⫽ .004. In trials without food (see Figure 2a), dogs stood
up significantly quicker when the owner left the room than when
it was being watched (see Table 2). But when the owner turned his
or her back on them, the dogs also stood up significantly quicker
than in Condition 1 when the owner looked at them and signaled
highest attention (Table 2). As in trials without food, in trials with
food (Figure 2b), the latency for disobeying the lying-down com-
mand was significantly shorter in Condition 5 than in Condition 1
In trials with food, the motivation of dogs to stand up (and then
take the food) was overall much higher than in trials without food.
They stood up (and then took the food, respectively) both more
often (with the exception of Condition 2) and more quickly in trials
in which food was used than when it was not (see Table 3).
It is worth noting that despite being strongly motivated by their
favorite food, the dogs obeyed the command to lie down in 41.25%
of all food trials, and they remained lying down the whole 60 s of
the trial. Therefore, to achieve a more sensitive comparison be-
tween conditions, we reanalyzed the data by considering only
those trials in which the dog disobeyed the command and ate the
food (see Figure 3; consumption trials). With this restriction, dogs
stood up significantly quicker when the owner read a book or
watched TV than when the owner looked at the dog (see Table 2).
There was no significant difference in performance between the
sexes in any of the conditions.
Taken together, the results corroborate those of Call et al. (2003)
by providing additional and to some extent complementary evi-
dence that dogs are able to discriminate attentional states in hu-
mans. Our results are comparable because in both studies dogs
Figure 1. Proportion of trials (a) without and (b) with food in which the dogs obeyed the command. The
figure shows percentage of trials in which they stayed lying down for the whole 60 s of the trial. The box
represents the interquartile range, which contains 50% of the values, and the bold lines indicate the median.
The error bars extend from the box to the highest and lowest values, excluding outliers, which are indicated
by black dots. *p ⬍ .05.
SCHWAB AND HUBER
showed they were sensitive to the attentional states of humans
when required to obey a command given by them. In both studies
the dogs disobeyed the human’s command more readily when the
human left the room than when he or she sat on a chair watching
them. In our study, this was the case in two types of trials, in those
with and those without food, because the dogs stood up more often
or more quickly when their owner left the room than when they
were watched. But also in trials without food of the back-turned
condition (Condition 4), which has no equivalent in Call et al.’s
(2003) study because they conducted their study with food, the
dogs stood up more quickly than when being watched. In addition,
with the temptation of their favorite food in front of them and in
cases when the dogs stood up and ate it, they did it more quickly
when the owner read a book or watched TV than when being
watched. This indicates that the dogs in our study responded
according to the different attentional states of their owners, cor-
roborating similar results of other dog studies, which we discuss in
Leaving the room seems to have a different quality than just
giving the least quantitative level of attention. Hare et al. (1998)
showed that being physically present is the first prerequisite for
communication. In their study dogs had to draw the attention of
a human to a location where food was hidden that the dogs
themselves could not reach. One of the dogs’ behaviors were
vocalizations they only produced if a human was present (Hare
et al., 1998). Discriminating front and back of humans was
tested by a human throwing a ball the dog should fetch, with the
restriction that the ball will only be thrown again if the dog
placed it inside the visual field of the human (Hare et al., 1998).
If the humans turned their back on the returning dogs, the latter
walked around and dropped the ball or, in very few trials, stayed
in front of the humans’ back side but then started to bark to get
the humans’ attention (Hare et al., 1998). This suggests that
dogs realize the asymmetry of humans’ front and back side and
understand the meaning of a human’s back as showing low attention
by the human.
Figure 2. Time (in seconds) until standing up in trials (a) without and (b) with food. The figure shows mean
duration of time dogs obeyed the command to lie down. The box represents the interquartile range, which
contains 50% of the values, and the bold lines indicate the median. The error bars extend from the box to the
highest and lowest values, excluding outliers, which are indicated by black dots. *p ⬍ .05.
