Assessment of Positive Emotions in Animals to Improve Their Welfare

INRA, UR1213 Herbivores, Site de Theix, F-63122 Saint-Genès-Champanelle, France.
Physiology & Behavior (Impact Factor: 2.98). 11/2007; 92(3):375-97. DOI: 10.1016/j.physbeh.2007.02.003
Source: PubMed


It is now widely accepted that good welfare is not simply the absence of negative experiences, but rather is primarily the presence of positive experiences such as pleasure. However scientific investigation of positive emotions has long been neglected. This paper addresses two main issues: first, it reviews the current state of scientific knowledge that supports the existence of positive affective states in animals and, second, it suggests possible applications of this knowledge that may enhance quality of life under animal management conditions. In the first part of the paper, recent advances in psychology and neuroscience are reviewed to provide pragmatic frameworks based on cognitive processes (such as positive anticipation, contrast and controllability) for further investigations of positive emotions in animals. Thereafter, the neurobiological bases of positive emotions are highlighted in order to identify behavioral and physiological expressions of positive experiences in animals. Monitoring both the autonomic nervous system (via heart rate and its variability) and the immune system could offer relevant tools to better assess emotional states in animals, complementary to classical adrenocortical measures. In the second part of the paper, useful strategies for enhancing positive experiences (such as physical, social and cognitive enrichment or putative genetic selection) are outlined. Then this paper emphasizes practical applications for assessing and promoting positive emotions that may help in providing animals with a better quality of life. Play, affiliative behaviors and some vocalizations appear to be the most promising convenient indicators for assessing positive experiences in laboratory and farm animals under commercial conditions.


Available from: Jan Langbein
Assessment of positive emotions in animals to improve their welfare
Alain Boissy
, Gerhard Manteuffel
, Margit Bak Jensen
, Randi Oppermann Moe
Berry Spruijt
, Linda J. Keeling
, Christoph Winckler
, Björn Forkman
, Ivan Dimitrov
Jan Langbein
, Morten Bakken
, Isabelle Veissier
, Arnaud Aubert
INRA, UR1213 Herbivores, Site de Theix, F-63122 Saint-Genès-Champanelle, France
Research Unit Behavioral Physiology, Research Institute for the Biology of Farm Animals, Dummerstorf, Germany
Faculty of Agricultural Sciences, University of Aarhus, Foulum, Denmark
Department Production Animal and Clinical Sciences, School of Veterinary Science, Oslo, Norway
Department Animal Science and Society, Veterinary Faculty, Utrecht, Netherlands
Department Animal Environment and Health, University of Agricultural Sciences, Uppsala, Sweden
Department Sustainable Agricultural Systems, University of Natural Resources and Applied Life Sciences, Vienna, Austria
Department Large Animal Science, University of Copenhagen, Copenhagen, Denmark
Research Institute of Agricultural Science, Stara Zagora, Bulgaria
Department Animal and Aquacultural Sciences, University of Life Sciences, As, Norway
Department Behavioral Sciences (DESCO), Faculty of Sciences, Tours, France
Received 30 March 2006; received in revised form 5 February 2007
It is now widely accepted that good welfare is not simply the absence of negative experiences, but rather is primarily the presence of positive
experiences such as pleasure. However scientific investigation of positive emotions has long been neglected. This paper addresses two main
issues: first, it reviews the current state of scientific knowledge that supports the existence of positive affective states in animals and, second, it
suggests possible applications of this knowledge that may enhance quality of life under animal management conditions. In the first part of the
paper, recent advances in psychology and neuroscience are reviewed to provide pragmatic frameworks based on cognitive processes (such as
positive anticipation, contrast and controllability) for further investigations of positive emotions in animals. Thereafter, the neurobiological bases
of positive emotions are highlighted in order to identify behavioral and physiological expressions of positive experiences in animals. Monitoring
both the autonomic nervous system (via heart rate and its variability) and the immune system could offer relevant tools to better assess emotional
states in animals, complementary to classical adrenocortical measures. In the second part of the paper, useful strategies for enhancing positive
experiences (such as physical, social and cognitive enrichment or putative genetic selection) are outlined. Then this paper emphasizes practical
applications for assessing and promoting positive emotions that may help in providing animals with a better quality of life. Play, affiliative
behaviors and some vocalizations appear to be the most promising convenient indicators for assessing positive experiences in laboratory and farm
animals under commercial conditions.
© 2007 Elsevier Inc. All rights reserved.
Keywords: Positive emotion; Positive mood; Cognitive appraisal; Play; Grooming; Vocalization; Animal welfare
1. Introduction .............................................................. 376
2. Biological processes underlying positive emotions .......................................... 377
2.1. Cognitive aspects of positive emotions ............................................ 377
2.1.1. Emotions and feelings ................................................ 377
2.1.2. Cognitive components of emotional processing ................................... 377
Physiology & Behavior 92 (2007) 375 397
Corresponding author. Tel.: +33 473 62 42 98; fax: +33 473 62 41 18.
E-mail address: (A. Boissy).
0031-9384/$ - see front matter © 2007 Elsevier Inc. All rights reserved.
Page 1
2.1.3. Emotion and mood: acute and long-term evaluations ................................. 378
2.1.4. What are positive emotions? ............................................. 379
2.2. Neurobiological bases of positive emotions: the emotional 433 brain ............................ 379
2.2.1. Actions on the reward system ............................................ 379
2.2.2. Dopamine reflects intensity of wanting while opioids determine what is wanted................... 380
2.3. Behavioral aspects of positive emotions ............................................ 381
2.3.1. Behavioral needs ................................................... 381
2.3.2. How can behavioral needs induce pleasure and why do they make the animal vulnerable? ............. 381
2.4. Physiological markers of positive emotions .......................................... 382
2.4.1. Sympathetic and parasympathetic balance ...................................... 382
2.4.2. Neuroendocrine activation .............................................. 382
2.4.3. Immunological activation ............................................... 383
3. Application of positive emotions in animal welfare .......................................... 383
3.1. Enhancing quality of life by promoting positive experiences .................................. 383
3.1.1. Positive anticipation .................................................. 383
3.1.2. Positive contrast .................................................... 384
3.1.3. Coping and controllability .............................................. 384
3.2. Enhancing long-term positive emotional states ......................................... 385
3.2.1. Concept of temperament ............................................... 385
3.2.2. Genetic background .................................................. 386
3.2.3. Ontogenetic influences ................................................ 386
3.2.4. Enriching the environment or just reducing stressful events? . . ........................... 386
3.3. Inducing positive emotions for improving health ....................................... 387
3.4. Using positive emotions for assessment and monitoring 1301 of animal welfare ....................... 387
3.4.1. Assessing and enhancing behavioral expressions of positive emotions ........................ 387
3.4.2. Developing an on-farm monitoring system to assess animal welfare ......................... 390
4. Conclusion .............................................................. 390
Acknowledgements ............................................................. 391
References ................................................................. 391
1. Introduction
In the last decades of the 20th century, there was a surge of
interest in animal sentience. Animal welfare scientists quickly
realized that welfare problems can be better addressed with an
understanding of how animals could feel. It is now widely
accepted that animals can feel pain and suffering, and methods
to assess pain and suffering have been developed. However,
there is still no agreement on how to assess positive experiences
although these are suggested to be a core component of good
welfare [1,2] Nevertheless, animals that have successfully
obtained commodities such as food or social contact are known
to display signs that are reminiscent of pleasure in humans.
Therefore, a central issue is to be able to assess whether and
under what circumstances animals experience positive emo-
tions. This issue is far more than merely theoretical; it also has
ethical and practical importance for animal welfare. It has been
pointed o ut that well-being is not simply the absence of negative
affects, but also (and even predominantly) the presence of
positive affects [3,4]. That means it is better to ensure joyful and
contented behaviors rather than focusing on behaviors that
represent needs that have to be fulfilled to avoid suffering [5].In
addition, the absence of signs of pleasure or positive affect may
be an indication on its own of a state of affective discomfort.
Indeed anhedonia, which is the inability to experience pleasure,
is one of the core symptoms of depression. Therefore, a new
challenge for animal welfare science is to better understand the
link between the ability to express positive emotions and a more
persistent positive affective state, such as happiness. Moreover,
applied ethology deals with a wide range of animal species,
having different emotional repertoires and different behavioral
patterns. Another challenge is thus to describe the range of
putative positive emotions in each of the laboratory and farm
animal species.
From an evolutionary perspective, emotions are considered
as adaptive programs designed through repeated encounters that
are intended to either direct other physiological programs or to
directly solve adaptive problems faced by a species over time.
One of the key factors explaining the phylogenetic success of
emotions is that they would favor adaptive cognition and action.
According to evolutionary psychology, emotions are considered
as super-ordinate mechanisms allowing the whole organism to
operate in a homogenous way when the individual is con-
fronting relevant triggering conditions or situations. They are
adaptations that have arisen in response to the adaptive problem
of mechanism orchestration [6]. Generally speaking, the sit-
uations where emotions play a role are those that recurred an-
cestrally, those that could not be dealt with without a governing
program, and those in which an error would have resulted in large
fitness costs [7]. However, given the very nature of emotional
self-experience, there is ultimately no way to know if animals
experience emotions similar to humans. However, behavior,
structure, and brain chemistry are similar in humans and in a large
number of animal species. It is therefore likely that they feel as we
376 A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 2
do, including not only well-studied negative emotions, but also
positive ones.
Study of emotions in animals has received a growing interest
in the past few decades, as testified by the emergence of a new
discipline referred to as Affective Neuroscience [8]. However,
despite the efforts of pioneering authors such as Colin Allen,
Jaak Panksepp, Michel Cabanac or Kent Berridge, relatively
little has been done to further the understanding o f positive
emotions by comparison to their negative counterparts such as
fear. Such an imbalance does not seem to be unique to animal
studies. Indeed, the study of emotions in humans suffers the
same bias. As in animal studies, the human psychological study
of well-being has long been dominated by stress studies. The
primary reason for this bias is presumably that the expressions
of negative experiences are far more intense than positive
affects and therefore easier to study. These latter experiences are
commonly considered as less significant since they are more
labile and their expression is more subtle. Nevertheless, in the
last few years, new disciplines such as Positive Psychology
have been growing in order to explore positive emotions as
primary components of subjective well-being [9,10].
The present paper does not claim to re-equilibrate the bal-
ance by itself but to contribute to promote the promise of
studying positive emotions in animals. It addres ses two main
issues. It is intended first to review the current state of knowl-
edge regarding the biological foundations of positive emotions,
and second to indicate some po ssible applications for improving
the quality of life of animals under management conditions. The
presentation of the biological background of positive emotions
successively deals with the cognitive, neurobiological, behav-
ioral, and physiological aspects that support the existence of
positive e motions in animals. Following this rev iew, the
practical applications in animal welfare are outlined. These
applications include useful approaches for enhancing positive
experiences and on-going more persistent affective states, as
well as a discussion of the most promising positive indicators
that may be used to monitor animal welfare in both laboratory
and farm animals.
2. Biological processes underlying positive emotions
There is abundant literature from both biology and psy-
chology on the concept of emotion. Whereas there is not a
general definition, an emotion ca n be defined as an intense but
short-lived affective response to an event, which is associated
with specific body changes [11]. The duration of an emotion is a
much-debated question, but briefness seems widely accepted.
Emotions refer to processes, which are likely to have evolved
from basic mechanisms that gave the animals the ability to
avoid harm/punishment or to seek valuable resources/reward
[12,13]. Contemporary approaches to study emotions vary ac-
cording to whether they emphasize a classification of the
various emotions (dimensional approaches such as pleasant/
unpleasant valence or weak/strong excitation), whether they
focus on the eliciting mechanisms (neural circuit models and
basic emotion models), or whether they emphasize the different
components of an emotion. In componential approaches, an
emotion is classically described as including a behavioral com-
ponent (a posture or an activity), an autonomic component (vis-
ceral and endocrine responses) and a subjective component
(emotional experience or feeling) [reviewed by [11]].
