Animals' emotions: Studies in sheep using appraisal theories

Abstract

Animal welfare concerns stem from recognition of the fact that animals can experience emotions such as pain or joy. Nevertheless, discussion of animal emotions is often considered anthropomorphic, and there is a clear need to use explanatory frameworks to under-stand animals' emotions. We borrowed appraisal theories developed in cognitive psychology to study sheep emotions. Emotions are viewed as the result of how an individual evaluates a triggering situation, following a sequence of checks, including the relevance of the situation (its suddenness, familiarity, predictability, and intrinsic pleasantness), its implications for the individual (including consistency with the individual's expectations), the potential for control, and both internal and external standards. We assumed that if the outcome of checks has an impact on the animal's emotional responses, then animals do not only show emotional responses but also feel emotions. We showed that sheep use similar checks to those used by humans to evaluate their environment, ie suddenness, familiarity, predictability, consistency with expectations, and control. Furthermore, this evaluation affects their emotional responses (behavioural responses, such as startle, ear postures, and cardiac activity). It is concluded that sheep are able to experience emotions such as fear, anger, rage, despair, boredom, disgust and happiness because they use the same checks involved in such emotions as humans. For instance, despair is triggered by situations which are evaluated as sudden, unfamiliar, unpredictable, discrepant from expectations, and uncontrollable, whereas boredom results from an overly predictable environment, and all these checks have been found to affect emotional responses in sheep. These results have implications for animal welfare: although a completely invariable and totally predictable environment should be avoided to prevent boredom, sudden events should probably be minimised, the animals should be offered the possibility to control their environment, and care should be taken to ensure a degree of predictability concerning the various events.

Full text

347
© 2009 Universities Federation for Animal Welfare
The Old School, Brewhouse Hill, Wheathampstead,
Hertfordshire AL4 8AN, UK
Animal Welfare 2009, 18: 347-354
ISSN 0962-7286
Animals’ emotions: studies in sheep using appraisal theories
I Veissier*, A Boissy, L Désiré and L Greiveldinger
INRA, UR1213 Herbivores, F-63122 Saint-Genès-Champanelle, France
* Contact for correspondence and requests for reprints isabelle.veissier@clermont.inra.fr
Abstract
Animal welfare concerns stem from recognition of the fact that animals can experience emotions such as pain or joy. Nevertheless,
discussion of animal emotions is often considered anthropomorphic, and there is a clear need to use explanatory frameworks to under-
stand animals’ emotions. We borrowed appraisal theories developed in cognitive psychology to study sheep emotions. Emotions are
viewed as the result of how an individual evaluates a triggering situation, following a sequence of checks, including the relevance of the
situation (its suddenness, familiarity, predictability, and intrinsic pleasantness), its implications for the individual (including consistency with
the individual’s expectations), the potential for control, and both internal and external standards. We assumed that if the outcome of
checks has an impact on the animal’s emotional responses, then animals do not only show emotional responses but also feel emotions.
We showed that sheep use similar checks to those used by humans to evaluate their environment, ie suddenness, familiarity,
predictability, consistency with expectations, and control. Furthermore, this evaluation affects their emotional responses (behavioural
responses, such as startle, ear postures, and cardiac activity). It is concluded that sheep are able to experience emotions such as fear,
anger, rage, despair, boredom, disgust and happiness because they use the same checks involved in such emotions as humans. For
instance, despair is triggered by situations which are evaluated as sudden, unfamiliar, unpredictable, discrepant from expectations, and
uncontrollable, whereas boredom results from an overly predictable environment, and all these checks have been found to affect
emotional responses in sheep. These results have implications for animal welfare: although a completely invariable and totally predictable
environment should be avoided to prevent boredom, sudden events should probably be minimised, the animals should be offered the
possibility to control their environment, and care should be taken to ensure a degree of predictability concerning the various events.
Keywords:animal welfare, appraisal theories, ear postures, emotion, heart rate, sheep
Introduction
Animal welfare concerns stem from recognition of the fact
that animals are sentient beings able to experience emotions
such as fear, pain, joy, contentment, etc. The reference to
animal sentience is explicit in the Amsterdam treaty of the
European Union (European Union 1997), which stipulates
that measures shall be applied “to ensure improved protec-
tion and respect for the welfare of animals as sentient
beings”. The European Convention for the Protection of
farm animals also states that “No animal shall be provided
with food or liquid in a manner (…) which may cause
unnecessary suffering or injury” (Council of Europe 1976).
Furthermore, animal emotions form the core of many scien-
tific definitions of animal welfare (Duncan 1993; Dawkins
2006; Veissier & Boissy 2007). Hence, identifying the range
of emotions animals are capable of experiencing appears
crucial to the design of measures ensuring their welfare.
The word ‘emotion’ comes from the Latin ‘emovere’, to
remove or shake, and ‘movere’, to move. An emotion can
be roughly defined as something that moves one’s body
and mind. Emotions are more often defined by their
components: the internal-psychological component (what
one feels), the neurophysiological component (how the
body responds, eg by stress responses), and the behav-
ioural component (what one shows to others, eg facial
expressions and movements). Emotions differ from sensa-
tions, which are only physical consequences (eg heat), and
from feelings, which designate only internal states with no
reference to external reactions. It is generally agreed that
animals have emotional responses (neurophysiological
responses, such as increased heart rate and changes in
heart-rate variability reflecting the balance between the
sympathetic and parasympathetic branches of the
autonomic nervous system [Després et al 2003] or release
of corticosteroids in blood; behavioural responses such as
startle or attempts to escape a situation) but the issue of
whether animals feel emotions (psychological component)
remains controversial (discussed by Duncan 2006).
