Bats are the only mammals that have evolved true flight. Microchiropteran bats have developed an additional specialization, echolocation, which allows them to exploit an aerial nocturnal niche accessible to few other vertebrates. For echolocation, bats emit highly stereotyped vocalizations that vary across species. The type of calls used by a given species is highly correlated with the niche in which it forages (e.g., Neuweiler, 1990). For example, bats that hunt for insects in the open use short-duration frequency-modulated (FM) calls while those that hunt in dense foliage use relatively long-duration constant-frequency (CF) calls. By listening to how the echoes of their vocalizations are modified by reflective objects in the environment, bats can localize stationary or moving objects in three dimensions and determine the objects' size, shape, texture, and other characteristics (for review, see Simmons, 1989).
Although the bat's nervous system follows the general mammalian plan in both its structure and function, it has undergone a number of modifications associated with flight and echolocation. The most obvious neuroanatomical specializations are seen in the cochleas of certain species of bats and in the lower brainstem auditory pathways of all microchiroptera. The brainstem specializations are especially prominent in the superior olivary complex and nuclei of the lateral lemniscus, where there appear to be patterns of specialization characteristic of the different families of bats, the structure of their echolocation calls, and the specific ecological niches that they occupy. There is also some evidence that there may be specializations in the subcortical auditory-motor pathways of echolocating bats.
ECHOLOCATION CALL DESIGN AND CORRESPONDING SPECIALIZATIONS IN PERIPHERAL AUDITORY SYSTEM
All bats that have been studied have cochleas that are unusually large relative to body weight (for review, see Kossl and Vater, 1995). This is in keeping with the importance of hearing for bats and consistent with the large size of every part of the central auditory system relative to the size of the brain and other sensory systems. In terms of their echolocation signals, all echolocating bats use short FM sounds as part of their vocal repertoire. In addition, some species of bats that hunt for insects in dense foliage use relatively long constant frequency calls followed by a short FM component. The mustached bat, Pteronotus parnellii, and the rufous horseshoe bat, Rhinolophus rouxi, fall into this latter, CF-FM category. Anatomical and physiological studies have demonstrated that both of these species, as well as other bats that use CF-FM calls, have evolved cochlear specializations that result in a greatly expanded representation of the frequency range near the frequency of the dominant second harmonic of the CF component of the call (CF2) (e.g., Suga et al., 1975; Henson, 1978; Henson et al., 1985; Kossl and Vater, 1985; Vater et al., 1985; Kossl, 1994). This region is sometimes referred to as the acoustic fovea (e.g., Bruns and Schmieszek, 1980). The cochlear specializations consist of thickened regions of the basilar membrane and abrupt discontinuities in its width, both of which alter its mechanical properties (Fig. 1). It has been suggested that the thickening of the basilar membrane provides a reflection zone for incoming waves, allowing standing waves to be set up in a broad region that acts as a passive resonator for the frequency of the CF2 (Kossl and Vater, 1995). The result is very narrow, amplitude-tolerant frequency tuning throughout the auditory fovea region, with Q10dB values ranging from 20 to as high as 400 (for review, see Covey and Casseday, 1995). It has been suggested that auditory nerve fibers or central neurons with narrow frequency tuning provide a precisely phase-locked representation of echoes in which the CF2 frequency is Doppler-shifted in a periodic pattern by insect wingbeats (Neuweiler and Vater, 1977; reviewed in Pollak and Park, 1995), allowing the bat to detect flying insect prey. Although cochlear specializations are most pronounced in CF-FM bats, there is some evidence that even bats that do not use exclusively CF-FM calls may possess subtle cochlear specializations that result in an expanded representation of a behaviorally relevant frequency range (Vater and Siefer, 1995).
The magnified frequency representation that results from cochlear specializations is maintained throughout all levels of the central nervous system (Zook and Leake, 1989). It is interesting to note that some bats, such as Eptesicus fuscus, have no obvious cochlear specializations; nevertheless, their central nervous system contains an expanded representation of frequencies that correspond to the predominant frequency range present in the quasi-CF calls that this species emits when searching for prey (Casseday and Covey, 1992). These differences among species suggest that an acoustic fovea has evolved for the same purpose in different bat species through different mechanisms. In some species, it is created by peripheral mechanical specializations, whereas other species have developed central specializations in neural connectivity that result in a high magnification factor and narrow frequency tuning within a specific range that is important for detection of flying insect prey.
