Cognitive Development 24 (2009) 1­12
Contents lists available at ScienceDirect
Cognitive Development
Children's understanding of communicative intentions in
the middle of the second year of life
Tiziana Aurelia,
, Paola Perucchinib
, Maria Gencoa
a
Department of Biomedical Sciences, University of Chieti-Pescara, Via dei Vestini 31, 66013 Chieti, Italy
b
Department of Educational Science, University Roma Tre, Italy
a r t i c l e i n f o
Keywords:
Pointing comprehension
Declarative/informative intention
Social understanding
Communication
Infancy
a b s t r a c t
Two tasks were administered to 40 children aged from 16 to 20
months (mean age = 18;1), to evaluate children's understanding of
declarative and informative intention [Behne, T., Carpenter, M., &
Tomasello, M. (2005). One-year-olds comprehend the communicative
intentions behind gestures in a hiding game. Developmental
Science, 8, 492­499; Camaioni, L., Perucchini, P., Bellagamba, F., &
Colonnesi, C. (2004). The role of declarative pointing in developing
a theory of mind. Infancy, 5, 291­308]. In the first task, children had
to respond to the experimenter who pointed at a distal object; in
the second task, children had to find a toy in a hiding game after
the experimenter indicated the correct location either by pointing
or by gazing. In the first task, most children responded to the declarative
gesture by "commenting" on the pointed object instead of just
looking; however, looking responses were more frequent than commenting
responses. In the second task, children chose the correct
location of the object significantly more frequently when the informative
gesture was the point than when it was the gaze; moreover,
there were significantly more correct choices than incorrect choices
in the point but not in the gaze condition. Finally, no significant relation
was found between tasks. Taken together, the findings support
the view that infants' developing understanding of communicative
intention is a complex process in which general cognitive abilities
and contextual factors are equally important.
 2008 Elsevier Inc. All rights reserved.
 Corresponding author.
E-mail address: aurelit@libero.it (T. Aureli).
0885-2014/$ ­ see front matter  2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.cogdev.2008.07.003
2 T. Aureli et al. / Cognitive Development 24 (2009) 1­12
Children's understanding of others as having mental states is assumed to have major implications
for social and communicative development (Astington, 1993; Astington & Gopnik, 1988; Whiten, 1991).
Research in this area has traditionally focused on epistemic mental states, with most of the evidence
showing that 4-year-olds are capable of understanding other's beliefs even when they are different
from their own and from the state of reality (Welman, Cross, & Watson, 2001). More recently, the focus
has widened to include attentional and intentional states, and results allowed researchers to identify
the beginnings of children's social understanding at an even earlier age. Twelve-month-olds reliably
follow the gaze direction of another person, although disagreement exists about how to interpret
this evidence, that is, whether it involves infants' recognition of another's attentional focus from its
onset (Carpenter, Nagell, & Tomasello, 1998; Tomasello, 1999), or leads to such recognition only later
in development (Carpendale & Lewis, 2004; Moore & Corkum, 1994). When selective attention is
considered, 18-month-olds recognize what is new and what is old for a person, therefore showing
an ability to take into account the other's mental focus (Tomasello & Haberl, 2003). With respect to
intentional states, 18-month-olds have been shown to be able to imitate the intended purpose of
an action not actually fulfilled (Meltzoff, 1995), as well as to distinguish intentional from accidental
actions (Carpenter, Akhtar, & Tomasello, 1998) and to infer the objects of the other's desire, even when
it differs from their own (Repacholi & Gopnik, 1997). In sum, current findings converge in crediting
children at the middle of the second year of life with an understanding of others as attentional and
intentional agents.
In the present study, we investigated early understanding of communicative intentions. Studies
on this topic are very scarce in the developmental literature, and most of them are focused
on pointing gesture. Traditionally, two different aspects of infants' pointing have been considered:
the referential aspect, as in Werner and Kaplan's representational theory (1963), and the intentional
aspect, as in Austin's speech acts theory (1962). According to the first view, pointing is
considered as a means for the expression of reference, resulting from a process of differentiation
between the child and the object and therefore as a first step toward true symbolization. According
to the second perspective, pointing is a performative behavior expressing children's communicative
intentions in the prelinguistic period and thus a transitional form from the perlocutory ­ not
intentional and not linguistic ­ to the locutory ­ intentional-linguistic ­ stage of language development.
More recently, the intentional aspect of pointing has been emphasized by socio-cognitive
research and various motives have been suggested for infants' pointing. One motive is to request
an object/action from an adult (imperative intention), and another motive is to share with an adult
attention and interest in an object (declarative intention; Camaioni, 1993; Camaioni, Perucchini,
Bellagamba, & Colonnesi, 2004). More recently, an informative motive has been proposed, that is, helping
an adult to find something the adult is looking for (Liszkowsky, Carpenter, Striano, & Tomasello,
2006).
If we turn to pointing comprehension, the referential perspective has been the most common
framework and infants have been studied with respect to their ability to identify the pointed objects
at different spatial locations, not unlike gaze-following ability. It was found that 9- and 12-month-olds
are pre-referential, since they are only capable of locating the object when it is close to the pointing
finger, whereas most 14-month-olds are referential, being successful in locating distant objects
(Lempers, 1979; Murphy & Messer, 1977; Morissette, Ricard, & Gouin Decarie, 1995). At 15 months,
infants completely ignore an interfering object when looking at the target and are capable of looking
at it also when it is outside their visual field, for example behind them (Butterworth & Grover, 1988,
1990).
