Cognition 109 (2008) 224–234

Contents lists available at ScienceDirect

Cognition
journal homepage: www.elsevier.com/locate/COGNIT

Chimpanzees know what others know, but not what they believe
Juliane Kaminski *, Josep Call, Michael Tomasello
Department of Developmental and Comparative Psychology, Max-Planck Institute for Evolutionary Anthropology, Deutscher Platz 6, 04103 Leipzig, Germany

a r t i c l e i n f o a b s t r a c t

Article history: There is currently much controversy about which, if any, mental states chimpanzees and
Received 13 June 2007 other nonhuman primates understand. In the current two studies we tested both chimpan-
Revised 29 May 2008 zees’ and human children’s understanding of both knowledge–ignorance and false belief –
Accepted 22 August 2008
in the same experimental paradigm involving competition with a conspecific. We found
that whereas 6-year-old children understood both of these mental states, chimpanzees
understood knowledge–ignorance but not false belief. After ruling out various alternative
Keywords:
explanations of these and related findings, we conclude that in at least some situations
Knowledge
False belief
chimpanzees know what others know. Possible explanations for their failure in the highly
Mental states similar false belief task are discussed.
Chimpanzees Ó 2008 Elsevier B.V. All rights reserved.
Theory of mind
Social cognition

1. Introduction ies, recent research has shown that infants at around the
first birthday even understand that others know things,
In 1978, Premack and Woodruff asked, ‘‘Does the chim- that is, that others’ actions are governed by things they
panzee have a theory of mind?” This question sparked saw some moments before (e.g., Moll & Tomasello, 2007;
much research, most immediately on human children with Onishi & Baillargeon, 2005; Tomasello & Haberl, 2003). In
a focus on their understanding of false beliefs (Wimmer & contrast, if one requires children to express their knowl-
Perner, 1983). More recently, research has focused on how edge in action (as in most of the studies cited above), they
young children understand the psychological states of oth- show no understanding of false beliefs – that others’ ac-
ers more generally, including everything from goals and tions are governed by things the child knows are not true
intentions to perceptions, knowledge, and beliefs. – until much later at around 4 years of age (see Wellman,
Human infants begin to understand that others have Cross, and Watson (2001), for a review and meta-analysis).
goals quite early, before the first birthday (e.g., Behne, Car- Importantly, in a direct comparison Wellman and Liu
penter, Call, & Tomasello, 2005; Gergely, Nadasdy, Csibra, (2004) found that children develop an understanding of
& Biro, 1995), and they understand others’ rational choices knowledge–ignorance before they develop an understand-
of means toward goals (intentions) soon after (Gergely, ing of false beliefs.
Bekkering, & Kiraly, 2002; Schwier, van Maanen, Carpenter, As for Premack and Woodruff’s original question about
& Tomasello, 2006). Infants understand that others see chimpanzees, there has been controversy from the begin-
things from around the first birthday as well (e.g., Brooks ning. Thus, Savage-Rumbaugh, Rumbaugh, and Boysen
& Meltzoff, 2002; Moll & Tomasello, 2004), and they under- (1978) presented data from their chimpanzees suggesting
stand that others have perspectives that differ from their that the Premack and Woodruff (1978) goal-understanding
own by at least the second birthday (Level 1: Moll & Tom- tasks could be solved through simple association. Subse-
asello, 2006). Of particular importance to the current stud- quent experiments on other mental states also yielded
negative results. Most prominently, Povinelli and Eddy
* Corresponding author. (1996) found that juvenile chimpanzees begged food from
E-mail address: [email protected] (J. Kaminski). a human gesturally even when he was blindfolded or had a

0010-0277/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.cognition.2008.08.010
 J. Kaminski et al. / Cognition 109 (2008) 224–234 225

bucket on his head, suggesting no understanding of visual dominant knew the food was there, even though he could
perception. Similarly Povinelli, Rulf Alyssa, and Biersch- never seechoice.1
wale Donna (1994) found that when chimpanzees saw There were several additional control conditions in the
two humans pointing to different locations to indicate two Hare et al. studies that ruled out various more conser-
the location of a single piece of hidden food – and one of vative, less mentalistic interpretations of these results.
those humans had watched the original hiding process However, one final conservative interpretation is the so-
whereas the other had not – they followed the two hu- called evil eye hypothesis. Perhaps subordinates believe
mans’ pointing gestures indiscriminately, suggesting no that any piece of food observed by a dominant is ‘contam-
understanding of the distinction between knowledge and inated’ – it is forbidden once the dominant has put the evil
ignorance. And Call and Tomasello (1999) found that eye on it – and so the only safe food is food that he cannot
whereas 5-year-old children passed a nonverbal false be- see and indeed has never seen. In a final study of Hare et al.
lief test readily, chimpanzees failed it. (2001), both the dominant and the subordinate watched
All of these data led researchers to the conclusion that the food being hidden behind one of the two barriers, as
chimpanzees and other nonhuman primates do not under- usual; the dominant’s evil eye was thus placed on it, and
stand the psychological states of others (Heyes, 1998; Povi- so on this interpretation the subordinate should avoided
nelli & Vonk, 2003, 2004; Tomasello & Call, 1997). That is, it at all costs. But then in one experimental condition only
nonhuman primates can predict others’ actions in many the subordinate watched the food being moved to a new
situations based on past experience (and perhaps some location (dominant’s door down), whereas in another con-
specialized cognitive adaptations), but they do not go be- dition they both watched it being moved. Subordinates
neath the surface to an understanding of the goals, percep- went for the food when only they alone had watched the
tions, knowledge, and beliefs that guide others’ actions. But moving process, not when both competitors had watched
as always, negative experimental results have many possi- the moving process. Subordinates thus clearly did not be-
ble interpretations, and there have always been a number lieve in any dominant evil eye, since they went for the food
of informal observations by fieldworkers suggesting that whose movement to a new location the dominant had not
perhaps chimpanzees and other nonhuman primates can witnessed (even though he had put his evil eye on it ear-
understand some mental states in some situations. Most lier). Nevertheless, one other more conservative explana-
prominently, Byrne and Whiten (1990) reported a number tion is still viable. It could be that chimpanzees have
of informal observations from fieldworkers on so-called learned the behavioral rule: if a dominant individual ori-
tactical deception, which might, in some interpretations, ents to a piece of food in a particular location, then that food
suggest some form of mental state understanding. must be avoided (see Povinelli & Vonk,2003, 2004, and also
Hare, Call, Agnetta, and Tomasello (2000) and Hare, Call, Heyes, 1998, for more on the behavioral rules approach).
and Tomasello (2001) noted that almost all of the experi- To be completely confident that chimpanzees sometimes
ments with negative results from the laboratory required know what others know, we must rule out this alternative
cooperative communication with humans (e.g., interpret- hypothesis.
ing a pointing gesture, requesting food, etc.), whereas In the current studies, we developed a new methodol-
many of the potentially positive informal observations ogy – again based on competition with a conspecific – that
from the wild involved competition with conspecifics. enabled us to pursue two goals. First, it enabled us to di-
They therefore devised experiments in which chimpanzees rectly compare the hypothesis that chimpanzees some-
competed with one another for food. Of particular impor- times know what others know to the new evil eye
tance in the current context, Hare et al. (2001) investigated hypothesis. Second, it enabled us to compare chimpanzees
chimpanzees’ understanding of knowledge. They placed a in both a test for knowledge–ignorance and a test of false
subordinate and a dominant chimpanzee into rooms on belief understanding using the same basic methodology.
opposite sides of a third room. Each had a guillotine door The general method was a ‘‘game” in which subject and
leading into this middle room which, when opened at the competitor took turns back-and-forth choosing from a
bottom, allowed them to see into the middle room – and row of three opaque buckets, some of which contained
to see the other individual looking under her door as well. food. In the key condition in the test for knowledge–igno-
There was one piece of food in this middle room, which the rance in Study 1, the task for the subject was to determine
subordinate could always see on her side of one of two bar- which bucket might still contain food after the competitor
riers. The dominant could never see the food at the mo- had chosen a bucket for himself – given that the subject
ment of choice, but in one condition she had witnessed had seen that competitor witnessing the hiding of one of
the hiding process a few moments before (her door was
open at that time and the subordinate could see this),
1
and in another condition not (because her door was down). Hare et al. (2000) focused on chimpanzees’ understanding of visual
perception. Karin-D’Arcy and Povinelli (2002) failed to replicate this study,
The doors for both individuals were then opened (subordi-
but the size of their testing area was too small, which affected the nature of
nates had a slight headstart so that they could not react to the competition. Braeuer, Call, and Tomasello (2007) replicated the original
the dominant’s behavior). The clear finding was that in the Hare et al. (2000) findings with a new set of chimpanzees using the correct
trials in which the dominant had not previously witnessed spacing, and they also demonstrated the crucial role of space in the process.
the food being hidden, subordinates went for the food; in Also, recent studies show that chimpanzee sometimes attempt to conceal
their approach to hidden food from a competitor, further evidence of an
the trials in which the dominant had witnessed the food understanding of visual perception (Hare, Call, & Tomasello, 2006; Melis,
being hidden some moments before, subordinates stayed Call, & Tomasello, 2006; see also Flombaum and Santos (2005), for similar
away. Subordinates seemingly knew whether or not the evidence for rhesus monkeys).
226 J. Kaminski et al. / Cognition 109 (2008) 224–234

