Lucy A. Bates,1Phyllis C. Lee,2,3
Norah Njiraini,2Joyce H. Poole,2
Katito Sayialel,2Soila Sayialel,2
Cynthia J. Moss2and Richard W. Byrne1,4
Do Elephants Show Empathy?
Abstract: Elephants show a rich social organization and display a
number of unusual traits. In this paper, we analyse reports collected
over a thirty-five year period, describing behaviour that has the
potential to reveal signs of empathic understanding. These include
coalition formation, the offering of protection and comfort to others,
retrieving and ‘babysitting’ calves, aiding individuals that would
otherwise have difficulty in moving, and removing foreign objects
attached to others. These records demonstrate that an elephant is
capable of diagnosing animacy and goal directedness, and is able to
understand the physical competence, emotional state and intentions
of others, when they differ from its own. We argue that an empathic
understanding of others is the simplest explanation of these abilities,
and discuss reasons why elephants appear to show empathy more
than other non-primate species.
Empathy is defined as the ability to share someone else’s feelings or
experiences by imagining what it would be like to be in their situation
(Cambridge English Dictionary), often referred to as ‘putting oneself
into another’s shoes’. Empathy is a component of human conscious-
ness (Thompson, 2001), and the ability to detect and respond appro-
priately to the emotions of others is a cornerstone of normal social
function. The recent discovery of a mirror system for emotional
Journal of Consciousness Studies,15, No. 10–11, 2008, pp. 204–25
1School of Psychology, University of St Andrews, Fife, KY16 9JP, UK.
2Amboseli Trust for Elephants, PO Box 15135, Langata 00509, Nairobi, Kenya.
3Department of Psychology, Stirling University, Stirling, UK.
4Corresponding author: Email: firstname.lastname@example.org
responses in humans has provided evidence for the neurological basis
of empathy (Jabbi, Swart and Keysers, 2007; Wicker et al., 2003), but
little is known about the evolution of this emotional mirror system and
to what degree it is shared by any other species. Macaque monkeys are
known to possess mirror neurons that react to the physical actions of
others when they match actions in the monkey’s own repertoire
(Gallese et al., 1996; Rizzolatti et al., 1996), but the analogous emotional
mirror system has not yet been identified in non-human animals.
It is important, however, for all animals to be able to detect and
respond to the content and context of conspecifics’ emotional dis-
plays. Inappropriate responses to another’s aggressive displays, fear
reactions or sexual advances would be maladaptive and potentially
fatal; that animals respond appropriately in these circumstances sug-
gests they have at least a rudimentary form of emotion recognition
system. But human abilities go beyond simply reading and responding
to an emotional display in the present: we can also model emotional
states and desired goals that influence others’ behaviour in the past
and future, and use this to plan our own actions. Do any animals share
these advanced abilities, and if so can they be understood as the result
of empathic responses to other individuals?
Simple forms of empathy, such as emotional contagion, have been
used to explain contagious yawning, scratching, and the behavioural
copying shown in the play and aggression of chimpanzees and Japanese
macaques (Anderson, Myowa-Yamakoshi and Matsuzawa, 2004; de
Waal, 2008; Parr, Waller and Fugate, 2005). Parr et al. argue that ‘this
type of emotional awareness functions to coordinate activity among
group members, facilitate social cohesion and motivate conciliatory
tendencies, and is likely to play a key role in coordinating social
behaviours in large-brained social primates’. However, behavioural
contagion is also evident in chickens, and all these phenomena can be
explained with simple models of response facilitation (Byrne, 1994;
Hoppitt, Blackburn and Laland, 2007). Nevertheless, there is evi-
dence that even a behaviour as seemingly simple as contagious yawn-
ing does relate to empathic understanding in humans, so behavioural
contagion may well be a precursor to, or simplified form of, sophisti-
cated empathic abilities (Lehmann, 1979). Chimpanzees have been
suggested to show higher levels of empathy, such as in the behaviour
described as ‘consolation’ whereby an uninvolved bystander reas-
sures one of the individuals involved in an agonistic interaction (de
Waal and Aureli, 1996).
It seems most appropriate to look for evidence of empathy in
species that live in coherent, coordinated social groups, where
DO ELEPHANTS SHOW EMPATHY? 205
individuals have both opportunity and reason to model the behaviour,
emotions and mental states of conspecifics. One such species is the
African elephant. Elephants are long-lived, slowly developing, large
brained mammals that live in closely bonded societies (Bates, Poole
and Byrne, 2008; Douglas-Hamilton and Douglas-Hamilton, 1975;
Moss, 1988). Female savannah elephants remain within the same
family group throughout their lives, and ‘allomothering’ (caring for
another’s offspring) for female associates is common and important
for calf survival (Lee, 1987a; Moss and Poole, 1983). Furthermore,
elephants are known to possess complex auditory, olfactory and visual
communication systems (Langbauer Jr, 2000; McComb et al., 2000;
Poole, 1998; Poole, 1999; Poole and Granli, 2003). Given these char-
acteristics, it seems reasonable to predict that elephants would benefit
from being able to identify the underlying emotions and desires of
others. In this paper, we explore whether African savannah elephants,
a species that has not shared a common ancestor with humans for at
least 103 million years (Murphy et al., 2001), exhibit any signs of
advanced empathic ability.
Elephants’ well-documented interest in the carcasses and bones of
dead conspecifics might be viewed as evidence of their empathic
nature, although the biological functions of these behaviours are yet to
be determined (Douglas-Hamilton et al., 2006; McComb, Baker and
Moss, 2006). Furthermore, elephants have recently been suggested to
pass mirror self-recognition tests (Plotnik, de Waal and Reiss, 2006),
and Gallup (1982) has linked the capacity for empathy with the ability
for mirror self-recognition. However, it should be noted that Plotnik’s
paper did not follow the accepted protocol for tests of self-recognition,
as defined by Gallup (1970), and earlier attempts to show mirror
self-recognition in elephants have failed (Povinelli, 1989), so this
remains a contentious claim.
Here we present observational reports of wild elephants that appear
to respond to the emotional states of others, i.e. cases where an
empathetic understanding may be implicated, as recorded during a con-
tinuous thirty-five-year study of African elephants. We consider the
mechanisms of each of these reported cases, in terms of how they
could be modelled cognitively.