Recently, infants' comprehension of the communicative intent of the gesture has begun to be considered.
Unfortunately, research in this area is even more scarce. The only two studies available parallel
those on pointing production, one focused on declarative and imperative intentions, and the other on
informative intention.
In the first study, Camaioni et al. (2004) used a longitudinal design to analyse infants' skills at 12 and
15 months of age. An experimental task was devised that differentiated imperative from declarative
pointing. Two different conditions were used, the first presented with proximal objects that were
pointed to by the adult for requesting (imperative condition) and the second presented with distal
objects that were pointed to for directing infants' attention (declarative condition). For imperative
T. Aureli et al. / Cognitive Development 24 (2009) 1­12 3
comprehension, the target response was infants' giving or refusing to give the adult the indicated
object; for declarative comprehension, it was the expression of some sort of comment about the object,
by smiling, vocalizing or gesturing. Twelve-month-olds were better in comprehending the imperative
than the declarative intention, but they were equally able to comprehend both 3 months later. A second
study by Behne, Carpenter, and Tomasello (2005) used a cross-sectional design to study 14-, 18- and
24-month-olds' responses to pointing as an informative intention. Infants had to find a toy hidden
in one of two containers, using the experimenter's informative gesture, either point or gaze. Infants
performed above chance in retrieving the toy at all ages; however, on the pointing trials, 18- and
24-month-olds performed better than the youngest group, whereas on the gaze trials 24-month-olds
performed better than both 14- and 18-month-olds.
If we examine these results in more detail, some interesting questions arise concerning the nature
of children's understanding of communicative intentions in the middle of the second year of life. With
respect to the first study, pointing comprehension was attributed to the infant when she/he smiled,
vocalized or gestured toward the pointed-to object. However, in the classical studies (Lempers, 1979;
Murphy & Messer, 1977; Morissette et al., 1995; Butterworth & Grover, 1988, 1990), a correct response
was assigned when the infant only looked at it. This methodological difference in the choice of the
dependent variable can be traced to the two perspectives on pointing discussed earlier, as declarative
gesture, or as referential gesture. According to the first perspective, Camaioni et al. measured pointing
comprehension in terms of infants' ability to recognize the other's interest toward the attended object,
whereas, in line with the second perspective, the other studies measured infants' ability to only identify
that object. Hence, an infant's comment or an infant's look was the target response, respectively. Of
interest here is the fact that in Camaioni et al.'s study, 15-month-olds responded to the adult's gesture
on about half of the valid trials (59%), whereas infants in Morissette et al.'s sample (1995) responded
much more frequently (on 75% of the trials). Given the different variables used as a target response,
it seems that infants are more likely to recognize the referential aspect of the adult's pointing rather
than the declarative aspect. However, some of the "referential" responses in the Morissette et al. study
would have been "declarative" in Camaioni et al.'s terms, if the infants' look had been associated
with a comment. Therefore, the question about the nature of infants' understanding of pointing is
left unanswered. In the present study, we used Camaioni et al.'s paradigm to identify both types of
responses, the "looking at the target" and the "commenting on the target" responses. The main goal
was to determine if children in the middle of their second year of life react to the adult's pointing
by commenting rather than by only looking, therefore showing an understanding of pointing as a
declarative gesture that conveys both attention and interest, and not as a referential act, conveying
only attention.
In Behne et al.'s study (2005), 18-month-olds were better in the point than in the gaze trials, in
contrast to 14-month-olds who were equally poor in both conditions and to 24-month-olds who were
equally good. Since gaze direction, in contrast to point, "is not typically used to express communicative
intentions in interactions with infants" (p. 493), we could suppose that children in the middle of the
second year are especially sensitive to the difference between a familiar and an unfamiliar gesture
and perform worse in gaze than in point conditions because of this unfamiliarity. With reference to
Tomasello, Call, and Gluckman's (1997) account of children's reactions to a gesture they had very
likely never before encountered, we hypothesized two different reasons that could explain children's
poorer performance in the gaze trials. First, children were not able to generalize from their previous
communicative experience to the actual context and therefore did not comprehend the meaning of
the experimenter's gaze. Second, children had difficulties in understanding how the unfamiliar gesture
effectively works in that context and therefore understood the communicative intention of the gaze
direction but were not able to locate the object that the gaze was directed to. The two types of difficulties
are reflected in children's failures mentioned in the Behne et al.'s study. In that study, children failed to
search successfully because, after looking at the cued container, they chose the other one or because,
after following the experimenter's gesture, they searched in other locations. Since understanding the
nature of children's failures can improve our understanding of their performance, we tested children
with an adapted version of Behen et al's task, with the aim of finding whether children perform worse
on the gaze than on the point trials because they are not able to infer the relevance of the adult's gaze
or because they are not able to identify its referent. With this aim, we analysed children's searching
4 T. Aureli et al. / Cognitive Development 24 (2009) 1­12
behavior after the cue presentation, as in the original study, and children's looking direction during
this presentation as a new variable.