SPECIALIZATIONS IN CENTRAL AUDITORY SYSTEM
In all echolocating bats, the brainstem auditory pathways are greatly enlarged relative to the rest of the brain (e.g., Glendenning and Masterton, 1998). Aside from this difference in relative size, the same cell groups and connections that are present in all mammals seem to have undergone relatively subtle modifications in bats rather than bats having evolved any new structures or pathways. In other words, the same neural structures that subserve hearing in general have simply been exploited for a specialized behavior in addition to their original generalized role of processing ambient sounds and communication signals.
The cochlear nucleus of bats is not fundamentally different from that of other mammals, consisting of the same three major subdivisions and the same cell types. Except for narrow frequency tuning in the acoustic foveal range of certain species, neurons in the cochlear nuclei of bats have response properties similar to those described in other mammals (e.g., Haplea et al., 1994). The most obvious adaptation common to all bats is simply the large size of the auditory nerve and ventral cochlear nucleus, but there are some differences in cytoarchitecture as well. The anteroventral cochlear nucleus (AVCN) of at least some species of bats does not contain large spherical cells, but instead contains a specialized population of large multipolar cells along the medial edge (Zook and Casseday, 1982). In animals with good low-frequency hearing, such as the cat, the large spherical cells of AVCN project to the medial superior olive (MSO), where their input presumably is important for setting up MSO neurons' sensitivity to interaural time differences (ITDs) (for review, see Cant and Benson, 2003). Therefore, it would be interesting to know whether AVCN contains a population of large spherical cells in species such as the Mexican free-tailed bat, which have high-frequency hearing but whose MSO contains a significant proportion of binaural responses and exhibits sensitivity to ITDs outside the physiological range (for reviews, see Grothe and Neuweiler, 2000; Grothe and Park, 2000).
In the echolocating bats that have been examined, the dorsal cochlear nucleus (DCN) is relatively small. In at least some species, it appears to have poor lamination, a characteristic shared with anthropoid primates, including humans (Heiman-Patterson and Strominger, 1985), seals, sirenians, and cetaceans (Johnson et al., 1994). The octopus cell area of the posteroventral cochlear nucleus (PVCN) tends to be prominent in bats and provides a significant projection to the inferior colliculus (IC), a feature not seen in nonecholocating mammals (for review, see Covey and Casseday, 1995). Some of the specializations that have been described in the cochlear nucleus of bats appear to be highly species-specific. For example, CF-FM bats have an expanded tonotopic representation of the acoustic fovea (Vater et al., 1985; Zook and Leake, 1989). In horseshoe bats, the octopus cell area of PVCN is highly organized into a characteristic horseshoe formation (Fig. 2) (Pollak and Casseday, 1989), the function of which is unknown. Also, in the horseshoe bats, the DCN has a unique structure, consisting of a laminated subdivision in which frequencies below the CF frequency are processed, and a nonlaminated subdivision in which the CF2 biosonar harmonic and higher frequencies are processed. The granule cell/cartwheel cell system is absent from the nonlaminated portion of DCN (Kemmer and Vater, 2001).
Superior Olivary Complex
As in other mammals, the anteroventral cochlear nucleus is the main source of projections to the superior olivary complex (SOC) (e.g., Covey and Casseday, 1995). In many respects, the cell types that project to the SOC of bats are comparable to those in other mammals, and the lateral superior olive (LSO) and medial nucleus of the trapezoid body (MNTB) in bats are structurally and functionally similar to those of nonecholocating mammals. However, as noted above, the AVCN of bats appears to lack the large spherical cells that in other mammals project to the MSO. Another unusual feature is that the PVCN provides a tonotopically organized projection to the LSO, at least in horseshoe bats (Vater and Feng, 1990; reviewed in Neuweiler, 2003).