Frameworks to study emotions
Emotions in animals were first described by Darwin in The
Expression of Emotions in Man and Animals (1872), where
he described emotions as stereotyped facial expressions and
Universities Federation for Animal Welfare Science in the Service of Animal Welfare

348 Veissier et al
bodily postures in specific contexts. Darwin observed simi-
larities between human and non-human animal expressions,
in line with the theory of continuity between species. His
work on animals’ emotions was criticised for approximately
100 years and labelled as anthropomorphic (Grassé 1977;
Jacobs 1998). Animals’ emotions were considered as falling
outside of the realm of science and therefore not to be studied
scientifically. Darwin’s book was re-edited in 2001, and now
seems to be the time to reconsider animals’ emotions as a
scientific topic. However, discussion of animals’ emotions is
still often considered anthropomorphic.
On the one hand, anthropomorphism carries the risk of
misinterpreting animals’ responses. For instance, we may
think that a pig wallowing in mud is happy (hence the
expression ‘happy as a pig in mud’) whereas the pig is
actually suffering from over-heating. This makes it
dangerous ground to naïvely and anthropomorphically
project ourselves directly onto animals. On the other hand,
as argued by de Waal (1999), the danger lies not only in
anthropomorphism but also in ‘anthropodenial’. Animals
and humans do share common characteristics, and anthro-
pomorphism may help us investigate the human-like char-
acteristics of animals as well as the animal-like
characteristics of humans. This led Wynne (2004) to
conclude that “Progress will surely be most rapid when we
adopt explanatory frameworks that are concrete and unam-
biguous”. Appraisal theories developed in cognitive
psychology to understand human emotions appear to
provide a strong candidate framework (Désiré et al 2002).
Human emotions are viewed as the result of how an indi-
vidual evaluates a triggering situation, beginning with an
evaluation of the situation per se and followed by the
possible responses to that situation (Arnold 1945; Lazarus
et al 1970; Scherer 2001). Several authors have listed items
on which the evaluation is based. Scherer (2001), for
instance, asserted that the evaluation is operated following a
sequence of checks grouped into four classes:
• The relevance of the situation, including the check for
novelty (suddenness, familiarity and predictability of a
situation), the check for intrinsic pleasantness, and the
check for the relevance of the situation for the individual’s
own goals;
• The implications of the situation for the individual,
including the check of the probability of the consequences
expected from that situation, and the check of consistency
with the individual’s expectations;
• The coping potential, including the check for coping possi-
bilities (offered by the environment, ie controllability) and
abilities (within the individual);
• The normative significance, including the check for
internal standards (if one fails to respond correctly, will that
affect his/her self-esteem?) and the check for external
standards (are there responses which are preferable or
compulsory according to the social group one belongs to?)
These checks do not necessarily need high cognitive
processes. Some of them are fairly automatic and subcon-
scious, especially within the first check of relevance, while
others are more complex (Kappas 2006). Each check
operates at several different levels according to the intensity
of the cognitive processes required: a sensorimotor level
that involves automatic processes, a schematic level
requiring the individual to memorise emotional experiences
and involving conditioned responses, and a conceptual level
that is voluntarily and consciously activated (eg comparison
between the real-world situation and conscious plans or
self-representation) (Leventhal & Scherer 1987). Within the
check of relevance, the check of suddenness seems to
require only sensorimotor processes while familiarity and
predictability require schematic processes. Assessing a
situation in terms of expectations and controllability
requires schematic processes, while the check of normative
significance is likely to require conceptual processes.
The outcome of the checks is responsible for the psycholog-
ical component of an emotion, which in turn affects physi-
ological and behavioural responses. Typical emotions such
as fear, anger or happiness are linked to the outcome of the
evaluation. Table 1 reports the links between several
different human emotions and the evaluations that triggered
them. For instance, fear is experienced by an individual
when he/she is exposed to an unpleasant event which is
sudden, unfamiliar, could not be predicted, and not consis-
tent with his/her expectations (ie the event does not match
what he/she was prepared for); rage (ie ‘hot anger’ as
opposed to ‘cold anger’) is experienced in similar situa-
tions, except that the individual’s evaluation is that he/she
can control this situation; happiness is triggered by an event
evaluated as slightly sudden, quite predictable, very
pleasant and consistent with expectations; and so on. The
internal component, ie the feeling, may well be essential to
emotions, but it does not imply that the individual is
conscious of his/her own emotions. It can be assumed that
there is a gradient of emotional responses, from the mere
expression of rather automatic responses to the experience
of emotional feelings and the consciousness of self-
emotional experiences, depending on the level of the
cognitive processes used to appraise the situation: simple
checks are likely to lead to automatic responses (eg startle
responses when confronted with a very sudden event),
checks requiring schematic processes are likely to lead to
proper emotional experiences (ie felt by the individual, as
reported verbally by people), with checks requiring concep-
tual processes leading to more conscious emotions.
Other appraisal theories either place emphasis on the nature
of the checks (Smith & Kirby 2001) or on the levels of
cognitive processes (Philippot et al 2004), or present
slightly different descriptions of checks (Frijda 1986;
Roseman 1991). The results presented in this paper will be
discussed in the context of the appraisal theory developed
by Scherer’s team (2001), as this theory appears to
encompass the other appraisal theories, and similar interpre-
tations would thus be obtained with other frameworks.