As a further purpose of the study, we examined intra-individual competence on both tasks. One
might expect that children who are successful on one task might be successful also on the other, because
both tasks measure infants' understanding of communicative intentions. Alternatively, children could
perform differently because the understanding of the declarative intention requires different abilities
with respect to the informative intention, that is, to share attention and interest on a referent in the
first case or to cooperate in a given game in the second case. Therefore, both tasks were administered
to the same group of children.
1. Methods
1.1. Participants
Forty children (20 girls) from a middle-sized Italian city were tested in a university laboratory room.
Children ranged in age from 16 to 20 months (M = 18;1, S.D. = 1;3, range = 16;1­20;4). Children were
recruited from a database of parents who were contacted by family paediatricians and had volunteered
to participate in studies on child development.
1.2. Design and procedure
Children came to the laboratory with their mothers. After a familiarization phase with the experimenter
in a comfortable and cheerful room, the dyad was brought into the testing room. Two tasks
were used, first the "hiding-game task" and, after a 10 min pause, the "pointing task". The entire session
lasted about 1 h and was videotaped. The room was equipped with two cameras remotely controlled
from behind a one-way mirror. One camera provided a frontal view of the setting, including the child,
the mother, the experimenter, the table and the experimental objects. The second camera provided an
upper view of the child's face and torso.
1.2.1. The pointing task
Following the original study (Camaioni et al., 2004), we assessed infants' comprehension of declarative
pointing using the following procedure. Each child was seated on the mother's lap at a large
table (200 cm × 100 cm × 80 cm). The experimenter sat on one side of the table (at 90 angle) facing
the child. Distal stimuli for the declarative condition were three static objects: a wall poster with the
storybook characters of Lilly and the vagabond, a Santa Claus puppet hanging on the other wall and a
coloured stuffed cat standing on a small piece of furniture. They were located on the right, front and
left side of the child, respectively. Three trials were repeated in a row for each of the three stimuli.
At the beginning of each trial, the experimenter made eye contact with the child, said "Hi [the child's
name]!" and then pointed to the poster/puppet/stuffed toy with a positive face, while alternating gaze
between the child and the stimulus. The experimenter looked silently at the child for 15 s, waiting for
a response. The procedure was exactly the same as in Camaioni et al.'s (2004) declarative condition.
1.2.2. The hiding-game task
The procedure was similar to that described in Behne et al. (2005). Each child sat on the mother's
lap at the same table, with the experimenter sitting in front of the child. All children participated in a
warm-up task to familiarize them with the game and the containers used for hiding. The experimenter
placed a pair of identical containers on the table, at either end of a wooden tray (76 cm × 96 cm) and at
a distance of roughly 70 cm from each other. She then showed children an attractive toy mouse, saying
the Italian equivalent of `Look, now I'll hide it', while she overtly placed the toy in one of the containers
and closed it. Then she invited the child to retrieve it by pulling the tray forward.
Subsequently, children participated in experimental trials in which the experimenter hid the toy
behind a screen, so the child was not able to see in which of the two containers the toy was placed.
At the beginning of each trial, a pair of empty containers was placed on the tray. The experimenter
presented a toy and if the child showed interest in it, she placed the screen in front of the child while
T. Aureli et al. / Cognitive Development 24 (2009) 1­12 5
saying the Italian equivalent of, `Look, now I'll hide it'. She then lowered the toy behind the screen,
quickly pushed both containers toward each other, hid the toy in one, and pushed the containers apart
again. After removing the screen, she established eye contact with the child and indicated the hiding
location giving a communicative cue.
As in Behne et al.'s (2005) task, two types of cues were used: (a) ostensive gaze: the experimenter
repeatedly turned her head, gazed alternately between the baited container and the child, while accompanying
the gesture with a typical facial expression, such as raised eyebrows (gaze condition); (b) point:
she pointed to the baited container without touching it, looking back and forth at the container and
at the child with a typical facial expression such as raised eyebrows (point condition). After giving the
cue, she pulled the tray forward. Across the two conditions, the experimenter's head turn, look duration,
and facial expression were the same; the only difference was the pointing gesture production in
the point condition. Three pairs of containers, each of different colour, material and shape were used
as hiding locations. Each pair of containers was used for one point trial and for one gaze trial. The
containers were positioned at equal distance from E (70 cm), so that the child could not grab both the
containers, and at equal distance from the child (70 cm.), so that the child could not reach them simply
by leaning forward.
Findings from a pilot study showed that children in the test phase were likely to lean toward the
containers immediately after removing the screen, in both the point and gaze conditions. This was
probably due to our having the containers on the table where the child sat and was not observed
in the original Behne et al. (2005) study probably because the containers were located on a table in
front of the child. To prevent children from immediately choosing a container, an additional step was
inserted before testing. The experimenter introduced the screen as in the test phase and pushed the
containers behind it, this time without placing the toy inside any container. After removing the screen,
she verbally invited the child to find the toy, which the child immediately did by opening one of the
containers. After nothing was found, the experimenter quickly put the toy into the other container
through a hole not visible to the child, then encouraged the child to open the baited container and
praised the child for the successful result. The trial was repeated twice. This step probably made the
task more challenging for the children than it was in the original study, because of the experience of
not finding the toy where they thought it to be. However, it fulfilled our purpose because in the test
phase all children waited a moment after the screen was removed, looked at the experimenter's face
and attended to her cue.
