CHAPTER 1
Dogs in historical perspective,
and conceptual issues of the
study of their behaviour
1.1 Introduction
This book is about the biological study of dog behav-
iour, based on the programme summarized so clearly
by Tinbergen in 1963. He, Lorenz and others have
always pointed out ç8yy the main contribution of eth-
ology is the biological analysis of animal behaviour
based on observations in nature. Unfortunately,
however, only a handful of mainstream ethologists
have applied these concepts to dog behaviour. In
contrast to sticklebacks, honeybees or chimpanzees,
not to mention a few tens of other species, dogs
received relatively little attention from ethologists or
comparative psychologists. It seems that these crea-
tures ('man's best friends') have somehow become
outcasts from mainstream science, for reasons that
are not obviously clear but which may be guessed.
Dogs are often referred to as 'artificial animals',
probably because their history of being 'domesti-
cated'. Here the image is that of a 'savage' stealing
a wolf cub from its mother (e.g. Lorenz 1954), which
then 'became' dog after many years and generations
in the hands of humans. Today most researchers
disagree with this simplistic view of dog domesti-
cation (e.g. Herre and Rohrs 1990), and it is much
less clear on what grounds the evolution of such
'real' and 'artificial' animals can be differentiated.
The kind of goal-directed selective breeding
implied by the category of 'artificial animal' prob-
ably started much later than has been assumed.
Logically, an 'artificial animal' cannot have a nat-
ural environment, so in order to allow the dog into
the club of 'real' animals we would have to find a
natural environment for it (Chapter 3, p. 42).
The study of dogs did not fit well with the
increasing influence of behavioural ecology, which
was partially initiated by the call for a more func-
tional approach to behaviour by Tinbergen (1963).
Obviously, dogs are not the best candidates for
studying survival in nature, mainly because most
present-day dogs live with humans and have access
to vets, and we do our best to save our companions
from the challenges of nature. In this sense dogs
can be regarded as being special (but not necessar-
ily 'artificial').
More surprisingly, interest in the study of dogs
did not emerge with the cognitive revolution in
ethology. Griffin (1984), one of the initiators of this
movement, seems to have carefully avoided refer-
ence to dogs in most of his works on this subject.
We are introduced to miraculous behaviour of ants,
starlings or dolphins, which we look at with admir-
ation, but similar behaviour in dogs is often regarded
as suspicious. To some extent this attitude is under-
standable, as early workers were often tricked by
so called 'dog artists' who showed remarkable
skills for 'talking' or 'counting' (e.g. Pfungst 1912,
Grzimek 1940-41). (Figure 1.1) After it was found out
that such apparently clever behaviour could be
explained by the dog responding to minute bodily
cues produced either consciously or unconsciously
by the owner or trainer (the Clever Hans effect, see
Pfungst 1907 and Chapter 2.5, p. 37), dogs were ban-
ished from laboratories for being unreliable subjects.
However, it seems that dogs are showing signs of
making a real comeback. Ethologists, comparative
psychologists, and many others are now working
hard to find a place for dogs in the biological study
igure 1.1 (a) Stuppke, a counting dog artist, was observed by Bernhard Grzimek, a German zoologist. Stuppke barked the number shown
to him. The remarkable talent of the dog was based on recognition of a 'start' and a 'stop' signal given by his master, Mr Pilz. (b) No wonder
that Stuppke could also read numbers with his eyes covered (photos taken from Grzimek 1940-41). (c) Oskar Pfungst (1912) reported on Don,
the talking dog (photo from Candland 1993, Oxford University Press).
The Dog at the Convent Door.
Figure 1.2 The 'cultural transmission' of dog anecdotes. Menault
(1869) reports the story of a dog that, after observing beggars
ringing the bell at the door of the convent and receiving some soup,
went to the door and pulled the string. The ability to learn by
observation of humans has only recently been demonstrated
experimentally (Chapter 8, e.g. Kubinyi eta/. 2001 36; Box 8.6).
of behaviour. This is difficult, but the steep increase
in research papers over the last 10 years already
shows the fruit of this work. Thus there is every
chance that dog ethology will revive.
1.2 From behaviourism to
cognitive ethology
Early researchers, including Darwin (1872),
regarded the dog as a special animal that is com-
parable to humans. Many people shared this
anthropomorphic attitude and it is not surprising
that dogs ended up at the top of the ladder repre-
senting intelligence and emotional behaviour in
animals (Romanes 1882a, b) (Figure 1.2). It did
not take long for the situation to change, and
dogs could not avoid their fate when under the
increasing influence of behaviourism they were
then treated as a sort of stimulus-response
automaton. The interest in wolves and social
behaviour in general has helped dogs regain a
foothold in the behavioural sciences, and this has
led to an ethologically oriented understanding of
dog behaviour. The history of the study of dogs
reflects the changes in our views of animals, and
although much time has passed and a lot of
knowledge has been gained, the basic questions
of present-day research are more or less the same
as they were 100 years ago.