the two pieces of food but not the other. To rule out the 2.1. Methods
modified evil eye hypothesis, there were control conditions
in which the subject chose first, in which case it should not 2.1.1. Participants
matter what the competitor had and had not seen. We then Ten chimpanzees (Pan troglodytes) participated in this
used this basic methodology in Study 2 to create a false be- experiment, eight females and two males ranging in age
lief task closely matched to the knowledge–ignorance task from 4 to 29 years. Six apes were nursery reared whereas
in terms of task demands and so forth – and, importantly, four were mother reared. All subjects were housed at the
still in a competitive paradigm with conspecifics, which Wolfgang Köhler Primate Research Center in the Leipzig
has so far not been done. In this study, the subject saw Zoo (Germany), where they lived with conspecifics in a so-
the experimenter mislead the competitor by seeming to cial group and had access to both indoor and outdoor areas.
place the food in one bucket but actually placing it in an- Subjects were tested in their indoor cages, were fed
other – the question being whether the subject could use according to their normal daily routine, and were not food
this information to predict the competitor’s choice. In both or water deprived at any time. Subjects had previously par-
of these studies we also tested human children, so that we ticipated or were currently participating in other studies,
could compare children’s and chimpanzees’ understanding so they were comfortable participating in tests.
of both knowledge–ignorance and false belief all in a single Twelve 6-year-old children also participated in this
experimental paradigm involving competition with a experiment, six boys and six girls ranging in age from
conspecific. 5;11 to 6;02 (year; months). Children were recruited from
kindergartens in a middle-sized German city. Children
2. Study 1: knowledge–ignorance were not informed of the purpose of the study and were
encouraged to compete against an informed adult to obtain
The basic idea of this study is that subjects take turns access to toys. Twelve human adults also participated in
choosing buckets and receiving their contents in the appa- the experiment, four females and eight males ranging in
ratus pictured in Fig. 1. It is a test of knowledge–ignorance age from 21 to 41 years. Prior to the study the adults were
because the subject has to choose before or after her not informed of the purpose of the study, but were told
knowledgeable or ignorant competitor has already chosen. that they should compete to get as many tokens as possi-
The subject’s choice could thus potentially be based on the ble. After the study was completed the participants got full
fact that she has previously witnessed her competitor see- information of the purpose of the study.
ing one of the pieces of reward, but not the other, being
hidden. The new evil eye hypothesis predicts that the sub- 2.1.2. Apparatus
ject should avoid the piece of reward her competitor direc- For the chimpanzees, a table (80 cm  93 cm) with
ted his behavior to (put her evil eye on in that specific three cups (10  12 cm) attached to a sliding board (91
location) irrespective of whether she chooses first or sec- cm  31 cm) was placed on a platform which was located
ond, as he behaved towards it the same in both cases. In just outside their enclosure (see Fig. 1). The platform was
contrast, a more mentalistic hypothesis would predict that placed between two Plexiglas panels in a testing booth
only when the subject chooses second should she avoid the (81 cm  110 cm). At the bottom of each panel were three
piece of reward that the competitor saw being hidden; holes (each 3.6 cm in diameter) arranged in a straight line.
when she chooses first it should not matter what the com- The holes were 29 cm apart, as measured from the center
petitor saw (unless she is considering her second turn after of one hole to the centre of the next. The food reward for
that – which we explain later). the chimpanzees were grapes, pieces of banana, or food-
pellets depending on the individual’s preference.
For the 6-year-olds, a platform (56 cm  46 cm) with a
sliding board attached (46 cm  19 cm) was placed on a ta-
ble. There were three chairs on three separate sides of the
table for the two competing individuals and the experi-
menter to sit on. Three cups (8 cm  9 cm) were placed
on the table to hide the toys. For the children the rewards
were toys. The toys consisted of regular children’s toys
small enough to fit into the cups.
For the adults, a platform (80 cm  93 cm) with a slid-
ing board attached (91 cm  31 cm) was placed on a table.
There were three chairs on three separate sides of the table
for the two competing individuals and the experimenter to
sit on. Three cups (9 cm  11 cm) were placed on the table
to hide the tokens. The tokens consisted of yellow card-
board pieces.