All children participated in three point trials and three gaze trials, with the order of cue presentation
counterbalanced across children. The side, left or right, of the hiding location was also counterbalanced
for each child and each cue, with the toy never hidden more than twice in succession on the same
side.
1.3. Scoring
Children's responses were coded from videotape and scored according to the following procedure.
1.3.1. The pointing task
Valid and non-valid trials were first identified. A trial was coded as "not valid" if the child did not
attend to the experimenter's gesture. The mean percentage of non-valid trials was 14%, very similar to
that reported by Camaioni et al. (2004). For the valid trials, the following responses were identified as
target behaviors:
Looking response, if the child only looked at the pointed object or alternated her/his gaze between the
object and the experimenter.
Commenting response, if the child, after looking, smiled/vocalized/pointed toward the stimulus, and/or
produced a proto-word/word relative to the stimulus, and/or re-enacted some characteristics of the
stimulus, each time looking at the experimenter.
No response, if the child never looked or commented, for example staring at the experimenter's face
or looking elsewhere.
6 T. Aureli et al. / Cognitive Development 24 (2009) 1­12
Only one type of responses was counted per trial. To assess inter-observer reliability, 20% of the sessions
were randomly selected from the videotape and coded by a second trained person. The computed
K coefficient was .70.
1.3.2. The hiding-game task
All trials were valid. As the child approached the first container and opened or tried to open it,
a correct response was scored if the child chose the container the experimenter was pointing to or
gazing at; choosing the other container was scored as an incorrect response. Just touching a container
without attempting to open it was scored as no choice.
As an additional measure, infants' direction of attention during the cue presentation was considered.
Children looking at the experimenter or at the intended target or alternatively between the two was
coded. Only one attentional behavior was counted per trial. To assess inter-rater reliability, 20% of trials
for each condition were selected and independently coded from the tape by a second trained person.
With respect to the coding of the correct/incorrect choice of the child, there was 100% agreement
between the two coders; with respect to the attention behavior the computed K coefficient was .75.
1.4. Measures
The measure used for the pointing task was the percentage of trials in which there occurred one
of the three target behaviors relative to the total number of valid trials. For the hiding-game task, the
mean percentage of correct and incorrect responses, as well as the percentage occurrence of each type
of attentional behavior, were calculated for each condition.
The pattern of individual responses was also considered. For the pointing task, a child was considered
to comprehend declarative intention if he/she performed at least one commenting response in
any trial and was considered to comprehend referential intention if he/she performed no commenting
responses but only looking responses in more than half of the valid trials. In the first case, since a
specific behavior ­ expressive, gestural or vocal/verbal ­ was required for coding comprehension, one
response was thought to be sufficient, whereas in the second case, since the target behavior was not
as straightforward in revealing child's intentionality, a more stringent measure was used. Finally, the
child was considered as not responding in the absence of the above criteria, that is if the child performed
no commenting responses or only looking responses in half of the trials or less. Thus, children
were classified into three groups: commenting, looking and not responding.
For the hiding-game task, the child was considered to be successful in the gaze or in the point
condition if he/she performed correctly in at least two of three trials in a given condition. Therefore
children were classified into two groups: successful and not successful.
2. Results
2.1. Preliminary analyses
Preliminary one-way ANOVAs on the mean percentage of target behaviors showed that gender was
not significant in either task, nor was the order of cue presentation in the hiding-game task. Moreover,
no significant differences were found in either task between younger and older infants, based on a
median split of the sample. Therefore these variables were not included in further analyses.
2.2. The pointing task
As shown in Table 1, the mean percentage of trials in which infants looked and commented was
significantly higher than the mean percentage of trials in which they did not respond (66% vs. 34%;
t(39) = 4.82, p < .001); likewise, the number of children classified as "looking" or "commenting" was
also significantly higher than the number of children classified as "not responding" (36 vs. 4, binomial
test p < .001). Both results confirmed the developmental trend found in previous research, that children
in the middle of the second year of life are prone as well as capable of responding to an adult pointing
to distal objects.
T. Aureli et al. / Cognitive Development 24 (2009) 1­12 7
Table 1
Means, standard deviations, and frequencies for infants' responses in pointing task
Type of response Frequency Percentage No. of children
Mean S.D. Range Mean S.D. Range
No response 2.7 1.7 0­7 34 20 0­78 4
Looking 3.7 2.0 0­8 50 43 0­100 9
Commenting 1.3 1.5 0­6 16 16 0­67 27
As also shown in Table 1, the mean percentage of trials in which children commented was significantly
lower than the mean percentage of trials in which they looked (16% vs. 50%, t(39) = 5.01,
p < .001). The reverse was found with respect to numbers of children: according to our criteria, 27 of
40 children (binomial test p < .04) were classified as "commenting" and only 9 children were classified
as "looking". Thus, at the individual level most children responded with a comment, whereas at
the group level most responses were attentional. With respect to the distribution of the responses by
order of trials, no responses were equally distributed in proportion across the first, the second and the
third trial (.33, .35 and .32, respectively); the same was true with respect to looking responses (.36, .35
and .29, respectively). However, commenting responses were higher in the first trial (49%) than in the
second (34%) or third (17%) trial.
2.3. The hiding-game task
Data are presented first with respect to children's performance on point and gaze trials and then
with respect to the direction of attention during the cue presentation.
