Downloaded from http://rsbl.royalsocietypublishing.org/ on July 28, 2017
Animal behaviour
rsbl.royalsocietypublishing.org
Preparatory responses to socially
determined, mutually exclusive
possibilities in chimpanzees and children
Research
Thomas Suddendorf, Jessica Crimston and Jonathan Redshaw
Cite this article: Suddendorf T, Crimston J,
Redshaw J. 2017 Preparatory responses to
socially determined, mutually exclusive
possibilities in chimpanzees and children.
Biol. Lett. 13: 20170170.
http://dx.doi.org/10.1098/rsbl.2017.0170
Received: 20 March 2017
Accepted: 18 May 2017
Subject Areas:
cognition
Keywords:
prospective cognition, episodic foresight,
mental time travel, uncertainty, developmental
psychology, comparative psychology
School of Psychology, University of Queensland, St Lucia, Queensland 4072, Australia
TS, 0000-0003-3328-7442
The capacity to imagine and prepare for alternative future possibilities is
central to human cognition. Recent research suggests that between age 2
and 4 children gradually begin to demonstrate a capacity to prepare for
two simple, mutually exclusive alternatives of an immediate future event.
When children were given the opportunity to catch a target an experimenter
dropped into an inverted Y-shaped tube, 2-year olds—as well as great
apes—tended to cover only one of the exits, whereas 4-year-olds spontaneously and consistently prepared for both possible outcomes. Here we
gave children, age 2 to 4 years, and chimpanzees a different opportunity
to demonstrate potential competence. Given that social behaviour is particularly full of uncertainty, we developed a version of the task where the
outcome was still unpredictable yet obviously controlled by an experimenter. Participants could ensure they would catch the target by simply
covering two tube exits. While 4-year-olds demonstrated competence, chimpanzees and the younger children instead tended to cover only one exit.
These results substantiate the conclusion that the capacity for simultaneous
preparation for mutually exclusive event outcomes develops relatively late in
children and they are also in line with the possibility that our close animal
relatives lack this capacity.
1. Introduction
Author for correspondence:
Thomas Suddendorf
e-mail: [email protected]
Electronic supplementary material is available
online at https://dx.doi.org/10.6084/m9.
figshare.c.3789799.
The capacity to imagine diverse future events and prudently adjust current behaviour accordingly is a hallmark of human cognition [1,2]. Given that there are
typically multiple ways in which future events might unfold it can be beneficial
to prepare for more than just one eventuality. So people make contingency
plans and hedge their bets as they navigate their environment. Yet little is
known about when these capacities develop in children [3–5] and to what
extent they are shared with non-human animals [6,7]. Recently, a minimalist
paradigm has been introduced to give young children and great apes an opportunity to demonstrate this most basic capacity [8]. Subjects were presented with
an inverted Y-shaped forked tube and the experimenter dropped a target into
the top that could be caught coming out of one of the two exits (missed targets
rolled out of reach). Most 4-year olds spontaneously and consistently prepared
for both possibilities by placing their hands under both exits. This unswerving
success demonstrates that they have the basic capacity to simultaneously prepare for alternative event outcomes. In contrast, all the 2-year olds as well as
the chimpanzees and orangutans covered only one of the exits on the first
trial—as did most 3-year olds—and even those who eventually covered both
exits often regressed to covering only one on subsequent trials.
Failures are difficult to interpret as there are usually reasons other than the
cognitive capacities in question that may have limited performance [6]. However, we can provide subjects with multiple opportunities to demonstrate the
& 2017 The Author(s) Published by the Royal Society. All rights reserved.
Downloaded from http://rsbl.royalsocietypublishing.org/ on July 28, 2017
2
rsbl.royalsocietypublishing.org
Biol. Lett. 13: 20170170
Figure 1. Parallel tube set-up as presented to children and chimpanzees.
capacity in ways that minimize other demands. For instance,
subjects in this task may have been confused about what causally happens at the juncture where the tube forks into two
exits (they could not know that an internal platform and
hidden screw allowed the experimenter to determine where
the target exited). Here we therefore aimed to provide a
simple, visible and unambiguous (yet still unpredictable)
causal mechanism that determined where the target would
exit. Furthermore, given that the social world harbours
many uncertainties, especially if other agents have plans of
their own, it is possible that the capacity is more readily
observable in such a context. Some research suggests that
while young children and apes show similar skills when dealing with physical problems, children display more
sophisticated capacities when dealing with problems in the
social world (e.g. [9]). Thus, one might expect young children
but not apes to improve in performance when the contingencies of the task are clearly controlled by another agent, rather
than some non-social mechanism, as in the previous study.
