Signal detection theory involves a level of abstraction unfamiliar in field studies
of animal communication. Mastering its implications, however, leads to some
strong predictions about the evolution of signals and responses and to some new
procedures for investigating animal communication.
A consequence of this approach to communication is the fundamental conclusion that a
receiver cannot independently adjust its probabilities of correct detection and
false alarm (PCD and PFA). The only exception is the limiting case in which
the output of a channel in the presence of a signal is perfectly distinct from the
output in its absence, so PFA = 0. Otherwise, no matter how the criterion
for response changes, any change in PCD is accompanied by a corresponding change in
PFA.
This compromise leads ultimately to a prediction that receivers evolve to optimize
the net utility of their responses. The optimum might lie anywhere between
extremes of gullibility or fastidiousness. In turn, signalers should evolve
to balance the often incompatible advantages of increased detectability of signals,
increased complexity of encoding, and restriction of signals to intended receivers.
A second consequence of signal detection theory is the fundamental distinction
between the detectability of a signal and the receiver's criterion for a response.
Detectability depends on the contrast of the signal impinging on the subject
and on the selectivity of the
| |
subject's perceptual channels. A receiver's criterion for response depends
on its attitude toward the output of its perceptual channels, as a result of a
decision to accept particular probabilities of false alarm and correct detection.
Because any receiver's responses to stimulation depend on both the detectability of
the stimulus and the criterion for response, a definitive interpretation of
responses requires attention to both. For a full interpretation of a
receiver's performance, it is necessary to include null presentations in experiments
in order to measure false alarms as well as correct detections.
Signal detection theory thus suggests new ways to design and to interpret
experiments that compare responses to stimulation. Although some practical
difficulties face any application of signal detection theory to field studies, none
seems insurmountable.
With this approach, we stand to learn more about (1) the adaptations for
communication in situations with high background stimulation, such as in
choruses or complex social groups or at long range, (2) the effects of
contrast, redundancy, reduced uncertainty, and familiarity on receivers'
abilities to detect and discriminate signals, and (3) the evolution of
exaggeration or dishonesty in signals as a consequence of the evolution of
receivers' performance. In all of these ways, signal detection theory can
advance our understanding of both the physiology and the evolution of
communication.
|