As seen in Table 2, the mean percentage of trials in which children choose correctly was significantly
higher in the point than in the gaze condition (73% vs. 45%, t(39) = 4.61, p < .001); moreover, in the
point condition this percentage was significantly higher than the mean percentage of trials in which
children choose incorrectly, t(39) = 5.38, p < .001. Instead, no significant difference was found between
the two percentages in the gaze condition. Individual data showed a pattern consistent with these
findings. In the point condition, 30 children of 40 (binomial test, p < .05) passed our success criterion
­ performing correctly on at least two trials ­ whereas only 15 children passed it in the gaze condition
(binomial test n.s.). This pattern was corroborated by children's performance on the first trial in each
condition: 82% of children (33 of 40, binomial test, p < .05) searched correctly in the first trial when
the cue was the point and 45% of them (19 of 40, binomial test n.s.) did so when the cue was the
gaze.
With respect to children's attention to the informative gesture, children followed the cue's direction
on 99% of trials in the point condition and on 25% of trials in the gaze condition. In the latter condition,
they looked at the experimenter's face on 75% of trials. Therefore children always recognized the adult's
gesture as an attention-directing cue when the cue was the point but recognized it very rarely when
the cue was the gaze. We analysed gaze trials further to examine infants' choices by the direction of
their previous looking. We found that 56% of the trials in which children followed gaze direction were
successful and 44% were not; likewise, 46% of trials in which children stared at the experimenter's face
Table 2
Means, standard deviations, and frequencies for infants' responses in point and gaze condition of the hiding-game task
Type of response Frequency Percentage No. of children
Mean S.D. Range Mean S.D. Range
Point condition
Correct choices 2.2 0.8 1­3 73.3 27.4 33­100 30
Incorrect choices 0.8 0.8 0­2 26.7 27.4 0­67 10
Gaze condition
Correct choices 1.4 0.9 0­3 45.0 29.1 0­100 15
Incorrect choices 1.7 0.9 0­3 55.0 29.1 0­100 25
8 T. Aureli et al. / Cognitive Development 24 (2009) 1­12
were successful and 54% were not. Therefore, children's choices in gaze condition were independent
of their previous looking.
2.4. Relations between declarative pointing, informative pointing and informative gazing
No significant correlation was found between the number of trials in which children commented
in the pointing task and the number of trials in which children chose correctly in the point condition
(r = .-14, n.s.) or in the gaze condition (r = .03, n.s.) in the hiding-game task. The correlation between
children's correct responses in point and gaze conditions (r = .08) was also not significant. At the individual
level, based on our criteria, only 9 children of 40 were successful in all three contexts; 10 children
were successful in both the informative and the declarative pointing, 3 children in informative pointing
and gazing and 3 in both the declarative pointing and the informative gazing. We also found that eight
children were only successful in the informative pointing and five in the declarative pointing, whereas
no child was successful in the informative gazing only. As a conclusion, both at the group and the individual
level, no reliable relation was found between children's responses in the three communicative
contexts. Also, informative gazing appeared the most difficult intention to recognize.
3. Discussion
This study examined children's understanding of communicative intentions as expressed by prelinguistic
gestures. Two tasks, based on work by Camaioni et al. (2004) and Behne et al. (2005), were used
to evaluate the comprehension of two different intentions--the declarative one, as expressed by the
pointing gesture, and the informative one, as expressed by the point and the gaze gesture.
With respect to declarative intention, almost all children performed in the pointing task by following
the direction of the adult's gesture to the target. However, a small group of children, after locating
the target, only looked at it, whereas a larger group also smiled, vocalized or gestured. Thus, the first
group responded to the adult by attending to the pointed object, whereas the second group by both
attending and commenting. With reference to the two perspectives discussed earlier on pointing as
a referential or a declarative gesture, we could argue that the first group recognized the attentional
relation between the pointer and a given object and the other group recognized an additional meaning
of this relation, that is, the pointer was also interested in that object. With respect to our research
question concerning children's understanding of the nature of others' pointing in the middle of the
second year of life, the majority of children recognized the declarative aspect of the gesture compared
to the minority who only recognized its referential aspect.
The distinction between the referential and the declarative aspect in pointing comprehension parallels
a recent point made by Liszkowsky, Carpenter, and Tomasello (2006) about reference and attitude
in pointing production. They found that 12-month-olds looked at the adult's face more frequently when
the adult responded to their pointing only by looking at the pointed object than when he also talked
and commented on it, therefore showing a dissatisfaction with referential responses (Liszkowsky,
Carpenter, Henning, Striano, & Tomasello, 2004). Moreover, infants pointed more when the adult
reacted to their pointing by showing both attention and interest than when the adult matched only one
of them, therefore expecting to share with the adult both the reference and the attitude (Liszkowsky,
Carpenter, & Tomasello, 2006). They concluded that infant's pointing is a full communicative gesture
from the ouset, conveying both reference and attitude (Liszkowsky, Carpenter, & Tomasello, 2006; see
also Tomasello, Carpenter, & Liszkowski, 2007).
In the present study, most children commented on the pointed object instead of simply looking at
it. Since a comment is the proper modality for responding to a person who shows interest in what he
or she is looking at, it can be inferred that also pointing comprehension involves children's recognition
of the other's attitude and not only the reference. However, one third of the children never commented
but merely looked; therefore, the recognizing of the other's attitude could be more difficult than the
recognizing of the reference. This result suggests that pointing comprehension, is not a full competence
from the outset, but develops during the second year of life from infants' understanding of the other's
reference to their understanding of the other's attitude regarding the reference.