We therefore decided to give children and chimpanzees the
opportunity to demonstrate a basic capacity to prepare for
two simple, mutually exclusive possible outcomes of a
future event in a social game in which the experimenter
simply dropped the target into one of two parallel tubes.
2. Material and methods
We tested 36 Australian children (19 females, 17 males): 12 2-year
olds (Mage ¼ 24.93, s.d. ¼ 1.09), 12 3-year olds (Mage ¼ 36.34,
s.d. ¼ 0.49), and 12 4-year olds (Mage ¼ 47.78, s.d. ¼ 0.64) at the
Early Cognitive Development Centre at the University of
Queensland. We also tested five chimpanzees at the Rockhampton Zoo. These included three subjects that had participated in
the original forked tube study (‘Cassie’ a 43-year-old male,
‘Holly’ a 26-year-old female and ‘Samantha’ a 31-year-old
female) and two new group mates who to our knowledge had
not participated in any prior psychological research (‘Leakey’ a
21-year-old female and ‘Alon’ an 8-year-old male). Subjects
were presented with a simple two tube apparatus (see figure 1)
and were given the opportunity to catch a target (a ball or a
grape) that the experimenter dropped into one of the tubes. A
mesh fence prevented subjects from merely following the position of the target with their own hand (except for Cassie, who
was tested in a different area and had free movement of his
hands). Instead, they had to place one or both hands through
the mesh under one or both tube exits. If they did not catch
the target it fell on a ramp and rolled out of reach. At the beginning of each trial, the experimenter held the item in a position
above and in between the two tubes before rapidly moving the
hand to either the left or the right tube and immediately dropping it inside. This action happened quickly enough that it was
not possible for the participants to simply wait for the experimenter to move and then react by covering the appropriate
exit to catch the target. Prior to the test phase subjects completed
a demonstration phase in which they could see the item fall from
the exits (but not catch it) in a pseudorandom order for six trials.
During the test phase, subjects had the opportunity to catch the
item for 12 trials, as it fell from the exits in a pseudorandom
order. Further methodological details are outlined in the
electronic supplementary material.
3. Results
All but one 2-year old prepared for only one possible outcome on the first trial, but with increasing age children
were more likely to cover both exits on this trial, Cochran–
Armitage x2ð1Þ ¼ 5:19, p ¼ 0.023. Children’s (and chimpanzees’)
overall performance is summarized in figure 2.
A series of generalized estimating equations (GEE) analyses was conducted to compare children’s performance
across age groups and all 12 trials (see electronic supplementary material for full details). The best fitting model contained
a significant linear effect of age on covering both exits, generalized-score (GS) x2ð1Þ ¼ 7:74, p ¼ 0.005, but it did not contain
a linear trial effect or an interaction between trial and age.
Older children were also significantly more likely to cover
both exits on at least one trial, Cochran– Armitage
x2ð1Þ ¼ 11:56, p ¼ 0.001. Note that on 47 of the total 432
trials children failed to cover either exit—typically instead
placing a hand directly underneath the experimenter’s pretrial hand position in between the two exits, such that they
would have caught the ball if the experimenter dropped it
Downloaded from http://rsbl.royalsocietypublishing.org/ on July 28, 2017
100
3
cumulative % of participants
80
70
first trial, sustained
60
50
later trial, sustained
40
any trial, regressed
30
no trial
20
ch
im
pa
nz
ee
s
os
4y
os
3y
2y
os
0
Figure 2. Cumulative percentages of children (across age groups) and chimpanzees categorized according to when they covered two exits across all 12 trials. Blue
bars indicate participants that covered two exits on the first trial and all subsequent trials. Green bars indicate participants that did not cover two exits on the first
trial, but did so on a later trial and subsequently maintained that response across trials. Yellow bars indicate participants that covered two exits at some stage but
regressed to covering one exit on a later trial. Red bars indicate participants that did not cover two exits on any trial.
here. However, as 40 of these cases came from four 4-year-old
children, this behaviour did not drive the overall pattern of
age-related performance increase (and in fact it may have
artificially deflated the performance levels of the 4-year olds).