2.1. Cognitive aspects of positive emotions
2.1.1. Emotions and feelings
Although it is claimed that animals are sentient creatures, it is
paradoxical that there is no exact knowledge of what their emo-
tional experiences are about. Rollin [14] calls attention to the
paradox that scientists are reluctant to attribute emotions, such as
anxiety or suffering, to animals, while at the same time consi-
derable experimental work is manifestly being done to stop pain
or to relieve anxiety in animals. Using the term emotion in animals
is often considered unscientific and implicating anthropomorphic
assumptions of human-like subjective experience.
As pointed out by LeDoux [15], a subjective emotional
experience, like the feeling of being afraid, results when we
become consciously aware of an emotion system of the brain,
like the defense system. In other words, to feel afraid it is
necessary to have access to consciousness that allows awareness
of the internal state. Damasio [16] claims that the consciousness
of an emotion corresponds to the knowledge of the ability to
experience emotions. Perceptive (or anoetic) consciousness is
usually defined in terms of the capacity to be aware of feelings,
sensations, thoughts and emotions [17]. There is a vigorous
debate as to whether feelings of c onsciousness are present in
non-human animals, and if so in which species [reviewed by
[8,18,19]]. The common assumption is that the more cogni-
tively complex an animal is, the more likely it is to be con-
scious. However, why should increasing cognitive complexity
necessarily be linked to the emergence of consciousness or vice
versa? The approach of using complex cognitive capabilities as
potential indicators of the presence of conscious experience in
animals has rightly been criticized [reviewed by [20]]. Recently,
Alexandrov and Sams [21] proposed a unified concept of
consciousness and emotions. It is not our purpose to discuss the
links between cognition and consciousness [reviewed by [22]],
but we want to stress the usefulness of cognitive science for
research on emotions in animals and thus for animal welfare.
Feelings require some cognitive abilities to establish temporal
and instrumental contingencies. These abilities allow anticipation
or prediction of events whereas emotional responses involve
coping with the situation. Learning temporal contingency refers to
the ability to relate the occurrence of one stimulus to another when
the two occur in succession, the first one becoming a predictor for
the second. Learning an instrumental contingency refers to the
ability to assess the consequences of one's response to the sit-
uation. In this context, affects provide the common currency
[23] with which animals can balance conflicting demands of
avoiding bad things and approaching better ones, and can evaluate
the priority to be given to one over the other [24].
2.1.2. Cognitive components of emotional processing
All the previously reported studies emphasize the relation-
ship between emotions and cognition. They describe how an
377A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 3
emotional system in the brain is operating and the minimum
mental operation it requires. Cognition has to do with infor-
mation processing, and refers to the range of processes involved
in the acquisition, storage, and manipulation of information
[25]. Whereas the relationship between emotions and cognition
has received little attention in animal studies, it has been ex-
tensively investigated in human psychology. For most psychol-
ogists, there cannot be any emotion without some form of
cognition [reviewed by 26], and several theories stress the
generative role of cognition in the expression of emotional
states. For Lazarus [27], an emotional stat e requires a primary
appraisal of the emotional stimulus, which is responsible for the
action planning and execution; the consequences of this action
are then evaluated through a secondary appraisal. Cognitive
theories of emotion contend that the individual's cognitive
appraisal of the stimulating event deter mines the quality of
emotion [2831] . Although labeled as cognitive, the appraisals
are not thought to result from complex analytical reasoning.
Instead, they are presumed to be relatively effortless, intuitive
and automatic evaluations that are sensitive to events related to
survival (e.g. threat) and opportunities (e.g. forming attach-
ments). Various forms of evaluative processings , ranging from
rapid to subtle, occur to differentiate the emotional experiences.
Appraisal may be influenced by innate automatic responses that
probably evolved over many generations (e.g. response to a
snake), and by learn ing and memory of previous encounters
with stimuli during the individual's lifetime [15].
A simple framework based on cognitive evaluation can be
employed for the study of emotions in animals. Events can be
appraised according to whether they are rewarding or punishing
[12,32]. Presentation of rewards leads to positive emotions such
as pleasure, while omission of rewards leads to frustrat ion or
anger. Likewise, Lazarus [27] suggested two simple steps of
appraisal. The first step appraises whether the current situation
alters the well-being of the individual (if not, there is no
emotional arousal). The second step appraises the meaning of
the alteration. When the alteration reduces well-being, it leads to
a negative emotion whereas when the alteration increases well-
being, it leads to a positive emotion. The behavioral and phys-
iological profiles recorded under these conditions could be used
as indicators of putative emotional states. When a specific
internal state is the only discriminating stimulus available to the
animal, it can use the said internal state as a cue to select the
appropriate operant response in order to get a reward. One might
say the animal perceives the internal representation of its state
within its central nervous system in a way that contributes to or
determines its actions. As a behavioral outcome the animal will
choose what seems most pleasurable [23]. Measures of cognitive
processes, such as anticipation, can provide information about
animal emotion. Indeed, anticipatory behavior measured in a
Pavlovian conditioning paradigm has been identified as a po-
tential indicator of emotion [33].
Recent developments in human psychology of emotions could
help to create a more complete framework for categorizing emo-
tions in animals. According to Scherer [34], emotions are deter-
mined on the basis of a limited number of elementary evaluative
criteria: (i) the novelty of the event, (ii) its intrinsic pleasantness,
(iii) its goal significance, (iv) the coping potential of the in-
dividual, and (v) the relevance to social norms. The novelty of the
event, which breaks down into suddenness, familiarity and
predictability, and its intrinsic pleasantness are used to evaluate
the relevance of the event. The goal significance concerns the
consequences of the event relative to the individual's expecta-
tions. The coping potential indicates how well the individual can
control and adapt to the event. Finally the relevance to norms
concerns how the response fits the personal and social norms.
Scherer [34] postulates that the type of emotion results from the
particular combination of elementary criteria used to evaluate the
eliciting situation. Scherer's model appears useful for trying to
describe in very objective terms the types of emotion that are
accessible to different animals; it makes it possible to infer an
animal's subjective component and to predict responses accord-
ing to the evaluation process [35,36]. Most of the elementary
criteria defined by Scherer [34] are likely to be perceived and used
by animals, and the range of emotions that animals can access
may be estimated by investigating to what degree they can
distinguish among elementary criteria and respond to them. By
presenting animals with test situations for which one (or several)
likely evaluative criterion has been experimentally manipulated,
and at the same time recording their resultant cardiac and
behavioral reactions, direct relationships between presumed
appraisal and measurable emotional outcomes can be established.
Using such an approach, Dési and collaborators [37] have
demonstrated that lambs produce differential emotional responses
to suddenness (startle and tachycardia) and unfamiliarity
(behavioral orientation and increase heart rate variability). In
addition, as would be assumed from the dynamic sequential
organization of appraisal in Scherer's model [34], the combina-
tion of suddenness and unfamiliarity criteria enhances behavioral
and cardiac responses that are specific to one or the other criterion
[38].Theapriorifr amework propo sed by Désiré and
collaborators [36], which directly considers the cognitive
component of emotional processes, helps in mapping behavioral
and physiological responses in animals to particular emotional
states, including positive ones. For instance, an event evaluated as
being pleasant, of moderate predictability, and not sudden should
trigger a positive emotion.
Whatever the a priori framework, it is now time to integrate
theories and techniques from cognitive science in order to
develop new approaches to understanding emotions in animals.
Emotional processing must now be regarded as comprised of
not only behavioral, physiological, and subjective components
but also a cognitive component.
2.1.3. Emotion and mood: acute and long-term evaluations
As was previously reported, cognitive processing is involved
in determining emotional experience. However, there is evidence
in humans that causal links between emotions and cognition
occur in both directions, with cognitive manipulations influenc-
ing felt emotion and emotional manipulations influencing
cognitive processing [39]. For instance, the emotional state can
lead to changes in cognitive functioning such as an increased
tendency to attend to threatening stimuli during fear, or an
enhanced memory for an unhappy event during sadness [40].
378 A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 4
Recently, Paul and collaborators [39] reviewed the variety of
alterations of cognitive processing due to emotions in humans.
They classify these cognitive biases into three categories: atten-
tion, memory and judgment biases. For instance, anxiety has
been argued to represent an attentional shift towards potential
threats. Likewise, negative emotions are associated with a heigh-
tened ability to memorize negative events. Judgment processes
may be affected by current emotions in a number of ways:
emotion may directly alter the judgment or risk assessment
process, or act indirectly through biases in the attention and
memory processes that pertain to it. Such emotional modulations
of cognitive processes can be interpreted to have adaptive value
by helping a fearful or anxious individual to attend to, to memo-
rize and to make judgments about threatening circumstances. The
few studies on animals show that such effects are not restricted to
humans. In rodents, the startle response induced by exposure to a
sudden event tends to occur faster and with greater amplitude
under negative emotional states [41]. Rats housed in conditions
designed to induce a persistent negative mood are less ready to
respond to an ambiguous stimulus as signaling the delivery of a
good event [42]. Likewise, rats housed under social stress or
isolation exhibit a reduced frequency of behavioral transitions
between the presentation of a conditioned stimulus predicting
sucrose reward and the arrival of that reward, indicating a
reduced anticipatory behavior due to judgment difficulties [43].
Developing similar approaches in farm animals could help to
develop new indicators of persistent emotional states (i.e.,bador
good moods), such as an enhanced expectation of positive events
[33]. All these studies demonstrate how cognitive approaches can
be used in animals to probe emotions (as short-term affective
experiences) and moods (as persistent affective experiences).
2.1.4. What are positive emotions?
Except for a few approaches, such as the ones developed by
Wemelsfelder and Farish [44] and by Burghardt [45] that
emphasize positive as well as negative states, studies of neg-
ative emotions are generally far more numerous than those of
positive emot ions. A number of signs can indicate positive
experience in animalsfor examp le, successful coping, re-
ward, and wanting resulting in goal-directed behavior. Positive
emotions can be separated into three temporal categories: i) past
(e.g. post-consummatory satisfaction), ii) present (e.g. pleasant
sensory activity), and iii) future (e.g. positive expectation,
anticipatory joy). Focusing on present positive emotions, three
categories of pleasure have been proposed in humans [4]: sen-
sory pleasures, higher pleasures (i.e. non-homeostatic sensory
experiences), and gratifications (i.e. inner intellectual experi-
ences), with at least the first and presumabl y the second cate-
gories being candidates for animal investigations.
Hedonism refers in its narrowest sense to pleasure or positive
affect [46]. Hedonism can also have a broader meaning referring
to a general affective disposition, either positive or negative.
The presen ce of behavioral and physiological signs of pleas ure
reminiscent of human eu phoria has been demonstrated in an-
imals. Such signs can be induced both as a result of natural
rewards and in models for addiction resulting from treatment
with rewarding drugs such as heroin, other opioids, or dopa-
minergic drugs [47]. Animals that are about to obtain food or are
engaged in play often have a typical behavioral pattern char-
acterized by short, abrupt, and quickly changing movements
(Section 3.4). The behavioral and physiological homology
between natural rewards and drugs is striking. The brain mech-
anisms involved in mediating rewards have been functionally
related to motivational aspects of natural behaviors and be-
haviors of addiction. Classically motivated behaviors can be
divided into two components: appetitive and consummatory
behaviors. Interestingly, the appetitive component has also been
neurobiologically dissociated from the consummatory aspects.
Scientific evidence demonstrates similarity in underlying mech-
anisms for rewards provided by addictive substances and by
natural commodities such as foods (Section 2.2). Preventing a
behavior that is mainly motivated by internal causal factors is a
proposed cause of poor welfare of animals kept in captivity
[4850]. Thus, one might propose an approach to assessing and
improving animal welfare that focuses on signs of pleasure
instead of post hoc signs of displeasur e or of other negative
emotional states. After having drawn the neurobiological basis
of positive affects, the next sections of this review should help to
better assess positive emotional states in animals by validating
behavioral and physiological expressions of positive emotions.