There are a number of literature reports showing that animals
respond to the suddenness of a situation, its novelty,
predictability, correspondence to expectations, or controlla-
bility (reviewed by Désiré et al 2002). Mammals appear able
© 2009 Universities Federation for Animal Welfare

Animals’ emotions: studies in sheep using appraisal 349
to use all these checks. However, whether they distinguish
these various checks and how the outcomes of checks affect
their emotional responses remains to be clarified. For
instance, it was demonstrated back in the 1970s that the
predictability and controllability of aversive events can
reduce their negative impacts in the long term (ie reduced rate
of gastric ulcers in rodents following exposure to electric
shocks) (Weiss 1972). However, whether this effect is due to
different emotional responses immediately after the aversive
event remains unclear. A comprehensive study based on
similar paradigms applied to the same species would greatly
help our understanding of how exactly the animal evaluates
its environment and responds to it emotionally.
Our research group has transposed the framework
proposed by appraisal theories in order to study animals’
emotions (Désiré et al 2004, 2006; Greiveldinger et al
2007, 2009). We postulated that if the outcome of checks
that require some cognitive processes has an impact on
emotional responses, then the animal not only has
emotional responses (which in that case could not be seen
as mere automatic responses) but also feels emotions. We
used sheep as models because they have moderate
cognitive abilities (Nicol 1996), and we assumed that if
such animals are able to use similar checks to humans,
then most of the other animals reared on farms or used by
humans for other purposes would also use these same
checks. We essentially investigated whether sheep could
feel emotions. We exposed sheep to situations designed so
as to make one check more prominent. This check could
require only automatic processes (here, the check of
suddenness) or schematic processes (familiarity,
predictability, consistency with expectations, controlla-
bility), in which case its use by the animals would support
evidence of the existence of emotional feelings. We also
initiated some work on the influence of social norms, the
way the animals perceive dominance relationships with
their social partners being assimilated to norms. This kind
of check is supposed to require a conceptual information
process and could thus lead to conscious emotions.
Nevertheless, the discussion presented in this paper is
focused on the existence of emotional feelings in sheep,
and does not extend to conscious feelings.
We accustomed sheep to being separated from their peers in
an experimental chamber where they received highly
palatable foods. The sheep were then exposed to various
events while they were eating. For instance, sheep exposed to
a very sudden event were compared to sheep for which the
same event was introduced more gradually; sheep that could
control a situation were compared to sheep that could not;
and so on. We monitored physiological responses (heart rate
and heart-rate variability) and behavioural responses (ear
postures, movements). The relevance of each check for sheep
will be presented in the next section, where we report the
results from some of the experiments we have carried out.
The relevance of checks for sheep emotions
Each section below starts by describing the various checks
used by humans according to the framework proposed by
Scherer or his collaborators (Scherer 2001; Sander et al
2005). We then explain how we tested the relevance of
checks to sheep before going on to analyse how such checks
affect sheep’s emotional responses.
Novelty of a situation
When facing an event, the first thing evaluated by the indi-
vidual is the novelty of that event. Novelty comprises three
features: the suddenness of the event (ie does the event
occur abruptly?), its familiarity (ie does the event match
with something the individual already knows), and its
predictability (ie what was the probability of this event
occurring, and/or does it occur regularly?)
We exposed sheep to events that were presented abruptly or
gradually, were familiar vs unfamiliar, and were predictable
vs unpredictable. While the sheep were eating, we manoeu-
vred a piece of textile behind the trough. This object fell
rapidly (88 cm s–1) in front of half the animals and fell
slowly (6 cm s–1) in front of the other half (Désiré et al 2004,
2006). In addition, the object was familiar for some animals
and new for the others. The sheep showed startle responses
and asymmetric ear postures when the object fell rapidly
and they looked more often at the object with their ears
oriented forward when the object was unfamiliar. In
addition, different physiological responses to suddenness
and unfamiliarity were observed: when the object fell
Animal Welfare 2009, 18: 347-354
Table 1 Humans’ emotions in relation to the outcome of their evaluation of a triggering situation (from Sander et al 2005).
Emotion Fear Anger Despair Rage Boredom Happiness Pride Shame Disgust
Suddenness High Low High High Very low Low Low
Familiarity Low Very low Low High Low
Predictability Low Medium Low Low Very high Medium Low
Pleasantness Low Low High Very low
Consistency with expectations Low Very low Low High High
Control Open High Very low High Medium
Social norms Low High Low

350 Veissier et al
rapidly, there was a high but transient increase in heart rate
whereas sheep reacted to unfamiliarity with increased heart-
rate variability. We again exposed sheep to a rapidly falling
object, in this case a wooden panel, and we trained some
sheep to predict the occurrence of the object fall event via a
light signal (Figure 1) (Greiveldinger et al 2007). The sheep
that benefited from this conditioning reacted less to the fall
of the object, ie startle responses were less frequent and the
heart rate increase was limited. Therefore, sheep are
sensitive to the suddenness, unfamiliarity and predictability
of their environment, and these aspects do affect their
emotional responses.
Discrepancy from expectations
Individuals form expectations about their environment. They
expect some events to occur and they also expect these
events to have specific consequences. For instance, if a
person knows certain colleagues well, he/she expects these
people to behave in accordance with what they usually do.