We accept that this approach is controversial for two reasons.
Firstly, the application of cognitive models is contentious. In princi-
ple, the acquisition of all non-innate behaviours can be explained as
classical or instrumental conditioning (Skinner, 1953), just as all such
behaviours can also be understood cognitively (Byrne and Bates,
2006; Tomasello and Call, 1997). It is often suggested that if
206 L.A. BATES ET ALII
behaviour can be explained as conditioning there is no need to apply
cognitive terminology. Conditioning explanations are purportedly
more ‘parsimonious’ because they do not attribute any mental pro-
cesses: indeed, it is often falsely imagined that using cognitive models
to describe behavioural complexity implies greater ‘intelligence’
than reliance on associative explanations. Unfortunately, conditioning
accounts of complex behaviours are typically derived in a post-hoc,
and therefore unfalsifiable, fashion (Byrne and Bates, 2006). We
favour cognitive terminology in this paper purely because it allows
predictive models to be developed that can subsequently be tested and
potentially falsified. The application of cognitive models does not in
itself imply anything about our expectation of the level of cognitive
skill to be found in elephants.
Secondly, this approach is controversial for its use of observational
records of behaviour, often collected ad libitum, instead of controlled
experimental trials. Several experimental paradigms have been devel-
oped to test mental-state and emotional-state recognition in primates
(Hare et al., 2000; Hare, Call and Tomasello, 2001; Parr, 2001), but
adapting these tests to non-primate species often proves difficult and
lacks ecological validity. Whilst few experimental tests of captive ele-
phant behaviour have succeeded in engaging the potential subjects,
there is a wealth of knowledge of wild elephant behaviour available
from long-term study sites. For this reason, we base our study on
observational field records taken from the longest running African
savannah elephant study site, in Amboseli National Park, Kenya. This
approach allows us to consider elephants’ empathic skills within their
particular ecological context: the problems faced by the elephants
here are real and relevant to their success.
Study site and population
The Amboseli ecosystem of southern Kenya is a semi-arid savannah.
The Amboseli elephant population has been studied continuously for
over thirty-five years by members of the Amboseli Trust for Elephants
(ATE), set up by Cynthia Moss and Harvey Croze in 1972 (see
www.elephanttrust.org). During this time, over 2200 elephants have
been identified and named (Moss, 2001) and at December 2006
the population stood at 1434 living elephants, divided into fifty-
eight family units, with approximately 300 independent males. All the
elephants in the population are habituated to the ATE project vehicles,
allowing observation of behaviour at close range.
DO ELEPHANTS SHOW EMPATHY? 207
Since 1975, three long-term researchers — Cynthia Moss (CM),
Joyce Poole (JP) and Phyllis Lee (PL), and three permanent research
staff — Norah Njiraini (NN), Soila Sayialel (SS) and Katito Sayialel
(KS), have provided the bulk of ATE elephant data collection, totalling
approximately 480 months of daily elephant observations between
them. Data collected during this time has included focal sampling of
mothers and infants (CM, PL), focal sampling of males (JP), focal
behaviour sampling of a family group (JP), scan sampling of infant
activity (PL), and ad libitum recording of interactions between elephants
(all). Additionally, research staff census family units using the Park in
order to maintain accurate demographic records of the Amboseli ele-
phant population. In all records cited, two-letter codes (e.g. AA, EB)
signify the names of family groups, whereas names or numbers (e.g.
Echo, M27) signify individuals.
Bates and Byrne (2007) have argued that analysis of observational
records, where each single record may be called an anecdote, can be a
useful scientific tool if approached systematically. Firstly, only writ-
ten reports, recorded at the time of occurrence by experienced observ-
ers, should be included in any such analysis. Reports used should
conform to a strict, pre-determined definition of the behaviour of
interest. Reports can then be separated according to the details of the
behaviour observed, and any categories where only single records
exist must be discounted as un-interpretable (McGrew, 2004). Cate-
gories of behaviour that provide multiple independent observations of
the event can then be examined to determine the minimum cognitive
apparatus necessary to allow such behaviour. This approach generates
testable and falsifiable hypotheses regarding the underlying cognitive
mechanisms of behaviour.
The original data-sheets and notes used to record all elephant obser-
vations were made available to LB and RB. We extracted 255 reports of
behaviour that potentially illustrated empathic responses to distressing
situations, according to the following working definition: ‘A voluntary,
active response to another individual’s current or imminent distress or
danger, that actually or potentially reduces that distress or danger.’
Six records described events that were recorded only once, and so
were deemed un-interpretable and were excluded from further analy-
sis. The remaining 249 records were assigned to one of the categories
detailed below, according to the type of socio-ecological problem
addressed by the behaviour.
208 L.A. BATES ET ALII
Coalitions — In aggressive or hazardous situations, two or more indi-
viduals (a and b) may work simultaneously and in a cohesive manner
against another individual or individuals (x).
Protection — When a young or injured individual is in a potentially
dangerous situation, but is unable to defend itself sufficiently, it may
receive protection from another elephant. Such ‘protection’ is distinct
from coalition behaviour, as here the individual being protected is
unable to protect itself, usually because it is too small or too sick.
Comfort — Interactions were deemed ‘comforting’rather than protec-
tive if the recipient was distressed but not in any actual danger. Awide
range of routine behaviour performed towards young calves by older
female elephants have been described as maternal or allomaternal
comfort (Lee, 1987a). Females may comfort calves in various ways:
by touching or cradling an immature animal with the trunk; touching
an immature with the body, or allowing the immature to lean on or
touch oneself; and by tolerating comfort suckling, whereby a calf
attempts to suckle from an adolescent or adult female that is not its
mother. Previously, allosuckling (suckling another’s calf) has been
discussed with respect to nutrition (Lee, 1987a), but as that is not our
focus we include all tolerated allosuckling attempts, irrespective of
whether the female is lactating.
Babysitting — When a calf is separated from its mother for a pro-
longed period, one or more other females may show interest in and/or
direct care towards the calf. This is termed babysitting, and is often
seen when a calf has been orphaned or has strayed from its family.