1.2.1 Dog heroes visit the laboratory
Dogs have long been the favourite heroes of
animal stories. Sharing our daily life with these
animals has offered endless opportunities to
observe or witness the varieties of dog-human
interactions. One famous collector of such stories
was George Romanes (1982a). His descriptions of
dogs provided evidence for often very intelligent
behaviour which prompted him to argue that
such performances should be explained by
human-like thinking mechanisms (Candland
1993).
Interestingly, Lloyd Morgan (1903), who was a
strong critic of the methods used by Romanes, did
not refrain from telling such stories when he
wanted to illustrate a particular behavioural phe-
nomenon. At one point he describes how his fox
terrier Tony grappled with the problem of how to
carry a stick with unequal weights at its ends. After
describing the dog's behaviour Morgan concludes
that he has seen little evidence for assuming that
the dog 'understood the problem'. Instead, during
repeated attempts to carry the stick the dog learned
the solution by trial and error. Thus 'intelligent'
behaviour on the dog's part could often be based
on relative simple learning processes. For Morgan,
stories provided opportunities for formulating
hypotheses and did not serve as explanations for
mental abilities. Nevertheless he did not deny that
dogs could have a mental representation for an
object, such as a bone.
Thorndike (1911) was among the first to develop
a method to objectively measure learning in ani-
mals. He put hungry cats and dogs into a box which
could be opened from inside by manipulating a
simple latch. Observing the animals repeatedly in
this situation, he found that it took them less and
less time to get out. In agreement with Morgan, he
also thought that the final 'intelligent' behavioural
solution was the result of a step-by-step process of
'trial and error' learning. Thus the systematic
observations of both Morgan and Thorndike
seemed to contradict the conclusions of Romanes,
who argued that, for example, cats and dogs have
someideaaboutthepropertiesof locks.Interestingly,
Thorndike noted a difference between dogs and
cats, because, despite being starved for some time,
dogs were much inferior in escaping. From his
descriptions it seems that, in comparison with the
cats, dogs were less inclined to get out, and they
were also very cautious in interacting with the
latch, which probably indicates a different social
relation between people and these dogs. Thus it is
less surprising that in the textbooks the fame of
representing Thorndike's concept of trial-and-error
learning was left to the cats. From further experi-
ments Thorndike did not find support for the
long-held view that dogs learn by imitation (see
Chapter 8.6, p. 191) because animals did not escape
any earlier from the box if they were shown how to
open the lock.
In 1904 Pavlov received the Nobel Prize for
Medicine for the physiological study of the digest-
ive system, for which dogs had served as subjects.
By this time he had noted that not just the presence
of food in the mouth but also other external stimuli
(the sound of the food put in the bowl or the
approaching experimenter providing the food)
have the potential to elicit salivation. For many
years after that dogs remained one of the most pre-
ferred subjects in the research that led to the devel-
opment of the conditioned reflex principle (Pavlov
1927), which was extended by Pavlov's pupils.
Pavlov was not only a good experimenter, however,
but also a good observer. Thus he noted early on
that there are marked individual differences among
the dogs, which could be also observed in their
response to the training (Teplov 1964). Dogs were
categorized as belonging to one of the classic tem-
perament types described by Hippocrates (san-
guine, choleric, phlegmatic, melancholic) (see also
Box 10.1). Even at that time Pavlov pointed out that
observed behavioural traits are the outcome of
complex processes having both genetic and envir-
onmental components, and he was probably the
first to suggest separating these two effects by rais-
ing dogs in different environments before subject-
ing them to training. The generality of Pavlov's
work on the conditioning reflexes provided the
basis for comparative work on dogs and humans.
Based on this experimental approach, dogs can be
regarded as the first animal models of human per-
sonality (Chapter 10, p. 221). This makes it less sur-
prising that in contrast to some other laboratories
Pavlov's researchers respected the individuality of
the animal. Most dogs were given names, and the
observation of their spontaneous behaviour in the
laboratory or outside was used as additional infor-
mation for understanding their reaction in the
training situations. Importantly, in contrast to
recent research on personalities, Pavlov and his
colleagues based their investigations on single
dogs and then generalized the results to other indi-
viduals belonging to the same personality type.
1.2.2 Dogs in the comparative
psychology laboratory
One cannot avoid being emotionally touched on
reading many of the papers published on dog
behaviour in laboratories working on a Pavlovian
model of learning. Professional scientists, often
having a good 'personal' relationship with these
dogs, often do not seem to realize what they are
doing. There is no way that anyone today could or
would do many experiments like these. The purpose
of reviewing these experiments is to show how the
lack of ethological thought can misdirect scientific
efforts.