Similarly, if a person knows a certain food, then he/she
expects this food to have a specific flavour or texture. If the
colleagues behave in a constant manner or if the food has the
same properties as usual, then the situation will be fully
consistent with the individual’s expectations. However, this
is not always the case: the colleague may have experienced
a very difficult situation affecting his/her mood, or the food
being eaten, despite having a similar appearance, may taste
different from usual. In these cases, there is a discrepancy
between what the individual expects and the actual situation.
To test whether sheep can form expectations about their
environment, we trained them to perform an operant task (to
cross a beam with their muzzle) to get a large food reward
or a small food reward, and then we shifted the amount of
reward (from large to small and vice versa) for half the
animals (Greiveldinger et al 2006). Sheep were particularly
sensitive to the decrease in the size of the reward: compared
to sheep that had always been trained with a small reward,
those that were shifted from a large to a small reward
seemed disturbed, adopting an asymmetric ear posture at the
time of the reward delivery and walking around the
chamber, while their heart rate increased and heart-rate vari-
ability decreased (suggesting a lower parasympathetic tone,
which is indicative of stress (Porges 1995; Després et al
2003). Sheep responses to an increase in the reward were far
less marked and were observed only when it occurred after
a previous decrease in reward. In this case, the number of
operant responses performed by the sheep decreased, but
there was no change in emotional responses (heart rate, ear
postures). Sheep can thus form expectations about their
environment. A discrepancy from their expectations has an
effect on their emotional responses if an actual situation is
less attractive than the expected situation.
Control potential
The potential for control evaluated by an individual corre-
sponds to the extent to which this individual perceives that
he/she is able to influence the event to which he/she is
exposed and/or act on the consequences of this event.
To test whether sheep were sensitive to their potential to
control a situation, we gave them intermittent access to
food: from time-to-time during a session, an air blower was
turned on above the trough (Greiveldinger et al 2009). Half
the sheep were trained to cross an infra-red beam to
terminate the air blow and resume access to food (Figure 2).
The other sheep were yoked to the previous sheep and thus
received exactly the same access to the food but without
controlling the air disturbance. Again, the two treatments
elicited different behavioural and physiological responses:
when the sheep were eating, their ears were horizontal and
pointed downward, whereas immediately after the air blow
started, the ears were oriented forward in sheep that could
control the event and backward in those that could not. In
addition, for the entire duration of the session, the sheep that
could not control the event bleated four times as frequently
as the sheep that could control it, and their heart rate was
higher. Therefore, a disturbing situation has distinct conse-
quences on emotional responses of a sheep depending on
whether or not the disturbance can be controlled.
Social norms
When facing a triggering event, an individual also evaluates
whether the responses available to him/her are compatible
with the norms of the social group to which he/she belongs
(ie is a given response desirable or compulsory?)
To transpose this check to animals, we used the spontaneous
dominance hierarchy of sheep. A dominance relationship is
established when aggressive encounters or the mere
presence of an emitter animal trigger submissive postures or
avoidance patterns in a receiver animal (Bouissou et al
2001). This emitter thus dominates the receiver. It is likely
that animals know their position relative to the other animals
in their group. For instance, a subordinate animal may
© 2009 Universities Federation for Animal Welfare
Figure 1
Experimental set-up for studying the impact of novelty. The sheep
were trained to eat in the feeding bowl. An object (in this case a
large panel with white and blue squares) was then made to fall
either rapidly or slowly above the feeding bowl. For some sheep,
the fall of the object was cued by a light signal.

Animals’ emotions: studies in sheep using appraisal 351
display submissive postures or avoidance even without
aggression from the dominant animal, but may display
aggressive behaviour toward a third animal it dominates. We
thus assumed that dominance relationships act as external
norms, reflecting how the animal adjusts its behaviour in
accordance with the social identity of its group partners.
We thus checked the hypothesis that the same disturbing
event may not have the same impact on a target animal
accompanied by another animal that it dominates vs that it
is dominated by. Forty-eight sheep were accommodated
together. Prior to testing, the dominance relationships were
analysed between each pair of animals via food competi-
tion tests (Erhard et al 2004). Briefly, two animals chosen
at random were introduced into a chamber where a highly
palatable food (concentrates) was placed in a small trough
that only enabled one animal to feed at a time. An animal
was considered to dominate its paired counterpart when it
spent at least five times more time eating than its counter-
part over two test sessions. Triads of animals were then
composed of a target sheep, a sheep that the target
dominates, and a sheep dominating the target. The target
sheep was then tested in the same experimental chamber as
the one used to test predictability, but was accompanied by
either its dominant or its subordinated counterpart. We
observed startle responses and a transient increase in heart
rate immediately following the fall of the object behind the
trough, as in the first experiment when the fall of the object
could not be predicted. However, there were also subtle
differences between the two conditions: when accompa-
nied with a dominant animal, the target animal under test
often looked at the accompanying animal and the increase
in heart rate was more marked, whereas it walked back
further from the trough when accompanied by a subordi-
nate. Therefore, the emotional responses of a sheep to a
disturbing situation vary according to the dominance
relations with the partners around, with more internal
reactions when the sheep is dominated by its partner and
more overt reactions when it dominates the partner.
Implications for sheep emotions
These studies confirmed that sheep evaluate events in their
environment according to their suddenness, familiarity,
predictability, and the consistency of these events with their
own expectations and the control they have over the event.
