516
GOUTEUX, THINUS-BLANC, AND VAUCLAIR
during the 50 trials of Experiment 6, Krill, X2(1, N = 60) = 0.05;
Crevet, X2(l, N = 60) = 0.54; p > .05. These results indicate that
no improvement of the monkeys' performance took place during
Experiment 8.
Discussion
The subjects were able to correctly locate the rewarded box
without ambiguity. By using the four different comer cues, the
monkeys were able to distinguish the correct and the geometrically
equivalent corer. The control condition shows that no other cue
than the corer cues was used by the subjects to reorient inside the
experimental apparatus. Thus, we can conclude that when the cues
are salient enough, whatever their position, they can be used by the
monkeys and help them to correctly reorient in the rectangular
room.
General Discussion
In the present experiments, we investigated the ability of young
rhesus monkeys to reorient, after they had been disoriented, in a
perfectly symmetrical rectangular environment, by using the geo-
metric and nongeometric features provided by the apparatus. In the
first series of experiments, we tested the capacity of 3 monkeys to
reorient when only the geometric features of the experimental
rectangular room were available (Experiment 1). In Experiments 2
and 3 we tested the orientation of the monkeys by using a large
nongeometric feature (either associated or dissociated from the
reward box). Our results clearly show that the monkeys can rely on
both geometric and nongeometric features to locate the target. In
the second series of experiments, we investigated the orientation
abilities of the monkeys to use small, distal, and proximal cues
either indirectly (Experiment 4) or directly (Experiment 5) asso-
ciated to the corers of the rectangular room. In both conditions,
the monkeys were unable to correctly locate the target. In effect,
they still relied only on the geometry of the apparatus. In the third
series of these experiments, we investigated the effect of the
landmark size on monkeys' orientation abilities. In Experiment 6,
the use of a small central cue was not salient enough to allow for
a nonambiguous orientation. In Experiments 7 and 8, a larger
central or four different corer cues provided sufficient informa-
tion for correctly orienting the monkeys and allowing them to find
the rewarded box without ambiguity.
Our results extend the earlier findings on rats (Cheng, 1986;
Margules & Gallistel, 1988), and nonhuman (Tinklepaugh, 1932)
and human primates (Hermer & Spelke, 1994, 1996; Wang et al.,
1999), suggesting that these species are able to reorient on the
basis of the geometric features of the environment after disorien-
tation. By testing nonhuman primates in a task similar to that used
by Cheng (1986) and Hermer and Spelke (1996), we provide
further evidence to the existence of a common orientation mech-
anism, which is based on the geometry of the environment, found
in other mammal species, and more specifically in a species
phylogenetically closer to humans than rats. As stated by Cheng
(1986), it seems that there is a unit in the mammalian brain that
specifically encodes the geometric properties of the environment,
such as the metric relations between different spatial elements.
This "modular organization" of spatial representations can be
understood in the Fodorian meaning of a "module," that is, a rigid
structure by which information is required to transit and is auto-
matically processed (Fodor, 1983). As an example, in each of our
experiments, the subjects were able to reorient correctly with
respect to the geometric relationships that were defining the cor-
rect comer (and also its rotational equivalent). Even when mon-
keys were unable to use nongeometric information (Experiments 4,
5, and 6), they still relied, at least, on the geometry of the apparatus
to find the geometrically correct comers.
Gallistel (1990) suggested that disoriented animals determine
their position and heading within a mapped environment by com-
puting the principal axes (or other shape parameters) of the cur-
rently perceived environment and compared this information with
their spatial representation. By comparing shape parameters, the
animal brain determines the spatial information required for mak-
ing its position and orientation with respect to the currently per-
ceived world, congruent with a position and orientation on its
internal spatial representation. If animals compute the heading in
this way, then the process of determining heading is effectively a
module according to Fodor's (1983) definition. Such a module
must be unaffected by sensory data that define properties of the
environment other than its macroscopic shape, no matter how
relevant those properties may be to the task. The hypothesized
computational process is impenetrable to perceptual properties
other than shape because shape parameters, such as principal axes,
are defined only by the shape of the environment.
Furthermore, the experiments by Hermer and Spelke with young
children (1994, 1996) suggest that, in the same way, children's
reorientation system is unaffected by all but geometric informa-
tion, even when nongeometric information is available. According
to Hermer and Spelke, the limits of this process are overcome
during human development. More specifically, Hermer and Spelke
assumed that the use of both geometric and nongeometric infor-
mation provided by the environment after disorientation could be
a uniquely human capacity for problem solving. More striking is
their hypothesis concerning the use of language for solving such
spatial problems. In several articles, these authors (Hermer &
Spelke, 1996; Spelke & Hermer, 1996; Hermer-Vasquez et al.,
1999) proposed that the geometric reorientation process found in
rats and children is preserved in adults and is impervious to
interference from concurrent language production, whereas the
ability to use nongeometric information is vulnerable to such
interference. Indeed, they found that performing a verbal shadow-
ing task during disorientation prevented the participants from
taking into account nongeometric information for orienting
(Spelke & Hermer, 1996).
Thus, language and more specifically spatial language that chil-
dren of 2-3 years begin to produce and use (MacWhinney, 1991)
appears to provide an especially useful medium for representing
conjunctions of spatial and nonspatial properties of the environ-
ment. For example, Hermer-Vasquez, Moffet, and Munkholm
(2001) showed a correlation between the ability of children to
produce and use phrases involving "left" and "right" when ver-
bally describing the hidden location (e.g., "the toy is in the comer
to the left of the blue wall") and the ability to correctly reorient in
the blue-wall condition. Whereas systems for representing the
environmental layout may be confined to capturing geometric
information, and systems for representing movable objects may be
confined to capturing local properties of objects or surfaces, lan-
guage can bring these sources of information together, specifying