Fig. 1. Experimental set up. Two subjects sat on opposite side of the table. 2.1.3. General procedure
The task was a back-and-forth task in which a subject and a competitor The general procedure was similar for all three groups.
took turns choosing from a row of three opaque buckets, some of which
To give each subject some experience with the general set
contained a reward.
 J. Kaminski et al. / Cognition 109 (2008) 224–234 227

up, each subject received several warm-up trials before the 2.1.3.5. Control: subject first and competitor
experiment started. In a warm up trial both individuals re- visible. Everything was identical to the subject first exper-
ceived all possible relevant information and they chose in imental condition, except that the subject could see the
turns. This was to see if individuals paid attention to the competitor watching while she was choosing. This was
general course of events, that is whether they picked the an additional test of the evil eye behavioral rule hypothesis
reward which was still there after they had seen their com- to make sure that the subject did not avoid the known
petitor choosing. All subjects received at least six of these piece because the competitor made some kind of threat
warm up trials. If an individual subject was not successful or claim on it, or that the subject could tell that the com-
in four or more of these trials it received an additional set petitor was going to choose the known piece by reading
of six trials. Human subjects were instructed not to talk her behavior.
during the entire session, and they were encouraged to Each subject was tested in a total of four sessions: two
compete. Chimpanzee subjects had learned prior to this with the competitor first experimental condition, and two
experiment to poke a finger through one of the holes of a with the subject first experimental condition. In each of
Plexiglas panel to request the container located in front these sessions the subject received six trials in the experi-
of the hole. For test trials, the experimenter (E) sat behind mental condition and three trials in each of the two main
the table with one individual located on the left side and control conditions (all known and nothing known). Within
one individual on the right side such that both subjects each session, order of conditions was randomized. Each
were facing each other (see Fig. 1). A trial started with E subject thus received 12 experimental trials in each exper-
showing the rewards to both individuals. After that E imental condition and 12 trials in each control condition,
started baiting the cups. E always baited two of the three summing to a total of 48 trials altogether. Half of the sub-
cups with food (chimpanzees), cardboard tokens (adult hu- jects started with subject first trials in the first session and
mans), or toys (children), always starting with the cup then received competitor first trials in the second session,
closest to her, then baiting the middle cup, and then the and vice versa in the third and fourth sessions. The other
furthest cup. subjects were given the opposite order. The additional con-
In all conditions in these test trials the subject could see trol condition for the chimpanzee subjects (subject first
the baiting of both cups (and so she also saw which cup and competitor visible) was presented as a follow up and
was empty). What the competitor saw varied depending included 12 trials per subject which were presented to
on condition, with his view blocked by an opaque occluder the subjects in one session.
when necessary. There were two experimental conditions The location of the two rewards in the experimental
and two control conditions: conditions (known and unknown) was randomized and
counterbalanced across trials. After the baiting was com-
2.1.3.1. Experimental: competitor first. The baiting of one pleted, the subject got the first or second choice depending
piece of reward (the known reward) was visible to the sub- on condition. To give an individual the opportunity to
ject and the competitor. The baiting of the second reward choose, E slid the table to one side and, after this individual
(the unknown reward) and the empty cup were visible had made a choice and received the reward, slid the table
only to the subject and not to the competitor. The compet- immediately to the other side. The first choice occurred in
itor started the task by choosing first. private so that neither individual could see the other (ex-
cept in the final control for the chimpanzees); thus, the
2.1.3.2. Experimental: subject first. Everything was identical competitor could not see the subject’s choice when she
to the Competitor First experimental condition, except that chose first and the subject could not see the competitor’s
the subject started the task by choosing first. choice when he chose first. After the first choice in each
trial the view of neither the subject nor the competitor
2.1.3.3. Control: nothing known. The baiting of both pieces was blocked and the table was slid back and forth between
of reward was visible only to the subject and not the com- individuals, allowing them to choose until both pieces of
petitor (and the subject could see this). In half the trials the reward were gone.
competitor chose first and in half the trials the subject
chose first. This control condition was to make sure that
subjects were not making their choices based on some 2.1.4. Scoring and reliability
source of information other than the competitor’s visual For the chimpanzee subjects a choice was considered
experience during baiting. made when she poked her finger through one of the Plex-
iglas holes, where each hole corresponded to the location
2.1.3.4. Control: all known. The baiting of both pieces of re- of one of the three cups. A poke consisted of inserting a fin-
ward was visible to both the subject and the competitor. In ger through one of the Plexiglas holes so that parts of the
half the trials the competitor chose first and in half the tri- finger were visible on the experimenter’s side. If the sub-
als the subject chose first. This control condition was to ject poked through two holes at one time the experimenter
make sure that subjects paid attention to the competitor’s considered the choice to have not yet been made, and she
choice. waited until the subject poked a finger through only one
To further ensure that behavior-reading could not ac- hole. All trials were videotaped with three cameras (one
count for chimpanzees’ behavior in these conditions, we camera filming the subject, one filming the competitor,
conducted an additional control for the chimpanzee sub- and one filming the overall setup). The videotapes were
jects only: later coded by the first author. A second coder blind to
228 J. Kaminski et al. / Cognition 109 (2008) 224–234

condition coded 20% of the trials for reliability purposes, Table 1

also from videotapes. Interobserver reliability was excel- Number of trials subjects aimed for the known, the unknown, and the
empty piece of reward in both conditions of Study 1
lent (Cohens Kappa = 0.92, n = 442). For the Human sub-
jects a choice was considered made when the subject Subject first Competitor first
clearly pointed to one of the cups. A second coder blind Known Unknown Empty Known Unknown Empty
to experimental condition coded 20% of the trials for reli- Chimpanzees 8 3 1 8 4 0
ability purposes from videotapes. Interobserver reliability 7 5 0 4 7 1
was again excellent (Children: Cohens Kappa 0.97, n = 5 6 1 5 7 0
156, Adults: Cohens Kappa = 0.98, n = 502). 5 7 0 3 7 2
7 5 0 3 4 5
6 6 0 4 5 3
2.2. Results 7 5 0 6 5 1
6 5 1 4 7 1
Fig. 2 presents the percentage of trials in which subjects 2 8 2 3 8 1
5 7 0 3 8 1
selected the reward that was known only to them (hidden)
as a function of the order in which they chose (Table 1 pre- Children 6 6 0 0 11 1
5 6 1 5 5 2
sents the raw individual data). Chimpanzees selected the
7 4 1 4 5 3
hidden food significantly more often when they chose sec- 4 8 0 2 8 2
ond than when they chose first (paired sample t-test: 3 7 2 1 10 1
t9 = 3.43, p = 0.007, Cohen’s d = 0.701). Moreover, they se- 5 7 0 3 9 0
lected the hidden food above chance levels when they 10 1 1 1 11 0
11 1 0 1 11 0
chose second (one sample t-test: t9 = 2.35, p = 0.043), and
6 6 0 6 5 2
8 of the 10 chimpanzees did so on the very first trial. In 12 0 0 2 9 1
contrast, they did not select the hidden food at above 9 3 0 3 8 0
chance levels when they chose first (one sample t-test: 5 6 1 3 8 1
t9 = 0.02, p = 0.98). Subjects did not change in the nature Adults 8 4 0 1 11 0
of their choices in the first versus the last half of the trials, 7 5 0 1 11 0
6 6 0 1 10 0
either when they chose second (paired sample t-test:
9 3 0 2 10 0
t9 = 0.110, p = 0.91, Cohen’s d = 0.054) or when they chose 5 7 0 2 10 0
first (paired sample t-test: t9 = 0.669, p = 0.52, Cohen’s 11 1 0 0 12 0
d = 0.309). 6 6 0 6 6 0
A similar analysis of the competitors’ behavior in the 9 3 0 2 10 0
9 3 0 5 7 0
two experimental conditions showed that they used the
9 3 0 0 12 0
predicted search strategy of getting the food that was 4 8 0 0 11 1
known to both partners. Overall, competitors significantly 8 4 0 3 9 0
selected the known food over the hidden food irrespective 8 4 0 1 11 0
of whether they chose first (one sample t-test: t7 = 6.39,
p < 0.0001) or second (one sample t-test: t7 = 4.64,
p = 0.002). Moreover, they selected the hidden piece The children selected the hidden toy significantly more
equally often when they chose first than second (paired often when they chose second than when they chose first
t-test: t7 = 1.85, p = 0.108, Cohen’s d = 0.671). (paired sample t-test: t11 = 3.74, p = 0.003, Cohen’s