A subjective survey of the literature shows that
by the 1920s rats and pigeons had become the main
subjects of research. Thus we might wonder why
some research programmes seemed to prefer dogs.
Having adopted a clearly anthropocentric pro-
gramme in looking for appropriate animal models
of human behaviour, we could reason that for some
features of human behaviour dogs seemed to offer
a more appropriate model. By doing this, these
researchers have implicitly acknowledged that
dogs are more similar to humans than are other
species. Indeed, in discussing dog behaviour they
often relied on comparison with humans (children),
assuming similar underlying mental mechanisms
(e.g. Solomon et al. 1968, see Box 1.3). Interestingly,
this argument was not extended to subjective
states. Thus the dogs' suffering in many of these
experimental procedures was never really a con-
cern.
Another important aspect of these experiments
was that the experimental context had very little, if
any, relevance to the natural behaviour of the dog,
and there was very little correspondence between
the experimentally manipulated variables and the
variables that may relate to a natural situation. The
presence of humans was also confusing for the
dog, because the good/positive social relationship
before and after the experiment was contradicted
by the role of humans in the training trials.
One aim of this research was to provide a behav-
ioural model for neurosis, or traumatic experience
(Lichtenstein 1950, Solomon and Wynne 1953). For
example, dogs were shut into an experimental
chamber and exposed to electric shock ('helpless-
ness': Seligman et al. 1965). After this experience
they were tested in a task in which they were given
the possibility of avoiding similar shocks by escap-
ing from the dangerous place. Many experiments
found that after such an experience the dogs did
not learn. They showed low responsiveness and
seemed 'to give up and passively accept' the shock
(Seligman et al. 1965). We might question the etho-
logical basis of this behaviour. Is there a natural
situation when dogs experience such pain? The
most likely, if not only, situation is when a domin-
ant conspecific inflicts a physically dangerous
attack finished off by a persistent bite. In such a
case the attacked animal's only chance is to show
all possible signs of submission with as little move-
ment as possible ('freezing'). Some of the dogs
might have associated painful experience with
their interactions with humans, which certainly
contributed to the dog's 'neurosis' apart from the
effect of their lack of control over the situation
(Seligman et al. 1965).
A better aspect of this period is that many early
studies provided a detailed description of the dogs'
behaviour, and it became obvious that their reac-
tions to the treatments were very variable. This
suggests that despite being 'laboratory dogs' ani-
mals differed in their previous experience, includ-
ing their relationship with the humans inside or
even outside the laboratory. A further important
lesson from these studies is that training methods
using painful punishments can have unforeseeable
(and mostly negative) consequences on the behav-
iour of dogs, either because of their genetic endow-
ment or their earlier experience with humans
(socialization).
These traditions of comparative psychology were
left behind when more ethologically inspired ques-
tions dominated laboratory research (Figure 1.3.).
Figure 1.3 Dogs under study, (a) A dog in a Pavlovian stand as
illustrated in Woodbury (1943). The dog is trained to recognize
differences in acoustic sound patterns, (b) An illustration from
Jenkins eta/. (1978) showing 'Dog 7'which after being conditioned
to the light stimulus (at the front) signalling food, displays a range
of social behaviours towards the light stimulus and the food tray
(behind the dog, not shown on the illustration).
In 1978 Jenkins and co-workers contrasted the
Pavlovian stimulus substitution theory (Pavlov
1934) with the ethological analysis of the dog
'begging' for food (Lorenz 1969). Pavlov's theory
assumed that the (conditioned) stimulus (e.g. light
or bell) signalling the food will actually replace the
original (unconditioned) stimulus (e.g. food); that
is, when it sees the light come on the dog displays
preparatory acts which reflect consummatory
actions towards the conditioned stimulus (e.g. lick-
ing, snapping at the light). In contrast Lorenz
argued that the conditioned stimulus acts as a
releaser for appetitive behaviours. Thus the dog
searches for the food or displays 'begging', as when
pups solicit food from older conspecifics. Jenkins
et al. (1978) trained dogs to approach a lamp which
signalled the presence of a food reward. In the
course of the training dogs showed very variable
behaviour, but nevertheless many social behaviour
patterns emerged, such as play signals, tail wag-
ging, barking, nosing. Thus we could argue that
dogs interpreted the experimental situation in a
social context with which they were familiar. For
them the conditional stimulus (light) was not just
signalling the arrival of food but it was also a social
stimulus. In this more natural context, 'request' for
food (from humans) is usually preceded by some
sort of signalling (e.g. tail wagging, barking) and
behaviour actions (e.g. nosing, pawing). These
motor patterns are derived from the species-specific
behavioural repertoire of the dog, which is later
modified during the period of socialization. The
social experience and habitual behaviour of the
individual dogs markedly influences the behav-
iour during these observations. The important con-
clusion is that 'one must examine how dogs react to
natural signals of food outside the laboratory set-
ting' (Jenkins et al. 1978)—one of the first signs of a
need for collaboration between comparative psy-
chologists and ethologists. Such an approach opens
up a new way of combining methods that rely on
controlled laboratory settings with those that
emphasize observations on natural behaviour,
including knowledge of the individual's previous
experience.
