behavior in taxa that are extremely social, making interpretation of call
development patterns difficult. Vocalizations may appear abnormal not
because acoustic input and motor flexibility are necessary for normal development but because of pathological behavior or inappropriate social
context.
The vocal flexibility required for effective echolocation in odontocetes
and chiropterans may have allowed vocal learning of social signals as well,
an idea also developed by Tyack (personal communication). Bats, whose
echolocation behavior is well-studied, modify many aspects of their vocal
output to facilitate navigation and prey capture (Griffin 1958, 1986;
Schnitzler and Henson 1980). They alter the structure of sonar signals to
provide appropriate information at each stage of searching for and
approaching an insect prey, changing duration, extent of frequency modulation, bandwidth, and repetition rate of pulses (e.g., Kalko and Schnitzler
1993; Kalko 1995). Some species also exhibit Doppler-shift compensation,
adjusting the frequency of sonar emissions to stabilize echo frequency
during flight (Schnitzler 1967). This flexibility allows bats to finely tune their
vocal behavior to its specific function, and this modification occurs very
rapidly, often in less than a second (for reviews, see Schnitzler and Henson
1980; Griffin 1986). Such vocal flexibility in echolocation pulses is probably
under fairly strong selection because of its role in foraging. Once vocal flexibility for echolocation evolved, social calls may have been readily learned,
requiring little modification to vocal production or neural processing mechanisms. Indeed, it appears that processing of social calls is done by the same
neurons involved in echolocation processing (Ohlemiller et al. 1996; Esser
et al. 1997). Thus, adapting the similarity of individual or group signatures
to social partners becomes an easy task.
Individual signatures function best when they are maximally different
from others, whereas group signatures function best when group mates
sound similar yet distinct from other groups. The common effect of vocal
learning is to modify the extent of similarity between individuals; however,
changes might be in opposite directions, depending on call function.
5.2. Are Models of Song Learning in Birds Relevant
to Call Learning?
5.2.1. Sensorimotor Model
The most widely considered model for the process of song learning in oscine
birds is the auditory template model (Marler 1976) that builds on the sensorimotor model developed by Konishi (1965). This idea suggests that juvenile songbirds are born with a crude neural template of their species’ song.
This crude template guides song memorization and attention to certain
song types during the early phases of song learning. During the memorization phase, young birds memorize songs and song elements that they hear,
4. Comparative Vocal Learning
201
development patterns difficult. Vocalizations may appear abnormal not
because acoustic input and motor flexibility are necessary for normal development but because of pathological behavior or inappropriate social
context.
The vocal flexibility required for effective echolocation in odontocetes
and chiropterans may have allowed vocal learning of social signals as well,
an idea also developed by Tyack (personal communication). Bats, whose
echolocation behavior is well-studied, modify many aspects of their vocal
output to facilitate navigation and prey capture (Griffin 1958, 1986;
Schnitzler and Henson 1980). They alter the structure of sonar signals to
provide appropriate information at each stage of searching for and
approaching an insect prey, changing duration, extent of frequency modulation, bandwidth, and repetition rate of pulses (e.g., Kalko and Schnitzler
1993; Kalko 1995). Some species also exhibit Doppler-shift compensation,
adjusting the frequency of sonar emissions to stabilize echo frequency
during flight (Schnitzler 1967). This flexibility allows bats to finely tune their
vocal behavior to its specific function, and this modification occurs very
rapidly, often in less than a second (for reviews, see Schnitzler and Henson
1980; Griffin 1986). Such vocal flexibility in echolocation pulses is probably
under fairly strong selection because of its role in foraging. Once vocal flexibility for echolocation evolved, social calls may have been readily learned,
requiring little modification to vocal production or neural processing mechanisms. Indeed, it appears that processing of social calls is done by the same
neurons involved in echolocation processing (Ohlemiller et al. 1996; Esser
et al. 1997). Thus, adapting the similarity of individual or group signatures
to social partners becomes an easy task.
Individual signatures function best when they are maximally different
from others, whereas group signatures function best when group mates
sound similar yet distinct from other groups. The common effect of vocal
learning is to modify the extent of similarity between individuals; however,
changes might be in opposite directions, depending on call function.
5.2. Are Models of Song Learning in Birds Relevant
to Call Learning?
5.2.1. Sensorimotor Model
The most widely considered model for the process of song learning in oscine
birds is the auditory template model (Marler 1976) that builds on the sensorimotor model developed by Konishi (1965). This idea suggests that juvenile songbirds are born with a crude neural template of their species’ song.
This crude template guides song memorization and attention to certain
song types during the early phases of song learning. During the memorization phase, young birds memorize songs and song elements that they hear,
4. Comparative Vocal Learning
201