Summary of Results Concerning Comparisons of Latencies Between Condition 1 and All Other
Without food With food Consumption
nZ pnZ pnZ p
210⫺0.868 .386 10 ⫺0.051 .959 9 ⴚ2.134 .033
310⫺0.357 .721 12 ⫺1.059 .289 7 ⴚ2.366 .018
4 11 ⴚ2.179 .029 12 ⫺1.138 .255 10 ⫺1.173 .241
5 13 ⴚ2.271 .023 16 ⴚ3.051 .002 10 ⫺1.172 .241
Note. Results are given for trials without and with food and for consumption trials in which the dogs stood up
before the 60 s of the trials elapsed. All given results are from Wilcoxon’s tests and are two-tailed. Bold results
indicate significant differences between the given condition and Condition 1.
DOGS RESPOND TO HUMAN ATTENTIONAL STATES
Significant results of facing versus back turning to returning
dogs were also found by Ga´csi et al. (2004), indicating that the
ability to discriminate these two bodily orientations of humans can
serve as a basis to recognize attention. Our results from the
back-turned condition provide additional support for the dog’s
ability to judge a human’s back as showing low attention by the
When the owner was reading a book or watching TV, thereby
turning the body toward the dog but only head or eyes away, we
assumed a stronger temptation for the dogs to disobey the com-
mand than when it was being watched. In fact, when they stood up
facing their favorite food, they showed shorter latencies in these
two conditions than in the looking condition (see Figure 3).
Giving, processing, understanding, and using cues are probably
context dependent. This, we believe, could explain some of the
differences found between the study by Call et al. (2003) and the
present one. In both studies the special feature of eye contact
resulted in dogs obeying the given command most often or for the
longest time. However, concerning the results of the food trials,
there are interesting deviations from Call et al.’s (2003) study. In
their study the dogs took more food pieces in all experimental
conditions than when being watched, but they took them more
quickly only in the eyes-closed condition, which has no equivalent
in the present study, but not when the experimenter played a
handheld computer or turned her back to the dog. In contrast, the
dogs in our study did not show significantly different numbers in
the frequencies of taking the food, except when the owner left the
room, but the dogs took the food more quickly in the read book and
watch TV conditions, with the former being equivalent to the
distracted condition in Call et al.’s (2003) study. These differences
might be explained by the experimental context. In the present
study every dog got its favorite food and had to lie down on the
floor. They were tested in the owners’ apartment, and during the
experiment the only human present was their owner and most
familiar person. All relationships between humans and dogs in this
study were determined by friendliness and familiarity, so it can be
assumed that the dogs were not afraid of their owners. They have
probably learned throughout their ontogeny that obeying a com-
mand is desired by their owners but that disobedience will not be
punished severely. In contrast, the dogs of Call et al.’s (2003)
study could not know the reaction of a stranger if they failed to
obey. Probably those dogs did not dare to take the food as often as
in the present study, especially in Condition 1 (or in the forbid or
eyes-open condition in Call et al.’s, 2003, study). So influences on
these results might have arisen from the unfamiliar versus familiar
experimental setting and the use of the same dry dog food versus
the favorite food of each dog contributing to the less frequent
taking of food in Call et al.’s (2003) study.
For both studies it also seems plausible that dogs used experi-
ences they have made throughout their lives to judge the atten-
tional state of the humans involved. If one takes into account that
for an understanding of attention, an individual “may learn many
additional things about the relation of their group mates’ visual
access to objects in the environment and its implications for their
(their group mates’) subsequent behavior” (Hare, Call, Agnetta, &
Tomasello, 2000, p. 784), also learning experiences are likely to
play an important role in making a connection between visual
access and behavior of others in different social contexts. Dogs
have probably experienced situations with humans turning away
their head or eyes or turning their back on the dogs, and the dogs
could have learned that in those situations the human hardly
noticed an undesirable behavior of the dog. In other words, they
could have learned that with this constellation of observable char-
Figure 3. Time (in seconds) until standing up in consumption trials. Only
those trials with food were considered in which the dogs stood up before
60 s of the trials elapsed. The figure shows mean duration of time dogs
obeyed the command to lie down. The box represents the interquartile
range, which contains 50% of the values, and the bold lines indicate the
median. The error bars extend from the box to the highest and lowest
values, excluding outliers, which are indicated by black dots. *p ⬍ .05.
