51 2
GOUTEUX, THINUS-BLANC, AND VAUCLAIR
whether the location of featural information (near or far from the
target location) is crucial for disambiguating geometrically equiv-
alent comers. For that purpose, the four landmark panels were
located at each comer, during the presentation phase (before dis-
orientation). For the test trials and before the disorientation, the
panels at the correct corner and at its diagonally opposite corner
were removed. Consequently, there was no distinctive featurai
information to the correct corner and to its diagonal opposite. Only
landmarks located at the other two corners could be used to
differentiate the baited comer from its diagonal opposite, as it was
the case for Cheng's (1986) experiments.
Method
Subjects. The same 3 subjects were used in this experiment. A delay
of 1 month occurred between Experiments 3 and 4.
Apparatus. The apparatus was the same as used in Experiment 1
(all-white condition) but a specific landmark was located at each corner of
the experimental room. Landmarks had a similar shape and size (32
cm x 23 cm) but different colors (red, yellow, blue, and green) and
different patterns (horizontal lines, crosses, circles, and squares) were used
(see Figure 3).
Procedure. The same procedure as in the previous experiments was
used, except that during the disorientation procedure, the two landmarks of
the geometrically correct corners were removed.
Results
Table 5 presents the number of first-choice searches performed
by the 3 subjects during the 50 test trials. For each monkey, the
data for the test were subjected to a chi-square one-sample test by
which we compared the observed distribution in the geometrically
appropriate and geometrically inappropriate categories to the the-
oretical frequency of an equal distribution of these two categories
(i.e., 50% of chance for each one). These results always reached
statistical significance, Orcas, X(1, N= 50) = 23.1; Krill, X2(1, N
= 50) = 15.7; Crevet, X2(1, N = 50) = 50.0; p < .001. A second
chi-square one-sample test computed on the data obtained in the
observed geometrically appropriate category compared with an
equal frequency of distribution of the two comers' choices of that
category (i.e., 50% of chance for Comers C and R) indicated that
the number of visits to the geometrically appropriate comers were
not statistically different, Orcas, X(1, N= 42) = 2.38; Krill, X2(1,
N = 39) = 1.24; Crevet, X2(1, N = 50) = 0.32; p > .05.
Data analysis of the 10 first trials for each subject are reported
in Table 2 (Experiment 4). The score for the 3 subjects for the 10
first trials shows no statistical differences with the results of the
entire experimental session (i.e., 50 trials), Orcas, X(1, N = 10)
= 1.82; Krill, X2(1, N = 10) = 2.45; Crevet, X2(1, N= 10) = 0.18;
p > .05. Thus, we can conclude that no obvious improvement of
the monkeys' performance takes place during the experimental
procedure.
Discussion
The results of Experiment 4 indicate that monkeys were not able
to use small-size distal information to retrieve the rewarded box.
However, they still relied on the geometry of the apparatus because
they searched as often in the correct box as in the geometrically
equivalent one. An alternative explanation is that both the small
size of the landmarks and their location, dissociated from the
baited comer, may have rendered the task too difficult. Thus, to
test for this hypothesis, we ran another experiment (Experiment 5)
where four different panels remained during the test phase (after
disorientation).
Experiment 5
Because the monkeys failed to retrieve the reward box by using
the featural information distant from the geometrically correct
location, in Experiment 5, we tested whether monkeys were able to
use four small different landmark panels to reorient and find the
reward. In that experiment, each corner was individualized by a
featural panel, during the whole experiment. If monkeys used the
local cues to reorient, then they should have increased their per-
formance by finding the correct box during their first search.
Method
Subjects. The same 3 subjects were used in this experiment. A delay
of 1 month occurred between Experiments 4 and 5.
4 5
Figure 3. Overhead view of the testing environment for Experiments 4 and 5. Each comer is identified by a
unique nongeometric colored pattern. In Experiment 4, the two patterns directly associated to the rewarded
comer and to the geometrically equivalent comer were removed during subjects’ disorientation. The correct
location (noted “C”), the rotational equivalent comer (noted “R”), and the geometrically inappropriate near
(noted “N”) and far (noted “F”) comers are shown in the figure.