2.2. Neurobiological bases of positive emotions: the emo-
tional 433 brain
2.2.1. Actions on the reward system
There is now substantial evidence that emotions are based on
neural activity in the limbic forebrain, i.e. the prefrontal, insular
and cingulate cortex, the hippocampus, the amygdala and the
septal nuclei. While the cortical areas do not have direct homo-
logues in non-mammalian vertebrates, the others are ubiquitous
in at least all tetrapods [51]. Two sensory systems have shortcut
access to the limbic system: i) olfactory nuclei projecting di-
rectly to the amygdala, and ii) afferents from the medial geni-
culate body, the thalamic relay of acoustic signals [52].In
mammals, all other sensory modalities also reach the limbic
centers, including the amygdala, after being processed in their
respective sensory cortical areas [53]. Emotional states are
distributed among limbic centers. Various emotions have been
attributed to neural activity and the action of several neurotrans-
mitters and modulators in these areas in humans and laboratory
animals. Basically, the main players for positive emotional
states, such as reward and appetitive behavioral motivations and
actions (Section 2.3), are the a mygdala complex and, most im-
portant, the nucleus accumbens septi as a terminal site of the
dopaminergic mesolimbic axis originating in the midbrain
ventral tegmental area (VTA) [reviewed by 54,55]. It is note-
worthy in this context that opioids act on the mesolimbic axis
either indirectly by stimulating dopaminergic VTA neurons or
directly by increasing dopamine at the level of the nucleus
accumbens, supporting the view that mesolimbic dopamine is
closely linked with rewarding stimulation and action [56,57],
including positive anticipatory behavior [33,58]. Berridge [59]
has suggested that, at least in the context of feeding behavior,
the positive emotional valences wanting (appetite, incentive
379A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 5
motivation) and liking (pleasure) can be dissociated, as they
are apparen tly mediated by different transmitter systems in the
brain, that is dopamine and opioids, respectively.
Amygdala activity has been correlated to negative emotions
such as fear and anxiety, where its role is well established [15,60].
In humans the amygdala complex is involved in the understand-
ing of emotional facial expressions [61] and in the perception of
perceived or expected negative stimuli as revealed by imaging
techniques [62]. However, there is now evidence that the amyg-
dala complex is also involved in positive emotions. Pleasurable
music has been found to activate the amygdala and the ventral
striatum as well as orbitofrontal centers in humans [63], and it was
recently reported that in monkeys single amygdala neurons res-
pond in anticipation of fruit reward [64]. Hence, it seems that
subpopulations of amygdaloid neurons exist that respond dif-
ferently to the emotional value of a stimulus. These findings
indicate that this area is likely not a center that mediates judgment
of negative events exclusively. Together with thalamic centers the
amygdala has also been found to be a relay of oxytocin-mediated
social recognition and attachment [65,66]. It now seems evident
that, in mammals, brain oxytocin together with vasopressin plays
an important role in positive social bonding, mainly, though not
exclusively, between mates and between mothers and their
offspring [67,68]. Besides acting at amygdaloid sites, oxytocin
also acts at the mesolimbic axis, the nucleus accumbens in par-
ticular, where it triggers dopamine release in the context of social
bonding [69]. The main structures, neuromodulators, sites of
origin, and actions are summarized in Table 1.
Within the appraisal framework, the hippocampal and para-
hippocampal systems have been found to be involved in the
detection and evaluation of new events that are embedded in
semantic, temporal or local relations [80,81]. The detection of
novelty requires perceptual identification and attempts to match
actual sensory input to stored items. If this matching attempt
remains unsuccessful, the situation is perceptually new and may
demand increased effort. Consequently, coping with enriched
environments requires activity and exercise, and such environ-
ments favor hippocampal neurogenesis and synaptic plasticity
[82,83]. This view is supported by the involvement of the
hippocampal system in the mediation of stress reactions [84],
including the startle response [85]. When there is nothing new
and demanding, however, the hippocampus exerts an inhibitory
influence on the stress axis. More and more neurobiological
evidence has now become available indicating that in fact two
aspects of reward systems must be distinguished. These aspects
relate to the previously mentioned appetitive and consumma-
tory behaviors, which are in turn linked to wanting and liking
respectively. Combined, these two aspects are involved in a
costbenefit analysis that subjects make in relation to rewards
and may reflect the effort that individuals are willing to exert to
collect rewards [86].
2.2.2. Dopamine reflects intensity of wanting while opioids
determine what is wanted
The meso-accumbens dopamine system is involved in the
disposition to act upon previously liked rewards, i.e. in wanting
[47]. Behaviorally, wanting is expressed as changes in transitions
of goal-directed behavioral patterns, such as locomotor and
investigatory behavior, prior to obtaining rewards. These patterns
are labeled as anticipatory or appetitive behavior [33].Depending
on species and test conditions, behavioral changes either increase
or decrease [rats: [43,58],mink:[87]]. These changes may be
related to the role of dopamine in integrating cue- and context-
associated details of the response to be displayed or in other words
integrating neuronal inputs from the hippocampus and amygdala
in the output of the nucleus accumbens [88].
The meso-accumbens dopamine system has a high degree of
plasticity, which underlies its role in the organization of the
efficacy of behavior. Highly rewarding incentives may sensitize
the system, since it seems adaptive in the long term that what was
once liked to a certain extent should be wanted in the future to
the same extent. A seemingly contradictory finding is that stress-
ors such as short-lasting social isolation or poor rearing conditions
[58] also increase anticipatory activity for sucrose as a reward due
to the fact that stress sensitizes this system [33,89]. Hence it seems
that under conditions of apparent loss of control animals must
increase their appetitive attempts to improve the perceived affec-
tive state.
Stimulation of mu-opioid receptors in the VTA, the origin of
the meso-accumbens dopamine system, enhances dopamine
release in the accumbens [reviewed by [90]]. It was shown in
rats that the expression of anticipatory activity to rewards is
blocked by the mu-opioid antagonist naloxone [33, 91].This
point was further demonstrated in mice lacking the mu-opioid
receptor in a schedule-induced food-anticipation behavior par-
adigm [90]. These data show that the meso-accumbens dopamine
system is under control of the opioid system, and suggest that the
opioid system is important in adjusting the activity of the meso-
accumbens dopamine system to challenges from the environment.
Opioids in a low dose stimulate the onset of motivated behavior.
They elicit feeding [92], social interaction, and play in young
animals [93], depending on the context. By their stimulation of
motivated behavior and simultaneous suppression of pain etc.,
they allow the animal to start obtaining commodities. The opioid
system is thus related to the consummatory, liking aspect of
motivation, whereas dopamine, which is the interface between
emotion and motion, would be a system mediating motivation for
reward in a broad sense, including relief from stress, rather than
reward per se.Incontrasttostatesofmotivation,whererewardis
sought and eventually gained, very little is known about the
neuronal basis of satisfaction. In mammals this state of fulfilled
Table 1
Structures and neuromodulatory substances involved in positive emotions
Origin Substance Site Effect References
VTA Dopamine NAc Expect reward [7073]
Act for reward [70]
NArc, Pit μ-Opioid VTA,
Facilitate dopamine response [74,75]
Pit Oxytocin NAc Facilitate dopamine response
Promote social behavior
CA Inhibit negative affect [7679]
CA: central amygdala; NAc: nucleus accumbens; NArc: nucleus arcuatus; Pit:
pituitary; VTA: ventral tegmental area; μ-Opioid: receptor agonist.
380 A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 6
needs and cessation of drive for further rewarding sensations
seems to be characterized by reduced activities in a considerable
network of forebrain structures, including the hypothalamus,
amygdala, and frontal and insular cortex [94]. Shedding more
light on the neuronal base of this important state of welfare will be
necessary in future in order to have the full picture of well-being.
2.3. Behavioral aspects of positive emotions
2.3.1. Behavioral needs
Being aware of the fact that various perspectives on motivation
exist, here we follow the perspective described in behavioral
neuroscience, which is strongly supported by ethological and
neurobiological evidence. In the literature on reward, addiction
and affect [47,19], a close link has been made between neuronal
structures and the components of motivation (e.g. appetitive or
consummatory) [33]. The function of reward systems could be to
weigh costs versus benefits of obtaining a commodity. The
outweighing of costs by benefits would coincide with the ex-
perience of pleasure. A number of needs are dealt with auton-
omously, such as temperature, heart rate etc. However behaviors
that need integration in space and time and that use previously
acquired sensory information require focused attention of the
highest emotional and cognitive brain centers. Selection of the
most pressing need is executed by reward systems based on
expectations. Ventral tegmental dopamine has been called the
interface between motivation and movement [95] and links the
specific requirement of a particular need with expected sat-
isfaction of that need and activates higher cognitive processing.
At least in humans one might say that the reward appraisal centers
in combination with the nature of the need are experienced as
affective consciousness. Affective consciousness probably occurs
in animals as well though it will depend upon the cognitive
capacity of the animal [19].
Two types of behavioral needs can be distinguished. First,
needs that induce behavioral patterns can be evaluated by their
direct physiological consequences for the underlying motiva-
tional system. For instance, hunger requires eating of food,
which affects blood glucose, leptin levels and other parameters
representing the metabolic status of the individual and in-
fluencing food regulation centers in the brain. Second, Jensen
and Toates [96] defined a behavioral need as a specific behav-
ioral pattern that the animal apparently has to perform, irres-
pective of the environment or of the animal's physiological
needs. In other words, needs that involve the display of typical
behaviors lack direct physiological consequences and thus also
lack direct feedback from regulatory physiological mechanisms.
It seems as though the biological relevance of regular per-
formance of those behaviors lies in their long-term benefits for
the individual animal or its offspring, e.g. behaviors concerning
reproduction, foraging and grooming. The motivation to per-
form these essential behavioral patterns is governed by the
display itself rather than to meet short-term physiological needs
[48]. The mechanism underlying the regular display can be
understood when it is assumed that the display itself, rather than
its immediate effects, has rewarding properties. Long-term ef-
fects are then a probable outcome of the display, but beyond
what can be monitored by the individual animal. The impact of
the display on reward guarantees its regular occurrence. The
absence of such rewarding behavior coincides with the absence
of an important source of reward [33].
2.3.2. How can behavioral needs induce pleasure and why do
they make the animal vulnerable?
Appetitive behaviors with low probability of success or with a
long time interval between onset and eventual consummatory
responses would benefit from a regulating mechanism like that
just described for behavioral needs. The animal can never stop
looking for food, safety, and reproductive opportunities, even
following unsuccessful attempts; in other words investing for the
future must have direct rewarding consequences. Behaviors,
which serve as an investment for collecting a necessary incentive,
e.g. food or social companionship, may be self-rewarding in order
to maintain these types of searching behaviors. Other examples
are hunting in predators, exploration in a number of species,
rooting in pigs, etc. Appetitive or seeking behaviors (motivational
states of wanting) have been associated with mesolimb ic
dopamine, whereas consummatory behaviors (liking or hedonic
motivational states) have been more associated with opioids
[97,19,47,33]. Social behavior [98,19], reproductive behavior
[98],play[99],autogrooming[100] and dust-bathing [101,102]
are supposed to have rewarding properties.
To be self-rewarding, a specific behavior shoul d activate the
mesolimbic system, for instance by the release of endorphins,
which then exert a positive feedback on this beha vior, resulting
in a longer lasting, on-going behavioral bout. Taking this process
into account, one would expect that opioids may induce such
behaviors, while opioid antagonists would counteract their
display. One would expect self-rewarding behaviors to interact
cross sensitization and cross tolerance with drugs of addiction
or rewarding stimuli such as food. The opportunity to display
may itself act as a reinforcer and can be used as an incentive in
conditioned place preference studies. Being deprived of oppor-
tunities to perform specific behaviours resulting from these
behavioral needs produces signs of withdrawal paralleling those
observed in deprivation of an addictive drug [19,23,47,95,98].
Exploration in rodents, when modeled by wheel-running
behavior, shows such interaction (cross sensitization) with
mesolimbic dopamine and cross tolerance with morphine [103],
an interaction similar to those observed for drugs of addiction.