T. Aureli et al. / Cognitive Development 24 (2009) 1­12 9
The developmental nature of pointing comprehension was already noted by Lempers (1979). In her
representational account, infants initially recognize the perceptual contiguity between the pointing
finger and the pointed object and later the symbolic link between the two, thus progressing from prereferential
to referential comprehension. A more communicative account of this process could revise
this model by adding declarative comprehension, so as to argue that "referential" children progress
even further from sharing the referent itself to also sharing interest in it. The proposal that referential
understanding of pointing can be distinguished from declarative understanding is also in accord with
Woodward and Guajardo's (2002) suggestion that "the understanding of the object-directed nature
of pointing provides the basis for understanding the communicative nature of the gesture" (p. 1062).
In particular, since they claimed that in the first year of life pointing is understood in a referential
manner, the declarative manner could be considered the following step in children's understanding of
the different components of pointing. In particular, "declarative" children would express their sharing
earlier, at a sensorimotor level, by smiles, vocalizations and gestures, as in our sample, and later at a
linguistic level by informative statements. To support this account, we are carrying out a microgenetic
study of the development of the pointing gesture from 9 to 24 months of age.
Although the number of children who commented was higher than the number who only looked,
the mean percentage of looking responses was significantly higher than the percentage of commenting
responses. Thus, even competent children, who were otherwise capable of sharing the declarative
function of pointing, were likely to respond by sharing the referential function. Since commenting
responses were performed on the first trial rather than on successive ones, it may be that children
in our experimental setting were not motivated enough to comment every time, having commented
once. We therefore agree with Tomasello, Carpenter, Call, Behne, and Moll (2005) that in order to
advance in joint attention experience, infants have not only to understand the other's goals, but also to
be motivated to share these goals. Although their claim obviously refers to the ontogenesis of infant's
social understanding and not to the infant's responses in a given experiment, the apparently conflicting
data we obtained ­ the high level of children's competence in understanding declarative pointing and
the low consistency of this level in their performance ­ show, in a quite novel way, the impact that
social motivation has on infant's responses to the adult's communicative acts, whether the interaction
occurs at home or in a laboratory.
With regards to informative intention, 18-month-olds performed more poorly in the gaze condition
than in the point condition; more precisely, they were at chance level in the former, contrary to the
latter. Children in the gaze condition did not reliably choose the correct location when searching for
the hidden object, and the number of reliable children was not significant. Therefore, our results were
consistent with the original Behne et al. (2005) study with respect to the point condition, but not
the gaze condition. In the latter case, we confirmed Povinelli, Reaux, Bierschwale, Allain, and Simon's
(1997) findings, showing that 2-year-olds did not perform above chance when the adult gazed at the
baited container, in contrast to when the adult pointed at it.
Null results in the gaze condition were not completely unexpected, given the poorer performance
also showed in this condition with respect to point condition in the Behne et al. (2005) study. Indeed,
analysing children's errors was our main interest in the present study. We supposed that any problem
children would have in the gaze condition was due to their difficulty either in understanding the
referent of the gesture or in inferring its relevance. In the first case, children would fail to locate the
target and in the second case to choose the target after locating it. In any case, children were expected
to follow the adult's gaze direction. Instead, children stared at the experimenter's face almost every
time, both when they were successful and when they were not. Such a result sounds very striking, from
since gaze following ­ the ability to look where someone else is looking ­ is a milestone in infants'
development from the end of the first year of life. However, could sound less striking if we consider
that infants' skills in following the other's gaze has been tested in paradigms in which the gaze is used
as an index of the other's attention to a given target and not as a communicative cue to inform about
the target, for example to signal the place to find it, as in the present paradigm. Thus, findings from
these two different conditions cannot be properly compared.
We hypothesized that two processes were responsible for this result. On the one hand, children
noticed that the adult's gaze was different from the usual because of the mimical expressions used by
the adult to emphasize the informative motive; on the other hand, children were not able to assign the
10 T. Aureli et al. / Cognitive Development 24 (2009) 1­12
informative motive for the perceived difference. So, they were a bit puzzled when facing the ostensive
gaze and probably stared at the experimenter`s face in an effort to solve this puzzle. In other words,
children recognized that the ostensive gaze was a special kind of gaze but did not recognize what of
its special quality was. In Tomasello et al.'s terms (2007), children did not understand the "common
ground" within which the adult's behavior was expressed, that the adult was informing them where
to find the hidden toy. Probably, according to the engagement hypothesis suggested by D'Entremont
and Seamans (2007), they were only aware that the adult was engaged with them in a relationship,
and hence attended to the adult's expression as a social act without inferring her informative intent.