This category essentially draws on the protective and comforting
behaviours of females in the specific situation of a calf that has been
separated from its mother for at least several hours.
Retrievals — When a calf has been temporarily separated from its
family group, older females from the family may act to return the calf
to its natal group. In the category of retrieval, we include fetching
individuals that have wandered away from their group, been left
behind or been kidnapped.
Assisting mobility — When an individual has fallen over, become
stuck in mud, water or other difficult terrain, or is unable to proceed
forward for any other reason, other elephants may assist so that the
stuck individual can resume travelling. Acts used to assist mobility
include picking up, pushing and pulling, using the trunk, tusks or feet.
DO ELEPHANTS SHOW EMPATHY? 209
Removing foreign objects — When an elephant has a foreign object
such as a veterinary dart or spear protruding from its body, another
individual or individuals may touch or attempt to remove the object.
Having categorized each record, we then attempted to determine the
minimum necessary cognitive attributions made by the elephant before it
performed the behaviour that aided the distressed individual (see Table 1).
Table 1: Empathic attributions potentially made by elephants
Attribution of: Understanding that: Example
Animacy Some entities can
Recognize that immobility
of elephants is anomalous
Goal directedness Behaviour can be directed
at specific ends
Expect a recurring
behaviour to normally lead
to the same outcome, e.g. a
male chasing an oestrus
female will mount her
Emotion Others have emotions and
that these can be different
to one’s own
Recognize another animal’s
Others have abilities and
vulnerability and that these
can be different to one’s
Realizing that a calf cannot
cross a cattle grid
Perspective Others perceive things and
that their perspective can be
different to one’s own
Recognize that another
individual cannot perceive a
danger from its location
Intentions Others can have wants and
needs that can be different
from one’s own
Realizing that another’s
aims have been thwarted,
e.g. that a calf is trying to
get out of a river but is
Knowledge-belief Others have beliefs and
knowledge that can be
different to one’s own
behaviour by computations
of their ignorance or false
beliefs e.g. working out that
one’s calf might not know
to avoid a poisonous fruit
Coalitions (seventeen cases)
The coalitions were always aimed at adults, and were always formed
by two or more adults grouping together to threaten or chase awayone
210 L.A. BATES ET ALII
or more other, unrelated adults. Coalitions were formed to instigate a
threat (nine times) but also in retaliation to threats from the adversary
(eight times). Four coalitions were formed by males helping other
males of a similar age. Thirteen coalitions were formed by adult
females, always with related females (mother-daughter pairs in seven
cases, close matrilineal relatives — siblings, aunts and nieces, cous-
ins — in five cases, and a more distant matrilineal relative in one
case). No mixed-sex coalitions were recorded, and most were targeted
against same-sex adversaries (thirteen cases). All coalitions recorded
were successful, in that the target of the coalition moved away from
Cognitive processes underlying coalitions — Coalitions are com-
monly discussed in ethological literature, but are rarely considered as
potentially empathic. In all cases included here, only one of the coali-
tion partners (a) was originally engaging in the agonistic behaviour
with the adversary (x): the partner who joined (b) could have remained
uninvolved, but after they joined the interaction, the single adversary
(x) moved away. The joining of (a) and (b) in a coalition effectively
eliminated the potential danger to (a). The decision of (b) to act along-
side (a) must ultimately be based on selfishness at the kin selection
level (or reciprocal altruism between males), but a causal explanation
of how (b) decides to act is also necessary. In cases where the behav-
iour was retaliatory, (b) may have acted because it perceived the threat
from (x) as directed to itself, that is, the coalition behaviour may
have been entirely coincidental. However, this explanation cannot be
applied to the instigation of an attack. At a minimum, this requires an
understanding of animacy of the other elephant, recognition of another’s
emotion (from threat and fear displays), and goal directedness of the
other, whereby (b) understands that the threat behaviours of others (a)
are directed at displacing another individual (x).
Protection (twenty-nine cases)
Most reports relate to protection of calves under one year old (twenty-
seven cases). The exceptions were one for a calf of five years, and one
for an adult female who had been speared and was immobile. Members
of a different family were harassing the latter, and her adult daughters
were observed to push the threatening individuals away. In sixteen of
the cases involving calves, it was the mother who acted. Only three of
these mothers were primiparous (first-time mothers); the thirteen others
were experienced mothers with at least one calf born previously. In the
twelve cases where an allomother acted, five were parous females
DO ELEPHANTS SHOW EMPATHY? 211
(females who have given birth at least once) and seven were young
nulliparous females (females who have never given birth).
In most cases (twenty-two), the protectors acted before any harm
had actually come to the individual, and so before it had actually given
any obvious vocal or visual signal of distress. In the remaining seven
cases, the calves were pulled away from a situation after they had been
attacked or received a fright. The ‘pre-emptive’ reports of protection
occur in five situation types:
Chasing predators away from newborns (two cases);
started feeling it. Just then Echo chased off a hyena and trumpeted.
Immediately the calf whirled around and went back to Echo and at the
same time Enid, Elspeth and Eudora came rushing over, their heads up
and ears out and they joined Echo in a broad front facing the direction
that the hyena went in. — CJM: 21 August 1994.
Stopping play fights between calves (four cases);
At 17h45, Ely and Esau start playing, chasing and trying to mount one
another. They’re head to head and Enid goes over to them and clearly
pushes Esau away with her tusks. She does it three times and is getting
in between them, breaking them up. Esau keeps going around her to get
back at Ely. — CJM: 10 November 1992.
Pushing individuals away from calves (five males, five females);
As she moves, Echo stops, rumbles, ear flaps and looks back. Karl arrives
from the east. He is sniffing towrds her. Echo lifts her tusks and pushes
him away from her (one day old) baby. Echo continues slowly, stopping
and starting, rumbling and looking back. — CJM: 08 May 1994.
Keeping calves away from young males and females (four cases);
Time 12:25. At the start of the watch, Susan is mud splashing and SU9 is
sitting in the mud at the edge. Two minutes into the sample, SU9 backs
away from a young male. Sally then pulls SU9 closer to her, under her
chin, at the approach of the young male. SU9 goes around her while
Sally and the young male spar. —CJM: 27 August 1980.
