Learning new manual skills by observation: Comparison of Human Infants and
Macaques
J. Fagard1, R. Breveglieri2, A. Bosco2, R. Esseily1, J. Nadel3, P. Fattori2
1
Laboratoire Psychologie de la Perception, Université Paris Descartes, CNRS UMR 8158, Centre
Biomédical des Saints Pères, 45 rue des Sts Pères, 75006 Paris, France ; E-mail: jacqueline.fagard@univparis5.fr
2
Dipartimento di Fisiologia Umana e Generale di Bologna, Piazza di Porta San Donato 2, 40126,
Bologna, Italia
3
Laboratoire Vulnérabilité, Adaptation et Psychopathologie, Université Pierre et Marie Curie - CNRS
UMR 7593, Hôpital de la Pitié-Salpêtrière, Pavillon Clérambault, 75013 Paris, France
* Corresponding author
1
Abstract
We investigated manual skill in three young macaques, and we explored to what
extent the observation of a human performing a task would help them successfully achieve a
task they had previously failed to perform. The macaques were compared with 8- to 18month-old human infants on seven object-retrieval tasks of increasing level of difficulty. In
the spontaneous condition, the macaques exhibited a level of skill comparable to the infants.
Regarding the ability to acquire new manual skills after observation, we found some change
of strategies in 12- and 15-month-old infants, but the rate of success increased significantly
only for 18-month-olds. After observation, each macaque exhibited one additional success,
but still failed to perform the majority of the tasks. None of these successes concerned tooluse tasks or 18-month-old tasks. This is the first study directly comparing human infants and
macaques on observational learning of new motor skills using the same tasks. We concluded
that captive macaques’ capacity to learn from observation of a human modeler is restricted to
relatively simple tasks, and is lower than that of 18-month-old human infants.
Keywords: manual skill, macaques, human infants, observational learning
2
Adapting to the environment requires sensori-motor skills. One way to learn new
sensori-motor skills is through practice, i.e. using visual and proprioceptive feedback from the
preceding attempt to correct errors during the next one. Another way, which may be less
costly, is by observation of other more skilful individuals. In social species, learning by
observation may be an important mechanism. In a previous study (Fagard, Esseily, & Nadel,
submitted), we observed an age-related increase in human infants' capacity to benefit from
observation of an adult when learning a new manual skill. The tasks were different for each
age group, from 8 to 18 months, and they were chosen so that each task would be slightly too
difficult to be spontaneously achieved successfully. In the study presented here, we
investigated to what extent macaques, compared with human infants, could benefit from
observing a human who demonstrated a new action in front of them. In order to compare the
effect of observational learning between monkeys and infants on tasks with the same relative
level of difficulty, we tested the macaques on all the manual tasks presented to 8- to 18-month
human infants. We first observed at what age level a macaque could be compared with human
infants in term of spontaneous success, and second, to what extent the observation of a human
adult would help the macaque when the task was too difficult to be spontaneously achieved
successfully.
There is a long tradition of imitation studies in human infants and in non human
primates, boosted recently by the increasing knowledge about the common neural pathways
that underlie the observation of an action and its actual execution (Iacoboni et al., 1999), the
so-call mirror-neuron system (Gallese, Fadiga, Fogassi, & Rizzolatti, 1996; Rizzolatti, Fadiga,
Gallese, & Fogassi, 1996), and by the observation of motor facilitation in motor evoked
potentials of the hands of individuals observing an experimenter grasping something (Fadiga,
Fogassi, Pavesi, & Rizzolatti, 1995): it is clear from all these studies that observing another
person performing an action influences the neuro-motor system of the observer.
In the study presented here, we were interested in observational learning rather than in
imitation: we refer to observational learning, as opposed to imitation, when the observed
action applies to an act that is absent from the participant’s motor repertoire (a novel act and
not only a novel object), and when the observer is not allowed to perform the modeled action
before a delay. Researchers concerned with imitation and observational learning usually make
a distinction between reproduction of means to reach the goal and emulation learning
3
(reproduction of the goal, understanding of the affordance of an object) (Byrne & Russon,
1998; Nagell, 1993; Tomasello, 1998; Tomasello, Davis-Dasilva, Camak, & Bard, 1987).
In humans, observational learning has been mostly studied in adults’ or children’s
learning of a new skill. These studies have shown a significant advantage of modeling over
practice-only condition for motor learning, with more or less gain as a function on where the
emphasis of modeling was (form versus outcome of action), of the age of the observer (see
Ashford, Davids, & Bennett, 2007’s meta analysis for a review), and, probably also of the
kind of skill to be learned. In human infants, there are many studies on imitation, in the
context of social cognition, such as recognizing the goal of the modeler (Gergely, Bekkering,
& Kiraly, 2002; Meltzoff, 1995). A few studies have focused on imitation of simple actions
(Abravanel, Levan Goldschmidt, & Stevenson, 1976; Meltzoff, 1988), or of more complex
means-end actions (Elsner & Aschersleben, 2003; Provasi, Dubon, & Bloch, 2001). These
studies showed that infants can understand specific relations between observed actions and
effects by the second year, when they start to benefit from observation of others’ actions to
organize their own.
