SPATIAL REORIENTATION IN MONKEYS
517
that an object or a surface with one set of properties stands in a
particular geometric relationship to another object or surface. In
sum, language acquired by children (starting at 2-3 years of age)
would allow them to perform as well as adults (at 5-7 years of age)
in a reorientation task, and no longer as rats.
Showing for the fast time that a mammal species, the rhesus
monkey, phylogenetically close to humans, is able to combine both
geometric and nongeometric information to reorient (Experi-
ments 2 and 3) leads us to cast doubt (a) on the "language
hypothesis" in the particular experimental setup we used, and (b)
on the proposal that using local features of the environment is a
specifically human ability. As a matter of fact, the joint use of both
geometric and landmark-based cues by rhesus monkeys suggests
that one or several types of spatial processing could either emerge
or have evolved and become more flexible with evolution. Such
flexibility could have evolved independently of specifically human
cognitive features such as symbolic representation, formal reason-
ing, and, of course, language.
One reason that could explain this evolution is that (a) the
foraging behavior in the wild of a monkey is certainly more
complex and more difficult than the foraging behavior of a rat, and
(b) increasing foraging demands led to elaborate cognitive capac-
ities (see Milton, 1981; Vauclair, 1996). It is clear that, for any
mammal, the use of geometric information to reorient in a natural
environment is safer because the macroscopic shape of the land-
scape does not change across seasons. In contrast, there are im-
portant changes in the nongeometric properties of the wild, for
example, the appearance of trees with or without leaves or snow-
fall and melting snow (Hermer & Spelke, 1996). However, for
more complex mammals such as nonhuman primates, the ability to
use also nongeometric information, such as the color of a mature
fruit, could certainly improve their foraging strategies and conse-
quently improve their survival chances.
Alternative explanations of the monkeys' performance in our
experimental situation can be advanced. Because each monkey had
a single rewarded box during each experiment, another possible
explanation of the results observed in Experiments 2 and 3 could
be that our subjects developed some specific strategies during
the 50 trials of each experiment to succeed in the task. Neverthe-
less, close examination of the first data shows no improvement in
performance across trials (see Table 2 for detailed results), indi-
cating that monkeys were relying on the geometric and nongeo-
metric properties of the environment from the first trials. Further-
more, preliminary results from a study conducted with another
primate species, the cotton top tamarin ( Miller, Gouteux, Delpolyi,
Santos, & Hauser, 2001) indicate that, in a similar task, these
monkeys succeeded within four trials in locating the target by
using the nongeometric properties of the environment. Though
there are differences between rhesus monkeys and cotton top
tamarins, we can reasonably conclude that animals from both
primate species are able to use geometric and nongeometric fea-
tures of the experimental environment in the very first trials.
Nevertheless, it could be that subjects used only nongeometric
information to find the correct comer when nongeometric infor-
mation was sufficient to distinguish the correct corner from its
geometrically opposite. In such a case, subjects could then use
self-referent spatial information, namely by distinguishing right
and left side when facing the landmark. However, according to our
results and to the modularity hypothesis, it would be surprising that
geometric information was not used, even when the environment
provided nongeometric cues. In effect, the use of geometry appears
to be a predominant element for spatial reorientation. Thus, in each
experiment we ran, monkeys used geometry to reorient even when
cues were available (but not used) in the experimental apparatus.
This finding led us to claim that when monkeys use nongeometric
spatial information, they certainly also use geometric information.
When more precise local information was provided (a small
angular cue in Experiments 4 and 5), the monkeys were not able to
correctly reorient by using this unambiguous spatial information.
In these cases, they were acting as if no spatial information other
than the shape of the experimental environment was present:
Monkeys were searching with equal frequencies in the correct and
geometrically correct locations. These results lead us to suggest
several alternative interpretations. First of all, the cues might not
have been noticed by the subjects (that would explain why they did
not use them to reorient). However, the analysis of videotapes
shows that in many trials the monkeys displayed investigatory
reactions to these comer cues. For example, they often took a look
at or touched them during trials. Thus, it seems unlikely that
monkeys had not paid attention to these cues.
Another explanation of the errors made during Experiments 4
and 5 could be due to a lack of motivation of the subjects.
However, rhesus monkeys were sufficiently motivated (and at-
tended) to search correctly in the geometrically appropriate comer,
indicating that they correctly encoded and used the geometry of the
room to reorient.
A more plausible explanation would be that these angular cues
were not considered by the monkeys as stable or nonmovable.
According to Biegler and Morris (1993, 1996), the relative stabil-
ity of landmarks is an important factor in the development of
spatial behaviors. It is generally thought that an object must be
considered as immovable in order to be used as a landmark in
navigation tasks (for an example with young children, see
Acredolo, 1990). In Experiment 4, the two geometric comer cues
were shown before disorientation but were removed afterward.
This experimental procedure may have led the monkeys to con-
sider the angular cues as movable and thus not reliable for reori-
entation. This kind of explanation is also suggested by Learmonth,
Newcombe, and Huttenlocher (1998) to explain Hermer and
Spelke's results with children. According to Learmonth et al.
(1998), the nongeometric cues in Hermer and Spelke's experi-
ments appear to be nonpermanent because children could see the
experimenter attaching the blue fabric to one of the white walls,
but this fabric was sometimes there and sometimes not there
(because the manipulation was performed within subjects). In
another study by Hermer and Spelke (1996), children were allowed
to play with a toy truck and a toy teddy bear, that is, with movable
objects that were used as landmarks by the experimenter. On the
basis of these data, Learmonth et al. have replicated Hermer and
Spelke's experiments with the hypothesis that children of the same
age would reorient by using more permanent appearing landmarks,
such as a door or a bookcase. Their data show that children
between 18 and 24 months can, to some extent, use the door and
the bookcase to reorient because they found the hidden toy reliably
more often than chance and search reliably more often at the
correct comer than at the rotationally equivalent comer.
Concerning the results of Experiment 5, the only explanation
that accounts for the lack of utilization of local cues for reorienting