9. Echolocation in Dolphins
401
and Patterson 1981). Another reason for the experiment was to study the
dolphin's decision-making process.
Helweg et al. (1996) examined the capabilities of an echolocating dolphin
to develop aspect-independent representation of acoustically nonsymmetrical targets from echolocation. Echoes from the foam targets measured in
a test pool at aspects of 0° to 90° in 10° increments are shown in Figure 9.23,
where 0° corresponded to the planar view. The experimental setup of
Roitblat et al. (1990) was used and a similar matching-to-sample paradigm
was incorporated to test the dolphin. The orientation of the sample and the
comparison targets was not controlled in the experiment so that the targets
were free to revolve in a random fashion from trial to trial. The dolphin
correct response level for the different targets were 81 % for the rectangle,
79% for the pyramid, and 17% for the cube. However, when the sample
target was the cube, the animal had a strong bias toward responding rectangle. This occurred 63% of the time, suggesting that the dolphin could
recognize the cube well above chance but had a biased response. The results
of this study support the speculation that dolphins identify objects by averaging, or integrating, information gleaned from successive echoes within an
echolocation scan.
Harley et al. (1996) examined the dolphin's ability to integrate visual and
echolocation information. A variety of different types of targets were used
in a matching-to-sample paradigm to test the ability to recognize targets
using echolocation only, vision only, and echolocation with vision and crossmodally between echolocation and vision. They found that the dolphin performed best when it was allowed to use both vision and echolocation. They
also found that with familiar targets, the dolphin could perform cross-modal
matching, suggesting that the dolphin has an object-based representation
system. However, the dolphin performed considerably better when the task
was to echolocate the sample and visually choose the matching target
(echoic-vision) than vice versa. For familiar targets, the dolphin's performance on one set of targets was 78% correct for the echoic-vision and 22%
for the vision-echoic condition. For another set of familiar targets, correct
performance was 99% for echoic-vision and 61 % for vision-echoic, and for
a third set of targets the correct performance was 67% for echoic-vision
compared to 39% for vision-echoic. When novel targets were used, the
dolphin's performance was 39% correct for both echoic-vision and visionechoic. Chance performance was 33%.
Investigation of cross-modal identification of targets was also performed
by Pack and Herman (1995). They used targets made out of short sections
of gray PVC pipes having outer diameters of 1.27, 1.91,2.54, and 3.81 cm.
The short sections of pipes were filled with sand and glued together so that
a variety of different target shapes were achieved. The dolphin was trained
to do a matching-to-sample task visually, echoically and cross modally. In
the experiment, the dolphin was presented with a sample target, followed
by two alternative targets, one of which was the same as the sample. Targets
401
and Patterson 1981). Another reason for the experiment was to study the
dolphin's decision-making process.
Helweg et al. (1996) examined the capabilities of an echolocating dolphin
to develop aspect-independent representation of acoustically nonsymmetrical targets from echolocation. Echoes from the foam targets measured in
a test pool at aspects of 0° to 90° in 10° increments are shown in Figure 9.23,
where 0° corresponded to the planar view. The experimental setup of
Roitblat et al. (1990) was used and a similar matching-to-sample paradigm
was incorporated to test the dolphin. The orientation of the sample and the
comparison targets was not controlled in the experiment so that the targets
were free to revolve in a random fashion from trial to trial. The dolphin
correct response level for the different targets were 81 % for the rectangle,
79% for the pyramid, and 17% for the cube. However, when the sample
target was the cube, the animal had a strong bias toward responding rectangle. This occurred 63% of the time, suggesting that the dolphin could
recognize the cube well above chance but had a biased response. The results
of this study support the speculation that dolphins identify objects by averaging, or integrating, information gleaned from successive echoes within an
echolocation scan.
Harley et al. (1996) examined the dolphin's ability to integrate visual and
echolocation information. A variety of different types of targets were used
in a matching-to-sample paradigm to test the ability to recognize targets
using echolocation only, vision only, and echolocation with vision and crossmodally between echolocation and vision. They found that the dolphin performed best when it was allowed to use both vision and echolocation. They
also found that with familiar targets, the dolphin could perform cross-modal
matching, suggesting that the dolphin has an object-based representation
system. However, the dolphin performed considerably better when the task
was to echolocate the sample and visually choose the matching target
(echoic-vision) than vice versa. For familiar targets, the dolphin's performance on one set of targets was 78% correct for the echoic-vision and 22%
for the vision-echoic condition. For another set of familiar targets, correct
performance was 99% for echoic-vision and 61 % for vision-echoic, and for
a third set of targets the correct performance was 67% for echoic-vision
compared to 39% for vision-echoic. When novel targets were used, the
dolphin's performance was 39% correct for both echoic-vision and visionechoic. Chance performance was 33%.
Investigation of cross-modal identification of targets was also performed
by Pack and Herman (1995). They used targets made out of short sections
of gray PVC pipes having outer diameters of 1.27, 1.91,2.54, and 3.81 cm.
The short sections of pipes were filled with sand and glued together so that
a variety of different target shapes were achieved. The dolphin was trained
to do a matching-to-sample task visually, echoically and cross modally. In
the experiment, the dolphin was presented with a sample target, followed
by two alternative targets, one of which was the same as the sample. Targets