Preventing calves from moving into dangerous areas (two cases);
At 9.30, the EB family go to the big wallow. Most of them get right in.
Eventually Echo and the baby came. Echo splashed, getting mud on the
baby too but backed away from the edge, looking and keeping the baby
with her. — CJM: 11 May 1994.
Cognitive processes underlying protection — The twenty-three cases
of ‘pre-emptive’ protection suggest that the protector is empathic in
the sense of predicting the distress that the calf will feel if the current
212 L.A. BATES ET ALII
situation is not stopped. As the calves were not necessarily experienc-
ing any distress, the protector could not always be responding simply
to the direct perception of pain signals or distress responses by the
calf. Instead, they must at least have been using their past experiences
with such interactions, e.g. they have learned that boisterous young
males are dangerous to young infants, or play fighting with larger
calves can cause distress. They must have recognized the situation as
conforming to that type, and taken action to prevent the distress from
occurring. Therefore, these cases provide evidence that the actor
could attribute animacy, goal directedness and emotional attribution
to another individual.
Comfort (129 records)
Lee (1987a) showed that mothers and allomothers frequently touch
young calves and maintain close proximity to them, although the rate
of touching and degree of closeness declines as the calf ages. In this
analysis, thirty-five records of reassuring immatures were extracted,
with each always directed to calves under two years old. In 61% of
cases (n=21) the calves were newborns or less than one month old
(76% less than twelve months old), and in 42% (n=15) the com-
fort-reassurance behaviour was preceded by the calf giving a distress
vocalization. In one case where no vocalization was recorded, the
calves (twins) were noted as standing with ‘heads up’in an alarm pos-
ture, and in 9% (n=3) of cases the calves initiated the physical contact
themselves, by touching their bodies against an allomother. In 45%
(n=16) of cases, calves received comfort without showing any obvi-
ous external behavioural cues.
Comfort allosuckling by calves was recorded on ninety-two occa-
sions. Only one of these records describes comfort suckling by an
immature older than two years (a six year old male was allowed to
comfort suckle from an adolescent female in his family). In all but one
case, the calf and older female were from the same family or bond
group. In forty-seven of the ninety-two cases, it was a parous female
who allowed another’s calf to suckle from her. Adolescents under the
age of thirteen were only recorded to allosuckle calves on fifteen
occasions. Two negative examples, where allosuckling attempts were
refused, concerned adult females who had calves of their own and
thus would have been lactating. One of these was from the same fam-
ily as the refused calf, and the other was not.
Cognitive processes underlying comfort behaviours — The frequency
of comfort behaviour suggests that elephants are very sensitive to the
DO ELEPHANTS SHOW EMPATHY? 213
emotional needs of calves. In approximately half of the cases, comfort
occurred when a calf was obviously distressed, with the distress most
often conveyed by auditory emotional communication signals. In
these cases, the mother or allomother has to recognize the emotional
signals of the calf and respond appropriately by offering comfort. In
principle, comfort might be offered because of an understanding of
wants and needs, in the absence of any expression of emotion. But in
these cases, the most parsimonious explanation is that the mother was
responding to the emotions of the calf.
Where comfort was shown to calves without any obvious external
behavioural cues from the calf, it is harder to determine the ontogeny
of the mother’s behaviour. However, we suggest that seeking physical
contact with calves may require no stimuli other than the appearance
and subsequent recognition of the young calf, and is most likely an innate
behavioural response of females towards elephant calves. Attraction to
young calves is thought to serve an important evolutionary function in
elephant families, increasing calf survival and enhancing the stability
of the group (Lee, 1987a).
Comfort suckling was always initiated by calves, and so required
little action on the part of the older female. In the two cases where
allosuckling was refused, the female evidently discriminated the calf
as not her own. In both reports the calf’s actions are described as ‘tried
to suckle’ suggesting it had at least raised its trunk towards the
female’s breast. Recognition of the goal towards which the calf’s
actions were directed, or attribution of goal directedness, is implied by
the active refusal of the adult.
Babysitting (twenty-one cases)
On six occasions, an unknown calf was seen with an un-related family
whose members showed some care of the calf. The calf did not survive
for more than a few weeks in any of these cases, although the precise
time taken for the calf to die depended upon its age and condition
when it was orphaned or otherwise separated from its mother. All
other cases of babysitting occurred within the calf’s natal family. Six
refer to orphaned calves; in four of these the calf only survived one to
two months after its mother’s death. The other two calves are cur-
rently still alive, after five and three years respectively. In nine cases
the separation was temporary, and the calf was safely reunited with its
mother after being cared for by babysitting family members.
Cognitive processes underlying babysitting — Babysitting is most
obviously explained as an over-extension of a female’s natural
214 L.A. BATES ET ALII
tendency to care for calves: occasionally females make the ‘mistake’
of targeting care at non-kin calves. Babysitting can be considered a
natural consequence of comforting and allomothering calves, and so
probably relies on the same cognitive attributions of animacy and, on
occasion, emotion, to the calf. The physical presence or absence of the
mothers may not be a critical variable: calves are allomothered in both
cases. However, babysitting is more common when calves are at a dis-
tance from the mother (Lee, 1987a). Rarely, babysitting may ensure a
calf’s survival even after its mother’s death, although in cases where
the calf has yet to be weaned, survival almost never results.
Retrievals (22 cases)
All individuals who had to be retrieved were under five years old.
Nine calves were retrieved from unrelated females or families, six
were alone or left behind before being reunited with their families,
five were pulled away from the ATE observation vehicle, and two
were retrieved from males to whom they had wandered too close.
They were all retrieved either by the mother acting alone, or by the
mother and another female family member. In nine cases the female
retrieved the calf following ‘lost call’ vocalizations by the calf, but in
the other cases no vocalizations were noted. Of the nine who were
with unrelated females, six most likely occurred because the calf wan-
dered over to the unrelated females, who then began to allomother
them. But on three occasions it appears that females from dominant
families actively attempted to remove a newborn calf from the subor-
dinate family and/or resisted the retrieval by the mother, as described
At 10.10, Freda and others move rapidly to the place where the EB’s are
resting and supplant them. Then they kidnap Ely (Echo’s newborn calf),
and he calls and cries out. Enid stays but Echo has run off and tries to
grab him back. The FB’s stick with Ely and at one point he is kicked and
knocked down. —CJM: 24 March 1990.