None of the chimpanzees covered both exits on the first
trial, or indeed on any of the 12 trials (figure 2). These subjects included Samantha and Holly, who had covered both
exits on at least some later trials in the previous study [7],
so there was no sign of a carry-over effect. Instead, the chimpanzees tended to cover one exit and so obtained a reward on
about half of the trials. However, the two subjects that lacked
experience with psychological testing frequently reached
towards the location where the experimenter was holding
the grape before the trial, instead of covering either of the
exits. We therefore administered extra trial blocks to these
two subjects after a further training phase. Subsequently,
both subjects regularly covered one exit and Alon even covered two exits on four occasions, although he regressed to
covering one exit on following trials (see electronic supplementary material for further details). Such a response
pattern was observed also in some of the children
(figure 2), as well as in the great apes and children that
participated in the previous study [7].
4. Discussion
The current study found that, in a social game, the capacity
for simultaneous preparation for mutually exclusive event
outcomes was evident in children around age 4. Few 3-year
olds and one 2-year old also demonstrated competence, but
the other children and chimpanzees did not. This pattern of
results is similar to the recent finding with the forked tube
task [8] and suggests that difficulties understanding the
causal mechanism were not responsible for poor performance
in the original study. Subjects in the current study were able
to observe why the target was going to come out of the exit it
did come out of (i.e. they saw the experimenter drop it into
the relevant tube), yet they had to prepare for the possibility
that the experimenter dropped it in either one of the tubes if
they wanted to be certain about catching it. However, presentation of the problem in this arguably more ecologically valid,
social context did not improve performance levels: neither on
the first trial nor subsequently; and neither for chimpanzees
nor for young children who in other contexts outperformed
chimpanzees on social tasks [9].
The fact that some children and, on subsequent later
trials, one chimpanzee covered both exits demonstrates that
they are physically capable of the optimal response, even if
they subsequently regressed to covering only one exit. This
may reflect subjects being weakly conditioned into using
the two-handed response through trial-and-error learning
rather than insightful preparation for two mutually exclusive
versions of the future. The first trial and consistent successes
of many older children, on the other hand, strongly suggest
they are capable of such deliberate preparation for an
uncertain future.
In contrast to our postulation that the current task may be
easier to pass than the original forked tube task [8] because
the causal mechanism is visible, unambiguous and social,
we found that the children performed worse (see electronic
supplementary material for a cross-study statistical analysis).
One possibility is that they may have perceived the task as a
social game in which they were required to guess, rather than
‘cheat’ by covering both exits. This seems unlikely, however,
considering that children’s conformity to implicit rules
against cheating generally increases with age [10], whereas
here we found that older children were more likely than
younger children to cover two exits. Although we cannot
say for certain why the children performed worse on this version of the task, one possibility is that young children’s
mental representations of others’ intentions (as in the current
task) may be stronger than their representations of physical
possibilities (as in the forked tube task), leading them to be
more committed to their initial prediction of the experimenter’s action. Another possibility is that children are
more likely to commit to a single possible outcome when
the mechanism controlling the outcome is obvious (cf. [3]).
More research is required to clarify the reasons for differential
performance levels, for instance using similar tasks where the
Biol. Lett. 13: 20170170
10
rsbl.royalsocietypublishing.org
90
Downloaded from http://rsbl.royalsocietypublishing.org/ on July 28, 2017
future events compared with both younger children and
our closest living animal relatives. The difference could lie,
for example, in the capacity to metarepresent: that is, in the
capacity to reflect on the fact that representations of possible
future events (e.g. experimenter drops ball into the left tube)
could be incorrect, and that therefore it is worth preparing for
other possibilities (e.g. experimenter drops ball into the right
tube) as well [7,8]. The present results are consistent with
more general developmental changes in future-oriented
behaviour around age 4 [13] and also with the proposal
that certain elements of foresight may be uniquely human [6].
Funding. This research was supported by an Australian Research
Council grant to T.S.
Acknowledgements. We thank the parents and children that participated
in this research as well as the staff at Rockhampton Zoo. We also
thank Adam Bulley, Karri Neldner and Phoebe Pincus for support
with the comparative testing.
References
1.
2.
3.
4.
5.
Michaelian K, Klein SB, Szpunar K (eds). 2016
Seeing the future: theoretical perspectives on futureoriented mental time travel. New York, NY: Oxford
University Press.
Suddendorf T. 2006 Foresight and evolution of the
human mind. Science 312, 1006– 1007. (doi:10.