In all species of bats that have been examined, it appears that the structure, connections, and response properties of neurons in the MSO have undergone considerable modification for echolocation, with somewhat different patterns seen in different species of bats. The MSO of bats differs from that of nonecholocating mammals in several respects. First, given the small head size and high-frequency hearing range of bats, there is a virtual absence of any interaural time differences that could be used for sound localization. Because the MSO is thought to function primarily in the analysis of interaural time differences, it might be expected that bats, like other small mammals such as mice, would have a very small MSO. However, this is not the case (e.g., Covey and Casseday, 1995; Grothe, 2000). Instead, the MSO of bats is typically a large, thick structure that contains many neurons arranged in a nonlaminar fashion (Figs. 3 and 4). This is in contrast to mammals with good low-frequency hearing in which MSO neurons are arranged in a lamina one neuron thick in the medial-lateral dimension. The general structure of MSO in bats is reminiscent of the barn owl's MSO analog, nucleus laminaris, which is relatively thick and nonlaminar compared to that of diurnal birds such as chickens that do not depend on hearing to capture prey. It has been suggested that this adaptation is related to the owl's need to perform fine temporal discriminations to localize prey in the dark (Kubke et al., 2002).
The structural features of the SOC differ across different families of bats and appear to be more closely related to phylogeny than to echolocation behavior or call design. Figure 5 shows examples typical of the anatomical design of the SOC in Pteronotus, a neotropical mormoopid bat that uses a CF-FM call, Rhinolophus, an Old World horseshoe bat that also uses a CF-FM call, and Eptesicus, a widely distributed vespertilionid bat that uses mainly FM calls. The main variations are in the relative sizes of LSO, MSO, and the superior paraolivary nucleus (SPN), and in the shape of MSO. The horseshoe bats have the most unusual structure in that MSO is differentiated into a dorsal and ventral subdivision, and the cells that are normally segregated to form SPN are embedded into the dorsal edge of the dorsal division of MSO (Casseday et al., 1988).
Electrophysiological evidence suggests that the MSO of bats may indeed be specialized for temporal pattern processing, but that the nature of the temporal information has more to do with identifying sounds on the basis of their temporal patterns than it does with analyzing interaural time differences to determine their location (Covey et al., 1991; Grothe et al., 1992, 2001). Unlike most mammals, in which MSO neurons are excited by sound at either ear and are sensitive to interaural time differences, the MSOs of all of the bat species that have been studied contain a high proportion of monaural neurons. Moreover, many of those neurons that are binaural are unusual in that they are inhibited rather than excited by sound at the ipsilateral ear. The proportion of monaural neurons varies across species, ranging from about three-quarters of MSO neurons in Peteronotus parnellii to about half in Eptesicus fuscus, and less than 20% in the Mexican free-tailed bat, Tadarida brasiliensis (Covey et al., 1991; Grothe et al., 1997, 2001). The relative strengths of projections from the contralateral and ipsilateral ears to the MSO complex of different bat species are consistent with the proportion and types of binaural neurons in those species, reinforcing the notion that the functional properties of MSO have been adjusted during evolution to perform analyses related to echolocation (Casseday et al., 1988; Covey et al., 1991; Grothe et al., 1994). Figure 6 summarizes the relative proportions of monaural neurons and different classes of binaural neurons in the bat species that have been examined to date.
The SOC is also the origin of efferent projections to the cochlea and cochlear nucleus (for review, see Huffman and Henson, 1990). Figure 7 shows the different patterns of cochlear efferents in two species of bats. The structure and organization of the olivocochlear efferents varies considerably across mammals, and across different species of bats (for review, see Covey and Casseday, 1995). In all mammals, the efferent projections originate to some extent in the periolivary nuclei, small groups of cells that lie outside the principal nuclei, but which are highly variable across species. In nonecholocating mammals and most bats, a population of cells near LSO or within the LSO itself gives rise to a lateral olivocochlear pathway that innervates the inner hair cells. A second group of larger neurons located more medially in the periolivary region forms a medial olivocochlear pathway that innervates the outer hair cells. It has been suggested that the medial olivocochlear system plays a role in attenuating neural responses to the bat's own vocalizations (Goldberg and Henson, 1998). The horseshoe bat is unusual in that the only neurons that project to the cochlea are located in a small area between LSO and MSO (Aschoff and Ostwald, 1987; reviewed in Kossl and Vater, 1995). This species does not appear to have a medial olivocochlear system, which is consistent with the finding that outer hair cells in the cochlea do not receive efferent innervation. The functional significance of the lack of efferents to outer hair cells in horseshoe bats is not known.