Observational learning has been observed in several species of non human primates,
for example in cultural behaviors restricted to small geographic areas in monkeys (de Waal,
1996), or in the spreading of tool-use throughout a population, particularly in chimpanzees
(Boesch, Marchesi, Marchesi, Fruth, & Joulian, 1994; Whiten et al., 1999), less frequently in
monkeys (Ducoing & Thierry, 2005). Learning by observation has also been observed in
object-reward associations in rhesus monkeys living in a semi-natural habitat (Meunier,
Monfardini, & Boussaoud, 2007). The basis for imitation seems present at birth in monkeys as
in humans, since, for example, rhesus macaques are capable of neonatal imitation, a capacity
that disappears in infant macaques after two weeks (Ferrari et al., 2006). There are examples
of imitation in older monkeys, such as tongue protrusion and hand actions in Macaca fuscata
(Kumashiro et al., 2003), or opening a box with the mouth or the hand as a function of the
experimenter’s modeling in marmosets (Voelkl & Huber, 2000).
It is sometimes argued that young infants are poor at learning affordances by
observation (Want & Harris, 2001), whereas non human primates are poor at imitating
(Fragaszy & Visalberghi, 1989; Visalberghi & Fragaszy, 1996), and better at reproducing the
goal of the observed action (Call, Carpenter, & Tomasello, 2005; Nagell, 1993; Whiten,
Custance, Gomez, Teixidor, & Bard, 1996). Few studies have directly compared human
children and chimpanzees on observational learning of the same tasks (Hopper, Lambeth,
4
Schapiro, & Whiten, 2008; Nagell, 1993). Hopper’s study showed that chimpanzees were
capable of emulation learning in a ghost condition in relatively simple tasks, and only
fleetingly, whereas 3- to 4-year-old human children showed long-term observational learning.
Another study showed that enculturated chimpanzees (raised in a human-like environment,
with everyday activities of the type human children engage in routinely) show a capacity for
real imitation which is not different from that of human children (Tomasello, Savage
Rumbaugh, & Kruger, 1993; see also Hayes & Hayes, 1952, and, for a similar observation in
capuchin monkeys, Fredman & Whiten, 2008). To our knowledge, there is no study directly
comparing human infants and macaques on observational learning (for a comparison between
human children and macaques on cognitive imitation, see Subiaul, Cantlon, Lurie, Holloway,
& Terrace, 2003). Since brain-imaging studies can be easily performed on non human
primates, it would be useful to know to what extent it makes sense to use monkeys as a model
for observational learning, as compared with humans.
We compared three macaques with 43 infants in which observational learning had
been studied on tasks with a comparable relative level of difficulty for the four age groups (8,
10, 12, 15, and 18 months). We did not know in advance which of the age-level tasks the
macaques would spontaneously succeed in accomplishing. Diamond compared human infants
and monkeys on one manual skill, the Object Retrieval task, in which the individual is
presented with an object located behind a transparent barrier, so that a detour reaching is
required for successful grasping (Diamond, 1990). This task is successfully achieved even by
very young infant monkeys, as long as they have intact dorsolateral prefrontal cortex,
responsible for the capacity to inhibit a direct reach. Thus, we first evaluated the macaques’
level of manual skill on all human infant tasks, starting with the 8-month-old up to the 18month-old tasks. Then, for each of the failed tasks, we observed the effect of observing a
human adult performing the task on the outcome during the trial following this observation.
These tasks required understanding different spatial relationships between the object to
be grasped and its environment, and figuring out the action to retrieve the object, which
involves one to three steps. For these, a developmental gradient has been demonstrated in
human infants (Bruner, Lyons, & Watkins, 1969; Diamond, 1981; Fagard & Jacquet, 1989;
Lockman, 2000). The relationships were: “behind, one step” (8 months), “inside without need
for opening, one step” (10 months), “inside after opening a lid, two steps” (12 months),
“inside after turning upside-down, two steps” (15 months), “beside with the need of a tool,
three steps,” (15 months), “inside with the need of the finger as a tool, two steps” (18
5
months), and “inside with the need of a tool and change in orientation, three steps” (18
months). The actions to be used for successful retrieval changed from making a detour (at 8
months), using bimanual simultaneous movements with only one hand being active (at 10
months), performing a two-step action with both hands (at 12 months), changing the
orientation of the container (at 15 months), using a tool for a three-step action (at 15 months),
using a tool with a change in orientation for a three-step action (at 18 months), performing a
two-step action using the finger as a tool (at 18 months).
Method
Participants
Macaques
Three young macaques weighing between 3 and 4 kg (Macaca Fascicularis), Monkey
C (age = 64 months), Monkey L (age = 39 months), and Monkey G (age = 40 months) were
tested. All three macaques received the same protocol as described below. Experiments were
carried out in accordance with National Laws on care and use of laboratory animals and with
the European Community Council Directive of 24 November 1986 (86/609/EEC), and were
approved by the Bioethical Committee of the University of Bologna.
Infants
Forty-three 8- to 18-month-old infants, divided in five age groups, were compared to
the macaques. There were nine 8-month-olds (6 boys and 3 girls), nine 10-month-olds (7 boys
and 2 girls), nine 12-month-olds (5 boys and 4 girls), nine 15-month-olds (7 boys and 2 girls),
and seven 18-month-olds (3 boys and 4 girls). The infants were recruited from the city of
Paris birth record. For some reasons, the boys' parents seemed more willing to participate in
the study than the girls', but the ratio did not differ significantly across age groups. For the
infants, parental consent was granted prior to the experiment which was conducted in
accordance with the ethical standards specified in the 1964 Declaration of Helsinki.