In addition, the way the animals evaluate their dominance-
subordination relationships with others, similarly to social
standards, also plays a role. In humans, emotions derive
from the outcome of these checks, leading to specific
responses and internal feelings (Rosemand & Evdokas
2004). We found that the behavioural and cardiac responses
of sheep (ear postures, startle, agitation, orientation, heart
rate, and heart-rate variability) also vary according to the
outcome of these checks. The fact that, in humans, similar
responses resulting from the outcome of checks are tightly
linked to the internal psychological component of emotions,
ie the internal feeling, and these checks are used by sheep
and modify their emotional responses, support the existence
of emotions in sheep. Sheep emotional responses are not
only modified by a simple check like that of suddenness but
also by checks that require a schematic process of appraisal
(familiarity, predictability, consistency with expectations,
controllability). Hence, we assume that sheep do not only
display emotional responses but can also experience the
‘feeling’ component of emotions. Nevertheless, whether
sheep are conscious of their emotions remains open to spec-
ulation. The fact that their relationship towards other sheep
accompanying them when they face triggering situations
affects their responses would add support to the assumption
of conscious feelings in sheep, but our results on this point
are not very clear-cut (we found only subtle variations
between sheep accompanied by a dominant vs a subordinate
animal) and need further confirmation. Furthermore,
specific paradigms should be used to test whether sheep can
remember their emotions. According to the framework used
by Sander et al (2005), sheep appear to have the potential to
feel at least the following emotions:
•Anger, since they are sensitive to suddenness, unpre-
dictability, controllability, and social norms;
• Rage, since they are sensitive to suddenness, unfamiliarity,
unpredictability, discrepancy from expectations, controlla-
bility, and social norms;
• Despair, since they are sensitive to suddenness, unfamil-
iarity, unpredictability, discrepancy from expectations, and
controllability;
• Boredom, since they are sensitive to suddenness, unfamil-
iarity, unpredictability, discrepancy from expectations, and
controllability.
Our team has not yet been able to robustly analyse how the
pleasantness of an event affects the emotional responses of
sheep. However, other teams have successfully conducted
choice tests on sheep (Dumont et al 1995; Rushen 1990). It
is clear that sheep have preferences for certain foods or
certain partners, including human beings (Tallet et al
Animal Welfare 2009, 18: 347-354
Figure 2
Experimental set-up for studying the impact of the controllability
of an event. Sheep were trained to eat in a feeding bowl. From
time-to-time during a session, an air blower was turned on, blow-
ing into the feeding bowl and preventing the sheep from eating.
Some sheep could terminate the airflow by nosing into an aper-
ture; and cutting an infra-red beam. Their yoked counterparts
received the same air blowing condition but could not control it.

352 Veissier et al
2005), and that they actively avoid (so probably dislike)
specific situations, such as being handled roughly.
Therefore, it is reasonable to believe that sheep can experi-
ence fear which is triggered by suddenness, unfamiliarity,
unpredictability, unpleasantness, and discrepancy from
expectations. It is also reasonable to believe that sheep can
experience disgust, which is triggered by unfamiliarity,
unpredictability and unpleasantness, as well as happiness,
which is triggered by suddenness, unpredictability, pleas-
antness and discrepancy from expectations.
In humans, emotional responses are specific to the
emotional feeling. For instance, competing athletes facing
the same situation (a win or a defeat) exhibit the same facial
expressions, regardless of whether they are sighted or blind
(ie cannot have seen this expression exhibited by others)
(Matsumoto & Willingham 2009). Sheep showed startle
responses and a transient heart rate increase when
confronted with sudden events, while their ears were
pointed backward when they lacked control over a
disturbing situation. Such responses might thus be specific
to fear and anger. However, we have tested too few situa-
tions involving one or several more salient checks to be able
to draw firm conclusions on emotional signatures in sheep.
To conclude that sheep can feel certain emotions, we first
postulated that if the outcome of the checks has an impact
on responses, then the animal not only displays emotional
responses but also feels emotions (see above). This
postulate is still open to debate, as it can be argued that the
analogy of cognitive treatments and emotional responses
between animals and humans is not enough to prove that
an animal’s cognitive processing of a situation they are
exposed to results in similar mental states as in humans
(see Discussion in Volpato et al 2007). Nevertheless, the
present findings broadly support the idea that animals
possess intuitive understanding, according to which it
would make no sense to deny mental states to non-human
animals (Bekoff 2008). Another body of findings that
would add support to the existence of emotions in animals
could be brought by functional imagery. Several authors
have studied the links between emotions and brain activity
in humans. For instance, the fronto-temporal areas are
activated by joy, sadness and disgust (Esslen et al 2004;
Delplanque et al 2005). Similar activations may be
observed in animals, at least those that, like vertebrates,
present homologous brains. Nevertheless, although this
sort of finding could provide further evidence for the
existence of emotions in animals, it would not offer any
sceptical mind with final proof. We strongly believe that it
is getting harder and harder not to accept the existence of
emotions in non-human animals, even if these emotions
are not as sophisticated as those of humans.
Most of the results presented in this paper were obtained in
situations that were rather aversive to animals, except when
we enlarged a reward to make it higher than expected.
According to Fraser and Duncan (1998), different evolu-
tionary processes seem to have selected negative vs positive
emotions: negative emotions are supposed to have evolved
in ‘need situations’, such as a threat to survival or reproduc-
tive success, whereas positive emotions are supposed to have
evolved in ‘opportunity situations’ where the action resulting
from the positive emotional state enhances but is not
essential to individual fitness. This may be the reason why
positive emotions have more between-animal variability and
are more difficult to highlight than negative emotions.