Comparison of Conditions With and Without Food Concerning Proportion of Trials in Which the
Dogs Obeyed the Command (Frequency) and Time Until Standing Up (Latency)
nZ pnZ p
112⫺2.565 .01 12 ⫺2.353 .019
211⫺1.841 .066 12 ⫺2.589 .01
313⫺2.229 .026 14 ⫺2.542 .011
414⫺2.127 .033 14 ⫺2.103 .035
513⫺2.727 .006 16 ⫺2.999 .003
Note. All given results are from Wilcoxon tests and are two-tailed. Bold results indicate significant differences.
SCHWAB AND HUBER
acteristics there are less consequences if they do not obey a
command. Naturally, these experiences could be increased through
The ongoing discussion of attributing mental states to others
mainly deals with the question of whether organisms that seem to
show this ability do or do not refer to observable behavioral cues.
This conflicting debate does not seem to be decisive for judging
attentional states of others because if attention is understood as
turning the mind to something, this often involves turning the
body, head, and especially the eyes to something (Go´mez, 1991).
And because even the most complex attributional processes are
necessarily supported by objective cues (Povinelli, Nelson, &
Boysen, 1990), mental states like attention are characterized by
overt signals (Sce´rif et al., 2004) and are directly perceivable in
behavioral cues (Baron-Cohen, 1991). Dogs have been shown to
be sensitive to attentional states of their owners by using such
signals provided by absence, front and back side, head orientation,
and eye orientation of their owners and by adapting their behavior
Anderson, J. R., Sallaberry, P., & Barbier, H. (1995). Use of experimenter-
given cues during object-choice tasks by capuchin monkeys. Animal
Behaviour, 49, 201–208.
Baron-Cohen, S. (1991). Precursors to a theory of mind: Understanding
attention in others. In A. Whiten (Eds.), Natural theories of mind:
Evolution, development and simulation of everyday mindreading (pp.
233–251). Oxford, England: Blackwell.
Baron-Cohen, S., & Swettenham, J. (1996). The relationship between SAM
and ToMM: Two hypotheses. In P. Carruthers & P. K. Smith (Eds.),
Theories of theories of mind (pp. 158 –168). Cambridge, England: Cam-
bridge University Press.
Burger, J., & Gochfeld, M. (1993). The importance of the human face in
risk perception by black iguanas, Ctenosaura similis. Journal of Herpe-
tology, 27, 426 –430.
Burger, J., Gochfeld, M., & Murray, B. G. (1991). Role of a predator’s eye
size in risk perception by basking black iguanas, Ctenosaura similis.
Animal Behaviour, 42, 471– 476.
Burger, J., Gochfeld, M., & Murray, B. G. (1992). Risk discrimination of
eye contact and directness of approach in black iguanas (Ctenosaura
similis). Journal of Comparative Psychology, 106, 97–101.
Burghardt, G. M. (1991). Cognitive ethology and critical anthropomor-
phism: A snake with two heads and hog-nose snakes that play dead. In
C. A. Ristau (Eds.), Cognitive ethology: The minds of other animals (pp.
53–90). Hillsdale, NJ: Erlbaum.
Burghardt, G. M., & Greene, H. W. (1988). Predator simulation and
duration of death feigning in neonate hognose snakes. Animal Behav-
iour, 36, 1842–1843.
Call, J., Agnetta, B., & Tomasello, M. (2000). Cues that chimpanzees do
and do not use to find hidden objects. Animal Cognition, 3, 23–34.
Call, J., Bra¨uer, J., Kaminski, J., & Tomasello, M. (2003). Domestic dogs
(Canis familiaris) are sensitive to the attentional state of humans. Jour-
nal of Comparative Psychology, 117, 257–263.
Call, J., Hare, B. A., & Tomasello, M. (1998). Chimpanzee gaze following
in an object-choice task. Animal Cognition, 1, 89 –99.
Ga´csi, M., Miklo´si, A
., Varga, O., Topa´l,J.&Csa´nyi, V. (2004). Are
readers of our face readers of our minds? Dogs (Canis familiaris) show
situation-dependent recognition of human’s attention. Animal Cognition,
7, 144 –153.
Gallup, G. G., Nash, R. F., & Ellison, A. L. (1971). Tonic immobility as a
reaction to predation: Artificial eyes as a fear stimulus for chickens.