Tasks
There were seven objects, each corresponding to one age group, except for the 15month-old and the 18-month-old human infants. We chose to present two objects to the two
older groups of human infants because we were less sure of what a "slightly too difficult" task
could be at that age, and we expected the older infants to be willing to perform more trials
6
than the younger ones. The seven objects were presented to the macaques, whereas each
human infant was only presented with the object (or the two objects) corresponding to his age
group. The objects were identical or almost identical for macaques and infants: however, for
infants what was to be retrieved was always a toy whereas for the macaques it was always a
piece of food (piece of apple, banana, or candy), except for one object (“Tube-out-of-thecontainer”).
1/ Detour Reaching (8-month-olds): the object was a rectangular transparent plastic
box (8.2 x 5.5 cm, 12.6 cm high), with only the two lateral sides open. A toy (infants) or a
piece of food (macaques) was placed right behind the front wall of the box. The subject had to
resist the drive to grasp the seen object straight ahead, and had to do so by making a detour.
2/ Tube-out-of-the-container (10-month-olds): the object was a wooden container (2.4
x 2.5 cm, 9.2 cm high) inside which was inserted a plastic tube with a bright red cap,
protruding from the container by 2.5 cm. The task required the participant to pull the tube out
of the container.
3/ Box (12-month-olds): the object was a box with a transparent lid hinged to it and
with a toy (infants) or food (macaques) visible inside. The box was slightly different for the
infants and for the macaques: for the infants, it was a small transparent plastic box. For the
macaques the food was hidden in a 11.9 x 15 cm, 5.7 cm high wooden box with a transparent
lid hinged to it. The task was to raise the lid with one hand while grasping the object with the
other.
4/ Bottle with a peg inside (raisin for macaques) (15-month-olds): the neck of the
bottle was too narrow (1cm½) for a finger to be inserted in it to retrieve what was inside, so
that to retrieve the peg (or the raisin) it was necessary to turn the bottle upside down. The
bottle itself was 4 x 8 cm.
5/ Rake (Tool-use, 15-month-olds): the rake (7 x 20 cm) was placed in the middle of
the table (30 x 20 cm plastic board for macaques), within reach, in front of a plastic block
(piece of food for macaques) which was placed out of reach, behind the rake. In order to grasp
the block (or food), the rake had to be used as a tool to bring it within reach.
6/ Finger-in-the-lid (18-month-olds): the object was a 12.8 x 5.5 cm, 8.5 cm high
transparent plastic box, with a lid which had a hole (1.9 cm diameter) in the middle, and a
plastic block (infants) or a piece of food (macaques) at the bottom of the box. The participant
had to open the lid using one finger of one hand, and grasp the toy (food) with the other hand.
7
7/ Stick (infants) or fork (macaques) (tool used vertically, 18-month-olds): the object
was a 5.5 x 7.2 cm, 8.5 cm high transparent plastic box, half covered with a piece of tape, and
a plastic block (infants) or a piece of food (macaques) inside the box. A 14 x 2cm wooden
stick (infants) or a plastic fork (2.6 x 18 cm) (macaques) was placed to the side of the box. In
order to get the plastic block, the infants must grab the stick which terminated with a piece of
“Velcro”, touch the block with the tool: the plastic block, on which the other half of the
“Velcro” had been glued, could then be retrieved. In order to retrieve the food, the macaque
had to grasp the plastic fork, hit the banana with the fork, and get it out of the box. A tape
covering part of the top prevented from grasping the block or the food with the hand.
Protocol
The protocol was similar for infants and macaques. The object was presented to the
participant for 30 sec, during which he could explore it. After this spontaneous trial, the
experimenter showed the participant how to perform the task, three times in a row. For that,
she was particularly cautious to check that the participant was looking at her throughout each
demonstration. This was followed by a two-minute delay. During the delay the infants were
shown other interesting objects, such as mechanical toys or finger puppets, for example; the
macaques were given raisins. After the two-minute delay, the object was given back to the
participant. Each infant only received the object (or objects) corresponding to his age group.
The macaques received the object in chronological order (starting with the object of the 8month-olds and ending with that of the 18-month-olds).
The infants were tested in a quiet room in presence of their parents (mother, father or
both) who were instructed not to intervene with their child during the whole experiment. The
infant sat on a high chair in front of a table, and the experimenter sat on the other side of the
table. The experimenter presented the object to the infant on the table for him to explore it
spontaneously, and then made the demonstration across the table.
The macaques sat in a primate chair in front of the experimenter. The objects were
presented on a plastic board, fixed (for some of them) at comfortable arm length in front of
the animal. The different objects were always presented in the same spatial position, and the
animal could see that a piece of food could be grasped (except for the “Tube out of the
container”).
8
Data collection
Infants: a digital video camera directed at the infant recorded the whole experiment. A
frame by frame analysis could then be made on all recordings.