* * *
100
Mean % of unknown rewards

80

60
Subject First
Competitor First
40

20

0
CHIMPANZEES CHILDREN (6y) ADULTS

Fig. 2. The mean percentage of experimental trials [±SD] in which subjects from each group chose the unknown piece across condition in Study 1.
 represents significant difference (p < 0.05).
 J. Kaminski et al. / Cognition 109 (2008) 224–234 229

d = 1.68). Moreover, they selected the hidden toy above not influenced by the competitors’ gazing or other behav-
chance levels when they chose second (one sample t-test: ior (paired sample t-test: t8 = 0.69, p = 0.51, Cohen’s
t11 = 5.06, p<0.0001) but not when they chose first (one d = 0.459).
sample t-test: t11 = 1.41, p = 0.19). Children did not change
in the nature of their choices in the first versus the last half 2.3. Discussion
of the trials when they chose second (paired sample t-test:
t11 = 1.05, p = 0.32, Cohen’s d = 0.335). When they chose These results suggest that, at least in some situations,
first, before the competitor, they came to prefer the known chimpanzees know what others know, in the sense of have
toy – paired sample t-test between first and second half: seen. Specifically, in the current study naive chimpanzees
t11 = 3.49, p = 0.005, Cohen’s d = 1.034 – and they chose correctly inferred what a competitor did some moments
the unknown toy below chance in the second half of trials, before based not on the competitor’s behavior, but on what
one sample t-test: t11 = 4.0, p = 0.002. This may suggest the competitor had seen earlier. Thus, subjects made their
that they were learning to anticipate their second chance, choice differently depending only on whether they were
after the competitor chose, and thus chose first the reward choosing first, before the competitor, or second, after the
that was ‘‘at risk”. competitor. When choosing before the competitor, they
The adults selected the hidden reward significantly felt free to choose either piece of food. When choosing after
more often when they chose second than when they chose the competitor, however, they tried to maximize their
first (paired sample t-test: t11 = 6.82, p < 0.001, Cohen’s gains by guessing which cup the competitor had chosen
d = 2.868). Moreover, subjects selected the hidden reward which was in turn based on what she had seen some mo-
above chance levels when they chose second (one sample ments before then. Interestingly, competitor experience
t-test: t11 = 7.34, p < 0.0001) and below chance levels when did not enhance any of the subjects’ performance in the
they chose first (one sample t-test: t11 = 2.71, p = 0.02). The task and had in fact the exact opposite effect for the chim-
adults in this condition were thus clearly anticipating their panzees (footnote 2). This is most likely because perform-
second chance by choosing first the reward that was ‘‘at ing as competitors required no cognitive engagement or
risk”. They did not change in the nature of their choices attention to the other at all – the competitor simply chose
in the first versus the last half of the trials, either when the only piece of food whose location he knew – and this
they chose second (paired sample t-test: t11 = 0.32, inattention to the other likely carried over into the main
p = 0.75, Cohen’s d = 0.078) or first (paired sample t-test: test so that they ignored the competitor’s experience.
t11 = 0.68, p = 0.51, Cohen’s d = 0.285). These results are not susceptible to the same alternative
In the all known control condition, in which everything explanation as those of Hare et al. (2001): the new evil eye
was visible to both individuals throughout, including the hypothesis. That is, in that study subjects could label a
hiding of both pieces of reward and the competitor’s choice piece of food as ‘‘dangerous” as soon as the competitor
(to make sure subjects paid attention to the competitor’s had looked at it (and as ‘‘not dangerous” if not), without
choice), subjects of both species (chimpanzees and hu- any inferences about what the other did and did not know
mans) performed well, as they found the reward success- at the moment of choice. In the current study, if the evil
fully above chance independently of whether they were eye hypothesis were correct subjects should have avoided
to choose first (one sample t-test: chimpanzees: the food piece that the competitor had seen in both exper-
t9 = 10.49, p < 0.0001, children: t11 = 22.99, p < 0.0001, imental conditions equally, but they did not. Moreover,
adults 100% successful) or second (one sample t-test: subjects should have avoided choosing the known piece
chimpanzees t9 = 7.83, p < 0.0001, children: t11 = 18.57, when the competitor was visible (subject first and compet-
p < 0.0001, adults t11 = 23.0, p<0.0001). This control condi- itor visible control), but again, they did not. Instead, chim-
tion shows that subjects were attentive, knew how the task panzees combined their knowledge of what the competitor
worked, and attended to the other’s choice. had seen with the timing of their own and their competi-
We conducted the Nothing Known control condition, in tors’ choices to maximize their food intake. One could ar-
which the competitor did not see the original hiding event gue that chimpanzees’ performance in this task, although
to ensure that concealing the view of an individual truly statistically significant, is not overwhelming in the sense
resulted in this individual being ignorant to the location that the effect is not as strong as it has been in other stud-
of the reward. In both conditions competitors did not find ies (e.g., Hare et al., 2001). However, note that chimpan-
the reward at above chance levels – demonstrating that zees in the current study had to memorize what had
individuals were not using any uncontrolled cues: first been hidden where (as they had no visual access to the
(one-sample t-test: chimpanzees: t7 = 0.43, p = 0.68, food while making their decision) and in addition, had to
adults: t5 = 1.024, p = 0.353) or second (one-sample t-test: inhibit grabbing for the food the competitor had seen. This
chimpanzees t7 = 1.13, p = 0.29, adults: t5 = 0.23, p = 0.83) makes this task more demanding than others, where the
respectively. (Since the competitor for the children was location of the food was perceptually available throughout
an informed experimenter this measure was only analyzed the time of choice.
for the chimpanzees and the adult humans). The fact that children and adult humans also showed
In the subject first and competitor visible control condi- this same basic pattern of results validates the task. How-
tion, which was a direct test of the new evil eye hypothesis, ever, there was one main difference with human adults.
chimpanzee subjects chose randomly (while the competi- When human adult subjects chose first, they preferentially
tor watched) between the piece that was known or un- chose the reward that the competitor had also seen, pre-
known to the competitor, thus showing that they were sumably in anticipation of their next turn after the
230 J. Kaminski et al. / Cognition 109 (2008) 224–234