Psychonomic Science, 23, 79 – 80.
Go´mez, J.-C. (1991). Visual behaviour as a window for reading the mind
of others in primates. In A. Whiten (Eds.), Natural theories of mind:
Evolution, development and simulation of everyday mindreading (pp.
195–207). Oxford, England: Blackwell.
Hare, B., Brown, M., Williamson, C., & Tomasello, M. (2002). The
domestication of social cognition in dogs. Science, 298, 1634 –1636.
Hare, B., Call, J., Agnetta, B., & Tomasello, M. (2000). Chimpanzees
know what conspecifics do and do not see. Animal Behaviour, 59,
Hare, B., Call, J., & Tomasello, M. (1998). Communication of food
location between human and dog (Canis familiaris). Evolution of Com-
munication, 2, 137–159.
Hare, B., & Tomasello, M. (1999). Domestic dogs (Canis familiaris) use
human and conspecific social cues to locate hidden food. Journal of
Comparative Psychology, 113, 173–177.
Itakura, S., & Anderson, J. R. (1996). Learning to use experimenter-given
cues during an object-choice task by a capuchin monkey. Current
Psychological Cognition, 15, 103–112.
Kaminski, J., Riedel, J., Call, J., & Tomasello, M. (2005). Domestic goats,
Capra hircus, follow gaze direction and use social cues in an object
choice task. Animal Behaviour, 69, 11–18.
McKinley, J., & Sambrook, T. D. (2000). Use of human-given cues by
domestic dogs (Canis familiaris) and horses (Equus caballus). Animal
Cognition, 3, 13–22.
., Polga´rdi, R., Topa´l, J., & Csa´nyi, V. (1998). Use of
experimenter-given cues in dogs. Animal Cognition, 1, 113–121.
., Polga´rdi, R., Topa´l, J., & Csa´nyi, V. (2000). Intentional
behaviour in dog– human communication: An experimental analysis of
“showing” behaviour in the dog. Animal Cognition, 3, 159 –166.
Povinelli, D. J., Bierschwale, D. T., & Cech, C. G. (1999). Comprehension
of seeing as a referential act in young children, but not juvenile chim-
panzees. British Journal of Developmental Psychology, 17, 37– 60.
Povinelli, D. J., Nelson, K. E., & Boysen, S. T. (1990). Inferences about
guessing and knowing by chimpanzees. Journal of Comparative Psy-
chology, 104, 203–210.
Rodd, Z. A., Rosellini, R. A., Stock, H. S., & Gallup, G. G. (1997).
Learned helplessness in chickens (Gallus gallus): Evidence for atten-
tional bias. Learning and Motivation, 28, 43–55.
Sce´rif, G., Go´mez, J. -C., & Byrne, R. W. (2004). What do Diana monkeys
know about the focus of attention of a conspecific? Animal Behaviour,
68, 1239 –1247.
Shapiro, A. D., Janik, V. M., & Slater, P. J. B. (2003). A gray seal’s
(Halichoerus grypus) responses to experimenter-given pointing and di-
rectional cues. Journal of Comparative Psychology, 117, 355–362.
Soproni, K., Miklo´si, A
., Topa´l, J., & Csa´nyi, V. (2001). Comprehension
of human communicative signs in pet dogs (Canis familiaris). Journal of
Comparative Psychology, 115, 122–126.
Soproni, K., Miklo´si, A
., Topa´l, J., & Csa´nyi, V. (2002). Dogs’ (Canis
familiaris) responsiveness to human pointing gestures. Journal of Com-
parative Psychology, 116, 1– 8.
Topa´l, J., Miklo´si, A
. & Csa´nyi, V. (1997). Dog– human relationship
affects problem solving behavior in the dog. Anthrozoo¨s, 10, 214 –224.
Vira´ny, Z., Topa´l, J., Ga´csi, M., Miklo´si, A
. & Csa´nyi, V. (2004). Dogs
respond appropriately to cues of humans’ attentional focus. Behavioural
Processes, 66, 161–172.
Received July 28, 2005
Revision received December 30, 2005
Accepted January 31, 2006 䡲
DOGS RESPOND TO HUMAN ATTENTIONAL STATES