Macaques: two cameras were used: one was an infrared digital camcorder put in front
of the animal so as to record the gaze toward the object and the experimenter. This was to
insure the animal was looking at the demonstration. The second camera was a regular digital
camera, placed on the right side of the monkey, in order to record his action on the object.
Data analyses
To compare the percentage of success as a function of group, we used Chi square
analyses. MANOVA were computed to compare the score (0,1) before and after observation,
for infants or macaques separately, or between infants and macaques, with trial as repeated
measures. Sometimes we also compared the score before and after observation task by task,
using a t-test.
Results
Spontaneous success in human infants
Since the tasks had been devised to be equivalent across age groups, we first ensured
that the rate of success did not differ significantly across age groups. This was made by
comparing the frequency of success as a function of age in the trial before demonstration.
There was a low percentage of success at eight months of age, and no success at eighteen
months for the Stick task (see Table 1). We tested whether the percentage of success differed
with age, first considering Bottle and Finger-in-the-lid for the 15- and 18-month-olds, then
considering tool use for both age groups (Rake and Stick). In both cases, a chi2 analysis
showed that the age-related difference was not significant. We thus considered that the
comparison of the effect of observation on learning could be made across ages on our tasks.
Insert Table 1 about here
Learning through practice in human infants
To see whether the participants were able to discover how to retrieve the object
through practice, we compared the frequency of success during the first 10 seconds of the 30second trial, with the next 10 seconds and with the last 10 seconds. This analysis of the 30second exploration showed that the infants did not succeed significantly better after 10 or 20
9
seconds of practice: although the majority of the few successes happened between 10 and 20
seconds after receiving the object, the difference in frequency of success between the three
time periods was not significant. On a total of 12 successes, two were obtained during the first
10 seconds, seven were obtained during the next 10 seconds, and three were obtained during
the last 10 seconds. In addition, and although statistics could not be performed on such a
small number of occurrences, it did not seem to differ between age groups.
Spontaneous success in macaques
As can be seen in Table 2, on the trial before demonstration there were only three successes
out of the 21 possible ones (3 macaques x 7 tasks): one macaque was successful at Detour
Reaching and at Bottle. One macaque was successful at the Box (the result was not clear on
Detour Reaching for this macaque: it was the very first object that we tested and the animal
threw the object away out of fear; the experimenter, for whom it was also the first object with
the first macaque tested, did not realize in time that the 30 seconds were not over). The third
macaque failed at all tasks in the spontaneous condition (see Table 3).
Insert Table 2 about here
Learning through practice in macaques
As for the infants, we checked whether the three successes occurred during the first 10
seconds of the 30-second trial, during the next 10 seconds and during the last 10 seconds, to
see if the macaques succeeded after some practice with the objects. One success occurred
before 10 seconds (Detour Reaching), another one occurred after 17 seconds (Box) and the
last one after 26 seconds (Bottle). Thus, we observed no tendency.
Insert Table 3 about here
Success before demonstration: Comparison of human infants and macaques
As can be seen in Figure 1, there were no striking differences between frequencies of
success before demonstration in macaques as compared with infants. However, the macaques
tended to be better on the task targeted toward the youngest infants (Detour Reaching, 8month-old task), but less successful at the tasks targeted toward the oldest infants, more
clearly so on one of the tasks of the 18-month-olds (Finger-in-the-lid). We computed a Chi
square to compare the rate of success between human infants and macaques for each task
separately. The difference between the two groups was never significant.
10
Insert Figure 1 about here
The strategies leading to failure were often similar in human infants and macaques, at
least for some of the objects: trying to grasp the object through the transparent wall or holding
the front wall between both hands, for Detour Reaching; bringing the tube into the mouth,
pulling slightly the tube with one hand without finishing the action, turning the object upsidedown, for Tube-out-of-the-container; placing the hands on the lid and exploring the box with
both hands, for the Box; bringing the bottle to the mouth, turning the bottle, for Bottle; trying
to reach through the transparent wall, exploring the box with both hands, for Finger-in-the-lid;
ignoring the rake and begging for the object, for Rake; neglecting the tool for the 18-monthold tool-use task. However, some behaviors were observed in macaques but not in human
infants, such as smelling objects, even the object without food, munching objects (wooden
container of the tube, bottle, rake, and fork), looking through the neck of the bottle with one
eye, throwing away the tool or eating pieces of it (we used several plastic forks for these
experiments!). Conversely, some behaviors were typical of human infants and not observed in
macaques, such as shaking the wooden container or the bottle repeatedly, trying to retrieve the
object from the box unimanually, or banging on the box repeatedly.
Learning by observation in human infants
To evaluate the influence of observation, we analyzed the change in success between
the trial before observation and the test trial (after observation). We computed an age x trial
(repeated measures) MANOVA on the success rate (0 or 1). When the 15-month- and the 18month-old tasks considered were respectively Bottle and Finger-in-the-lid, it showed no age
effect, no effect for trial, and no significant age x trial interaction. When the tasks of the15and 18-month-olds considered were the tool-use tasks (respectively Rake and Stick), it
showed no main effect for age, a significant effect for trial (F(1,37) = 17.9; p<.001), and a
significant age x trial interaction (F(4,37) = 8.7; p=.0001; see Table 1). A post-hoc LSD test
showed that the age effect was due to the 18-month-olds: the difference in success between
before observation and test trial following observation was significant only at 18 months, thus
for Stick (p<.00001). In addition, when a t-test for matched samples was calculated
independently for each of the seven tasks, between the trial before and the trial after
observation, it showed a significant effect for Finger-in-the-lid (t (6) = -2.8, p<.05).