1.2.3 Naturalistic experiments
Especially during the first half of the last century,
dogs were popular subjects for investigators who
rejected arbitrary laboratory observations. This
work, which culminated just before the Second
World War, was mostly carried out in Germany and
the Netherlands. These researchers continued the
tradition of Morgan and others recognizing the
importance of controlled (more or less) experi-
ments, but they wanted to rely, to a greater extent,
on the natural behaviour of dogs. Many of them
were pupils or followers of Kohler (1917/1925),
who emphasized the role of 'insight' in solving new
problems, and Uexkull (1909), who stressed the
importance of recognizing the features of the nat-
ural environmental (Umwelt) of the animal under
study. Importantly, both Kohler and Uexkull had a
marked influence on early ethological thought
(Lorenz 1981), thus to some extent Buytendijk and
Fischel (1936), Sarris (1937), Fischel (1941), Grzimek
(1941) and others can be regarded as forerunners of
present-day dog ethologists. Although most of their
experiments were performed in the laboratory or in
Figure 1.3 Dogs under study, (a) A dog in a Pavlovian stand as
illustrated in Woodbury (1943). The dog is trained to recognize
differences in acoustic sound patterns, (b) An illustration from
Jenkins eta/. (1978) showing 'Dog 7'which after being conditioned
to the light stimulus (at the front) signalling food, displays a range
of social behaviours towards the light stimulus and the food tray
(behind the dog, not shown on the illustration).
In 1978 Jenkins and co-workers contrasted the
Pavlovian stimulus substitution theory (Pavlov
1934) with the ethological analysis of the dog
'begging' for food (Lorenz 1969). Pavlov's theory
assumed that the (conditioned) stimulus (e.g. light
or bell) signalling the food will actually replace the
original (unconditioned) stimulus (e.g. food); that
is, when it sees the light come on the dog displays
preparatory acts which reflect consummatory
actions towards the conditioned stimulus (e.g. lick-
ing, snapping at the light). In contrast Lorenz
argued that the conditioned stimulus acts as a
releaser for appetitive behaviours. Thus the dog
searches for the food or displays 'begging', as when
pups solicit food from older conspecifics. Jenkins
et al. (1978) trained dogs to approach a lamp which
signalled the presence of a food reward. In the
course of the training dogs showed very variable
behaviour, but nevertheless many social behaviour
patterns emerged, such as play signals, tail wag-
ging, barking, nosing. Thus we could argue that
dogs interpreted the experimental situation in a
social context with which they were familiar. For
them the conditional stimulus (light) was not just
signalling the arrival of food but it was also a social
stimulus. In this more natural context, 'request' for
food (from humans) is usually preceded by some
sort of signalling (e.g. tail wagging, barking) and
behaviour actions (e.g. nosing, pawing). These
motor patterns are derived from the species-specific
behavioural repertoire of the dog, which is later
modified during the period of socialization. The
social experience and habitual behaviour of the
individual dogs markedly influences the behav-
iour during these observations. The important con-
clusion is that 'one must examine how dogs react to
natural signals of food outside the laboratory set-
ting' (Jenkins et al. 1978)—one of the first signs of a
need for collaboration between comparative psy-
chologists and ethologists. Such an approach opens
up a new way of combining methods that rely on
controlled laboratory settings with those that
emphasize observations on natural behaviour,
including knowledge of the individual's previous
experience.
1.2.3 Naturalistic experiments
Especially during the first half of the last century,
dogs were popular subjects for investigators who
rejected arbitrary laboratory observations. This
work, which culminated just before the Second
World War, was mostly carried out in Germany and
the Netherlands. These researchers continued the
tradition of Morgan and others recognizing the
importance of controlled (more or less) experi-
ments, but they wanted to rely, to a greater extent,
on the natural behaviour of dogs. Many of them
were pupils or followers of Kohler (1917/1925),
who emphasized the role of 'insight' in solving new
problems, and Uexkull (1909), who stressed the
importance of recognizing the features of the nat-
ural environmental (Umwelt) of the animal under
study. Importantly, both Kohler and Uexkull had a
marked influence on early ethological thought
(Lorenz 1981), thus to some extent Buytendijk and
Fischel (1936), Sarris (1937), Fischel (1941), Grzimek
(1941) and others can be regarded as forerunners of
present-day dog ethologists. Although most of their
experiments were performed in the laboratory or in
pg 16 a 21
pg 16 a 21