Nuclei of Lateral Lemniscus
One of the most striking anatomical specializations in echolocating bats is seen in certain parts of the nuclei of the lateral lemniscus, a collection of cell groups that receive input from the cochlear nucleus and/or SOC and provide a major input to the inferior colliculus (for reviews, see Covey, 1993a; Covey and Casseday, 1995; Oertel and Wickesberg, 2002). The dorsal nucleus of the lateral lemniscus (DNLL) is similar to that of other mammals in its structure, connections, and basic functional properties (Covey, 1993b; Markowitz and Pollak, 1994), so will not be discussed here. In all of the bat species that have been examined, the more ventral cell groups, comprising the intermediate and ventral nuclei of the lateral lemniscus (INLL and VNLL), are greatly enlarged, so much so that they cause a prominent bulge in the side of the brainstem. In most mammals, the INLL and VNLL consist of a heterogeneous mixture of cell types loosely scattered within the ascending fibers of the lateral lemniscus; in the bat, however, the different cell types are largely segregated, resulting in a clear differentiation into three major subdivisions. The INLL is composed mainly of elongate neurons and is cytoarchitecturally similar to the INLL in nonecholocating mammals. The VNLL is clearly subdivided into two very different structures that are not seen as separate in nonecholocating mammals (Zook and Casseday, 1982; Covey and Casseday, 1986). The location of the different subdivisions of VNLL varies according to species as summarized in Figure 8. As is the case with the structure of MSO, the structure of the VNLL appears to be related more to phylogeny than to echolocation call design or foraging behavior.
The most striking subdivision of VNLL is a highly organized and homogeneous group of small round neurons that are morphologically similar to spherical bushy cells in the cochlear nucleus and the bushy cells that are intermingled with other cell types in the VNLL of nonecholocating mammals. Because of the tightly stacked columnar arrangement of cells in this area, it has been called the columnar nucleus of VNLL (VNLLc). Each row of cells within the VNLLc receives a highly focused projection from a relatively broad frequency region in the contralateral cochlear nucleus and projects in turn to a correspondingly broad frequency region in the inferior colliculus. Low- to high-frequency ranges are represented from dorsal to ventral within the VNLLc (Covey and Casseday, 1986).
A row in VNLLc could be thought of as a “frequency funnel” that integrates input from a broad frequency range, then redistributes its output to a broad frequency range in the inferior colliculus (Covey and Casseday, 1986). Figure 9 shows a row of neurons in VNLLc that have been labeled by retrograde transport of HRP from a comparatively large injection in the inferior colliculus, illustrating the precise nature of the projection from the rows of VNLLc. It has been suggested that the neurons that make up a row in the VNLLc integrate input from a broad frequency range in order to enhance timing precision (Covey and Casseday, 1986, 1991).
The second subdivision of VNLL is composed mainly of multipolar neurons and for that reason has been called the multipolar cell region of VNLL (VNLLm) (Covey and Casseday, 1986). The cell types that make up VNLLm resemble those that are intermingled with spherical bushy neurons in nonecholocating mammals. There is no obvious simple systematic tonotopic organization in the VNLLm. Instead, the frequency organization appears to take the form of concentric shells or some other complex configuration (Covey and Casseday, 1986, 1991).
Electrophysiological studies of the INLL, VNLLc, and VNLLm in two different species of bats show that nearly all neurons in these structures are monaural, excited by sound at the contralateral ear and unaffected by sound at the ipsilateral ear (Metzner and Radtke-Schuller, 1987; Covey and Casseday, 1991). This is consistent with the pattern of projections to these nuclei, all of which receive their major excitatory input from both divisions of the ventral cochlear nucleus and inhibitory input from the ipsilateral MNTB, which in turn receives its input from the contralateral cochlear nucleus (Zook and Casseday, 1985; Covey and Casseday, 1986; Huffman and Covey, 1995). Neurons in INLL, VNLLc, and VNLLm have different morphological features, which brain slice recording studies suggest are correlated with different intrinsic properties (Wu, 1999; Irfan et al., 2005). Neurons in different regions also have different patterns of synaptic terminals on the soma and dendrites (Vater et al., 1997) and receive different ratios of excitatory and inhibitory input from lower brainstem auditory structures (Huffman and Covey, 1995). All of these factors together act to create different response properties in each subdivision of INLL and VNLL.