11
In addition, we looked whether the strategies of the infants leading to failure were
different after observation of the adult. In other words, we wondered whether the infants
changed their behaviors even when the latter did not lead to success. We observed a few
changes only for Rake: some infants tried to reach for the object with the rake with clumsy
movements across the table.
Learning by observation in macaques
As for infants, to evaluate the influence of observation we analyzed the change in success
between the before observation trial and the test trial (after observation). Table 2 shows that
observing a human performing the task changed failure into success for three instances only:
Detour Reaching for Monkey C, Tube-out-of-the-container for Monkey L, and Bottle for
Monkey G. A MANOVA on the score of success as a function of condition (x 2, before and
after demonstration) and for tasks (x 7) indicated no significant effect for condition, no
significant effect for task, and no task x condition interaction. On three of the most difficult
tasks (Rake, targeted to 15-month-olds, and Fork and Finger-in-the-lid, targeted to 18-montholds), there were zero successes even after demonstration. Two out of the three changes after
demonstration occurred on tasks on which at least one macaque had been successful before
demonstration (Detour Reaching, targeted to 8-month-olds, and Bottle, targeted to 15-montholds; see Table 3).
As done for infants, we also checked whether there was a change in behavior after
observation of the demonstration followed by failure. We only observed a few changes:
Monkey C pulled the tube partially out of the container after demonstration, Monkey L kept
the rake in one hand while begging for the piece of food with the other hand, Monkey L kept
his hands on top of the box for Finger-in-the-lid, and Monkey G munched the fork.
Learning by observation: Comparison of human infants and macaques
To compare the influence of observation on success between human infants and
macaques, we conducted a MANOVA on the rate of success, with group as independent
variable (x 2), and trial as independent variable with repeated measures (x 2; before and after
demonstration). This was calculated for each task separately. We observed no group or trial
effect and no significant interaction for the following tasks: Object Retrieval, Tube-out-of-the
container, Box, Bottle, and Rake. For the 18-month-old tasks the lack of variance kept us
from computing a MANOVA. We computed a Chi square to compare the rate of success after
12
demonstration between human infants and macaques for these two tasks. The difference was
significant for both tasks (x2 (1) = 10; p <.01). Thus18-month-old infants learned from
observation whereas macaques did not (see Figure 2).
Insert Figure 2 about here
Discussion
We devised tasks with the goal that they would show a comparable level of difficulty
for infants of different age, in order to compare the effect of observational learning in human
infants across ages, and with macaques. To do so, we changed the object-environment
relationship (behind, inside, etc.) so that the retrieval becomes increasingly difficult
(involving detour, bimanual movement, two-step action, tool-use, etc.). There was some
variation in the frequency of spontaneous success for the different age groups. However, we
succeeded in devising tasks which were never spontaneously successfully completed by more
than 50% of the infants, for each age group, and for which the age effect on the rate of success
was not significant when all age groups were compared. We thus considered that we could
compare the effect of observational learning across ages on these tasks. The same tasks were
presented to the macaques.
When infants succeeded spontaneously, they did not do it more toward the end of the
30-second trial. This indicates that when infants did not know from the start what they could
do with the object, they did not discover it through manipulation, at least within this time
period.
There were very few spontaneous successes by the three macaques at the different
tasks. None of the 18-month-old tasks were successfully completed and the tool-use task of
the 15-month-olds was also failed. There was only one success (out of three) at each of the
three following tasks: Detour Reaching (8-month-old task), Box (12-month-old task), and
Bottle (15-month-old task). Thus, discovering how to get a piece of food behind a transparent
wall, how to raise the transparent lid of a box with one hand while retrieving the piece of food
from inside with the other hand, and how to put a bottle upside down in order to get a raisin
out of it, was something possible without training at least for some of the macaques. In term
of spontaneous success, the macaques were not noticeably different from 8- to 18-month-old
infants, even if they tended to be better that 8-month-olds and slightly less skilful than older
infants, in particular 18-month-olds. The failure on the tool-use tasks within such a short
period of practice was not unexpected. Tool-use has been observed in macaques, but after
13
many trials and errors when in captivity (Beck, 1976), or in free or semi-free habitat where
practice is not limited (Ducoing & Thierry, 2005; Sinha, 1997; Tokida, Tanaka, Takefushi, &
Hagiwara, 1994). The failure on the 10-month-old task (Tube-in-the-container) is probably
due to the fact the no food reward was associated with its achievement. It was interesting to
note, though, that after Monkey L succeeded at this task, he seemed interested in doing it
repeatedly.
We observed some common failure behaviors between human infants and macaques
(trying to grasp objects through transparent walls, turning the object upside-down, mouthing
the tube, begging for the far-away object or food), together with species-specific behaviors
(banging or shaking objects repeatedly for human infants, smelling objects for macaques). It
would be interesting to check with macaque infants whether the repetitive banging, for
example, is a real species-specific characteristic (alternately, it could be due to the young age
of human infants).
