506
GOUTEUX, THINUS-BLANC, AND VAUCLAIR
the subject was rotated several times without vision. For instance,
in Cheng's (1986) experiments, rats searched for food previously
hidden in one of the four corners of a rectangular apparatus. After
animals were familiarized with the experimental environment, they
were removed from the apparatus, disoriented within a closed box,
and returned to the apparatus to search for food. In this reorienta-
tion task, the rat had to reestablish its position and heading before
it engaged in goal-directed behavior. To be oriented again, the rat
could rely on the shape of the experimental apparatus, on the
patterns, on the odors (if any), or on the brightness of the walls.
Assuming that the oriented rat initially stored an internal repre-
sentation (Tolman, 1948) of the food location (e.g., that the food
was hidden near a particular cue or odor) and that the disoriented
animal had to reorient itself to retrieve food (see also Gallistel,
1990; MacNaughton, Knierim, & Wilson, 1994), Cheng reason-
ably concluded that the position where the rat searched for the
reward indicates which spatial element the rat used to reorient
itself.
In Cheng's (1986) experiments, rats showed a high rate of
search both at the correct corner (the reward corner) and at the
rotationally equivalent opposite corner, and these two corners on
the same diagonal are defined by the same geometric relation
within the experimental apparatus (length and width). This search
pattern was constant during all experiments conducted by Cheng
(1986, 1987), despite the availability of many cues (including
obvious salient cues, such as strong distinctive odors, and large
differences in contrast and luminosity) that could easily differen-
tiate the two symmetrical locations. This finding suggests that rats
were able to reorient in accord with the geometric relations pro-
vided by the environment. However, the rats seemed to be unable
to take into account the nongeometric properties of that environ-
ment. Cheng concluded that the failure to reorient along nongeo-
metric information stemmed from limits that are specific to the
reorientation process. For Cheng (1986), reorientation in rats de-
pended on a "geometric module," a task-specific, encapsulated
system (see Fodor, 1983).
However, some of the results obtained by Cheng suggest that the
greatest number of searches were conducted in the correct location,
even when no significant difference was noted between the two
geometrically correct corners. This tendency to retrieve the correct
location could reflect a weak ability to use nongeometric informa-
tion but could also result from an incomplete disorientation pro-
cedure of the rats. To distinguish between these two possibilities,
Margules and Gallistel (1988) used a more rigorous disorientation
procedure. Rats were first exposed to the reward location in an
exposure box and then moved into another similar box (test box)
that was differently located and oriented compared to the first one.
This ingenious disorientation procedure led Margules and Gallistel
to the same conclusion drawn by Cheng, namely that rats failed to
use the nongeometric information to correctly locate the target.
These findings, and similar findings from other investigators
(Biegler & Morris, 1993, 1996; Dudchenko, Goodridge, Seiterle,
& Taube, 1997), provide further empirical data, showing that rats
reoriented in accord with the shape of the environment but seemed
to neglect nongeometric properties of the environment.
These findings stand in contrast with those gathered in the
animal physiology literature, showing that rats can easily reorient
and retrieve a goal by using different nongeometric information
provided by the environment (see, e.g., Benhamou & Poucet,
1998; Knierim, Kudrimoti, & MacNaughton, 1995). Differences
between the experimental environments, between procedures and
motivational factors, could account for the discrepancy observed in
the data (Hermer, 1997). For example, in studies by Cheng (1986),
Hermer and Spelke (1994, 1996), or Biegler and Morris (1993,
1996), subjects were tested in apparatus with an informative shape
(such as a rectangle). In contrast, other studies often placed rats in
an environment with a minimal distinctive geometry (most often
an enclosed cylinder). Another difference that may account for
discrepant data is the level of familiarity with the test situation.
Most often, in the studies about geometry, the subjects spend a
relatively short time in test environments, whereas more classical
studies have tested rats repeatedly in situations that become famil-
iar through learning. It is thus likely that animals may use non-
geometric information only when geometric information is mini-
mized and when the cues have become stable and familiar.
However, the use of geometric coding might have an adaptive
significance for an animal.
Thus, Hermer and Spelke (1996) have provided an interesting
ecoethological explanation for the predominant use of geometric
information for reorientation in a natural environment. The mac-
roscopic shape of the landscape does not change across seasons. In
contrast, there are important changes in the nongeometric proper-
ties of the environment (e.g., the appearance of trees with or
without leaves, or snowfall and melting). Thus, it is safer for
animals to rely on geometric information that is not modified by
seasonal changes. In a recent study (Gouteux, Vauclair, & Thinus-
Blanc, 1999), using a reaction to spatial change procedure, we
examined the spontaneous exploratory activity of four young ba-
boons to determine the type of spatial coding (purely geometric or
related to the identity of the objects) that was implemented while
the animals were getting acquainted with an object configuration.
During a habituation phase, all the subjects were individually
familiarized with the initial spatial configuration (in an outdoor
enclosure) made by the four objects of the configuration. We
observed a decrease in the duration of contacts, indicating that the
baboons got progressively familiarized with the initial situation.
After habituation, animals were tested for their exploratory reac-
tions (contact duration and order of spontaneous visits) to spatial
changes brought about to the initial object configuration. Two
kinds of spatial changes were made: a modification of (a) the shape
of the configuration (by displacing one of the four different ob-
jects), and (b) the spatial arrangement without changing the initial
shape (exchanging the location of two objects). In a second ex-
periment, the four objects were identical, and a modification of the
geometry was performed. Finally, in the third experiment, a sub-
stitution of a familiar object with a novel one was performed
without changing the object configuration. Results showed that
baboons strongly reacted to the geometric modifications of the
configuration because they massively and selectively reexplored
only the displaced object. In contrast, the baboons were less
sensitive to changes of the local features that did not affect the
initial spatial configuration. In this latter case, no specific spatial
reexploration of the displaced objects was noticed. However, the
results also suggest that the geometric encoding requires that the
various elements that define the geometry of the explored space
were not identical and that each of the various elements specifies
a location to induce a selective reexploration directed toward the
displaced objects.