competitor had chosen. By choosing in this way, adult sub- 3.1. Methods
jects diminished the competitor’s chances of success while
simultaneously increasing their own chances on their next 3.1.1. Participants
turn. Neither 6-year-old children nor chimpanzees fol- Eight chimpanzees participated in this experiment, six
lowed this strategy, suggesting that they were not thinking females and two males ranging in age from 4 to 28 years.
prospectively. However, with some experience, the 6-year- The chimpanzees were randomly recruited from the same
old children developed the same strategy as the adults. groups as the ones participating in Study 1. Two apes were
This could be due to their becoming more experienced nursery reared whereas six were mother reared. Six of the
with the general course of events in the task, but it could subjects had participated in the previous study while two
be due to their facing a more predictable competitor (an were naïve. Twenty 3-year-old human children partici-
‘‘informed” adult) and therefore a more predictable out- pated in this experiment as well, 10 males and 10 females
come than the chimpanzees. ranging in age from 3;04 to 3;09. Twenty 6-year-old chil-
Interestingly, Hare et al. (2000) did document the use of dren also participated, 10 males and 10 females ranging
a similar ‘‘grab the food at risk now” strategy in dominant in age from 5;10 to 6;03. Children were recruited in the
chimpanzees when there was one piece of food that both same manner as Study 1, and were not informed about
the dominant and subordinate chimpanzee could see and the purpose of the study. None of the children had partic-
another piece that only the dominant could see. However, ipated in Study 1.
in that study chimpanzees did not take turns but attempted
to get as much food as possible in the same turn – and the 3.1.2. Apparatus
food was always visible. Thus, targeting the piece that was The apparatus for each group was the same as in Study
‘at risk’ was a strategy that allowed dominant chimpanzees 1. However, an additional small table was placed next to
to maximize their present gains, not their future ones. the subject such that only the subject had access to the
Although there is some recent evidence that apes plan cup (8.5  13.5 for the chimpanzees and 8 cm  9 cm for
for future needs (Mulcahy & Call, 2006), this negative result the children) that was placed on top of it. For each group
is perhaps more surprising in children because even 4-year- there were two types of rewards present during each trial,
olds can already make some claims about future events, a high quality reward (a piece of banana for the chimpan-
and by the age of five to six they have already developed zees and a toy for the children) and a low quality reward (a
some knowledge of future hypotheticals even when these piece of apple for the chimpanzees and a small wooden
include weighing several possibilities (Beck, Robinson, Car- block for the children). The high quality reward was always
roll, & Apperly, 2006). Although it is true that over time placed in one of the cups on the sliding table apparatus,
children, unlike chimpanzees, developed a preference for and the low quality reward was always placed on the table
the ‘at risk’ reward when they chose first, it is unclear next to the subject. The idea was that if the subject was not
whether this was a consequence of differential reinforce- sure if the high quality reward was still available when it
ment or prospective strategic planning. It is conceivable was her turn to choose, she could choose the ‘‘safe” option
that chimpanzees’ and children’s apparent lack of sponta- on the table.
neous prospective strategic planning is related to the likeli-
hood of getting the reward. Note that in the current study,
3.1.3. General procedure
picking the reward at risk does not lead to another reward
The general logic of the paradigm was similar to Study
with certainty but it simply increases the subject’s subse-
1, but slightly different. A trial started with E showing
quent chances for a reward on the next turn. Perhaps chim-
the low quality reward to the subject and placing it under
panzees and children may be less skilful than adults at
the cup on the table beside her; choosing this ‘‘safe” option
computing such probabilities. This is clearly a question
was possible on every trial (presumably when the subject
for future research, but in either case, the advantage of
was unsure if the high quality reward on the sliding table
the adult humans in this experiment may not be in social
was still available). After that, E placed the high quality re-
cognition but rather in some non-social cognitive skill such
ward under one of the three cups on the sliding table. E
as projecting into the future or computing probabilities.
manipulated the cups in a constant order, always starting
with the cup closest to her, then the middle cup, and then
the furthest cup. In all conditions the subject and the
3. Study 2: false belief
competitor both saw the initial baiting of the high quality
reward (initial baiting) on the sliding table. After this, E re-
In this study, using the same basic experimental set up
baited the reward (final baiting) either by lifting it and
as Study 1, the subject saw the experimenter mislead the
then placing it back in its initial location (lift) or by lifting
competitor by seeming to place a single piece of reward
it and placing it in a different location (shift). Whether or
in one bucket but actually placing it in another – and the
not the competitor had information about this final baiting
subject then had to predict the competitor’s choice of the
depended on experimental condition.
incorrect bucket based on his false belief (by knowing
which piece remained when it was her turn to choose).
3.1.3.1. Known lift. After the initial baiting the experi-
This is the first study to test chimpanzees’ and children’s
menter lifted the reward and placed it back in the initial
understanding of false beliefs in a competitive task with
location with both individuals having visual access to this
a conspecific, and the positive findings from Study 1 enable
manipulation.
us to, in a sense, control for basic task demands.
 J. Kaminski et al. / Cognition 109 (2008) 224–234 231