Cross sensitization between food and drugs of abuse has also
been shown, and this phenomenon could be counteracted by
an opiate antagonist [104]. Social behavior, play and self-groom-
ing can be induced by endorphins and counteracted by naloxone
A drawback of this efficient mechanism regulating the
motivation of those behaviors that represent animals' behavioral
needs is that self-rewarding behaviors may run out of control
when dopaminergic systems are sensitized by stress [105107].
If some stressors amplify animals' wanting for rewards when no
external rewards are immedi ately available, they may instead
look for compensation by performing self-rewarding behavior
[108,109]. Thus, depending on the species and circumstances,
stressed individuals may display excessive locomotion, sexual
381A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 7
activity, or self-grooming. Because highly motivated behaviors
are so rewarding, being deprived of them is a severe stressor. At
the same time some of them do not have an immediate high
priority because they have a long-term payoff. Play for instance
only occurs when other needs are met. Therefore, occurrence of
play behavior or typical signs of satisfaction indicate that the
animal is not deprived of important sources of pleasure and that
other needs are being met. That is why such behaviors are good
candidates for welfare indicators (Section 3.4).
2.4. Physiological markers of positive emotions
2.4.1. Sympathet ic and parasympathetic balance
The autonomic nervous system (ANS) is an important
efferent of the limbic system. It is primarily amygdala output,
acting via hypothalamic centers, that influences autonomic
brainstem nuclei controlling the activity of sympathet ic and
parasympathetic systems. The relative level of activity of these
two systems is an indicator of an organism's general
physiological state (action or digestion). Though this relative
activity depends on homeostasis, it is also determined to a great
degree by current emotions. Hence, measures of heart rate (HR)
are broadly accepted as a valid index of stress in many
vertebrate species as they give an impression of the activity of
the ANS [110112]. Interpretations have often been based on
the assumption that HR reflects the activity of the sympathetic
branch of the ANS and therefore can be used as an indicator of
the sympatho-adrenomedullary stress response. However, the
heart is also under parasympathetic control, and HR at every
point in time is the result of non-additive effects of the inter-
action between the two branches of the ANS [113]. Thereby, the
effects of emotional perceptions on cardiac activity are me-
diated via direct innervations of the sino-atrial node of the heart
by sympathet ic and parasympathetic fibers [114,115].
As a consequence of these permanent regulatory functions,
HR is never constant but varies from beat to beat, hence the
concept of heart rate variability (HRV). HRV is mediated by either
an increase/decrease of sympathetic tone or an increase/decrease
of parasympathetic tone. Hence, analyzing HRV helps to identify
which branch of the ANS is actually mediating heart rate and to
assess the sympatho-vagal balance of an organism [116].The
assumption underlying the assessment of HRV is that, as a pe-
ripheral measure of ANS activity, it serves as a proxy for vagal
tone, vagal reactivity and sympatho-vagal balance that otherwise
could only be measured by hard-to-perform invasive techniques
[115,117]. A certain degree of HRV is a characteristic of any
healthy vertebrate [118].
In recent years, cardi ac vagal tone assessed by HRV-analysis
has received considerable attention as a psychophysiological
marker of emotion regulation, and of certain aspects of psycho-
logical adjustment in humans [reviewed by 119,120]. Most of
the work has focused mainly on negative emotions and emo-
tional disorders, which have been referred to as anger-prone-
ness, irritability, irritable distress, and negative mood. However,
cardiac vagal tone may also be an indicator of positive emo-
tional states. Fox [121] found a near significant correlation
between infant HRV and positive reactivity during a peek-a-boo
procedure with mothers and strangers. McCraty and colleagues
[122,123] reported higher HRV in human adults after self-
induced positive emotional states. Recently, Lee and collabora-
tors found higher levels of HRV and reduced HR in humans
after the intake of a traditional herbal remedy [124] and during
Qi therapy, a Chinese external therapy [125]; this effect was
correlated with more pleasant and calm emotions compared to
the placebo group. Rainville and collaborators [126] found
evidence that distinct patterns of peripheral physiological ac-
tivity (e.g. HRV) are associated with different emotions like
sadness or happiness. Recently, patterns of HRV reduction have
been studied in farm animals in reaction to stressors [127129],
behavioral disorders [130] and in the context of cognitive ap-
praisal [37]. For a broader discussion of this topic see also the
review on HRV by von Borell and collaborators in this issue.
The potential elements and patterns of autonomic activity
have not been exhaustively examined. Moreover, it is possible
that emotions or components of emotions could be differentiated
on the basis of sympathetic and parasympathetic innervations of
the viscera rather than on visceral responses per se [131]. How-
ever, taken together the data indicate that monitoring activity of
the autonomic nervous system, essentially through heart rate and
its variability as suggested especially by human studies, may
deliver suitable approaches to assess positive emotions in
animals when combined with behavioral records [36].
2.4.2. Neuroendocrine activation
The physiological responses of the autonomic nervous and
neuroendocrine systems trigger neuronal processes in the brain
that influence the way animals cope with the situation. For
instance, sustained high cortisol levels are associated with a
higher probability of engaging in passive coping (i.e. being
reactive), whereas high catech olamine levels are associated with
a higher probability of engaging in active coping [132].However,
transient activations of the stress axesi.e. the adrenocortical axis
(HPA) and the ANSare regularly coincident with any emotion-
ally triggered activity and cannot be easily differentiated for
positive and negative emotions. They might as readily be
activated for actions of defense as for reward. Accordingly, in
a comprehensive study in pigs it has been demonstrated that
individuals with a high success in a dominance intruder test
displayed dynamic peripheral responses of the HPA that were
virtually the same as in losers [133].Roughlythesamewas
revealed in the ANS, however with a tendency toward a more
rapid decrease after fighting in the high success animals. Further,
winners tended to have a higher noradrenalin/adrenalin ratio
which, in general, could reflect a less passive behavioral pattern
[134]. Hence, intrinsic processes (
e.g. fluctuations of hormone
concentrations, induced spontaneously or by chronic emotional
stress) may cause changes in mood, while sensory inputs may be
able to release emotions.
Although experimental evidence is still scarce concerning
changes in physiological parameters that accompany positive
emotional experiences, some recent studies with human subjects
reveal promising leads to follow in animal studies. Taking
advantage of recent developments in salivary assay methods,
physiological studies of human saliva suggest that its protein
382 A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 8
composition could be used as a monitoring tool for the assess-
ment of some emotional/mood states. Indeed, while negative
emotions are associated with a decrease in 20200 kD proteins,
in positive emotions an increase in these proteins is observed
[135]. It has also been found that salivary alphaamylase (sAA)
levels in humans are increased in response to physiological and
psychological stress [136]. Furthermore, it is suggested that the
level of this enzyme could be considered as an index of soothing
effect or relaxation in humans as sAA levels are significantly
decreased by so othing/pleasant video viewing [137].The
detailed mechanism of the amylase decrease remains unclear.
However, developing such measures in animals could offer a
relevant tool to assess positive emotional states in animals.
2.4.3. Immunological activation
An important literature exists regarding the emotional
influences on immune functions, supported by both human
and animal studies. Indeed, not only is there an extensive
literature on the complex interactions between the HPA-axis and
the immune system [138], but also emotional limbic activity has
been found to have important effects on immunity [139].In
pigs, positive and negative psychological experiences affected
immune system parameters antagonistically, depending on the
success of coping [140] . Hence, immune status may be another
indicator of long-lasting emotional state (or mood), and immu-
nological parameters may be indicators of frequent positive (or
negative) emotions [141]. However, wi th the present state of
knowledge, they are difficult to interpret since the various
immune factors constitute a multidimensional vector whose
components are mutually interacting and can be easily in-
fluenced by non-obvious infections. Unfortunately, experimen-
tal evidence linking positive emotional experiences to the
immune system is lacking. It is therefore risky to infer that
positive emotional stimuli would enhance immune defenses.
However, a recent study found that pleasant stimuli in human
(i.e. humoristic story-telling) induced an increase in secretory
immunoglogulin-A (s-IgA) [142]. This confirms results from a
pioneering study by Dillon and collaborators [143] in which the
salivary IgA concentration was specifically increased after sub-
jects viewed a humorous video sequence. Since s-IgA is con-
sidered an indicator of overall immune function, such measures
in laboratory and farm animals could offer new insights for the
understanding of emotional influence over immunity.
Conversely with brain-to-immune influences, growing
evidence reveals that the immune system influences brain func-
tions, including structures involved in emotional processes. In
addition to various effects on emotion-relat ed neurotransmitters
(i.e. serotonin or noradrenalin), immune activation is known to
induce HPA activation [138]. When these data are combined
with the fact that immune and stress responses have been argued
to share common evolutionary defensive properties, immune
activation is increasingly considered as a stress response by
itself [144]. Moreover, some studies show that immune acti-
vation has emotional effects, such as anxiety- or depression-like
behaviors [141,145].
Nevertheless, it has recently been shown that rats made sick
through the peripheral administration of endotoxin are still able
to response positively (i.e. hedonic facial responses) to palatable
taste solutions such as sucrose [146]. However, when subjected
to mixed bittersweet solutions, these rats display more aversive
and fewer hedonic responses than do controls [146]. Therefore,
immune activation cannot be considered to induce anhedonia as
had been reported in earlier studies. Rather, these data may
illustrate alliesthesia, i.e. modification of the affective value
(i.e. valence) of a specific stimulus related to the subject's
physiological status [147]. In other words, it has been proposed
that instead of causing a negative emotional state, immune
activation would enhance reactivity of subjects to salient
negative features of a given situation, thus motivating their
defensive repertoire [148,149]. Inflamm atory markers could
therefore be used as possibl e indicators of a transitory
vulnerability of subjects to exter nal sources of stress, and
immune activation could be used as a tool to evaluate the
plasticity of the valence (e.g. from pleasant to aversive) of
specific features of each subject's environment [150].In
conclusion, even if a lot remains to be done to fully validate
their use and relev ance, heart rate variability as well as alpha-
amylase and immunoglobulin-A levels could be interesting
physiological parameters for assessing positive emotional states
in animals, at least as complements to classical adrenocortical
and/or inflammatory measures.
3. Application of positive emotions in animal welfare
Increasing research in the area of positive emotions, in com-
bination with press ure from society that animals should have a
good life and not just a not-so-bad
life, is leading to a greater
aspiration to apply some of the knowledge described in the first
part of this review. The most likely practical applications fall into
four broad areas: i) promoting positive experiences as a way to
give animals a generally better quality of life; ii) enhancing long-
term positive emotional states; iii) highlighting the link between
positive emotional states and improving health, and iv)
including criteria of positive welfare in on-farm monitoring
3.1. Enhancin g quality of life by promoting positive experiences
Here we outline new approaches that might be useful in
animal husbandry for inducing positive emotions in animals
throughout their lives. Three approaches can be defined-signaling
or predicting a reward in advance, giving a higher reward than
expected, and enabling the animals to cope with or to control the
3.1.1. Positive anticipation
Investigations of what animals find positively reinforcing
and of the animal's behavioral expression during anticipation or
expectation of the rewarding result provide a basis for assessing
positive emotional states in animals [33]. The period between
the signal and the arrival of the reward is the period in which
behavioral activity related to dopaminergic activity can be seen
(Section 2.3), i.e. anticipatory behavior. Anticipatory behaviors
are consi dered to represent biologically significant preparatory
383A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 9
behaviors [151]. When anticipating food rewards, rats [58],
mink [152] and pigs [153] show increased locomotor activity
and frequent behavioral transitions, i.e. anticipatory hyperac-
tivity. More specifically, the behavioral categories reported in
rats during anticipation of sucrose rewards include locomotion,
grooming, alertness, arousal, scanning, attention, exploration
and running [58]. In pigs, signaled positive stimuli such as extra
space, food, or straw lead to increased activity due to orientation
toward the place where the reward is offered and frequent
behavioral transitions, most often combined with play markers,
i.e. hopping, scampering, pivoting, pawing, flopping, and head
tosssing [153]. During anticipation of feed rewards, farmed
mink showed a general increase in locomotor activity, and they
spent more time moving in/out of the nest box, less time in the
nest box, more time nose-p oking around the feeding area, and
more time standing and scratching at the cage door [152,154].