Observation was clearly effective in human infants at 18 months of age for both tasks.
Before, there was some effect, as indicated by slightly more success at 10 months and at 15
months for Bottle, and by some changes in strategies after observation in 12- and 15-montholds, but without significant change in terms of success. It is interesting to note that for tooluse, observation did not help much 15-month-old infants, as it did for 18-month-olds. These
results indicate that observation becomes more effective as the infant grows up. They agree
with previous studies showing that understanding the relation between action and effects by
observational learning becomes more effective after 12 months of age (Elsner &
Aschersleben, 2003; Provasi et al., 2001). In addition, for a same age group, observation was
more or less effective depending on the task, whether for example, it involved tool use or not.
The macaques did not learn much through observation of a human. For the three
macaques altogether, there were only three examples out of the 18 failures (in spontaneous
condition) for which observation was followed by success. In addition, each macaque
improved for one task only, and for a different one at that for the three macaques. Observation
did not help for the most difficult tasks, and did not help for tool use, even for the 15-monthold task. As for the infants, we observed a few changes in some of the behaviors leading to
failure.
We tested a small sample of Macaques but it is noteworthy that the results are quite
consistent from one individual to the other. These results indicate that captive macaques are
able to learn something from observing a human adult modeling a task on which they failed.
14
However, this capacity for observational learning is very limited, restricted to relatively
simple tasks and not the same for all macaques, and it was not observed for the two tool-use
tasks. Thus, the capacity for observational learning of the macaques was lesser than that of a
18-month-old infant. Observation of a human using a tool, or using the finger as a tool, did
not help them as it did for 18-month-old infants or as it does for chimpanzees (Nagell, 1993).
One reason why observation was not more helpful for the macaques could be that
modeling was not performed by a conspecific However, it is not clear that a conspecific
demonstration would have been sufficient to lead to success. There is no strong consensus on
the macaques’ capacity to learn tool use from observing mates, and on the cultural
transmission of tool use. Although some researchers highlight the relative great capacity of
macaques for observational learning, at least compared with other cercopithecines (Beck,
1976), others argue that cultural transmission of tool use is rare in macaques (Ducoing and
Thierry, 2005): for example, in their study, Ducoing and Thierry showed that Tonkean
macaques do not learn how to use a pole to retrieve a reward after observing other macaques
performing the task. Similarly with another species of monkeys, Visalberghi (1987) observed
that capuchin monkeys who did not spontaneously understand how to use the tool to crack
nuts did not learn by observing their two proficient cage-mates. Former studies showing that
macaques are capable of observational learning used longer observational periods (Myers,
1970; see Beck, 1976, for a review of these former studies).
Another reason for the macaques’ failure to learn by observation how to use tools
could be that they failed to recognize the functionality of the tools. For example, it was
observed in capuchin monkeys that after they discovered, through practice, new means to get
a piece of food, using sticks in a complex condition, they kept making the same errors
throughout the trials (Visalberghi & Trinca, 1989), unlike chimpanzees (Visalberghi,
Fragaszy, & Savage, 1995).
Besides the difference in cognitive capacities, a difference in social training might also
explain part of the limitation of macaques’ observational learning. Learning by observation
requires the capacity to join attention, and to guess the intention of the modeler (Carpenter,
Nagell, & Tomasello, 1998). For example, Japanese macaques who demonstrated joint
attention through communicative eye gaze and pointing gestures were better at imitating
human actions than the monkeys who did not exhibit joint attention (Kumashiro et al., 2003).
The lack of social experience encouraging social interaction in macaques, and the lack of
triadic interaction around objects do not prepare them to learn by observation, as it does with
15
human infants. The greater capacity of imitation of chimpanzees (Tomasello et al., 1993) or of
capuchin monkeys (Fredman & Whiten, 2008) raised in a human-like environment with
everyday activities of the type human children engage in routinely supports this hypothesis.
16
Aknowledgments:
This work was supported by a European grant (MATHESIS, No IST-027574)
17
References
Abravanel, E., Levan Goldschmidt, E., & Stevenson, M. B. (1976). Action imitation: The
early phase of infancy. Child Development, 47(4), 1032-1044.
Ashford, D., Davids, K., & Bennett, S. J. (2007). Developmental effects influencing
observational modelling: a meta-analysis. J Sports Sci, 25(5), 547-558.
Beck, B. B. (1976). Tool use by captive pigtailed macaques. Primates, 17(3), 301-310.
Boesch, C., Marchesi, P., Marchesi, N., Fruth, B., & Joulian, F. (1994). Is Nut Cracking in
Wild Chimpanzees a Cultural-Behavior. Journal of Human Evolution, 26(4), 325-338.
Bruner, J. S., Lyons, K., & Watkins, D. (1969). The growth of human manual intelligence : II.
Acquisition of complementary two-handedness: Center for Cognitive Studies, Harvard
university. Unpublished manuscript.
Byrne, R. W., & Russon, A. E. (1998). Learning by imitation: A hierarchical approach.
Behavioral and Brain Sciences, 21(5), 667-721.
Call, J., Carpenter, M., & Tomasello, M. (2005). Copying results and copying actions in the
process of social learning: chimpanzees (Pan troglodytes) and human children (Homo
sapiens). Animal Cognition, 8(3), 151-163.