Further research is needed into the positive side of the
emotional scale to get an overall view of animal welfare.
We have not explored all the possible aspects of the
appraisal theories. The use of checks such as the relevance
of a situation for the animal’s goals, the probability of
consequences of a situation, an animal’s perceived
abilities to engage in control, or even the role of internal
standards (if any exist) have not yet been investigated, but
we believe that sophisticated conditioning paradigms
could make these investigations possible.
All of our experiments have been conducted in sheep.
Several recent studies confirm that pigs are sensitive to the
predictability of aversive events, which affects their vocal
responses (Duepjan et al 2008), and to the variability of
food rewards (De Jonge et al 2008). These findings suggest
that the checks of predictability and discrepancy from
expectations are also valid in other contexts and other
species than the sheep tested in our research group. A
comparison between species from different phyla using a
similar rationale would certainly provide critical insight into
the kind of emotions they can feel, and this would in turn
help to refine regulations adopted to ensure animal welfare.
In conclusion, by using the framework of appraisal
theories, the experiments reported in this paper give
support to the hypothesis that non-human animals do
experience emotions. Sheep seem able to experience a
wide range of emotions, including fear, anger, rage,
despair, boredom, disgust, and happiness.
Animal welfare implications
Although the findings reported in this paper were obtained
in artificial experimental situations, they have nevertheless
produced generic knowledge that can easily be transposed
to the actual living conditions of animals, including farm
animals. For instance, sudden noises or movements can
occur in any farm environment, and animals that are offered
new foods, mixed with new animals, or moved to a new
barn face unfamiliar situations. These situations are all
known to affect animal behaviour. Cows, for instance, are
very sensitive to shouting and dislike novel situations
(Rushen et al 1999; Herskin et al 2004). Predictability also
appears to be an important characteristic of farming envi-
ronments. It is common practice to feed or milk animals at
the same time(s) every day. If this regularity is broken for
any reason, it can be expected to disturb the animals. For
instance, it is well known to veterinarians that pigs can
develop gastric ulcers when feeding is delayed or when the
usual signals of feeding (noise in pipes) are not followed by
food delivery. Similarly, calves that receive their milk at
regular times during the day seem disturbed when feeding is
delayed (Johannesson & Ladewig 2000). There has been a
longstanding practice of tying large farm animals when they
© 2009 Universities Federation for Animal Welfare

Animals’ emotions: studies in sheep using appraisal 353
are housed indoors (eg as is the case with cows in winter),
which means the animals had little control over their envi-
ronment. More possibilities for control are offered by more
modern systems, such as loose housing and automatic
feeding systems where the animals need to go and operate a
device to get their food. Similarly, there are facilities
running milking robots that enable cows to decide to go to
milking. This level of control appears beneficial to the
animals: pigs given control over food delivery show
enhanced healing abilities (Ernst et al 2006).
There have also been studies describing the long-term
effects of the predictability and controllability of aversive
events. In both humans and non-human animals, unpre-
dictability can enhance negative emotional experiences,
such as fear (Adkin et al 2006; Armfield 2006; Carlsson
et al 2006) and induce negative cognitive bias whereby
neutral situations are more likely to be perceived as
negative (Harding et al 2004). This may lead to wide-
reaching disorders such as anxiety (Zvolensky et al 2000),
depression (Anisman & Matheson 2005) or neurosis
(Mineka & Kihlstrom 1978). Similarly, repeated uncontrol-
lability can induce chronic stress, as suggested by rats
unable to control the termination of an electric shock, which
develop more gastric ulcers than their yoked counterparts
able to terminate the shock (Weiss 1972; Milde et al 2005).
Similarly, restriction in movements, which can be viewed as
largely limiting the control the animal can exert, is known
to facilitate stereotypies and apathy in sows, which suggest
low welfare (Broom 1987; Terlouw et al 1991). The uncon-
trollability of positive events like food delivery may also
have negative consequences, leading for instance to hypoal-
gesia in mice (Tazi et al 1987) or learned helplessness in
hens (Haskell et al 2004), both of which are signs attributed
to chronic stress. Our findings suggest that the mechanism
underlying such long-term effects may be the repetition of
negative (short-term) emotions.
Findings such as these led Bassett and Buchanan-Smith
(2007) to recommend that the environment given to animals
should be predictable and controllable, allowing anticipa-
tion to develop fully. However, an overly predictable envi-
ronment may result in boredom (Van Rooijen 1984;
Wemelsfelder 1993). Therefore, initiatives to ensure animal
welfare need to find a balance between complete unpre-
dictability vs complete predictability of the environment,
while a degree of control always seems beneficial.
References
Adkin AL, Quant S, Maki BE and McIlroy WE 2006
Cortical responses associated with predictable and unpre-
dictable compensatory balance reactions. Experimental Brain
Research 172: 85-93
Anisman H and Matheson K 2005 Stress, depression, and
anhedonia: caveats concerning animal models. Neuroscience and
Biobehavioral Reviews 29: 525-546
Armfield JM 2006 Cognitive vulnerability: a model of the etiol-
ogy of fear. Clinical Psychology Review 26: 746-768
Arnold MB 1945 Psychological differentiation of emotional
states. Psychological Reviews 52: 35-48
Bassett L and Buchanan-Smith HM 2007 Effects of pre-
dictability on the welfare of captive animals. Applied Animal
Behaviour Science 102: 223-245
Bekoff M 2008 Why “good welfare” isn’t “good enough”: mind-
ing animals and increasing our compassionate footprint. ARBS
Annual Review of Biomedical Sciences, Theme Topic on ‘Unraveling
Animal Welfare’ 10: T1-T14
Bouissou MF, Boissy A, Le Neindre P and Veissier I 2001
The bovines among other bovids: social structure of wild bovines.