Farmed silver foxes anticipating a food reward or playing
incentive were more active, had more erect ears and higher body
carriage, and spent more time in front of the cage compared to
animals that were anticipating aversive stimuli [155]. The
predictability of the rewarding stimulus plays a role in
behavioral expression during anticipation. Silver foxes antici-
pating a positive predictable reward (food or play) had more
upright ears, whereas for a positive unpredictable reward the
ears were more backwards rotated and flat [155].
Sensitivity to dopamine is adapted to the impact of reward
and the needs of the animal in the manner that is related to
previous experience. Thus, the threshold of the animal to react
to a stimulus reflects the sensitivity of the system. In addition,
the amount of activity is related to the needs of animals. For
instance, if housed in a poor cage environment, rats were more
sensitive to signaled positive rewards compared to rats housed
in an enriched environment [58]. However, there is a risk of
misinterpretation here as the absence of a reaction to a reward
may indicate that the animal is in a state either of full sat-
isfaction or of anhedonism, and these are opposite states in
terms of welfare. In addition, the intensity of reaction to a small
reward has to be calibrated for each species and situation. In
silver foxes, anticipatory behavior consisted of high levels of
locomotor stereotypical behaviors such as running back and
forth in the cage environment [155]. Whether stereotypies
during positive anticipation are an expression of enhanced ex-
pectation shaped by the environment into repeated movements
or a stereotyped behavior due to previous and/or current frus-
tration [156] remains to be seen.
3.1.2. Positive contrast
In a typical positive contrast experiment an animal is first
trained to perform a task by giving it a reinforcer of a certain
size. The animal is then given a larger reinforcer than they had
been given before. If the animal changes its behavior so that its
response is faster/more vigorous than that of a control group
given the larger reinforcer from the start, a successive positive
contrast (SPC) has occurred. In the same way, an animal that has
been trained using a large reinforcer, and then shifted to a small
one, might react with a decrease in respon se, bringing it lower
than an animal that has always received a small reinforcer,
demonstrating a successive negative contrast (SNC) [reviewed
by [157]]. SPC and SNC can be found for both appetitive and
consummatory responses. It is the difference between the
expected value and the actual value that influences a subject's
behavior rather than the exact level of reinforcement. If the
animal has a previous experience of a higher value reinforcer,
the reaction to a given level of reinforcement is much less
pronounced. Any contrast effect is often temporary and dis-
appears over time [e.g. [158]]. The fact that the appearance of an
unexpected reward makes such a difference to the behavior of
the animal has led several authors to suggest that positive
contrast corresponds to the emotion elation in humans, while
negative contrast might be related to frustration or disappoint-
ment [e.g. [159,160]].
Overall there are more studies showing a negative than a
positive contrast. Negative contrast seems to be a more robust
phenomenon and is easier to detect. This is also possibly due to a
ceiling effect; for instance, an animal that is running at top speed
for a reinforcer is unable to increase its speed even if the
reinforcement is increased. Unfortunately this result has meant
that the majority of theories of the contrast effect have focused on
explaining the negativecontrast effectexclusively. Although some
of the theories involve emotions e.g. elation or disappointment as
described above, most are alternative non-emotional theories
such as generalization decrement [161], incentive averaging [162]
or exploration [Elliott, 1928 cited by [157]]. There are never-
theless a number of more specialized emotional theories that do
not rely on a general emotional drive, e.g. Amsel's frustration
theory [163] or Gray's disinhibition theory [29].Positivecontrast
has been found in few species of mammals and birds [164166].
For instance, in his original study, Crespi
[159] reported that rats
for which the quantity of reward has just been increased run faster
towards the food reward than do rats having always received the
large quantity of reward. However, the changes in the behavioral
responses are generally more pronounced for the negative than for
the positive contrasts [164166]. Due to the relatively small
number of studies, further research is needed to confirm the value
of a positive contrast model in the study of potential positive
emotions and their behavioral expressions.
3.1.3. Coping and controllability
The possibility of controlling the environment and coping
successfully with challenges may be another source of positive
emotions. Despite some degree of stress being necessary in the
initial state of coping to activate alertness and metabolism [167],
successful actions with a positive outcome make the animal
master of the environment. Successful coping stops the initial
stress reactions, independent of the positive or negative value of
the stressor, while un controllable and, hence, persistent negative
stressful stimul i lead to the well-known detrimental effects of
chronic stress [84,168170]. A prerequisite of controllability is
predictability, i.e. the environmental reactions are (mostly)
contingent on the subject's actions. If the environment where a
species lives is sufficiently constant during a long period of
time, successful behavioral coping strategies may well be
acquired during the process of evolution and, then, are innate at
the individual level. If there is no innate coping behavior, an
384 A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 10
animal may try to learn how to deal with a challenge. Success
depends on a number of prerequisites. First, the animal has to
have the motor patterns necessary to be able to deal with the
challenge with at least some success; eventually these patterns
can even be shaped to deal with it better. Second, there must be a
kind of random action selector to try out what could be helpful
when dealing with the challenge. Third, in the central nervous
system there must be the ability to link the sensory and
perceptual cues that are coincident with successful behaviors
and the challenging context. Fourth, there must be contiguity
between the challenge and the behavioral action [171]. If any of
these conditions remain unfulfilled the animal will be unable to
learn how to cope, a failure that will eventually lead to chronic
stress. Hence, inactivity displayed by an animal may be caused
by two emotional states: total equilibrium, when everything is at
or close to optimum, or learned helplessness, if the subject has
the experience that there is no possibility to escape the negative
environment [172]. The result is an inactive, depression-like
chronic stress state that is often accompanied, inter alia,by
permanently increased levels of the stress horm one cortisol.
Social and physical challenges can be perceived as positive
experiences when exists an adequate coping behavior. Hence,
frequent challenges that can always be successfully mastered and
eventually enable the animal to reach a desired and rewarding
goal may be suitable means to regularly evoke positive emotions.
Using an approach of this type, a program of research on the
effects of rewarded cognitive processes has been developed [173].
In the studies conducted thus far, each animal out of a group of
pigs had to learn an individual acoustic signal as a call to work for
food by pressing a button. The demands of this situation were
threefold: The animals had to be attentive for their individual
acoustic summons, they had to localize its source, and they had to
respond to a button with their nose disc in order to be rewarded
with a small portion of food. It was shown that the pigs were
willing and quite able to cope with these demands. Using this
approach it will be possible to analyze the physiological and
behavioral effects of a complex but predictable environment pre-
senting positive challenges with which individuals are able to cope.
3.2. Enhancing long-term positive emotional states
3.2.1. Concept of temperament
The emotional sensitivity or temperament of an anim al has
an important influence on its welfare, and so predisposition to
some positive emotional states is a good premise for inves-
tigating long-term positive emotional states in animals. The
terms and concepts used for defining long-term emotional states
differ in the degree of structure they provide, from indi vidual
differences to the concept of personality.
The distinction according to the species and the age of the
individual between temperament (for animals and human in-
fants) and personality (for children and adult humans) has not
been maintained consistently in the literature [174]. Tem-
perament can be defined as the characteristics of individuals that
describe and account for consistent patterns of feeling and
behaving [175]. Whereas dimensions of personality have gen-
erally been discussed in terms of either coping style or fearfulness,
there is currently no method for assessing a propensity to feel and
express positive emotions. However, by identifying long-term
positive emotional states we can examine whether characteristic
behavioral patterns could be linked to particular neuroendocrine
states and if they correlate within individuals and over time. From
such studies we might be able to predict responses of an animal in
one situation based upon its responses in another.
There is no universally accepted definition of individual
variation or individual differences. Erhard and Schouten [176]
structured individual variation in animal behavior using the
Eysenck [177] description of personality. This approach orga-
nizes the different aspects of personality into three levels
state, trait,andtype. The term state or mood is used for the
behavior that an individual performs at a specific moment in time
and in a specific situation. If an individual is repeatedly found to
be in similar states in similar situations, we can make assumptions
about the underlying pers onality trait (e.g. fearfulness or
happiness). Regarding the third level, when several trait
dimensions are linked in such a way that an individual's position
on one dimension predicts its position in another, the individual
can be categorized by their position in a type dimension.
This approach brings the study of individual differences in
non-humans closer to what is done in humans [39,176,178].
Different studies have been carried out to evaluate emotional
states in farm animals. For instance, from a combination of the
responses to fear and learning tests, three emotional states were
established in dairy sheep: non emotive ewes (calm, fear
sustainable, reacting adequately in learning tests), emotive ewes
(nervous, fear susceptible, reacting inadequately in learning
tests) and intermediate ewes [179]. It was further established
that the emotional state influenced various adaptive behaviors.
The non-emotive ewes revealed a stable maternal care, whereas
the anxiety increased in the emotive ewes on the second and
third days postpartum [180]. Furthermore, by contrast to the
emotive sheep, non-emotive animals usually successfully col-
lect different rewards (appetitive and consummatory behaviors)
ensuring successful reproduction and higher production [181].
Personality may have different dimensions. This idea was
supported by Kilgour and collaborators [182] showing different
components of personality in cattle and sheep adapted from
human studies: general agitation and avoidance. The hypothesis
that calm animals produce better quality milk than nervous
animals was supported by the result for protein concentration
[183]. Merino ewes of calm temperament can provide better
quality milk, indicating that temperament should be considered
when selecting dairy animals. Martin and collaborators [184]
propose a strategy aiming to maximize offspring survival by a
combination of management, nutrition and genetic selection for
temperament. From a divergent selection on behavioral reac-
tivity, calm ewes are found to be better mothers than nervous
ewes [185]: they spend more time with thei r lambs, have a short
flight distance when disturbed and return to their lambs faster
than nervous ewes. In addition to postnatal lamb survival,
temperament may influence other aspects of the reproductive
process, all of which might be improved by genetic selection for
calm temperament. These include the length of the estrous cycle
[186], ovulation rate [187], the proportion of ewes mated, and
385A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 11
sexual behavior [188] . Finally, temperam ent seems to affect
other aspects of production such as growth rate [189191] im-
mune function [191],milkyield[192] and meat quality [193,
194]. All of these outcomes could be used to improve productivity
while giving the industry an ethical quality [reviewed by [184]].
3.2.2. Genetic background
Large differences in emotional responses have been reported
between breeds in various farm species [195]. In dairy cattle, sire
had a significant effect on the reactivity of cows in the milking
parlor [196]. Similarly, sheep sired by Romney sires were more
reactive when placed in a fearful situation than ewes sired by
Suffolk or Colum bia rams [197]. Genetic analysis revealed that
such emotional responses were due to direct genetic differences
and less to direct maternal influence or to heterosis. The esti-
mates of heritability of fear in domestic animals seem to be
sufficiently high to allow further selection on fearfulness in
attempts to improve herbivore management [reviewed by [198]].
In addition, the development of marker-assisted selection can be
useful for psychological states such as fearfulness. Identifying
quantitative trait loci for fear has recently been extended to cattle
[199] So far , QTL for reactivity to humans have been identified
using linked markers in cattle [200,201]. Such linked markers may
be used for marker-assisted selection within sire families once the
relationship between the marker and the gene has been determined.
The interactions between the genotype of animals and their
environment must be seen as a critical determinant of animal
welfare that could reconcile production with ethical concerns [198].
3.2.3. Ontogenet ic influences
Abundant data demonstrate that stimulation occurring during
early life may markedly modulate the development of subs-
equent emotion processing in a positive direction. It has been
demonstrated that some postnatal stimulation have opposite
effects to those induced by prenatal stress. For example, adult
rats that have recei ved substantial positive sti mulation during
infancy, such as environmental enrichment or regular handling,
are less emotionally reactive to various challenging situations
[202204]. Handling animals during infancy also leads to
improved endocrine reactivity in adulthood. For instance, early
handling alters neuroanatomical development in rats by in-
creasing the n umber of hippocampal glucocorticoid receptors or
by decreasing CRF mRNA implicated in endocrine regulations
occurring under stress [205,206]. Extensive research on the
influences of early regular handling has been conducted in domes-
tic animals. For instance, early handled ungulates are generally
less fearful of humans than their non-handled counterparts, a
pattern shown in goats [207],horses[208],cattle[209],sheep
[210] and pigs [211]. Experiences that occur during adulthood
may also influence emotional responsiveness, particularly
through classical processes of learning and habituation. The
exposure of adult animals to enriched environments or handling
procedures has been shown to decrease subsequent fear, although
the reported effects are generally less marked than those induced
by the same treatments occurring during infancy [212].
Taken together these findings indicate that early stimulation
and later experiences interact with genetic background through-
out the life of the individual to continually modulate emotional
reactivity by shaping tendencies to feel secure. Although not all
aspects of personality in animals are clear, we can state that
emotional traits do exist in animals since emotional reactivity
was found to be consistent across time and situation [176,213].
Although most of the studies have been done to reduce negative
emotions, the study of emot ional traits may also provide good
tools to better enhance long- term positive emotional experi-
ences. For instance, social play behavior was more often seen in
mink selected for low fearfulness than in those selected for high
fearfulness [214] .
3.2.4. Enriching the environment or just reducing stressful
Environmental enrichment is often proposed as the way to
enhance positive experiences in farm and laboratory animals.
Unfortunately, most studies on environmental enrichment are
actually only adding resources or features to an impoverished
setting [215], suggesting the term
housing supplementation
might be better for such practices [216]. The beneficial effects of
supplementation are usually represented by a reduction in the
indicators of poor welfare (e.g. fewer stereotypies, less aggres-
sion) rather than an increase in indicators of good welfare. It is
argued that the term enrichment should be reserved for en-
vironments that are truly enriched beyond basic needs [217],and
it is this meaning of the term that is adopted here. This enhanced
quality of life could be achieved by physically enriching an
already varied environment with new resources or by cognitive
means. The methods for achieving such enrichment are presented
throughout this review, i.e. by creating a situation where there is
anticipation of a positive reward, by offering more space to
promote play, and by providing opportunities for positive contrast
situations, for improving coping abilities, and for information
gathering. Some concrete examples of where these methods have
been tried are given by Moe and collaborators [155] and Ernst and
collaborators [173], and other examples include the work by
Wood-Gush and Vestergaard on seeking novelty and its relation to
play [218], problem solving in cattle [219] and stroking animals
[220]. Nevertheless, in all cases it is necessary to confirm that the
enrichment is having the desired effect, which is not as easy as
might be proposed. At present, the enhancing effect is usually
taken to be an increase in those behavior patterns scientists think
are positive. However this approach entails a risk of circularity in
reasoning, i.e. we change an environment to promote exploration
and play, and then say it is enriched because we see more ex-
ploration and play.
Another practical application of positive emotions in enhanc-
ing overall quality of life is that they can be used to prevent or
reduce the development of negative emotions, so leading to a net
benefit. This effect may be achieved by associating an event, e.g.
taking a blood sample, with a positive one, e.g. getting a food
reward, [221] or by making negative events, e.g. a veterinary
procedure, less negative beforehand by habituating the animal to
being handled [222]. Although these are examples of reducing
negative emotions rather than enhancing positive ones, the net
benefit to welfare is positive. If this argument can be used to
reduce the effects of short duration negative events, the question
386 A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 12
of whether positive events can be used to offset negative ones over
the course of a lifetime can be raised. Although theoretically
justifiable using a preference utilitarian philosophical approach,
we propose that this summing up of earlier positive experiences is
not applicable to animals living mainly in the present as do most
farm animals. Neither should we justify poor long-term housing
and management because we sometimes give animals short-term
positive experiences. If we want to maximize welfare, then the
aim must be to give the animals as good quality of life as possible
all the time. Since it has been shown that anticipatory behavior
followed by the consum mato ry act has beneficial effects
compared to the provision of the mere reward alone, there is the
possibility to manage animals in a way which stimulates the
activation of these systems, while at the same time giving as good
housing and management as we can. The caveat is of course that a
certain degree of sensitization is necessary for induction of
pleasure but not to the extent that the period of need for reward
outweighs the satisfaction. This balance should be explored
further. Promoting play in young animals and stroking animals (or
alternatively promoting social licking) are also probably feasible
methods to promote positive emotions in practice and so enhance
overall quality of life in systems that are usually regarded as
acceptable from an animal welfare point of view. Finally, in
relation to the active behavioral expressions of positive
emotions (Section 3.4), like play, allogrooming and exploration,
another class of behavioral expressions of positive emotions
should be mentioned here, namely those expressing full comfort
and satiation, i.e. the fulfillment of biological needs.
3.3. Inducing positive emotions for improving health
Although research in this area is just developing, there are
already some data showing that positive emotions may improve
health. Pioneering researchers suggest that there are health
benefits in humans of humor and laughter as positive-emotions
promoters, especially in coping with a diagnosis of cancer [223].
Although the scientific evidence for this claim is inconclusive
[224], humor is being increasingly used as a mean of reducing
stress and making patients feel better [225].Obviously,thereare
no such studies in animals, but as it is documented in the next
section, play and tickle-induced ultrasonic vocalizations have
been identified in rats and interpreted as sharing evolutionary
relations with primitive laughter in human [226]. Nevertheless,
poor health is a well-accepted indicator of poor welfare in
animals, and improved conditions for animals usually lead to
improved health. It is logical to suggest, therefore, that giving
increased opportunities for positive experiences might also have a
positive effect on animal health. There is evidence that pigs given
the opportunity to show successful adaptation by rewarded
cognitive processes recover more quickly from a standard biopsy
[221] and have improved carcass quality [227].However,a
central question is whether the effect is a promotion of actual
health and fitness or a reduction of chance in health degradation.
The terms of the dilemma are the same as for environmental
enrichment. That is to say, are we really improving health or are
we just counteracting some of the negative health effects of the
way we keep animals? In laboratory animals, signaling rewards
can significantly reduce anhedonia [228]. More specifically,
signaled rewards counteracted the impairment of anticipatory
behavior in socially stressed rats, further supporting the view that
eliciting positive affective states has beneficial health effects.
However, even in this example the effect is in reducing otherwise
negative health effects rather than inducing positive health effects
per se. In any event, understanding these mechanisms will im-
prove our ability to understand the benefits of positive emotions
and also suggest new clinical treatments based on these
behavioral mechanisms.
3.4. Using positive emotions for assessment and monitoring
1301 of animal welfare
3.4.1. Assessing and enhancing behavioral expressions of
positive emotions Play behavi or. Play behavior includes elements of
well-defined functional behavior such as fleeing, fighting,
sexual, or predatory behavior as well as specific play behaviors.
However, in play the functional elements are exaggerated,
repeated, more variable, and without the corresponding
consummatory act [229232]. The function of play behavior
has been, and still is, the main focus in the scientific study of
this behavior. In early studies the function of play behavior was
thought to be exclusively long-term benefits. However, studies
testing functional hypotheses of play behavior found little
evidence of long-term effects. Thus, current theories focus more
on short-term benefits to the juvenile and less on long-term
benefits, although the two may not be mutually exclusive [232].
Proposed functions of play behavior are training of the skeletal
muscles [229] (later refined to include permanent modifications
in brain and skeletal muscles [233]), self-assessment of physical
and social abilities [234], and training of flexible kinematics and
emotional responses to the unexpected events, a hypothesis that
predicts both immediate and long-term effects of play behavior
The fact that play behavior may be suppressed in harsh and
unfavorable environmental conditions together with the pro-
blems of early studies to prove any long-term function of play
behavior suggested that play behavior was merely a sign of
surplus energy without any motivation of its own. However,
there are now several sources of evidence to suggest that play is
a rewarding activity [232]. First, animals actively seek out play
partners and solicit play behavior [229]. Second, the opportu-
nity to play may be used as a reward in place preference con-
ditioning experiments [236]. Finally, thwarting of play will lead to
a rebound when more favorable conditions occur [237].
Conditions intuitively associated with bad wel fare, e.g.
insufficient food supply, suppress play behavior in deer [238]
and domestic pigs
[239]. Castration eliminates play behavior in
lambs [240] and domestic pigs in a semi-natural environment
did not play during periods of cold weather [241]. Thus, the
presence of the motivation to play may indicate a state of good
welfare, i.e. that the primary needs of the animals are met.
In calves the two most common types of play are social play
and locomotor play. Locomotor play emerges first and is seen as
387A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 13
early as a few hours after birth, and the presen ce of the mother
has been found to stimulate this type of play behavior [242].
Under na tural conditions locomotor play behavior occurs within
the first weeks of life [243,244 ]. Increasing the space allowance
for group-housed calves has been found to increase the
occurrence of play behavior [245]. Furthermore, calves kept
with little space showed a large rebound of locomotor play
indicating that these calves had built up a larger motivation
during the period of confinement [245,246]. In piglets play
behavior includes both locomotor and social play behavior,
although it may be difficult to distinguish between play fighting
and genuine fights [197]. In piglets more locomotor play
behavior was seen in larger farrowing crates compared to
smaller ones [247]. In addition, different types of farrowing
crates placed in pens of equal space affected play behavior
[248]. Locomotor play behavior is reduced by mixing of piglets
[249] and by social isolation [250,251] . In mink the additional
provision of a water bath to mink kits stimulated more play
behavior, both solitary locomotor play and social play (e.g.
chasing and rough-and-tumble play ) [87]. Likewise, the
provision of a wooden block in the home cage of mink stim-
ulated accelerated running, interrupted by sham attacks of the
object [214]. As outlined above play behavior is a good
candidate for an indicator of posit ive emotions and good wel-
fare. However, caution must be taken when developing play
behavior as a positive welfare indicator since social play can end
in a fight, as in piglets [248]. This problem may be specific for the
juvenile pig, where dominance plays a role from birth, whereas in
cattle the tendency for social play to develop into real fights does
not increase until the animals reach puberty [252].However,for
all species some elements of play behavior may be better
indicators than others, and one way of developing the area further
could be to identify and validate so-called play markers as signs
of true play in the different animal species [241].Suchplay
markers may be play signals, which communicate a playful mood
to potential play partners. Play signals have the advantage that
they are often conspicuous and easy to distinguish. Present
evidence on play behavior applies only to juveniles of mammalian
farm animal species. In these species it will be important to
develop and validate measures of play as welfare indicators, both
in animals reared for meat or for replacement (weaned piglets,
growing pigs, calves, lambs, and goats), and in foals. Affiliative behavior. Affiliative behavior is character-
ized by maintaining proximity, providing food, protection or
allogrooming between specific individuals. In farm animals,
affiliative behavior has been much less well studied than social
competition and it has only rarely been investigated with regard
to affective states. However, terms such as affiliative,
sociopositive or preferential relationships already suggest
a beneficial effect for the animals involved.
In general, positive effects of affiliative behavior can be
attributed to improved group cohesion, building or strengthening
of bonds between group mates and reduced aggression [253].
Preferential relationships between group members have been
described for cattle [254],horses[255],poultry[256] and pigs
[257]. Such relationships occur between genetically related
[254,258] as well as unrelated animals [259,260]; the latter has
recently been shown for horses, donkeys, cattle and sheep [261].
Familiarity appears to be one of the main factors in establishing
cohesion [262264]. Stress responsiveness is lower in familiar
animals [265267] whereas mixing of unfamiliar animals is
accompanied by an increase in agonistic interactions in pigs
[268],cattle[269],poultry[270] and horses [271].Affiliative
behavior may therefore play a major role in achieving a positive
mood in animals. It should however be noted that external threats
might increase the proximity between animals. Therefore, inter-
animal distance cannot be taken as a direct measure of positive
emotion level without knowing more about the threats perceived
by the animals.
Allogrooming plays a key role with regard to so-called
sociopositive interactions. It is common among non-human
primates [33]. In farm animals, allogrooming is seen in cattle,
horses, pigs and, rather rarely, in sheep. Allogrooming seems to
serve both hygienic (body care) and socia l functions. With
regard to the latter, allogrooming is thought to play a major role
in reinforci ng social bonds and in reducing tension in groups of
animals [100,272]. Such an arousal-reduction effect suggests
that allogrooming may provide long-term benefits. However,
there is also evidence that allogrooming is rewarding in the
short term. Soothing effects in terms of a reduction in heart rate
have been demonstrated in primates [273,274] and cattle [272].
