Electric Fields and Electroreception
327
When the sphere is close, the electric image is small and the amplitude in the
center region is high (Fig. 3A). If the same sphere is located at a larger distance,
image width increases and image amplitude decreases (Fig. 3B). However, an
image very similar to the one evoked by the faraway small sphere can be produced
by a larger sphere of a higher resistance located at a closer distance (Fig. 3C).
How can a fish distinguish between these two possibilities?
We considered it very unlikely that electric fish would not possess a means of
determining object distance. Fish have to rely on active electrolocation during
their nocturnal lives. They have to find food, locate obstacles, determine their
distances to other fishes, and many similar tasks, all of which appear to be
impaired without some means of distance determination. Therefore, we conducted
behavioral training experiments to resolve this issue.
Individual G. petersii were trained to discriminate between two objects, each
placed at a different distance behind a gate in the training tank. In order to obtain a
food reward, the fish had to swim through that gate behind which the
corresponding object was placed further away, compared with the object behind
the other gate. First, we used two identical objects for training and fish learned in
only 2 weeks to choose the correct object, no matter how large the absolute
distance to the objects was. However, this performance did not demonstrate that
fish could measure distance. With two identical objects, distance discrimination
could have been based on pure amplitude or image width measurements.
In a second step, we presented two objects of different sizes or shapes or made
of different materials. Now fish could no longer solve the distance discrimination
task by measuring just image amplitude. Surprisingly, however, the fish needed no
extra training to correctly measure the distances of these new objects. They
immediately continued to choose the object located further away, as long as a
certain threshold value of interobject distances was exceeded. This demonstrated
that G. petersii can indeed measure object distance and that distance
discrimination is independent of object size, shape or impedance (von der Emde
et a!. 1998).
To fmd the mechanism which the fish use for distance perception, we
measured the electric images that our training objects projected onto the fish's
skin (Fig. 6). We searched for image parameters that correlated only with object
distance and were independent of other object properties. Ultimately, we found
such a parameter combination: the ratio of the amplitude decline of the image
amplitude at the image's edge (the slope of the image) and the maximal image
amplitude in the center. The slope is a measure of the fuzziness of the image, or of
how much the image is in focus. The slope/amplitude ratio decreases if an object
moves away, and it does not depend (with one exception) on the type of object
under investigation.
The electric image measurements also revealed that metal spheres yielded
slope/amplitude ratios that were smaller than those of other objects placed at the
same distance. Because smaller slope/amplitude ratios correspond to larger object
distances, we hypothesized that if the fish indeed used this ratio to determine
327
When the sphere is close, the electric image is small and the amplitude in the
center region is high (Fig. 3A). If the same sphere is located at a larger distance,
image width increases and image amplitude decreases (Fig. 3B). However, an
image very similar to the one evoked by the faraway small sphere can be produced
by a larger sphere of a higher resistance located at a closer distance (Fig. 3C).
How can a fish distinguish between these two possibilities?
We considered it very unlikely that electric fish would not possess a means of
determining object distance. Fish have to rely on active electrolocation during
their nocturnal lives. They have to find food, locate obstacles, determine their
distances to other fishes, and many similar tasks, all of which appear to be
impaired without some means of distance determination. Therefore, we conducted
behavioral training experiments to resolve this issue.
Individual G. petersii were trained to discriminate between two objects, each
placed at a different distance behind a gate in the training tank. In order to obtain a
food reward, the fish had to swim through that gate behind which the
corresponding object was placed further away, compared with the object behind
the other gate. First, we used two identical objects for training and fish learned in
only 2 weeks to choose the correct object, no matter how large the absolute
distance to the objects was. However, this performance did not demonstrate that
fish could measure distance. With two identical objects, distance discrimination
could have been based on pure amplitude or image width measurements.
In a second step, we presented two objects of different sizes or shapes or made
of different materials. Now fish could no longer solve the distance discrimination
task by measuring just image amplitude. Surprisingly, however, the fish needed no
extra training to correctly measure the distances of these new objects. They
immediately continued to choose the object located further away, as long as a
certain threshold value of interobject distances was exceeded. This demonstrated
that G. petersii can indeed measure object distance and that distance
discrimination is independent of object size, shape or impedance (von der Emde
et a!. 1998).
To fmd the mechanism which the fish use for distance perception, we
measured the electric images that our training objects projected onto the fish's
skin (Fig. 6). We searched for image parameters that correlated only with object
distance and were independent of other object properties. Ultimately, we found
such a parameter combination: the ratio of the amplitude decline of the image
amplitude at the image's edge (the slope of the image) and the maximal image
amplitude in the center. The slope is a measure of the fuzziness of the image, or of
how much the image is in focus. The slope/amplitude ratio decreases if an object
moves away, and it does not depend (with one exception) on the type of object
under investigation.
The electric image measurements also revealed that metal spheres yielded
slope/amplitude ratios that were smaller than those of other objects placed at the
same distance. Because smaller slope/amplitude ratios correspond to larger object
distances, we hypothesized that if the fish indeed used this ratio to determine