In bats, there is evidence that, in addition to being anatomically specialized, the VNLLc is functionally specialized in ways that would aid in analyzing biosonar signals. Neurons in this region have no spontaneous activity, and they respond to the onset of sound with a single action potential, becoming strongly adapted thereafter so that they do not respond at any time during a continuous sound. The response latency is tightly correlated with the time of sound onset, varying by only a few tens of microseconds over multiple trials. Most remarkably, however, these neurons' latency is resistant to large variations in sound amplitude and frequency, remaining virtually constant throughout the neuron's entire frequency response area, which is typically broader than those of neurons in most other auditory structures (Covey and Casseday, 1991).
During echolocation, the distance to an object is determined by measuring the time between the emitted signal, the amplitude of which can be well over 100 dB SPL, and the returning echo, which is considerably attenuated by an amount that depends on the distance, size, and material of the object, as well as other factors (for review, see Simmons, 1989). Most auditory neurons' response latency varies significantly as a function of sound amplitude, so the timing between their response to the emitted signal and their response to an echo would provide an inaccurate estimate of the distance to the reflective object. The constant latency properties of VNLLc neurons, on the other hand, make them uniquely suited to convey accurate and unambiguous information about timing, and hence about distance.
Other clues that the VNLLc of bats is specialized for echolocation come from studies using tones that are modulated in amplitude or frequency in a periodic way. VNLLc neurons are unresponsive to sinusoidal amplitude modulations, a pattern that mimics the signal that would occur if an echo were reflected from an insect with beating wings (Huffman et al., 1998b). Thus, VNLLc neurons would not respond to every wingbeat cycle, but only to the onset of the modulated echo, again providing an unambiguous signal for calculating distance. In contrast, when a sinusoidally frequency-modulated signal is presented, VNLLc neurons do respond to every modulation, but only once per cycle, following the downward frequency sweep component. They also respond to single downward FM sweeps, but not to upward FM sweeps (Huffman et al., 1998a). Eptesicus fuscus, the species in which these studies were done, uses a downward FM sweep as an echolocation signal, so it appears that VNLLc neurons are not only accurate timing devices, but also specialized to respond to echolocation signals, screening out signals with upward frequency components such as other bats' communication calls or ambient noises.
There is also evidence that other parts of the nuclei of the lateral lemniscus may be specialized for processing echolocation sounds. The INLL of Pteronotus parnellii contains neurons that respond selectively to combinations of sounds corresponding to specific pairs of harmonics in the echolocation signal (Portfors and Wenstrup, 2001). These neurons are similar to the combination-sensitive neurons that have been described in the auditory cortex (O'Neill, 1995), medial geniculate (Wenstrup, 1995), and inferior colliculus (Mittmann and Wenstrup, 1995) of Pteronotus, except that the majority are suppressed rather than facilitated by a specific combination of tones in a specific time relation.
The inferior colliculus (IC) of bats is remarkably similar to that of other mammals in its basic structure except for its large size relative to the brain as a whole. In most bats that have been examined, the IC is enlarged so that it covers part of the superior colliculus and protrudes all the way to the dorsal surface of the brain caudal to the occipital cortex (Fig. 10). The pattern of afferent projections to the IC is not fundamentally different from that in nonecholocating mammals, although, as mentioned earlier, it does receive a significant input from the PVCN and a uniquely organized system of projections from the VNLLc.
In most species of bats, the tonotopic organization of the IC follows the usual mammalian pattern, with low frequencies represented dorsally and high frequencies ventrally. Figure 11 compares the tonotopic organization in the IC of three different species of bats. Eptesicus fuscus and Antrozous pallidus use mainly FM calls, have no cochlear specializations, and have a straightforward dorsal-to-ventral tonotopic progression. