2.3.3.1. Categorization and Classification of Conspecific Vocalizations
One way that experience-dependent plasticity can influence the structure
and form of perceptual representations is through the emergence of categorical boundaries between various sets of stimuli. At the behavioral level,
categorical perception is observed as a nonlinear relationship between
stimulus variation along some dimension and a corresponding behavior.
Changes along a stimulus dimension that span the boundaries between categories are easy to detect, whereas equal-magnitude changes within a category are more difficult to detect. Such behaviors suggest the obvious
hypothesis that similarly categorical (i.e., nonlinear) neural responses
should be observed when particular neurons are exposed to fixed levels
of variation both within and across category boundaries. In general, the
complexity of most natural stimuli suggests that categorical neural
responses are likely to arise at higher levels within forebrain hierarchies.
An example of this is the recent data showing differential responses of
neurons in the prefrontal cortex of monkeys to different categories of visual
objects (Freedman et al. 2001). However, in cases where categorical, perceptual behavior corresponds closely to relevant dimensions of peripheral
tunings (e.g., frequency), categorical responses may first arise in peripheral
structures.
Not all categories of stimuli will necessarily meet the behavioral definitions required for strict demonstrations of categorical perception. More
general category formation can be described at varying levels of abstraction (Herrnstein 1990). For instance, some categories of stimuli may simply
reflect rote collections of arbitrary objects, whereas exemplars in other categories may share common physical features or be the predicate of an
abstract rule. For instance, classes of predators, food items, or potential
mates may form natural categories. These general forms of categorization
appear to be nearly ubiquitous among vertebrate species (Herrnstein 1990),
and some forms may occur among invertebrates as well (e.g., Wyttenbach
et al. 1996). Moreover, the fact that learning influences most categorical
behavior suggests an amazingly high degree of plasticity in the neural
processes underlying these capabilities.
Behavioral tests are required to determine the location of category
boundaries within natural stimulus sets. Here, the use of vocalization repertoires holds a distinct advantage because much of the behavioral work on
communication systems has been directed at the determination of behaviorally relevant classes of stimuli. The use of repertoires of natural stimuli
has been developed most prominently in the analysis of bird and primate
auditory systems, where neuronal response variation across different categories of vocalizations has been tested. For example, in European starlings,
a species of songbird, neurons throughout the field L (analogous to the
primary auditory cortex in mammals), the caudal medial neostriatum
(NCM), and caudal ventral hyperstriatum (cHV) respond to conspecific
7. Neuroethology of Vocal Communication
353
One way that experience-dependent plasticity can influence the structure
and form of perceptual representations is through the emergence of categorical boundaries between various sets of stimuli. At the behavioral level,
categorical perception is observed as a nonlinear relationship between
stimulus variation along some dimension and a corresponding behavior.
Changes along a stimulus dimension that span the boundaries between categories are easy to detect, whereas equal-magnitude changes within a category are more difficult to detect. Such behaviors suggest the obvious
hypothesis that similarly categorical (i.e., nonlinear) neural responses
should be observed when particular neurons are exposed to fixed levels
of variation both within and across category boundaries. In general, the
complexity of most natural stimuli suggests that categorical neural
responses are likely to arise at higher levels within forebrain hierarchies.
An example of this is the recent data showing differential responses of
neurons in the prefrontal cortex of monkeys to different categories of visual
objects (Freedman et al. 2001). However, in cases where categorical, perceptual behavior corresponds closely to relevant dimensions of peripheral
tunings (e.g., frequency), categorical responses may first arise in peripheral
structures.
Not all categories of stimuli will necessarily meet the behavioral definitions required for strict demonstrations of categorical perception. More
general category formation can be described at varying levels of abstraction (Herrnstein 1990). For instance, some categories of stimuli may simply
reflect rote collections of arbitrary objects, whereas exemplars in other categories may share common physical features or be the predicate of an
abstract rule. For instance, classes of predators, food items, or potential
mates may form natural categories. These general forms of categorization
appear to be nearly ubiquitous among vertebrate species (Herrnstein 1990),
and some forms may occur among invertebrates as well (e.g., Wyttenbach
et al. 1996). Moreover, the fact that learning influences most categorical
behavior suggests an amazingly high degree of plasticity in the neural
processes underlying these capabilities.
Behavioral tests are required to determine the location of category
boundaries within natural stimulus sets. Here, the use of vocalization repertoires holds a distinct advantage because much of the behavioral work on
communication systems has been directed at the determination of behaviorally relevant classes of stimuli. The use of repertoires of natural stimuli
has been developed most prominently in the analysis of bird and primate
auditory systems, where neuronal response variation across different categories of vocalizations has been tested. For example, in European starlings,
a species of songbird, neurons throughout the field L (analogous to the
primary auditory cortex in mammals), the caudal medial neostriatum
(NCM), and caudal ventral hyperstriatum (cHV) respond to conspecific
7. Neuroethology of Vocal Communication
353