Grooming stimulation produced simil ar effects in horses
[275,276]. In pigs, heart rate variability was used to assess the
effect of grooming simulation [277]. A short-term increase in
vagal tone appeared to be contradictory to the behavioral
reactions of pigs suggesting relaxation, a finding that was
interpreted as positive strain; regular grooming, however,
increased parasympathetic activity. Increased β-endorphin
levels have also been found in primates with intensive grooming
relationships [278].
Based on the above-mentioned findings, affiliative behaviors
appear to be promising indicators of long-term positive affective
states in farm animals. Allogrooming should b e further
investigated as a key indicator in cattle, horses and pigs. In
cattle, social licking is mainly carried out at the head, neck or
shoulder of the animals [252]. It is performed spontaneously,
after solicitation, or following agonistic interactions. Social
licking may also be terminated by agonistic behavior mostly
performed by the licked animal. Solicitation of social licking
[279] underlines the rewarding function of such behavior, at
least for the receiver. Whereas the overall benefits for the
receiver are obvious, it is less clear to what extent the actor is
positively influenced. While in horses mostly reciprocal
grooming (mainly consisting of biting and nibbling the group
mate's neck) between bonded pairs is observed [280], in cattle
Sato [281] found that all animals were groomed but that only
75% performed grooming. Reports on the importance of the
social position of the groomer are contradictory [252,282,283].
However, subordinate animals might experience performance
of the behavior as stressful if licking has been initiated by
dominant animals, although they may benefit from a less tense
social environment in the long term. The use of allogrooming as
an indicator in welfare assessment protocols may be limited
388 A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 14
since, for example, in horses social grooming is influenced by
season [260], social structure [264] and reproductive state [284]
and therefore needs further investigation with regard to
reliability issues. In wild boars, regular nose-to-body contacts
[285] suggest that allogrooming forms part of the pig's normal
behavior. Meynhardt [285] also states that the receiver shows
relaxation-related behaviors, a finding that has been confirmed
by a grooming simulation study [277]. However, high levels of
allogrooming activity may not necessarily reflect positive states
of the animal. Pigs in barren environments showed more
allogrooming than did those in enriched housing systems that
allowed for rooting and exploration [286,287]. On the one hand,
increased allogrooming may disturb the receiver. On the other
hand, redirected abnormal behaviors may develop such as tail-
or ear-biting [255]. Theref ore, the level of allogrooming
indicative of positive affective states in pigs must be determined
before it can be used as a welfare indicator. With regard to
animal species showing little allogrooming activity, such as
sheep and goats [288], or species in which other affiliative
behaviors have to be taken into account (poultry), potential
alternative indicators of positive affective states would have to
be identified and validated. With these caveats affiliative
behaviors and especially allogrooming is a promising candidate
for a type of behavior indicative of positive emotions . Self-grooming. Self-grooming is the maintenance of
one's own body surface. In mammals this maintenance is
achieved through licking, scratching, and rubbing of the fur, as
well as through wallowing and bathing. In birds it is through
preening and bathing in water or sand. These various forms of
self-grooming maintain the body surface by removal of dirt and
ectoparasites as well as by thermoregulation.
However, self-grooming may also occur as a displacement
activity with an arousal-reduction or relaxation effect [100].In
birds, self-grooming increases in response to a thwarting or a
conflict situation [289]. Likewise in mammals, self-grooming is
elicited in reaction to a novel or stressful situation [290292].
Self-grooming is also reported to be higher in animals kept in
barren environments. For example, self-grooming occurred less
among rabbits kept in cages with a shelter as compared to
rabbits kept in barren cages [293]. Self-grooming in animals
kept in barren or restrictive environments may be reduced either
by the provision of hay, e.g. in rabbits [294], or by increasing
the space allowance, e.g. in dogs [295] or in dairy calves
[296,297]. Likewise, the occurrence of self-grooming is de-
creased under group-housed farming, e.g. in calves [298300]
and horses [275]. When kept in groups the animals engage in
allogrooming, and the higher level of self-grooming in
individually housed animals may be due to the lack of allo-
grooming that may have an arousal-reducing effect (Section
3.4.1 affiliative behaviors). Thus, as outlined above, the re-
lation between the performance of self-grooming and positive
emotions is much less clear-cut than it is for allogrooming. The
performance of self-grooming may occur as a displacement
activity in stressful situations and barren environments. How-
ever, self-grooming may also be performed to substitute for
social grooming; it is possible that the performance of self-
grooming reduces arousal as has been suggested for social
grooming. If these effects of self-grooming can be demonstrat-
ed, it would be a potential indicator for positive emotions. Vocalizations. Specific vocalizations could be con-
sidered to some extent as direct expressions of positive
immediate experiences in animals. While vocalizations have
long been used as markers of emotions in animals [301], they
have usually concerned negative experi ences (e.g. distress, fear,
aggression, defeat). However, focusing on data obtai ned in rats,
Panksepp and Burgdorf [226] have recently argued that some
short chirping ultrasonic vocalizations centered around 50-kHz
could signal positive affects. This conclusion is supported by
the fact that such vocal signals are produced in positive
contexts, such as sex, winning fights, play, or manual tickling,
and are inhibited by negative situations [226]. Intriguingly, it is
proposed that these vocalizations could reflect a positive state in
rats analogous to primitive laughter that accompanies social
play in human infants. Not only does such a proposal offer a
framework to further study evolutionary roots of positive
emotions (and especially laughter) in humans, but it could also
be extended to other species (e.g. domestic species). For
example, some vocalizations of domestic cats, such as purring,
are well-known to be produced in situations of positive valence,
such as motherkitten interactions, contacts with familiar
partner, or during tactile stimulation with inanimate objects as
when rolling and rubbing [302]. Therefore, purring can be
generally considered as an indicator of pleasure in cats.
Similar data could be obtained in farm animals. Indeed, low-
pitched bleating in sheep has been associated with some
positive-valence situations, as they are produced by males as
an estrus female is approaching or by lactating mothers while
licking and nursing their lambs [288]. Such an approach could be
highly beneficial for animal welfare as farm animals' vocaliza-
tions could offer easily measurable markers of positive emotions
since they are audible sounds for humans and can be monitored
automatically [303]. Overall there are findings supporting the
hypothesis that at least play behaviors, affiliative behaviors, and
some vocalizations can be perceived as indicators of positive
affective states in animals. Information gathering. Information gathering or
exploration might be relevant for positive affective states for
two reasons. The first reason is that it is a behavior that most
species of animals are motivated to perform, i.e. they are
fulfilling a behavioral need. The second reason is that
exploration is a behavior that is closely related to and affected
by fear. Information gathering is a behavior that seems to be
self-reinforcing to some extent. It doesn't satiate in the same
way as many other behaviors do, and this proper ty has led some
authors to suggest that it might actually be a behavior that is
continually on-going and is only interrupted when other, more
immediate needs, are present [304]. There are two forms of
exploration: the inquisitive exploration in which the animal is
looking for a change and the inspective exploration in which the
animal responds to a change [305]. If the information primacy
interpretation is correct, then exploration, especially inquisitive
389A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 15
exploration, can be regarded as both an indicator of no other
immediate needs and a continuous pleasurable activity in itself
(since the animal keeps doing it). The maintenance of a behav-
ior might not in itself warrant calling it reinforcing, but com-
bining its flexibility with its self-reinforcing properties makes it
hard to avoid that supposition.
In inspective exploration, which typically is investigation
in response to a novel object, there is a gradual approach to
the novel object similar to that seen in other approac h/avoid-
ance situations involving, for example, hunger instead of cu-
riosity. In many of the motivational models of exploration, the
response to the novel object is a result of curiosity and fear. The
fear either interacts with curiosity, with exploration being the
behavioural outcome of that interaction [e.g. [306]], or al-
ternatively fear and exploration are seen as lying along the same
axis [e.g. [305,307]]. This close relationship between explora-
tion and fear is further supported by the finding that personality
research on animals tends to find a negative correlation between
the reaction of the animal in situations that provoke fear and
those that measure exploration [e.g. [308]]. However, some
caution is needed when interpreting these results since in some
cases only a few situations have been measured, and these may
make it inevitable that the two motivational systems are in-
timately linked.
3.4.2. Developing an on-farm monitoring system to assess
animal welfare
Consumers are increasingly concerned with the welfare of
animals contributing to the products they buy, and there have been
several initiatives to develop products marketed according to
animal welfare standards. In the main, welfare assessment
schemes have consisted of checks that the resources available to
the animal are sufficient and of good quality. Nevertheless, there
is a consensus among researchers that while this approach is a
good way to minimize the risk of poor welfare, the actual assess-
ment of animal welfare must depend on animal-based measures
[309]. Unfortunately, as already stated earlier in this section, there
are as yet no feasible animal-based measures indicative of good
welfare. Instead there are measures of poor welfare, such as injury,
body condition, fear etc., and it does not necessary follow that in
the absence of these indicators an animal's welfare is good. It
merely means that we have no avoidance that welfare is poor.
Qualitative assessment methods have been proposed as a
way to charact erize global expressive affects in animals using a
free choice profiling approach [44,310]. Among the terms
chosen by the observers are those usually associated with pos-
itive states, e.g. relaxed, friendly, affectionate, playful,
and social. There is difficulty though in validating these
potential welfare indicators since new measures are usually
validated agains t existing measures. On the other hand, we
know that play is a likely candidate for being associated with
a positive affective state and, furthermore, play behavior is
probably the indicator of positive emotions that is most easily
readable by non-professional observers (Section 3.4.1 play
behavior). Another approach to find animal-based indicators of
positive affects is to watch for the occurrence of affiliative
behaviors that an animal will only perform when it is feeling
safe, such as allogrooming (Section 3.4.1 affiliative behav-
ior). Such an approach may be related to the identification of
the terms relaxed or affectionate used by the observers in
the Wemelsfelder and collaborators [44] study. Finally, further
studies on vocalizations in farm animals could lead to the
identification of reliable vocal markers of positive states in these
domestic species (Section 3.4.1 vocalizations). These markers
would represent easy and convenient tools to assess positive
experiences in farm animals. Although only valid measures
should be included in monitoring systems, it may be that these
are the most valid measures currently available to us if we want
to include animal-based measures of positive affective states
under commercial conditions.
4. Conclusion
Over the past few years the growing interest in positive
emotions in neuroscience and psychology has led to the
development of new areas of investigation, both theoretical and
experimental. Successful expression of positive emotions typical-
ly requires a temporal sequence: a rewarding environment, eli-
citing a cognitive appraisal in the central nervous system and
followed by physiological and behavioral reactions. Repeated or
steady positive emotional experiences commonly lead to, and are
often referred to, as a global state of happiness. The co nc ept of
happiness has always been considered one of the greatest
philosophical quandaries and this has often detracted from
scientific investigations. However, such a co nc ept would be
very valuable for th e stud y of a nimal welfare since it could
represent th e affective bac kgro und r efle cting m axim um we ll-
being. Whereas pleasure is generally assessed in the context
of sensory stimulation, such as the hedonic impact of taste,
happiness refers to a longer and less i ntense internal state of
feeling, which we argue is closer to the concept of well-
being or quality of life. Although the concept of happi ness
has bee n studied ve ry little, there are som e theoretical at-
tempts in the evolutionary approach that integrates this
concept [311]. So me rec e nt wor k sug ge st s tha t hap pine ss
could be defined as consisting of both positi ve emotions and
positive activities [4,312]. Animal studies have accumulated a
large amount of valuable data r egarding sensory pleasure or
evaluating the rewarding prope rties of many agents, but there
is a lack of data concerning more persistent affective states in
animals. Hence, we must rely on theoretical and human stu-
dies to provide stimulating clues to the question of happiness
in animals and the developing study of positive emotions must
be considered as a first step in addressing the question of stable
and less expressive affective states (i.e. mood) in animals. Is there
such a state as happiness in animals? If so, how can it be reliably
assessed? Under what conditions is it maintained? What are its
relationships with cognitive abilities?