Cognitive processes underlying retrievals — Again, this behaviour
can be viewed as an extension of the allomothering ‘caretaking’ ten-
dencies of female elephants. Retrievals are usually led by the mother,
so she must differentiate her calf from any others present and respond
to its absence. Retrieval behaviour sometimes relies on an attribution
of emotion to the calf, at least when the mother is responding to vocal
distress signals, and potentially involves an attribution of goal direct-
edness to the adult females who ‘kidnap’ calves.
DO ELEPHANTS SHOW EMPATHY? 215
Assisting mobility (28 cases)
All but one of the immobile individuals were under two years old
when helped. The exception concerns a mother leading her two-year-
old and six-year-old daughters to a cattle-grid where there was a break
in the electric fence that was separating them from her. The mother
had to break through the electric fence, using her tusks to snap the
wire, and then enter the enclosure to lead her calves to the safe place to
exit (NN: 10 September 2006). The following day, the same individu-
als were again stuck in the enclosure. The youngest could not be
encouraged to cross the cattle grid for fifteen minutes, eventually run-
ning over it only when the mother walked back on to it herself, facing
the calf, and reversed slowly (NN: 11 September 2006).
Four records describe leading a calf to terrain that was easier for it
to negotiate, such as a less steep part of a riverbank or over a cattle
grid. Nine records refer to events where a calf had fallen over or could
not get up and was helped to stand, usually by the mother but in one
case by an unrelated adult male. In all cases, the calf was helped to its
feet by the adult using its trunk to lift it, and gently sliding the foot
underneath the calf.
The remaining fifteen records refer to calves who had fallen into
ditches or who could not climb in or out of mud wallows, rivers or
other water channels. The calf was pushed (three times) or pulled
(nine times) out, or an adult dug the sides of the bank with her tusks,
which decreased the incline (three times). In thirteen cases, the calf
was helped by the mother, in the other six cases the calf was helped by
other female family members. The passage below describes the action
of one such allomother:
IB’s are crossing Snipe River. Infant struggles to climb out of bank after
its mother. An adult female is standing right next to it, and moves closer
as the infant struggles, it does not push it out with its trunk, but it digs its
tusks into the mud behind the calf’s front right leg which acts to provide
some anchorage for the calf, who then scrambles up. — LB and NN: 21
Cognitive processes underlying mobility assistance — In all cases
described here, elephants must have made attributions of animacy and
physical competence to the individual that was immobile. Attributing
emotion is also highly likely in some circumstances, such as the
eleven cases observed where a calf gave a distress bellow vocalization
(seven when it was stuck in a water channel or ditch, and four when it
had fallen over), although emotion attribution is also possible when
216 L.A. BATES ET ALII
no vocalizations were given, for example from visual signals of
distress or frustration.
Hart, Hart and Pinter-Wollman (2008) discuss elephants helping
disabled conspecifics as evidence that elephants attribute mental states:
they consider helping implies attributing a mental state of ‘disable-
ment’, an example of ‘targeted empathic helping’. We suggest that
mental state attribution may not be necessary to understand these data;
an understanding of the physical state of the stuck individual is often
sufficient, with no need to understand intentions or knowledge. How-
ever, two scenarios included in our database do require an attribution
of intention by the helper to the calf.
Firstly, the three cases when calves were pushed out of a water
channel, such as in the example described above, seem to require an
understanding of the calf’s intention. In these cases, the adult respon-
ded to the signs of frustration or distress in the calf, but the act of help-
ing it increased the physical distance between calf and adult. In
contrast, pulling calves out of a channel does not necessarily require
understanding the calf’s intention to climb out of the bank. The
mother or allomother was already out of the channel, and — as with
retrievals — upon seeing its distress at the physical separation, she
acted to bring the calf near to her. Similarly, digging the sides of a
bank to allow a calf an easier ascent does not require an attribution of
intent as in all cases the adult’s behaviour could be argued to be self-
ish, whereby the females dug the bank to also ease their own climb,
although in most cases the adult could easily step out of the bank.
The second scenario that required an attribution of intention
describes a female leading a mother and calf to a shallower bank:
At 11.10ish Ella gives a ‘lets go’rumble as she moves further down the
swamp except Elspeth and her calf, born 2000, and Eudora (Elspeth’s
mother). At 11.25 Eudora appears to ‘lead’ Elspeth and the calf to a
good place to enter the swamp — the only place where there is no
mud. — JP: 3 June 2000.
In this case, Eudora had no need to enter the swamp at the point to
which she led her daughter and granddaughter, her behaviour was
adjusted specifically to the problem faced by the calf.
Removing foreign objects (three cases)
Only three records contribute to this category, although there are addi-
tional verbal reports that have not been included here as they do not
conform to the criteria set out in Bates and Byrne (2007). One
DO ELEPHANTS SHOW EMPATHY? 217
describes an adult (M324) pulling a tranquillizing dart out of another
male (M319) that had been darted by a vet prior to treatment for a
After [M319] was darted, another male — M324 — approached M319
and kept on touching the dart. M324 then pulled the dart out and
dropped it on the ground, and kept touching where the dart was. M324
was then pushing M319, then when M319 went down, M324 left. — SS:
4 September 2006.
The actor, M324, dropped the dart as soon as he had pulled it out, sug-
gesting he was not interested in the dart itself, but rather in removing it
from the other elephant.
The second observation in this category describes a six-year-old
male investigating a spear that was penetrating the back of another
juvenile (Matrix, born 2000). The spear had entered just above the hip
on the left hand side of the elephant, and came out about four inches
further back, one inch below the spine, and was lodged in the wound.
Matrix did not appear to be limping when walking, and was keeping
pace with her mother Marjorie.
Matrix is swishing her tail onto the wounds almost constantly, and both
the entry and exit wounds are oozing puss, but we can’t see any blood.
