SPATIAL REORIENTATION IN MONKEYS
507
However, the inability to conjointly rely on geometry and non-
geometric properties of the environment to reorient is not present
in all vertebrates. Some avian species have been tested during a
reorientation task adapted from Cheng's (1986) initial experiment.
Results clearly demonstrated that chickens (Vallortigara, Zanfor-
lin, & Pasti, 1990) and pigeons (Kelly, Spetch, & Heth, 1998), like
rats and humans, use geometry to reorient. Surprisingly, these
species were also able to use nongeometric spatial information to
perfectly reorient. Hermer and Spelke (1994, 1996), in a series of
studies conducted with human children, have examined the use of
a geometric module by toddlers. Children (aged from 18 to 24
months) saw a desired toy that was being hidden in one of the
comers of a rectangular homogeneous experimental chamber. Af-
ter disorientation, the participants were asked to retrieve the toy. In
one of the experiments, the chamber contained no distinctive
landmark. In another one, a nongeometric feature (blue wall) that
broke the symmetry of the experimental apparatus was added.
Hermer and Spelke found that, when no information other than the
shape of the environment was available, children searched equally
often in the correct and in the rotationally equivalent corner, and
more frequently in these two corners than in the other two remain-
ing comers. When nongeometric information (a blue wall or a pair
of toys placed in the room) was added, children still divided their
searches between the two rotationally equivalent corners and
seemed to ignore the added salient cues.
Hermer and Spelke concluded that young children do not use
information other than the shape of the experimental environment
to reorient, even when more salient nongeometric information was
available and could help them to locate the correct corner. In the
same study, Hermer and Spelke also tested human adults with the
same experimental setup. Unlike children, adults were able to use
both geometric and nongeometric information to optimize their
search. Similar results have been found by Wang, Hermer, and
Spelke (1999) in a square room; children (18-24 months old) used
a distinctive geometric cue, but not a colored wall, to locate the
hidden object, even though they had been familiarized with the
colored wall over multiple training sessions.
These studies suggest that young children reorient by a process
that is encapsulated, task-specific, and common to other mammals,
whereas adults reorient in a more flexible way. To explain this
source of flexibility, Spelke and Hermer (1996) argued that lan-
guage is necessary to penetrate the geometric module and to allow
for reorientation by integrating landmarks. More precisely, these
authors proposed that the age at which children begin to success-
fully locate the target by using geometric and nongeometric infor-
mation (at about 6-6.5 years of age; Hermer, unpublished data
quoted by Spelke and Hermer, 1996) approximately corresponds to
the age at which children begin to produce sentences that would
uniquely specify object location and orientation (MacWhinney,
1991), such as "near the blue wall" or "to the right/left of the blue
wall." Moreover, some interesting data show that when human
adults perform the reorientation task previously mentioned, al-
though they have to perform concurrent tasks (e.g., verbal shad-
owing), they are unable to use nongeometric information, but they
still succeed in using geometric information (Hermer-Vasquez,
Spelke, & Katsnelson, 1999). According to Spelke and Hermer
(1996), language seems to be necessary to bind the geometric
module and other, nongeometric modules. Consequently, the use
of both geometric information and local features in the environ-
ment would constitute a specificity of adult humans' spatial cog-
nitive abilities.
To our knowledge, no study has attempted to investigate this
particular question in species other than rats (Cheng, 1986), birds
(Vallortigara et al., 1990; Kelly et al., 1998), or human adults and
children (Hermer & Spelke, 1994, 1996). From a comparative
point of view, it seems that the use of geometry to reorient is
shared by several mammals. Along this perspective, we tested
rhesus monkeys in the present experiments with the goal of ex-
amining whether nonhuman primates use a reorientation mecha-
nism that is based on the geometry of the environment (Experi-
ment 1) and whether such a mechanism is compatible with the use
of nongeometric cues (Experiments 2 and 3). Because adult hu-
mans are able to linguistically encode the relevant information in
the reorientation experiment, the relationship between spatial and
object information can be encoded in other ways as well. Could a
nonlinguistic animal solve tasks that require the integration of
information from two distinct aspects of core knowledge? In
another way, by testing nonhuman primates, these experiments
allow us to test the language hypothesis, put forward by Hermer
and Spelke, explaining why in a reorientation task, only human
adults are able to conjointly use geometric and nongeometric
spatial information. In addition, assuming that rhesus monkeys
were able to use these two kinds of spatial information, we
investigated to what extent monkeys could take into account local
cues during their reorientation (Experiments 4 and 5) and also
whether they were sensitive to the size and location of such cues
in the experimental apparatus (Experiments 6, 7, and 8).
Part 1
We investigated the spatial reorientation abilities of nonhuman
primates by using the geometric and nongeometric properties
provided by a rectangular room. For that purpose, three experi-
ments, adapted from the original experiment by Cheng (1986),
were conducted with 3 rhesus monkeys. The first experiment was
designed to test the reorientation of the monkeys when the only
available spatial information was the shape of the apparatus. In the
next two experiments, we investigated the use of geometric and
nongeometric information, either associated to the rewarded box
(Experiment 2) or dissociated from it (Experiment 3) in a task
performed within the same rectangular room. The order of these
experiments was counterbalanced across subjects.
Experiment 1
Method
Subjects. Three experimentally naive young rhesus monkeys (Macaca
mulata) were individually studied. The group included 2 males (Orcas and
Krill) and I female (Crevet). The subjects were 1.9 (SD = 0.4) years of age
at the beginning of the experiment. All monkeys were born in captivity at
the Primatology Unit of the CNRS in Marseille, France, and they were
housed in individual indoor cages during the experiments. They received
food pellets and vegetables once a day in the afternoon.
Apparatus. Subjects were tested in a rectangular chamber (4.0 m
long X 2.5 m high X 2.0 m wide), filled with 3 cm of sawdust and perfectly
homogeneous with respect to visual cues (see Figure 1). The apparatus was
housed within a larger dark room with no obvious source of noise. It was
composed of a metallic frame entirely recovered by white wooden panels.
Two doors with an access hatch, opening at the center of each of the two