3.1.3.2. Known shift. After the initial baiting the experi- the cups in which the reward had never been located
menter lifted the reward and placed it to a new location (empty cup).
with both individuals having visual access to this Choices were coded as in Study 1 and again a second co-
manipulation. der who was blind to experimental condition coded 20% of
the trials for reliability. Interobserver reliability was excel-
3.1.3.3. Unknown lift. After the initial baiting the experi- lent (chimpanzees: Cohens Kappa = 1.0, n = 59; 3 year old
menter lifted the reward and placed it back in the initial children: Cohens Kappa = 0.93, n = 120; 6 year old chil-
location with only the subject having visual access to this dren: Cohens Kappa = 0.87, n = 120).
manipulation.
3.2. Results
3.1.3.4. Unknown shift. After the initial baiting the experi-
menter lifted the reward and placed it in a new location Recall that in this study subjects always chose second.
with only the subject having visual access to this The analysis was thus a 2  2 ANOVA on the percent of
manipulation. choices of the high-quality reward with the factors:
Whenever necessary, an opaque occluder was held up Manipulation (shift or lift reward) and Witnessing (com-
to block the view of the competitor. After the baiting was petitor did or did not know about the manipulation). As
completed the table was always first slid to the competi- can be seen in Fig. 3, chimpanzees chose the high-quality
tor’s side and the competitor was the first to choose. The cup (where they last saw the reward) more often when
subject never witnessed the competitor’s choice. After their competitor had not witnessed the final baiting than
the competitor made her choice, the table was slid to the when he had, main effect (F1,7 = 14.99, p = 0.006, Partial
subject’s side and it was the subject’s turn to choose. Each g2 = 0.68), thus replicated the findings on knowledge–
subject received six trials per condition with conditions ignorance from Study 1. There were no other significant
presented in a randomized order, summing up to 24 exper- effects, and in particular chimpanzees did not care whether
imental trials. Additionally, each subject received six moti- the reward was shifted in position or not (and this did not
vation trials. Trials were presented in two sessions with 15 interact with Witnessing). In the motivation trials in which
trials each. Motivation trials were interspersed with exper- the chimpanzees could see the other individual choosing
imental trials and had the same general structure as exper- the high-quality reward before their own choice, they
imental ‘‘Shift” trials but here baiting and choice of each of clearly chose the alternative low-quality cup at above
the competing individuals were visible throughout the chance levels (one sample t-test: t7 = 4.16, p = 0.004) show-
trial. These trials were conducted to assure that the sub- ing that they followed the general course of events and
jects followed the general course of events and understood accepted the low-quality reward as an alternative.
the idea of the less valuable reward as an alternative if no We conducted a second repeated measures ANOVA ana-
high quality reward was left. lyzing the overall choice pattern of the chimpanzees. As
chance probabilities for choosing the empty cup (the cup
in which the reward had never been) was different in the
3.1.4. Scoring and reliability
shift (chance probability of choosing an empty cup is
All trials were videotaped. The videotapes were later
33.33%) than in the lift conditions (chance probability of
coded by the first author. We scored the number of trials
choosing an empty cup is 66.67%), we ran this analysis
in which subjects selected each of the three following op-
after correcting the data by subtracting the value expected
tions: (1) the cup where it had last seen the high-quality
by chance from the observed value divided by the value ex-
reward after the second manipulation (high quality cup),
pected from chance. This analysis showed that the chim-
(2) the alternative cup (low-quality cup), and (3) one of
panzees overall preferred some choices over the other as

100 * *
Mean % of high quality reward

80

Kn Shift
60 Kn Lift
Unkn Shift

40
Unkn Lift

20

0
CHIMPANZEES CHILDREN CHILDREN
3-YEARS 6-YEARS

Fig. 3. The mean percentage of trials [±SD] in which subjects from each group chose the high-quality cup (where they had last seen the reward) in the
different conditions of Study 2.  represents significant difference (p < 0.05).
232 J. Kaminski et al. / Cognition 109 (2008) 224–234

the ANOVA did show a significant effect of Choice no evidence of understanding false beliefs. Specifically,
(F2,14 = 19.54, p<0.001, Partial g2 = 0.736). Post-hoc com- chimpanzees chose the high-quality cup on the sliding
parisons revealed that the chimpanzees preferred the apparatus (i.e., where they last saw the high-quality re-
high-quality cup over the empty cup (p = 0.001) as well ward) more often when their competitor had not seen its
as the low-quality cup (p = 0.005). They chose the low- final placement than when he had seen its final placement,
quality cup and an empty cup with equal frequency thus showing an appreciation of the competitor’s knowl-
(p = 1.0), perhaps reflecting a high-risk strategy in favor edge state. But they did not differentiate between two crit-
of the high-quality reward. ical cases in which the competitor did not see the final
The 3-year old children chose the high-quality cup at placement: the case in which the reward was placed back
low levels overall (see Fig. 3). The 2  2 repeated measures in its original location (only lifted), so that the competitor
ANOVA on the percent of choices of the high-quality re- was likely to know where it still was, and the case in which
ward with the factors: Manipulation (shift or lift toy) and its location was shifted, so that the competitor could not
Witnessing (competitor did or did not know about the know where it was. They did not choose in a way that took
manipulation) showed a significant effect for the main fac- advantage of their competitor’s likely false belief when he
tor Manipulation (F1,19 = 6.91, p = 0.017, Partial g2 = 0.267). did not see the shifting of the high-quality reward’s loca-
Overall children chose the high-quality reward more when tion. When chimpanzees did not choose the cup where
the reward had only been lifted than when it had been they last saw the high-quality reward, they chose an empty
shifted – with no attention to the known–unknown manip- cup on the apparatus and the alternative cup on the table
ulation as there was no significant effect of the other main equally often – perhaps reflecting a high-risk strategy in fa-
factors or and no interaction between the two main factors. vor of the high-quality reward in preference to the safe,
In the motivation trials again the three year olds chose the low-quality reward.
alternative low-quality reward at above chance levels (one In contrast, the 6-year-olds not only appreciated knowl-
sample t-test: t19 = 23.13, p < 0.0001) indicating that they edge–ignorance, but also false beliefs. That is, they clearly
accepted the alternative reward as an option, but clearly differentiated the cases in which the competitor did and
in the test trials they were not tuned in to the competitor’s did not see the second, final placement of the reward.
experience. But, in addition, given that the competitor did not see the
The 6-year-old children were affected by both manipu- final placement, they also chose the high-quality cup on
lations (see Fig. 3). The 2  2 ANOVA again with the two the apparatus more often when its location had been
main factors Manipulation (shift or lift toy) and Witnessing shifted (so that the competitor had a false belief about its
(competitor did or did not know about the manipulation) final location) than when it was simply placed back in its
showed a significant effect of the main factor Witnessing original location (only lifted). They knew when the com-
(F1,19 = 116.64, p < 0.0001, Partial g2 = 0.86). Overall the 6- petitor would be fooled into having a false belief about
year-olds chose the high-quality reward more when the the high-quality reward’s location, and they took advan-
competitor had not seen the final baiting than when she tage of this. The behavior of the 6-year-olds would thus
had seen it, thus demonstrating their sensitivity to knowl- seem to validate this task as tapping into some kind of
edge–ignorance, just like the chimpanzees. The other main understanding of false beliefs.
factor, Manipulation, did not have a significant effect. We tested 3-year-old children in hopes of replicating
However, there was a significant interaction between the the chimpanzee pattern of understanding knowledge–
two main factors (F1,19 = 16.54, p = 0.001, Partial g2 = ignorance but not false beliefs. However, this did not hap-
0.465) indicating that the knowledge state of the compet- pen as the behavior of the 3-year-old children was mainly
itor affected the subjects behavior based on how the re- guided by the movement of the high-quality reward during
ward had been manipulated. A comparison between the the final baiting. They actually went for the reward slightly
two conditions in which the competitor was not knowl- more when it was simply lifted and placed back into the
edgeable showed that the children chose the high-quality same location than when it was shifted, regardless of the
cup more when the reward had been shifted than when competitor’s visual access. This may reflect their poor
it had only been lifted (paired sample t-test: t19 = 2.4, memory for the different events in the task rather than
p = 0.027, Cohen’s d = 0.728), whereas the opposite was their understanding of the social situation itself. Moreover,
true in the two conditions in which the competitor was the 3-year-old children had a strong preference to go for
knowledgeable (paired sample t-test: t19 = 2.98, p = 0.008, the low-quality, safe alternative cup on the table, perhaps
Cohen’s d = 0.689). This shows that the children under- reflecting their general uncertainty over the location of
stood that the competitor was less likely to have taken the high-quality reward on the apparatus.
the high-quality reward when he had not seen its location The behavior of the chimpanzees in this study is compa-
being shifted, that is, when he held a false belief about its rable to other false belief studies, which have provided no
location. In the motivation trials again the 6-year-olds convincing evidence of false belief attribution (Call & Tom-
chose the alternative low-quality reward well above asello, 1999; Krachun, Carpenter, Call, & Tomasello, sub-
chance levels (one sample t-test: t19 = 53.1, p < 0.0001. mitted for publication). The difference is that in the
current two studies chimpanzees showed skill in almost
3.3. Discussion exactly the same task when they simply had to appreciate
whether the competitor was knowledgeable or ignorant.
In the current study chimpanzees showed once again an This means that their failure in the false belief version of
understanding of knowledge–ignorance, but they showed this second study is not due to their inability to deal with
 J. Kaminski et al. / Cognition 109 (2008) 224–234 233