Carpenter, M., Nagell, K., & Tomasello, M. (1998). Social cognition, joint attention, and
communicative competence from 9 to 15 months of age. Monogr Soc Res Child Dev,
63(4), i-vi, 1-143.
de Waal, F. B. (1996). Macaque social culture: development and perpetuation of affiliative
networks. J Comp Psychol, 110(2), 147-154.
Diamond, A. (1981). Retrieval of an object from an open box : the development of visualtactile control of reaching in the first year of life. Society for Research in Child
Development Abstracts, 3(78).
18
Diamond, A. (1990). Developmental time course in human infants and infant monkeys, and
the neural bases of, inhibitory control in reaching. Ann N Y Acad Sci, 608, 637-669;
discussion 669-676.
Ducoing, A. M., & Thierry, B. (2005). Tool-use learning in Tonkean macaques (Macaca
tonkeana). Anim Cogn, 8(2), 103-113.
Elsner, B., & Aschersleben, G. (2003). Do I get what you get? Learning about the effects of
self-performed and observed actions in infancy. Conscious Cogn, 12(4), 732-751.
Fadiga, L., Fogassi, L., Pavesi, G., & Rizzolatti, G. (1995). Motor facilitation during action
observation: a magnetic stimulation study. J Neurophysiol, 73(6), 2608-2611.
Fagard, J., Esseily, R., & Nadel (submitted). Learning through practice versus learning by
observation in infants.
Fagard, J., & Jacquet, A. Y. (1989). Onset of Bimanual Coordination and Symmetry versus
Asymmetry of Movement. Infant Behavior and Development, 12, 229-236.
Ferrari, P. F., Visalberghi, E., Paukner, A., Fogassi, L., Ruggiero, A., & Suomi, S. J. (2006).
Neonatal imitation in rhesus macaques. PLoS Biol, 4(9), e302.
Fragaszy, D. M., & Visalberghi, E. (1989). Social influences on the acquisition of tool-using
behaviors in tufted capuchin monkeys (Cebus apella). J Comp Psychol, 103(2), 159170.
Fredman, T., & Whiten, A. (2008). Observational learning from tool using models by humanreared and mother-reared capuchin monkeys (Cebus apella). Animal Cognition, 11(2),
295-309.
Gallese, V., Fadiga, L., Fogassi, L., & Rizzolatti, G. (1996). Action recognition in the
premotor cortex. Brain, 119 ( Pt 2), 593-609.
Gergely, G., Bekkering, H., & Kiraly, I. (2002). Rational imitation in preverbal infants.
Nature, 415(6873), 755.
19
Hayes, K. J., & Hayes, C. (1952). Imitation in a Home-Raised Chimpanzee. Journal of
Comparative and Physiological Psychology, 45(6), 450-459.
Hopper, L. M., Lambeth, S. P., Schapiro, S. J., & Whiten, A. (2008). Observational learning
in chimpanzees and children studied through 'ghost' conditions. Proceedings of the
Royal Society B-Biological Sciences, 275(1636), 835-840.
Iacoboni, M., Woods, R. P., Brass, M., Bekkering, H., Mazziotta, J. C., & Rizzolatti, G.
(1999). Cortical mechanisms of human imitation. Science, 286(5449), 2526-2528.
Kumashiro, M., Ishibashi, H., Uchiyama, Y., Itakura, S., Murata, A., & Iriki, A. (2003).
Natural imitation induced by joint attention in Japanese monkeys. Int J Psychophysiol,
50(1-2), 81-99.
Lockman, J. J. (2000). A perception-action perspective on tool use development. Child Dev,
71(1), 137-144.
Meltzoff, A. N. (1988). Infant imitation and memory: nine-month-olds in immediate and
deferred tests. Child Dev, 59(1), 217-225.
Meltzoff, A. N. (1995). Understanding the Intentions of Others - Reenactment of Intended
Acts by 18-Month-Old Children. Developmental Psychology, 31(5), 838-850.
Meunier, M., Monfardini, E., & Boussaoud, D. (2007). Learning by observation in rhesus
monkeys. Neurobiol Learn Mem, 88(2), 243-248.
Myers, W. A. (1970). Observational learning in monkeys. J Exp Anal Behav, 14(2), 225-235.
Nagell, K. (1993). Processes of social learning in the tool use of chimpanzees (Pan
troglodytes) and human children (Homo sapiens). Journal of Comparative
Psychology, 107(2), 174-186.
Provasi, J., Dubon, C. D., & Bloch, H. (2001). Do 9-and 12-month-olds learn means-ends
relation by observing? Infant Behavior & Development, 24(2), 195-213.
20
Rizzolatti, G., Fadiga, L., Gallese, V., & Fogassi, L. (1996). Premotor cortex and the
recognition of motor actions. Brain Res Cogn Brain Res, 3(2), 131-141.
Sinha, A. (1997). Complex tool manufacture by a wild Bonnet macaque, Macaca radiata.
Folia Primatologica, 68(1), 23-25.
Subiaul, F., Cantlon, J., Lurie, H., Holloway, R., & Terrace, H. (2003). A re-evaluation of
human and macaque "imitation": Human children and rhesus macaques do not
qualitatively differ in a copying task. American Journal of Physical Anthropology,
204-204.