In: Keeling L and Gonyou H (eds) Social Behavior in Farm Animals
pp 113-145. CAB International: Oxon, UK
Broom DM 1987 Applications of neurobiological studies to farm
animal welfare. In: Wiepkema PR and Van Adrichem PWM (eds)
Biology of Stress in Farm Animals: An Integrative Approach pp 101-
110. Martinus Nijhoff Publishers: Dordrecht, The Netherlands
Carlsson K, Andersson J, Petrovic P, Petersson KM,
Ohman A and Ingvar M 2006 Predictability modulates the
affective and sensory-discriminative neural processing of pain.
NeuroImage 32: 1804-1814
Council of Europe 1976 European Convention for the Protection of
Farm Animals Kept for Farming Purposes, No. 87 6 pp. European
Treaty Series: Brussels, Belgium
Dawkins MS 2006 Quality of life, good welfare and the animal’s
point of view. Proceedings of the Quality of Life: the Heart of the
Matter. UFAW: Wheathampstead, Herts, UK
De Jonge FH, Ooms M, Kuurman WW, Maes JHR and
Spruijt BM 2008 Are pigs sensitive to variability in food
rewards? Applied Animal Behaviour Science 114: 93-104
de Waal FBM 1999 Anthropomorphism and anthropodenial:
consistency in our thinking about humans and other animals.
Philosophical Topics 27: 255-280
Delplanque S, Silvert L, Hot P and Sequeira H 2005 Event-
related P3a and P3b in response to unpredictable emotional stim-
uli. Biological Psychology 68: 107-120
Désiré L, Boissy A and Veissier I 2002 Emotions in farm ani-
mals: a new approach to animal welfare in applied ethology.
Behavioural Processes 60: 165-180
Désiré L, Veissier I, Després G and Boissy A 2004 On the
way to assess emotions in animals: Do lamb evaluate an event
through its suddenness, novelty or unpredictability? Journal of
Comparative Psychology 118: 363-374
Désiré L, Veissier I, Després G, Delval E, Toporenko G
and Boissy A 2006 Appraisal process in sheep: synergic effect of
suddenness and novelty on cardiac and behavioural responses.
Journal of Comparative Psychology 120: 280-287
Després G, Boissy A, Désiré L, Le Neindre P and Veissier
I 2003 Validation of the measure of sympatho-vagal effect in lambs
through autonomic blockades and heart rate variability indexes.
Journal of Animal and Veterinary Advances 2: 615-619
Duepjan S, Schoen PC, Puppe B, Tuchscherer A and
Manteuffel G 2008 Differential vocal responses to physical and
mental stressors in domestic pigs (Sus scrofa). Applied Animal
Behaviour Science 114: 105-115
Dumont B, Petit M and D’hour P 1995 Choice of sheep and
cattle between vegetative and reproductive cocksfoot pastures.
Applied Animal Behaviour Science 43: 1-15
Duncan IJH 1993 Welfare is to do with what animals feel. Journal
of Agricultural and Environmental Ethics 6: 8-14
Duncan IJH 2006 The changing concept of animal sentience.
Applied Animal Behaviour Science 100: 11-19
Erhard HW, Fábrega E, Stanworth G and Elston DA
2004 Assessing dominance in sheep in a competitive situation:
level of motivation and test duration. Applied Animal Behaviour
Science 85: 277-292
Animal Welfare 2009, 18: 347-354

354 Veissier et al
Ernst K, Tuchscherer M, Kanitz E, Puppe B and
Manteuffel G 2006 Effects of attention and rewarded activity on
immune parameters and wound healing in pigs. Physiology and
Behavior 89: 448-456
Esslen M, Pascual-Marqui RD, Hell D, Kochi K and
Lehmann D 2004 Brain areas and time course of emotional pro-
cessing. NeuroImage 21: 1189-1203
European Union 1997 The Amsterdam treaty modifying the
treaty on European Union, the treaties establishing the European
communities, and certain related facts. Official Journal C340: 1-144
Fraser D and Duncan IJH 1998 ‘Pleasures’, pains’ and ani-
mal welfare: toward a natural history of affect. Animal Welfare
7: 383-396
Frijda NH 1986 The Emotions 528 pp. Cambridge University
Press: Cambridge, UK
Grassé PP 1977 Evolution of Living Organisms 480 pp. Academic
Press: New York, USA
Greiveldinger L, Veissier I and Boissy A 2006 Predictability
and expectations influence emotional responses in lambs.
Proceedings of the Proceedings of the Congress The Quality of Life: The
Heart of the Matter. UFAW: Wheathampstead, Herts, UK
Greiveldinger L, Veissier I and Boissy A 2007 Emotional
experience in sheep: Predictability of a sudden event lowers sub-
sequent emotional responses. Physiology & Behavior 92: 675-683
Greiveldinger L, Veissier I and Boissy A 2009 Behavioural
and physiological responses of lambs to controllable vs. uncon-
trollable aversive events. Psychoneuroendocrinology 34: 805-814
Harding EA, Elizabeth ES and Mendl M 2004 Animal behav-
iour: cognitive bias and affective state. Nature 427: 312
Haskell MJ, Coerse NCA, Taylor PAE and McCorquodale
C 2004 The effect of previous experience over control of access
to food and light on the level of frustration-induced aggression in
the domestic hen. Ethology 110: 501-513
Herskin MS, Kristensen A-M and Munksgaard L 2004
Behavioural responses of dairy cows toward novel stimuli pre-
sented in the home environment. Applied Animal Behaviour Science
89: 27-40
Jacobs ME 1998 Mr. Darwin Misread Miss Peacock’s Mind: A New
Look at Mate Selection in Light of Lessons From Nature 248 pp.