At 10.35 Matrix mud splashes with Marjorie, and initially directed all
sprayings to the wound area ...11.07Matrixisdustingthewound. She
does not throw dust anywhere else on her body, just on the wound. 11.12
Winona’s male calf (born 2000) moves out of the swamp near to Matrix.
He is standing just behind her. He touches the spear (the exit side) with
his trunk three times. As he did this Matrix stopped feeding and stood
still. 11.15 The 2000 male moved off, and Matrix stepped forward and
continued feeding. No vocalisations were heard. — LB and KS: 24 Feb-
Matrix or Marjorie were not seen to investigate the wound with their
trunks at any point, and there was apparently no attempt by the juve-
nile male to pull the spear out of Matrix. Verbal reports of other inci-
dents suggest that sometimes elephants do attempt to pull out spears
that protrude from other’s bodies, however.
The final report in this category describes an experienced matriarch
removing rubbish from her calf’s mouth.
Echo is with Esprit (born 2005) behind the Safari Lodge. Echo picks up
a plastic bag then drops it again almost immediately. Esprit, standing
right next to Echo, then picks it up and starts to put it in her mouth. Echo
immediately picks the bag off her, holds it in her own trunk-curl for sev-
eral seconds, then drops it and moves on. Esprit moves with her. — LB
and NN: 22 February 2006.
218 L.A. BATES ET ALII
Interestingly, elephants frequently carry vegetation on their bodies,
and conspecifics apparently pay little interest to these natural objects;
there are no reported observations of individuals removing vegetation
from another’s body, although they may play with and toss around
vegetation from their own bodies before discarding it. Similarly, during
experimental trials of a study that involved presentation of brightly
coloured cloths to the elephant groups, several different individuals
picked up a white cloth in some trials, and draped it over their bodies
whilst walking along. On these occasions, we never observed others
attempting to remove the cloth (Bates et al., 2007).
Cognitive processes underlying removal of foreign objects —Inthe
cases presented here, the actors apparently recognized the dart, spear
and plastic bag as foreign objects that should not be in contact with the
others’ bodies, and in two cases removed them. Minimally, this act
could be achieved with knowledge of what is ‘normal’ for elephant
bodies, with action taken to investigate or remove the visible objects
that do not conform to this template. This explanation only works if
clumps of vegetation carried on an elephant are seen as normal, which
is plausible given how often researchers see this. However, it does not
explain why no investigations were made of the cloths carried on oth-
ers’ backs or tusks during the experimental trials. Such a minimal
explanation of ‘norm’ attribution gives no significance to the fact that
male M319 was visibly ill before he was darted, and in the case of
Matrix, puss was seeping from the wounds: the potential health state
of the individual should be irrelevant. We would therefore have to pre-
dict that elephants would remove foreign objects from the bodies of
any elephant, healthy or sick, irrespective of the nature of the foreign
object. However, this prediction is apparently not supported, as it is
only in cases where there is a visible injury or sickness, or the object
presents a potential hazard, that another elephant has felt compelled to
touch the foreign object.
It is not possible to draw strong conclusions from only three
observations, but these reports suggest that elephants can distinguish
dangerous from benign foreign objects, perhaps using emotional and
behavioural cues from the wounded animal, and it is this recognition
that prompts the action to remove them. In addition, removal may be
altruistic, as the interaction with a foreign object that is associated
with illness, injury, or danger in another could be potentially danger-
ous to the actor as well.
DO ELEPHANTS SHOW EMPATHY? 219
Based on the evidence presented here, as summarized in Table 2, we
would argue that elephants routinely recognize animacy and goal-
directedness; that is, elephants recognize certain characteristic aspects
of normal elephant behaviour, and have expectations about the out-
comes usually achieved by such behaviour. Furthermore, there is
strong evidence that elephants are able to recognize accurately and
respond appropriately to a range of emotions of other elephants, usu-
ally but not exclusively kin. Elephants therefore understand that other
elephants are animate agents that can perform directed behaviours and
experience autonomous emotions, which they can recognize.
Table 2: Summary of behaviours observed and the implications for
Behaviour Context Requirement Empathic
Recognition of threat from
third parties to allies
Recognition of danger to
Recognition that another has
Recognition of physical
distress of calf
Recognition of emotional
distress of calf
Recognition of identity of
Babysitting Related calves Recognition that calf is not
with its mother
Recognition that calf is not
with its mother
Retrievals Calf left alone Remembering that calf
should be present
Recognition of calf and that
it should be present
that drew it
Recognition of calf and that
it should be present
220 L.A. BATES ET ALII
Behaviour Context Requirement Empathic
Leading Recognition that calf cannot
negotiate certain terrain
Recognition that calf cannot
Pulling out of
Recognition that calf
distressed because lacks
ability to join mother
Pushing out of
Recognition that calf wants
to get out of ditch but lacks
Recognition that mother’s
efforts will be insufficient to
overcome calf’s physical
Recognition that object is
unusual and dangerous
It is important to realize that the 249 records discussed here almost
certainly under-represent the frequency with which elephants engage
in these sorts of empathic behaviour. Behaviours such as comforting
and seeking physical contact with calves can be observed every few
minutes in Amboseli, and others such as coalitions, protection and
assisting with mobility probably also occur on a daily basis. Because they
are now so familiar to observers, these actions are now rarely recorded
except during focal sampling or when they occur in circumstances
that are noteworthy for another reason, although they were recorded
more frequently in the first ten to fifteen years of the project. Given
the potentially controversial nature of this paper, we have been careful
to only include reports that fully conform to the criteria laid out by
Bates and Byrne (2007). This stringency reduced our sample size, but
we feel it was necessary. Even with this reduced sample of observa-
tions, we were still able to uncover some evidence, from behaviour in
the category of ‘assisting mobility’, that elephants understand the
physical competence and intentions of others, where these differ from
their own. We anticipate that structured observations will in the future
strengthen this conclusion.
DO ELEPHANTS SHOW EMPATHY? 221
We therefore take it that elephants do indeed show certain kinds of
empathy. Empathy can operate at a number of levels, from the sim-
plest level of ‘contagion’, to a more sophisticated level described by
de Waal (2008) as ‘sympathetic concern’. The latter is illustrated in
our data by instances in which elephants offer protection and comfort
to the calves of others, ‘babysit’ them or retrieve them from harm. The
highest level of empathy de Waal describes, ‘empathic perspective
taking’, is characterized by ‘targeted helping’ towards needy individ-
uals. In our data, this was shown in several cases in which calves were
helped to overcome mobility problems.