Tokida, E., Tanaka, I., Takefushi, H., & Hagiwara, T. (1994). Tool-using in Japanese
macaques: Use of stones to obtain fruit from a pipe. Animal Behaviour, 47(5), 10231030.
Tomasello, M. (1998). Emulation learning and cultural learning. Behavioral and Brain
Sciences, 21(5), 703-+.
Tomasello, M., Davis-Dasilva, M., Camak, L., & Bard, K. (1987). Observational learning of
tool-use by young chimpanzees. Human Evolution, 2, 175-183.
Tomasello, M., Savage Rumbaugh, S., & Kruger, A. C. (1993). Imitative learning of actions
on objects by children, chimpanzees, and enculturated chimpanzees. Child
development, 64(6), 1688-1705.
Visalberghi, E. (1987). Acquisition of nut-cracking behaviour by 2 capuchin monkeys (Cebus
apella). Folia Primatologica, 49(3-4), 168-181.
Visalberghi, E., & Fragaszy, D. M. (1996). Pedagogy and imitation in monkeys: Yes, no, or
maybe? In D. R. Olson (Ed.), The handbook of education and human development:
New models of learning, teaching and schooling (pp. 277-301). Oxford, England UK:
Blackwell Publishers Inc.
21
Visalberghi, E., Fragaszy, D. M., & Savage, R. S. (1995). Performance in a tool-using task by
common chimpanzees (Pan troglodytes), bonobos (Pan paniscus), an orangutan
(Pongo pygmaeus), and capuchin monkeys (Cebus apella). Journal of Comparative
Psychology, 109(1), 52-60.
Visalberghi, E., & Trinca, L. (1989). Tool use in capuchin monkeys: Distinguishing between
performing and understanding. Primates, 30(4), 511-521.
Voelkl, B., & Huber, L. (2000). True imitation in marmosets. Anim Behav, 60(2), 195-202.
Want, S. C., & Harris, P. L. (2001). Learning from other people's mistakes: Causal
understanding in learning to use a tool. Child Development, 72(2), 431-443.
Whiten, A., Custance, D. M., Gomez, J. C., Teixidor, P., & Bard, K. A. (1996). Imitative
learning of artificial fruit processing in children (Homo sapiens) and chimpanzees
(Pan troglodytes). Journal of Comparative Psychology, 110(1), 3-14.
Whiten, A., Goodall, J., McGrew, W. C., Nishida, T., Reynolds, V., Sugiyama, Y., et al.
(1999). Cultures in chimpanzees. Nature, 399(6737), 682-685.
22
Table 1: Success as a function of age in infants (%): comparison before and
after demonstration
Age group
8-mo
10-mo
12-mo
15-mo
18-mo
Before
demonstration
11.11
37.5
50
44.4 (Bottle)
14.3 (Rake)
42.9 (Finger)
0 (Stick)
After
demonstration
11.1
55.6
44.4
77.8 (Bottle)
28.6 (Rake)
100 (Finger)
100 (Stick)
23
Table 2: Success as a function of task level (in months)(N, out of three), in macaques:
comparison before and after demonstration
Task level
8 mo
10 mo
(Detour
(TubeReaching) container)
12 mo
(Box)
15 mo
18 mo
Before
demonstration
1
0
1
1 (Bottle)
0 (Rake)
0 (Finger)
0 (Fork)
After
demonstration
2
1
1
2 (Bottle)
0 (Rake)
0 (Finger)
0 (Fork)
24
Table 3: Outcome of the trials before and after observation for three macaques at
the first session
Monkey C
Session 1
Detour
Monkey L
Spontaneous After Demo Spontaneous
Monkey G
After Demo
Spontaneous
After Demo
Fail.
Succ.
Succ.
Succ.
Fail.
Fail.
Fail.
Fail.
Fail.
Succ.
Fail.
Fail.
Box
Succ.
Succ.
Fail.
Fail.
Fail.
Fail.
Bottle
Fail.
Fail.
Succ.
Succ.
Fail.
Succ.
Rake
Fail.
Fail.
Fail.
Fail.
Fail.
Fail.
Finger
Fail.
Fail.
Fail.
Fail.
Fail.
Fail.
Fork
Fail.
Fail.
Fail.
Fail.
Fail.
Fail.
Reaching
Tube
Container
25
Figure Captions
Figure 1: Frequency of success before demonstration as a function of task (and age for
infants): comparison human infants – macaques (none of the group differences were
significant)
Figure 2: Frequency of success after demonstration as a function of task (and age for
infants): comparison human infants – macaques
26
Percentage of Success
100
80
60
Infants
40
Macaques
20
0
Det. Reach.
(8)
Tub. Cont.
(10)
Box (12)
Bottle (15)
Rake (15)
Finger (18)
Stick/Fork
(18)
Tasks (Infants' age in m onths)
Figure 1
27
Percentage of success
100
80
p<.01
60
p<.01
Infants
Macaques
40
20
0
Det. Reach.
(8)
Tub. Cont.
(10)
Box (12)
Bottle (15)
Rake (15)
Finger (18)
Stick/Fork
(18)
Tasks (Infant's age in m onths)
Figure 2
28