the task demands in general. The strategic behavior of the capacity and it emerges in human ontogeny relatively
6-year-old children further validates the task as a test of independent of particular types of individual experience.
false belief understanding. Second is the possibility that chimpanzees have the capac-
ity for understanding false beliefs, but the ‘‘pull of the real”
is even stronger for them than it is for young children. It is
4. General discussion well-known that children’s skills of inhibition and execu-
tive control influence their performance on false belief
The current findings provide evidence that chimpan- tasks (Moses, 2001), and it is also the case that chimpan-
zees understand when others are knowledgeable and igno- zees are less skilled at inhibition and executive control
rant – in the sense of what those others have and have not than are young children (e.g., Vlamings, Uher, & Call,
seen in the immediate past – at least in some situations. In 2006). There are no studies correlating these two sets of
both Study 1 and Study 2 we found that chimpanzees (and skills in chimpanzees, however. Third is the possibility that
human children and adults) preferentially selected a piece children’s development of a fully representational theory
of reward that their competitors had not seen being hidden of mind, including false beliefs, is dependent on several
over one he had seen being hidden. The previous studies of years of linguistic communication – and of course chim-
Hare et al. (2000, 2001) had a number of different control panzees are not evolved for this. There is much evidence
conditions that ruled out very low level explanations of for the role of language in the development of false belief
the chimpanzees’ behavior in the basic competitive para- understanding, including the findings that deaf children
digm. But the behavioral rule as manifest in the new evil who do not learn sign language in the normal way are
eye hypothesis – that chimpanzees would avoid food that much delayed in this task (Peterson & Siegal, 2000) and
another had directed her behavior to at its current location that children who are given special training in certain
– was still a possibility. The current Study 1 eliminated this kinds of linguistic discourse pass the task earlier than those
possibility by showing that under certain conditions they who are not given such training (Lohmann & Tomasello,
do not do this. It is of course possible that some other 2003; see Astington, (2001), for a review). At the moment,
behavior reading explanation (Povinelli & Vonk, 2003; we are unable to choose among these three alternative
Povinelli & Vonk, 2004), could be devised to explain the re- hypotheses, and there are very likely other potential
sults of both previous and current results, but the diversity hypotheses as well.
of findings makes this extremely unlikely. Finally, we found no evidence that either chimpanzees
Study 2 found that in the same experiment in which or 6-year-old children could use their understanding of
they discriminated knowledge and ignorance in their com- knowledge–ignorance to anticipate some steps ahead in
petitor, chimpanzees did not distinguish their competitor’s the task of Study 1, as the human adults did. Although a
true belief from his false belief. If the experimenter tricked number of nonhuman species are able to plan for future
the competitor by moving the high-quality piece of reward events (Clayton & Dickinson, 1998; Mulcahy & Call,
(and the subject saw this), the subject gained no further 2006), humans have developed this ability to a much
advantage over the case in which the competitor simply greater degree (Roberts, 2002; Suddendorf & Corballis,
did not see the hiding at all. In contrast, 6-year-old chil- 1997; Tulving, 2005). And, in the current context at least,
dren did gain an additional advantage when they saw that it would seem that developing the capacity to apply this
the competitor had a false belief over the case in which he ability in mental state understanding takes some ontoge-
was simply ignorant. This finding is consistent with the netic time.
experiment of Call and Tomasello (1999), who also found The field of primate social cognition has made great ad-
that children, unlike chimpanzees and orangutans, pro- vances in the last decade. We have now moved beyond the
vided positive evidence of false belief attribution in a coop- simple question of ‘‘do they or don’t they” have a theory of
erative experimental paradigm. Moreover, our current mind, yes or no? Research with both young children and
chimpanzee results are consistent with the data of Hare various nonhuman primates, especially chimpanzees, has
et al. (2001) in a competitive situation in which chimpan- shown that mental state understanding does not come all
zees once again provided evidence consistent with the at once, but rather it emerges differently for different men-
knowledge–ignorance distinction but not with false belief tal states at different stages of ontogeny – and this may
attribution. In that study, chimpanzees behaved in the even differ across species. We must therefore investigate
same way regardless of whether their competitor was social understanding across species specifically for differ-
uninformed (knowledge–ignorance) or misinformed (false ent mental states on their own terms: goals, intentions,
belief) about the location of the reward. visual perception, auditory perception, attention, perspec-
To fully explain chimpanzees’ success in understanding tive, desire, beliefs, and others. In our view, current re-
knowledge–ignorance, but failure in understanding false search demonstrates that chimpanzees have a basic
belief, we would need a comprehensive theory of the perception-goal psychology in which they understand the
ontogeny of false belief understanding – which does not goal-directed actions and perceptions, and even knowl-
exist. However, there would seem to be three obvious edge (in the sense of what another has seen in the imme-
hypotheses. First is simply that the understanding of be- diate past), of others. At the moment there is no evidence
liefs requires a fully representational theory of mind in a that they understand others’ false beliefs, that they have
way that the understanding of other mental states does a fully representational theory of mind, but that may
not, and chimpanzees simply do not have this fully repre- change if in the future researchers can devise an improved
sentational theory of mind. Humans have evolved this experimental methodology.
234 J. Kaminski et al. / Cognition 109 (2008) 224–234