However, children's looking behavior did not influence their choices, because they were correct
or not whether looking at the target or at the experimenter's face. This result strongly corroborates
Behne et al.'s claim (2005) that successful search involves more than the ability to follow another
person's gesture to a target; rather, it implies the ability to infer why the adult was gesturing to that
target. Therefore, what matters in the gaze condition is children's ability to share a common ground
with the adult. We suggest that successful children were those who recognized that the experimenter
had communicative intentions toward them and used this inference about the nature of the whole
context to understand the nature of a specific behavior. This interpretation extends to an earlier age
Tomasello et al.'s (1997) account of children's understanding of novel signs in the third year of life,
saying that also children's understanding of unfamiliar gestures in the second year depends on their
ability to infer the communicative meaning of the gesture from the communicative context. If this
account of both failures and successes in gaze condition is correct, the abilities required to recognize
the informative intention when expressed by the gaze seem to be at a higher level than those required
to recognize the same intention when expressed by the point. In the latter case, the gesture used in
the experimental context was one which children are familiar with and was used by the experimenter
for conveying an intention frequently associated to that gesture. Therefore, although children have
to recognize the common ground provided by the hiding game both in point and gaze conditions,
pointing makes the recognition task easier. Contrary to the suggestion that recognition of informative
pointing is a reliable indicator of children's differentiated understanding of communicative intentions
(Tomasello & Carpenter, 2007; Moll & Tomasello, 2006), we argue that recognition of informative gaze
is a better candidate to assess this level of social understanding.
As a final result, the tasks were not reliably related to each other, whatever measure was used. In
particular, no reliable relationship was found between gaze and point conditions in the informative
task; nor was one found between the declarative and informative tasks. It seems that the abilities
required for passing the tasks were different, depending on contextual features. When the contexts
were different with respect to the gesture, as in the first case, children were required to make inferences
from the entire situation when the gesture was the unfamiliar gaze, whereas they had only to perform
a social practice when it was the familiar point. In the latter case, when the difference was due to the
intention underlying the gesture, it could be that the declarative motive is managed by young children
independently from the informative motive, as far as the first is aimed at sharing an attitude toward a
common focus and the second at cooperating toward a common outcome.
Taken together, our results suggest that infants' understanding of communicative intention is a
process much more gradual, differentiated and context-bound than implied by Tomasello et al.'s (2007)
rich interpretation. In particular, it seems not to be a multilayered knowledge from the beginning. Three
findings in particular support this conclusion. First, only a small proportion of children who passed
the informative task when the cue was the point passed the same task when the cue was the gaze.
If, according to the Tomasello et al. (2007) account of Behne et al.'s results, the reason for the success
of children in the point condition was their full understanding of the communicative nature of the
adult's behavior within a common ground, they should understand the adult's informative intention
also when the gesture was the gaze. This was not the case. Second, children presented with the gaze
gesture showed different kinds of failures. If the understanding of communicative intention was a
full-fledged skill from its outset, the only errors caused by facing an unfamiliar gesture would be due
to problems in generalizing this competence. As in one type of error reported by Behne et al. (2005),
children failed because, although understanding the communicative relevance of the gesture, they did
not understand how the gesture worked in that particular situation. Instead, more frequently, children
did not understand the communicative intention itself, as happened in the other type of errors in
T. Aureli et al. / Cognitive Development 24 (2009) 1­12 11
Behne et al.'s study and, even more, in children's looking behavior in our study. Third, no relationships
were found across tasks. This result suggests that the understanding of the other's communicative
intent is not a single competence but involves different abilities, related, according to our evidence, to
some specific contextual features.
To conclude, different levels in the early understanding of communicative gestures were found in
our study. With respect to declarative pointing, some children were able to respond at a referential level
and others at a declarative level; with respect to the informative gesture, most children were capable
of recognizing the point and less than half of recognizing the gaze; moreover, children who understood
the informative gaze were more likely to understand declarative pointing, but the reverse was not the
case. Results of this type support the idea that the understanding of communicative intentions is a very
gradual process, in which earlier forms of communication are shared by infants before they become
able to share more mature forms (Carpendale & Lewis, 2005). In order to account for this process, efforts
should be made to identify what aspects of the intentional nature of a communicative context children
appreciate. However, research examining the understanding of others' mental states at a prelinguistic
age is hindered by the difficulty in objectively verifying claims (Leavens, Hopkins, & Thomas, 2004).
Assuming an operational perspective and detailing child behaviors, as was attempted in the present
study, could be a way to obtain evidence leading to a better understanding of this phenomenon.
Acknowledgments
This study was made possible by grants from the University G. D'Annunzio and the University
Roma Tre. We thank the mothers and the infants who participated to the study. We also wish to thank
the three anonymous reviewers and especially Diane Poulin-Dubois for her helpful comments on a
previous draft of this article. Portions of these data were presented at the 37th Annual Meeting of the
Jean Piaget Society (Amsterdam, 2007).
References
Astington, J. W. (1993). The child's discovery of mind. Cambridge, MA: Harvard University Press.
Astington, J. W., & Gopnik, A. (1988). Knowing you've changed your mind: Children's understanding of representational change.
In J. W. Astington, P. L. Harris, & D. R. Olson (Eds.), Developing theories of mind (pp. 193­206). Cambridge: Cambridge University
Press.
Austin, J. L. (1962). How to do things with words. London: Oxford University Press.
Behne, T., Carpenter, M., & Tomasello, M. (2005). One-year-olds comprehend the communicative intentions behind gestures in
a hiding game. Developmental Science, 8, 492­499.
Butterworth, G. E., & Grover, L. (1988). The origins of referential communication in human infancy. In L. Weiskrantz (Ed.), Thought
without Language (pp. 5­21). Oxford: Oxford University Press.