Nature Books: Goshen, IN, USA
Johannesson T and Ladewig J 2000 The effect of irregular
feeding times on the behaviour and growth of dairy calves. Applied
Animal Behaviour Science 69: 103-111
Kappas A 2006 Appraisals are direct, immediate, intuitive,
and unwitting and some are reflective. Cognition and Emotion
20: 952-975
Lazarus RS, Averill JR and Opton EM Jr 1970 Towards a
Cognitive Theory of Emotion pp 207-232. Academic Press: New
York, USA
Leventhal H and Scherer K 1987 The relationship of emotion
to cognition: a functional approach to a semantic controversy.
Cognition and Emotion 1: 3-28
Matsumoto D and Willingham B 2009 Spontaneous facial
expressions of emotion of congenitally and noncongenitally blind
individuals. Journal of Personal Social Psychology 96: 1-10
Milde AM, Arslan G, Overmier JB, Berstad A and
Murison R 2005 An acute stressor enhances sensitivity to a
chemical irritant and increases (51)Credta permeability of the
colon in adult rats. Integrative Physiological and Behavioral Science
40: 35-44
Mineka S and Kihlstrom JF 1978 Unpredictable and uncon-
trollable events: A new perspective on experimental neurosis.
Journal of Abnormal Psychology 87: 256-271
Nicol CJ 1996 Farm animal cognition. Animal Science 62: 375-391
Philippot P, Baeyens C, Douilliez C and Francart B 2004
Cognitive regulation of emotion. In: Philippot P and Feldman RS
(eds) The Regulation of Emotion pp 73-100. Laurence Erlbaum
Associates: New York, USA
Porges SW 1995 Orienting in a defensive world: mammalian
modifications of our evolutionary heritage. A polyvagal theory.
Psychophysiology 32: 301-318
Roseman IJ 1991 Appraisal determinants of discrete emotions.
Cognition and Emotion 5: 161-200
Roseman IJ and Evdokas A 2004 Appraisals cause experienced
emotions: experimental evidence. Cognition and Emotion 18: 1-28
Rushen J 1990 Use of aversion-learning techniques to measure
distress in sheep. Applied Animal Behaviour Science 28: 3-14
Rushen J, De Passillé AM and Munskgaard L 1999 Fear of
people by cows and effects on milk yield, behavior, and heart rate
at milking. Journal of Dairy Science 82: 720-727
Sander D, Grandjean D and Scherer KR 2005 A systems
approach to appraisal mechanisms in emotion: emotion and brain.
Neural Networks 18: 317-352
Scherer KR 2001 Appraisal considered as a process of multi-
level sequential checking. In: Scherer KR, Schorr A and Johnstone
T (eds) Appraisal Processes in Emotion: Theory, Methods, Research
pp 92-120. Oxford University Press: Oxford, UK
Smith CA and Kirby LD 2001 Towards delivering on the
promise of appraisal theory. In: Scherer KR, Schorr A and
Johnstone T (eds) Appraisal Processes in Emotion pp 121-140.
Oxford University Press: New York, USA
Tallet C, Veissier I and Boivin X 2005 Human contact and
feeding as rewards for the lamb’s affinity to their stockperson.
Applied Animal Behaviour Science 94: 59-73
Tazi A, Dantzer R and Le Moal M 1987 Prediction and con-
trol of food rewards modulate endogenous pain inhibitory sys-
tems. Behavioral Brain Research 23: 197-204
Terlouw EMC, Lawrence AB and Illius AW 1991 Influences
of feeding level and physical restriction on development of stereo-
typies in sows. Animal Behavior 42: 981-991
Van Rooijen J 1984 Impoverished environments and welfare.
Applied Animal Behaviour Science 12: 3-13
Veissier I and Boissy A 2007 Stress and welfare: two comple-
mentary concepts that are intrinsically related to the animal’s
point of view. Physiology & Behaviour 92: 429-433
Volpato GL, Gonçalves-de-Freitas E and Fernandes-de-
Castilho M 2007 Insights into the concept of fish welfare.
Diseases of Aquatic Organisms 75: 165-171
Weiss JM 1972 Psychological factors in stress and disease.
Scientific American 226: 104-113
Wemelsfelder F 1993 The concept of animal boredom and its
relationship to stereotyped behaviour. In: Lawrence AB and
Rushen J (eds) Stereotypic Animal Behaviour: Fundamentals and
Applications to Welfare pp 65-95. CAB International: Oxon, UK
Wynne CD 2004 The perils of anthropomorphism. Nature
428: 606
Zvolensky MJ, Eifert GH, Lejuez CW, Hopko DR and
Forsyth JP 2000 Assessing the perceived predictability of anxi-
ety-related events: a report on the perceived predictability index.
Journal of Behavior Therapy and Experimental Psychiatry 31: 201-218
© 2009 Universities Federation for Animal Welfare