Acknowledgements Melis, A., Call, J., & Tomasello, M. (2006). Chimpanzees (Pan troglodytes)
conceal visual and auditory information from others. Journal of
Comparative Psychology, 120(2), 154–162.
We are grateful to the Zoo keepers from the Wolfgang Moll, H., & Tomasello, M. (2004). 12- and 18-month-old infants follow
Köhler Primate Research Center for their help with the gaze to spaces behind barriers. Developmental Science, 7(1), F1–F9.
Moll, H., & Tomasello, M. (2006). Level 1 perspective-taking at 24
chimpanzees. In addition, we also thank Hanna Petschauer,
months of age. British Journal of Developmental Psychology, 24(3),
Kristin Liebal, Eva Leermann, Marlen Kaufmann, Anke Rad- 603–613.
ke, and Yvonne Rekers for help with data collection and Moll, H., & Tomasello, M. (2007). How 14- and 18-month-olds know
what others have experienced. Developmental Psychology, 43(2),
reliability coding. Alicia Melis provided valuable com-
309–317.
ments on an earlier version of this manuscript. Moses, L. J. (2001). Executive accounts of theory-of-mind development.
Child Development, 72(3), 688–690.
Mulcahy, N. J., & Call, J. (2006). Apes save tools for future use. Science, 312,
References 1038–1040.
Onishi, K. H., & Baillargeon, R. (2005). Do 15-month-old infants
Astington, J. W. (2001). The paradox of intention: Assessing children’s understand false beliefs? Science, 308, 255–258.
metarepresentational understanding. In B. F. Malle, L. J. Moses, & D. A. Peterson, C., & Siegal, M. (2000). Insights into theory of mind from
Baldwin (Eds.), Intentions and intentionality: Foundations of social deafness and autism. Mind and Language, 15(1), 123–145.
cognition (pp. 85–103). Cambridge, MA, US: The MIT Press. Povinelli, D. J., Rulf Alyssa, B., & Bierschwale Donna, T. (1994). Absence of
Beck, S. R., Robinson, E. J., Carroll, D. J., & Apperly, I. A. (2006). Children’s knowledge attribution and self-recognition in young chimpanzees
thinking about counterfactuals and future hypotheticals as (Pan troglodytes). Journal of Comparative Psychology, 108(1), 74–80.
possibilities. Child Development, 77(2), 413–426. Povinelli, D. J., & Eddy, T. J. (1996). What young chimpanzees know about
Behne, T., Carpenter, M., Call, J., & Tomasello, M. (2005). Unwilling versus seeing. Monographs of the Society for Research in Child Development,
unable: Infants’ understanding of intentional action. Developmental 61(3), 1–152.
Psychology, 41(2), 328–337. Povinelli, D. J., & Vonk, J. (2003). Chimpanzee minds: Suspiciously
Braeuer, J., Call, J., & Tomasello, M. (2007). Chimpanzees really know what human? Trends in Cognitive Sciences, 7(4), 157–160.
others can see in a competitive situation. Animal Cognition, 10(4), Povinelli, D. J., & Vonk, J. (2004). We don’t need a microscope to explore
439–448. the chimpanzee’s mind. Mind and Language, 19(1), 1–28.
Brooks, R., & Meltzoff, A. N. (2002). The importance of eyes: How infants Premack, D., & Woodruff, G. (1978). Does the chimpanzee have a theory of
interpret adult looking behavior. Developmental Psychology, 38(6), mind? Behavioral and Brain Sciences, 1(4), 515–526.
958–966. Roberts, W. A. (2002). Are animals stuck in time? Psychological Bulletin,
Byrne, R. W., & Whiten, A. (1990). Tactical deception in primates: The 128(3), 473–489.
1990 database. Primate Report, 27, 1–101. Savage-Rumbaugh, E. S., Rumbaugh, D., & Boysen, S. (1978). Sarah’s
Call, J., & Tomasello, M. (1999). A nonverbal false belief task: The perfor- problems in comprehension. Behavioral and Brain Sciences, 1,
mance of children and great apes. Child Development, 70(2), 381–395. 555–557.
Clayton, N. S., & Dickinson, A. (1998). Episodic-like memory during cache Schwier, C., van Maanen, C., Carpenter, M., & Tomasello, M. (2006).
recovery by scrub jays. Nature, 395(6699), 272–274. Rational imitation in 12-month-old infants. Infancy, 10(3), 303–311.
Flombaum, J., & Santos, L. (2005). Rhesus monkeys attribute perceptions Suddendorf, T., & Corballis, M. C. (1997). Mental time travel and the
to others. Current Biology, 15(5), 447–452. evolution of the human mind. Genetic, Social, and General Psychology
Gergely, G., Bekkering, H., & Kiraly, I. (2002). Rational imitation in Monographs, 123(2), 133–167.
preverbal infants. Nature, 415(6873), 755. Tomasello, M., & Call, J. (1997). Primate cognition. New York, NY, USA:
Gergely, G., Nadasdy, Z., Csibra, G., & Biro, S. (1995). Taking the intentional Oxford University Press.
stance at 12 months of age. Cognition, 56(2), 165–193. Tomasello, M., & Haberl, K. (2003). Understanding attention: 12- and 18-
Hare, B., Call, J., Agnetta, B., & Tomasello, M. (2000). Chimpanzees know month-olds know what is new for other persons. Developmental
what conspecifics do and do not see. Animal Behaviour, 59(4), Psychology, 39(5), 906–912.
771–785. Tulving, E. (2005). Episodic memory and autonoesis: Uniquely human? In
Hare, B., Call, J., & Tomasello, M. (2001). Do chimpanzees know what H. S. Terrace & J. Metcalfe (Eds.), The missing link in cognition: Evolution
conspecifics know? Animal Behaviour, 61(1), 139–151. Jan 2001. of self-knowing consciousness. New York: Oxford University Press.
Hare, B., Call, J., & Tomasello, M. (2006). Chimpanzees deceive a human Vlamings, P., Uher, J., & Call, J. (2006). How the great apes (Pan troglodytes,
competitor by hiding. Cognition, 101(3), 495–514. Pongo pygmaeus, Pan paniscus, and Gorilla gorilla) perform on the
Heyes, C. M. (1998). Theory of mind in nonhuman primates. Behavioral reversed contingency task: The effects of food quantity and food
and Brain Sciences, 21(1), 101–134. visibility. Journal of Experimental Psychology: Animal Behavior
Karin-D’Arcy, M., & Povinelli, D. J. (2002). Do chimpanzees know what Processes, 32(1), 60–70.
each other see? A closer look. International Journal of Comparative Wellman, H. M., Cross, D., & Watson, J. (2001). Meta-analysis of theory-of-
Psychology, 15(1), 21–54. mind development: The truth about false belief. Child Development,
Krachun, C., Carpenter, M., Call, J., & Tomasello, M. (submitted for 72(3), 655–684.
publication). Children Read Minds but Apes Read Reaches in a Wellman, H. M., & Liu, D. (2004). Scaling of theory-of-mind tasks. Child
Competitive, Nonverbal False Belief Task. Development, 75(2), 523–541.
Lohmann, H., & Tomasello, M. (2003). The role of language in the Wimmer, H., & Perner, J. (1983). Beliefs about beliefs – Representation
development of false belief understanding: A training study. Child and constraining function of wrong beliefs in young childrens
Development, 74(4), 1130–1144. understanding of deception. Cognition, 13(1), 103–128.