Butterworth, G. E., & Grover, L. (1990). Social cognition in infancy: Joint visual attention, manual pointing and preverbal communication
in human infancy. In M. Jeannerod (Ed.), Attention and performance XIII (pp. 605­624). Hillsdale, NJ: Erlbaum.
Camaioni, L. (1993). The development of intentional communication: A re-analysis. In J. Nadel & L. Camaioni (Eds.), New
perspectives in early communicative development (pp. 82­96). London: Routledge.
Camaioni, L., Perucchini, P., Bellagamba, F., & Colonnesi, C. (2004). The role of declarative pointing in developing a theory of
mind. Infancy, 5, 291­308.
Carpendale, J., & Lewis, C. (2004). Constructing an understanding of mind: The development of children's social understanding
within social interactions. Behavioral and Brain Science, 27, 96­97.
Carpendale, J., & Lewis, C. (2005). How children develop social understanding. Oxford: Basil Blackwell.
Carpenter, M., Akhtar, N., & Tomasello, M. (1998). Fourteen- to 18-month-old infants differentially imitate intentional and
accidental actions. Infant Behavior and Development, 21, 315­330.
Carpenter, M., Nagell, K., & Tomasello, M. (1998). Social cognition, joint attention, and communicative competence from 9 to 15
months of age. Monographs of the Society for Research in Child Development, 63, 4.
D'Entremont, B., & Seamans, E. (2007). Do infants need social cognition to act socially? An alternative look to infant pointing.
Child Development, 78, 723­728.
Lempers, J. (1979). Young children's production and comprehension of nonverbal deictic behaviors. The Journal of Genetic
Psychology, 135, 93­102.
Leavens, D. A., Hopkins, W. D., & Thomas, R. K. (2004). Referential communication by chimpanzees. Journal of Comparative
Psychology, 118, 48­57.
Liszkowsky, U., Carpenter, M., Henning, A., Striano, T., & Tomasello, M. (2004). Twelve-month-olds point to share attention and
interest. Developmental Science, 7, 297­307.
Liszkowsky, U., Carpenter, M., & Tomasello, M. (2006). Reference and attitude in infant pointing. Journal of Child Language, 33,
1­20.
Liszkowsky, U., Carpenter, M., Striano, T., & Tomasello, M. (2006). 12- and 18-month-olds point to provide information for others.
Journal of Social Cognition and Development, 7, 173­187.
12 T. Aureli et al. / Cognitive Development 24 (2009) 1­12
Meltzoff, A. N. (1995). Understanding the intentions of others: Re-enactment of intended acts by 18-month-old children.
Developmental Psychology, 31, 838­850.
Moll, H., & Tomasello, M. (2006). Coppertatin & human cognition: The Vygotskian intelligence hypothesis. Philosophical Transactions
of the Royal Society Series B, 2006, 1­10. doi:10.1098/rstb.2006.2000
Morissette, P., Ricard, M., & Gouin Decarie, T. (1995). Joint visual attention and pointing in infancy: A longitudinal study of
comprehension. British Journal of Developmental Psychology, 13, 163­175.
Moore, C., & Corkum, V. (1994). Social understanding at the end of the first year of life. Developmental Review, 14, 349­372.
Murphy, C. M., & Messer, D. J. (1977). Mothers, infants and pointing: A study of gesture. In H. R. Schaffer (Ed.), Studies in
mother­infant interaction (pp. 325­354). London: Academic Press.
Povinelli, D. J., Reaux, J. E., Bierschwale, D. T., Allain, A. D., & Simon, B. B. (1997). Exploitation of pointing as a referential gesture
in young children, but not adolescent chimpanzees. Cognitive Development, 12, 327­365.
Repacholi, B., & Gopnik, A. (1997). Early understanding of desires: Evidence from 14 and 18-month-olds. Developmental Psychology,
33, 12­21.
Tomasello, M. (1999). The cultural origins of human cognition. Cambridge, MA: Harvard University press.
Tomasello, M., Call, J., & Gluckman, A. (1997). Comprehension of novel communicative signs by apes and human children. Child
Development, 68, 1067­1080.
Tomasello, M., & Carpenter, M. (2007). Shared intentionality. Developmental Science, 10(1), 121­125.
Tomasello, M., Carpenter, M., Call, J., Behne, T., & Moll, H. (2005). Understanding and sharing intentions: The origins of cultural
cognition. Behavioral and Brain Sciences, 28, 675­691.
Tomasello, M., Carpenter, M., & Liszkowski, U. (2007). A new look at infant pointing. Child Development, 78, 705­722.
Tomasello, M., & Haberl, K. (2003). Understanding attention: 12- and 18-month-olds know what's new for other persons.
Developmental Psychology, 39, 906­912.
Welman, H. M., Cross, D., & Watson, J. (2001). Meta-analysis of theory-of-mind development: The truth about false belief. Child
Development, 72, 655­684.
Werner, H., & Kaplan, B. (1963). Symbol formation. NY: Wiley.
Whiten, A. (Ed.). (1991). Natural theories of mind: Evolution, development and simulation of everyday mind reading. Oxford: Basil
Blackwell.
Woodward, A. L., & Guajardo, J. J. (2002). Infants' understanding of the point gesture as an object-directed action. Cognitive
Development, 17, 1061­1084.