Few animal species have been suggested to show such a high level of
empathy. There seem to be three possible reasons for that: a genuine
lack of ability, a lack (in the animals) of much functional utility to show-
ing such empathy, or an antipathy (in the research community) towards
labelling animal behaviour as empathic. Because of the inherent diffi-
culty in measuring animals’ emotions and other mental states, animal
behaviour researchers are understandably reluctant to appeal to empa-
thy as a causal explanation: preferring alternatives that do not impute
emotional state understanding, often going no further than a functional
account. Intriguingly, this sparse and intellectually hygienic approach
contrasts with how (often the same) researchers typically talk about
their subjects’ behaviour amongst themselves! Invoking advanced cog-
nitive capacities such as empathy is often the easiest way to describe an
event to someone who did not themselves observe it, but it is another
matter to prove their existence to a sceptical audience.
Quite apart from these differences of interpretation, there are perhaps
few species where individuals would benefit from empathic responses,
whether or not they have the ability to show them. For many mammals,
the only social period of life is a relatively brief period of dependence
on the mother, and maternal solicitude is shown in routinely protective
behaviour that can be explained by simpler means than empathy. Much
the same applies to birds, although manybird species are pair bonded so
both parents may supply care and protection. Only with species where
several generations live socially and the group may contain both kin
and non-kin with various degrees of affiliation, as with many anthro-
poid and some strepsirrhine primates, are researchers able in principle
to detect the signs of empathic responses. In these cases, the ability to
discriminate among individuals according to differences in their physi-
cal abilities, their knowledge and their needs and wants, may pay in
evolutionary terms. It is, then, no surprise that it is in primates (de Waal,
1996, 2008) and in elephants that empathy has been detected. The level
of empathic responsiveness of elephants seems exceptionally high, but
222 L.A. BATES ET ALII
fair comparison is difficult. For example, the relatively slowly develop-
ing infants of most primates would still be held close to their mothers at
the age when elephant calves are often the subjects of helping by other
It is obvious from the records we have discussed here that elephants
are very sensitive to the distress of others, and remarkably capable of
anticipating and preventing such distress. These reports mostly concern
the potential distress of calves, and it is mostly relatives who react. As
de Waal (2008) argued, empathy is the causal mechanism underlying
directed altruism. In all animal species studied, altruism is most com-
mon along kin lines, but the high reproductive ‘value’of an elephant
calf to its mother and other kin (Lee, 1987b) particularly increases the
adaptive value of any effective care by elephants. Nevertheless, as evi-
denced by a few of our records, adult elephants do sometimes help indi-
viduals who are not related to them; thus, not all empathic behaviour
may be linked to kinship, and may represent a more generalized response
to distress (Douglas-Hamilton et al., 2006). Some of the many instances
of altruism described in other animal species may also depend on empa-
thy; but as yet it remains unclear if helping in any other species requires
empathic comprehension of the distressing event. Other than elephants
and primates, species likely to give convincing evidence of empathy are
those, such as social carnivores, that show female philopatry (i.e. where
females stay in the natal home area and adolescent males move away to
breed) and altruistic and co-operative helping behaviours. For example,
meerkats, which show many kinds of altruistic helping behaviour
(Clutton-Brock et al., 2000; Clutton-Brock et al., 2002), would be
interesting to consider from this perspective, but as yet no evidence of
empathy has been reported.
We would like to thank the Leverhulme Trust for financial support
(Grant F/00 268/W). Our study of the cognition of the African ele-
phant could not take place without the continual collaboration of the
Amboseli Trust for Elephants. We thank the Office of the President,
the Kenya Wildlife Service and the Amboseli Park wardens for allow-
ing ATE to study the elephants of Amboseli National Park.
Anderson, J.R., Myowa-Yamakoshi, M. and Matsuzawa, T. (2004), ‘Contagious
yawning in chimpanzees’, Proceedings of the Royal Society (B): Biology Let-
ters Supplement,271, pp. S468–70.
DO ELEPHANTS SHOW EMPATHY? 223
Bates, L.A. and Byrne, R.W. (2007), ‘Creative or created: Using anecdotes to
investigate animal cognition’, Methods,42, pp. 12–21.
Bates, L.A., Poole, J.H. and Byrne, R.W. (2008), ‘Elephant cognition’, Current
Biology,18, pp. R544–R546.
Bates, L.A., Sayialel, K.N., Njiraini, N., Moss, C.J., Poole, J.H. and Byrne, R.W.
(2007), ‘Elephants classify human ethnic groups by odor and garment color’,
Current Biology,17 (22), pp. 1938–42.
Byrne, R.W. (1994), ‘The evolution of intelligence’, in Behaviour and Evolution,
ed. P.J.B. Slater and T.R. Halliday (Cambridge: Cambridge University Press),
Byrne, R.W. and Bates, L.A. (2006), ‘Why are animals cognitive?’, Current Biol-
ogy,16, pp. R445–R448.
Clutton-Brock, T.H., Brotherton, P.N.M., O’Riain, M.J., Griffin, A.S., Gaynor, D.,
Sharpe, L., Kansky, R., Manser, M.B. and McIlrath, G. (2000), ‘Individual con-
tributions to babysitting in a cooperative mongoose, Suricata suricatta’, Pro-
ceedings of the Royal Society of London (B),267, pp. 301–5.
Clutton-Brock, T.H., Russell, A.F., Sharpe, L.L., Young, A.J., Balmforth, Z. and
McIlrath, G.M. (2002), ‘Evolution and development of sex differences in coop-
erative behavior in meerkats’, Science,297 (5579), pp. 253–6.
de Waal, F.B.M. (1996), Good Natured: The Origins of Right and Wrong in
Humans and Other Animals (Boston MA: Harvard University Press).
de Waal, F.B.M. (2008), ‘Putting the altruism back into altruism: The evolution of
empathy’, Annual Review of Psychology,59, pp. 279–300.
de Waal, F.B.M. and Aureli, F. (1996), ‘Consolation, reconciliation, and a possible
cognitive difference between macaque and chimpanzee’, in Reaching into
Thought: The Minds of the Great Apes, ed. A.E. Russon, K.A. Bard and S.T.