Supported by the recent interest of neuroscience and psy-
chology, further research into the field of positive emotions and
more persistent positive affective states in animals represents a
new avenue for increasing our understanding of animal welfare
processes. Such development could provide new insights in
establishing the bases of real positive well-being in animals.
390 A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 16
Research in positive affective states in animals is clearly an
ambitious task, but there are promising returns. Promoting pos-
itive experiences can be a way to improve health in animals and
more generally to give them a better quality of life. Likewise,
including criteria of positive affective states in farm or laboratory
monitoring schemes can improve welfare assessment beyond the
traditional focus on mere absence of disease and distress.
The authors are greatly indebted to the anonymous referees
for their stimulating advice and to Dr. Dwight Krebhiel for
correcting the English of the manuscript.
This review is based on the conclusions of the Task
Force entitled Positive emotions of the COST Action 846
Measuring and monitoring farm animal welfare supported by
the EU Commission. The text represents the authors' views and
does not necessarily represent a position of the Commission who
will not be liable for the use made of such information.
[1] Duncan IJH. Animal welfare in terms of feelings. Acta Agric Scand
[2] Fraser D. Science, values and animal welfare: exploring the inextricable
connection. Anim Welf 1995;4:10317.
[3] Bradburn NM. The structure of psychological well-being. Chicago:
Aldline Publishing Company; 1969.
[4] Seligman MEP, Csikszentmihalyi M. Positive psychology: an introduc-
tion. Am Psychol 2000;55:514.
[5] Lawrence A. Consumer demand theory and the assessment of animal
welfare. Anim Behav 1987;35:2935.
[6] Tooby J. The emergence of evolutionary psychology. Emerging syntheses
in science. Proceedings of the founding workshops of the Santa Fe
InstituteSanta Fe, NM: The Santa Fe Institute; 1985.
[7] Duchaine B, Cosmides L, Tooby J. Evolutionary psychology and the
brain. Curr Opin Neurobiol 2001;11:22530.
[8] Panksepp J. Affective neuroscience. The foundations of human and
animal emotions. New York: Oxford University Press; 1998.
[9] Diener E, Lucas RE. Subjective emotional well-being. In: Lewis M,
Haviland-Jones JM, editors. Handbook of emotions. 2nd ed. New York:
The Guilford Press; 2000.
[10] Linley PA, Joseph S, Harrington S, Wood AM. Positive psychology: past,
present, and (possible) future. J Posit Psychol 2006;1:316.
[11] Dantzer R. Les émotions. Paris: Presses Universitaires de France; 1988.
[12] Panksepp J. Evolution constructed the potential for subjective experience
within the neurodynamics of the mammalian brain. In: Eckman P,
Davidson RJ, editors. The nature of emotion: fundamental questions.
Oxford: Oxford University Press; 1994. p. 33699.
[13] Rolls ET. Precis of the brain and emotion. Behav Brain Sci 2000;23:
[14] Rollin BE. The unheeded cry: animal consciousness, animal pain and
science. Ames: Iowa State University Press; 1989. 330pp.
[15] LeDoux JE. Emotion: clues from the brain. Ann Rev Psychol 1995;46:
[16] Damasio AR, Grabowski TJ, Bechara A, Damasio H, Ponto LL, Parvizi J,
et al. Subcortical and cortical brain activity during the feeling of self-
generated emotions. Nat Neurosci 2000;3:104956.
[17] Block N. How can we find the neural correlate of consciousness? Trends
Neurosci 1998;19:4569.
[18] Berridge KC, Winkielman P. What is an unconscious emotion? The case
for unconscious licking. Cogn Emot 2003;17:181211.
[19] Panksepp J. The neurobiology of positive emotions. Neurosci Biobehav
Rev 2006;30:17387.
[20] Dawkins MS. How can we recognise and assess good welfare? In: Broom
DM, editor. Coping with challenge: welfare in animals including humans.
Berlin: Dahlem University Press; 2001. p. 6376.
[21] Alexandrov YI, Sams ME. Emotion and consciousness: ends of a
continuum. Brain Res Cogn 2005;25:387405.
[22] Mendl M, Paul ES. Consciousness, emotion and animal welfare: insights
from cognitive science. Anim Welf 2004;13:S1725.
[23] Cabanac M. Pleasure: the common currency. J Theoret Biol 1992;155:
[24] Schneirla T. An evolutionary and developmental theory of biphasic pro-
cesses underlying approach and withdrawal. In: Jones M, editor. Sym-
posium on motivation. Nebraska: University of Nebraska Press; 1959.
[25] Shettleworth SJ. Cognition, evolution, and behavior. Oxford: Oxford
University Press; 1998.
[26] Scherer KR. Toward a dynamic theory of emotion: the component
process model of affective states. Geneva Stud Emot Commun 1987;1:
[27] Lazarus RS. On the primacy of cognition. Am Psychol 1984;39:12412.
[28] Lazarus RS. Progress on a cognitivemotivationalrelational theory of
emotion. Am Psychol 1991;46:81934.
[29] Fridja NH. Emotion, cognitive structure, and action tendency. Cognit
Emot 1987;1:11543.
[30] Scherer KR. On the nature and function of emotion: a component process
approach. In: Scherer KR, Ekman P, editors. Approaches to emotion.
Hillsdale: Lawrence Erlbaum; 1984. p. 293317.
[31] Scherer KR. Appraisal theory. In: Dalgleish T, Power M, editors.
Handbook of cognition and emotion. John Wiley and Sons Ltd.; 1999.
p. 63763.
[32] Gray J. The psychology of fear and stress. Cambridge: Cambridge
University Press; 1987.
[33] Spruijt BM, van den Bos R, Pijlman FT. A concept of welfare based on
reward evaluating mechanisms in the brain: anticipatory behavior as an
indicator for the state of reward systems. Appl Anim Behav Sci
[34] Scherer KR. Appraisal considered as a process of multi-level sequential
checking. In: Scherer KR, Schorr A, Johnstone T, editors. Appraisal
processes in emotion: theory, methods, research. New York: Oxford
University Press; 2001. p. 92120.
[35] Dantzer R. Can farm animal welfare be understood without taking
into account the issues of emotion and cognition? J Anim Sci
[36] Désiré L, Boissy A, Veissier I. Emotions in farm animals: a new approach
to animal welfare in applied ethology. Behav Process 2002;60:16580.
[37] Désiré L, Veissier I, Després G, Boissy A. On the way to assess emotions
in animals: do lambs evaluate an event through its suddenness, novelty or
unpredictability? J Comp Psychol 2004;118:36374.
[38] Désiré L, Veissier I, Després G, Delval E, Toporenko G, Boissy A. Ap-
praisal process in sheep: interactive effect of suddenness and unfamiliarity
on cardiac and behavioural responses. J Comp Psychol 2006;120:2807.
[39] Paul ES, Harding EJ, Mendl M. Measuring emotional processes in
animals: the utility of a cognitive approach. Neurosci Biobehav Rev
[40] Matthews A, MacLeod C. Cognitive approaches to emotion and
emotional disorders. Ann Rev Psychol 1994;45:2550.
[41] Lang FR, Staudinger UM, Carstensen LL. Perspectives on socio-
emotional selectivity in late life: how personality and social context do
(and do not) make a difference. J Gerontol 1998;53:2130.
[42] Harding EJ, Paul ES, Mendl M. Cognitive bias and affective state. Nature
[43] von Frijtag JC, Reijmers LG, van der Harst JE, Leus IE, van den Bos R,
Spruijt BM. Defeat followed by individual housing results in long-term
impaired reward-and cognition-related behaviors in rats. Behav Brain Res
[44] Wemelsfelder F, Hunter TEA, Mendl MT, Lawrence AB. Assessing the
whole animal: a free choice profiling approach. Anim Behav 2001;62:
391A. Boissy et al. / Physiology & Behavior 92 (2007) 375397
Page 17
[45] Burghardt GM. The genesis of animal play. MIT Press; 2005.
[46] Booth DA. Learned ingestive motivation and the pleasures of palate.
The Hedonics of Taste. Hillsdale: Lawr ence Erlbaum Associates; 1991.
p. 2958.
[47] Berridge KC, Robinson TE. Parsing reward. Trends Neurosci 2003;26:
[48] Friend T. Recognizing behavioral needs. Appl Anim Behav Sci
[49] Dawkins MS. From an animal's point of view: motivation, fitness, and
animal welfare. Behav Brain Sci 1990;13:161.
[50] Rushen J, Lawrence AB, Terlouw CEM. The motivational basis of
stereotypies. In: Lawrence AB, Rushen J, editors. Stereotypic animal
behavior: fundamentals and applications to welfare. Wallingford: CAB
International; 1993. p. 4164.
[51] Bruce LL, Neary TJ. The limbic system of tetrapodes: a comparative
analysis of cortical and amygdala populations. Brain Behav Evol 1995;46:
[52] Farb CR, LeDoux JE. NMDA and AMPA receptors in the lateral nucleus
of the amygdala are postsynaptic to auditory thalamic afferents. Synapse
[53] Aggleton JP. The contribution of the amygdala to normal and abnormal
emotional states. Trends Neurosci 1993;16:32833.
[54] Koob GF, Sanna PP, Bloom FE. Neuroscience of addiction. Neuron
[55] Kelley AE. Neural integrative activities of nucleus accumbens sub-
regions in relation to learning and motivation. Psychobiology 1999;27:
[56] Mirenowicz J, Schultz W. Preferential activation of midbrain dopamine
neurons by appetitive rather than aversive stimuli. Nature 1996;379: 44951.
[57] Burgdorf J, Panksepp J. The neurobiology of positive emotions. Neurosci
Biobehav Rev 2006;30:17387.
[58] van der Harst JE, Baars AM, Spruijt BM. Standard housed rats are more
sensitive to rewards than enriched housed rats as reflected by their
anticipatory behavior. Behav Brain Res 2003;142:1516.
[59] Berridge KC. Food reward: brain substrates of wanting and liking.
Neurosci Biobehav Rev 1996;20:125.
[60] Davis M, Rainnie D, Cassell M. Neurotransmission in the rat amygdala
related to fear and anxiety. Trends Neurosci 1994;17:20814.
[61] Adolphs R, Tranel D, Damasio AR. The human amygdala in social
judgment. Nature 1998;393:4704.
[62] Ueda K, Okamoto Y, Okada G, Yamashita H, Hori T, Yamawaki S. Brain
activity during expectancy of emotional stimuli: an fMRI study. NeuroRe-
port 2003;14:515.
[63] Blood AJ, Zatorre RJ. Intensely pleasurable responses to music correlate
with activity in brain regions implicated in reward and emotion. Proc Natl
Acad Sci 2001;98:1181823.
[64] Wilson FAW, Rolls ET. The primate amygdala and reinforcement: a
dissociation between rule-based and associatively-mediated memory
revealed in neuronal activity. Neurosciences 2005;4:106172.
[65] Ferguson JN, Young LJ, Insel TR. The neuroendocrine basis of social
recognition. Front Neuroendocrinol 2002;23:20024.
[66] Bielsky IF, Young LJ. Oxytocin, vasopressin, and social recognition in
mammals. Peptides 2004;25:156574.
[67] Winslow JT, Insel TR. The social deficits of the oxytocin knockout
mouse. Neuropeptides 2002;36:2219.
[68] Carter CS. Oxytocin and sexual behavior. Neurosci Biobehav Rev 1992;16:
[69] Carter CS. Is there a neurobiology of good welfare? In: Broom DM,
editor. Coping with challenge: welfare in animals including humans.
Berlin: Dahlem University Press; 2001. p. 1130.
[70] Roitman ME, Wheeler RA, Carelli RM. Nucleus accumbens neurons are
innately tuned for rewarding and aversive taste stimuli, encode their
predictors, and are linked to motor output. Neuron 2005;45:58797.
[71] Wise RA. Forebrain substrates of reward and motivation. J Comp Neurol