1126/science.1129217)
Beck SR, Robinson EJ, Carroll DJ, Apperly IA. 2006
Children’s thinking about counterfactuals and future
hypotheticals as possibilities. Child Dev. 77, 413–
426. (doi:10.1111/j.1467-8624.2006.00879.x)
Robinson EJ, Beck S. 2000 What is difficult about
counterfactual reasoning? In Children’s reasoning
and the mind (eds P Mitchell, KJ Riggs), pp. 101–
119. Hove, UK: Psychology Press/Taylor & Francis.
Robinson EJ, Rowley MG, Beck SR, Carroll DJ,
Apperly IA. 2006 Children’s sensitivity to their own
6.
7.
8.
9.
relative ignorance: handling of possibilities under
epistemic and physical uncertainty. Child Dev. 77,
1642 –1655. (doi:10.1111/j.1467-8624.2006.
00964.x)
Suddendorf T. 2013 The gap—the science of what
separates us from other animals. New York, NY:
Basic Books.
Redshaw J. 2014 Does metarepresentation make
human mental time travel unique? Wiley Interdiscip.
Rev. Cogn. Sci. 5, 519–531. (doi:10.1002/wcs.1308)
Redshaw J, Suddendorf T. 2016 Children’s and apes’
preparatory responses to two mutually exclusive
possibilities. Curr. Biol. 26, 1758 –1762. (doi:10.
1016/j.cub.2016.04.062)
Herrmann E, Call J, Hernandez-Lloreda MV, Hare B,
Tomasello M. 2007 Humans have evolved
specialized skills of social cognition: the cultural
10.
11.
12.
13.
intelligence hypothesis. Science 317, 1360–1366.
(doi:10.1126/science.1146282)
Kochanska G, Murray K, Coy KC. 1977 Inhibitory
control as a contributor to conscience in childhood:
from toddler to early school age. Child Dev. 68,
263–277. (doi:10.2307/1131849)
Henrich J, Heine SJ, Norenzayan A. 2010 The
weirdest people in the world? Behav. Brain Sci. 33,
61– 83. (doi:10.1017/S0140525X0999152X)
Nielsen M, Haun D. 2016 Why developmental
psychology is incomplete without comparative and
cross-cultural perspectives. Phil. Trans. R. Soc. B
371, 20150071. (doi:10.1098/rstb.2015.0071)
Suddendorf T, Redshaw J. 2013 The development of
mental scenario building and episodic foresight.
Ann. NY Acad. Sci. 1296, 135–153. (doi:10.1111/
nyas.12189)
Biol. Lett. 13: 20170170
Ethics. This experiment was approved by the UQ Animal Ethics Committee (PY332/13) and the School of Psychology Ethics Committee
16-PSYCH-4-120-AH, and the work was carried out in accordance
with the relevant guidelines. Informed consent was obtained from
children’s parents (as was permission for reproducing the photo).
Data accessibility. The dataset supporting this article has been uploaded
as part of the electronic supplementary material.
Authors’ contributions. T.S. and J.R. designed the experiment. T.S. tested
the chimpanzees; J.C. tested the children. All authors contributed
to the writing of the manuscript, approved the final version, and
agree to be held accountable for the content.
Competing interests. We have no competing interests.
4
rsbl.royalsocietypublishing.org
outcome is certain rather than uncertain. Perhaps there are
other contexts in which subjects can perform well and so
more studies with different versions of such tasks are
needed to give children and animals further opportunities
of demonstrating competence if they have it. Although we
cannot conclude that chimpanzees and young pre-schoolers
are not capable of conceiving of and preparing for mutually
exclusive possibilities per se, repeated failure on diverse versions of such tasks makes such an explanation increasingly
likely.
Similarly, having established competence in older preschoolers across several studies and contexts increases the
likelihood that by that stage of development this becomes
part of normal cognitive maturation. However, thus far we
have only tested children in Australia who may be regarded
as ‘WEIRD’ (Western, Educated, Industrialized, Rich and
Democratic) [11], and work with children from other cultural
backgrounds is required before conclusions about human
universality are warranted [12].
In conclusion, the current results show that many 4-yearold children are capable of preparing for two mutually
exclusive alternative actions of a social agent, just as they
are capable of preparing for multiple possible outcomes of
a future event caused by a hidden, non-social mechanism
[7]. We could find only limited evidence for this capacity in
younger children and no evidence in a sample of chimpanzees. These results support the possibility that there is a
genuine difference in how older children consider uncertain