Parker (Cambridge: Cambridge University Press), pp. 80–110.
Douglas-Hamilton, I., Bhalla, S., Wittemyer, G. and Vollrath, F. (2006), ‘Behav-
ioural reactions of elephants towards a dying and deceased matriarch’, Applied
Animal Behaviour Science,100 (1–2), pp. 87–102.
Douglas-Hamilton, I. and Douglas-Hamilton, O. (1975), Among the Elephants
(New York: Viking Press).
Gallese, V., Fadiga, L., Fogassi, L. and Rizzolatti, G. (1996), ‘Action recognition
in the premotor cortex’, Brain,119, pp. 593–609.
Gallup, G.G., Jnr (1970), ‘Chimpanzees: Self-recognition’, Science,167, pp. 86–7.
Gallup, G.G., Jnr (1982), ‘Self-awareness and the emergence of mind in primates’,
American Journal of Primatology,2, pp. 237–48.
Hare, B., Call, J., Agnetta, B. and Tomasello, M. (2000), ‘Chimpanzees know
what conspecifics do and do not see’, Animal Behaviour,59, pp. 771–85.
Hare, B., Call, J. and Tomasello, M. (2001), ‘Do chimpanzees know what con-
specifics know?’, Animal Behaviour,61, pp. 139–51.
Hart, B.L., Hart, L.A. and Pinter-Wollman, N. (2008), ‘Large brains and cognition:
Where do elephants fit in?’, Neuroscience and Biobehavioural Reviews,32 (1),
Hoppitt, W., Blackburn, L. and Laland, K.N. (2007), ‘Response facilitation in the
domestic fowl’, Animal Behaviour,73, pp. 229–38.
Jabbi, M., Swart, M. and Keysers, C. (2007), ‘Empathy for positive and negative
emotions in the gustatory cortex’, NeuroImage,34 (4), pp. 1744–53.
Langbauer Jr, W.R. (2000), ‘Elephant communication’, Zoo Biology,19, pp.
Lee, P.C. (1987a), ‘Allomothering among African elephants’, Animal Behaviour,
35 (1), pp. 278–91.
224 L.A. BATES ET ALII
Lee, P.C. (1987b), ‘Nutrition, fertility and maternal investment in primates’, Jour-
nal of Zoology,213, pp. 409–22.
Lehmann, H.E. (1979), ‘Yawning: A homeostatic reflex and its psychological sig-
nificance’, Bull. Meninger Clin.,43, pp. 123–36.
McComb, K., Baker, L. and Moss, C. (2006), ‘African elephants show high levels of
interest in the skulls and ivory of their own species’, Biology Letters,2, pp. 26–8.
McComb, K., Moss, C., Sayialel, S. and Baker, L. (2000), ‘Unusually extensive
networks of vocal recognition in African elephants’, Animal Behaviour,59, pp.
McGrew, W.C. (2004), The Cultured Chimpanzee: Reflections on Cultural Primatology
(Cambridge: Cambridge University Press).
Moss, C. (2001), ‘The demography of an African elephant (Loxodonta africana)
population in Amboseli, Kenya’, Journal of Zoology,255, pp. 145–56.
Moss, C.J. (1988), Elephant memories: Thirteen years in the life of an elephant
family (London: Elm Tree Books).
Moss, C.J. and Poole, J.H. (1983), ‘Relationships and social structure of African
elephants’, in Primate Social Relationships: An Integrated Approach,ed.R.A.
Hinde (Oxford: Blackwell Scientific Publications), pp. 315–25.
Murphy, W.J., Eizirik, E., O’Brien, S., Madsen, O., Scally, M., Douady, C.J.,
Teeling, E., Ryder, O.A., Stanhope, M.J., de Jong, W. and Springer, M.S.
(2001), ‘Resolution of the early placental mammal radiation using Bayesian
phylogenetics’, Science,294, pp. 2348–51.
Parr, L.A. (2001), ‘Cognitive and physiological markers of emotional awareness
in chimpanzees (Pan troglodytes)’, Animal Cognition,4(3–4), pp. 223–9.
Parr, L.A., Waller, B.M. and Fugate, J. (2005), ‘Emotional communication in
primates: Implications for neurobiology’, Current Opinion in Neurobiology,
15 (6), pp. 716–20.
Plotnik, J.M., de Waal, F.B.M. and Reiss, D. (2006), ‘Self-recognition in an Asian
elephant’, Proceedings of the National Academy of Sciences,103, pp. 17053–7.
Poole, J. (1998), ‘An exploration of the commonality between ourselves and ele-
phants’, Etica and Animali,9, pp. 85–110.
Poole, J. (1999), ‘Signals and assessment in African elephants: Evidence from
playback experiments’, Animal Behaviour,58, pp. 185–93.
Poole, J.H. and Granli, P.K. (2003), ‘Visual and tactile signals of African savanna
Povinelli, D.J. (1989), ‘Failure to find self-recognition in Asian elephants (Elephus
maximus) in contrast to their use of mirror cues to discover hidden food’, Jour-
nal of Comparative Psychology,103, pp. 122–31.
Rizzolatti, G., Fadiga, L., Fogassi, L. and Gallese, V. (1996), ‘Premotor cortex and
the recognition of motor actions’, Brain Research,3, pp. 131–41.
Skinner, B.F. (1953), Science and Human Behaviour (New York: Macmillan).
Thompson, E. (2001), ‘Empathy and consciousness’, Journal of Consciousness
Studies,8(5–7), pp. 1–32.
Tomasello, M. and Call, J. (1997), Primate Cognition (New York: Oxford Univer-
Wicker, B., Keysers, C., Plailly, J., Royet, J., Gallese, V. and Rizzolatti, G. (2003),
‘Both of us disgusted in my insula: The common neural basis of seeing and feel-
ing disgust’, Neuron,40 (3), pp. 655–64.
DO ELEPHANTS SHOW EMPATHY? 225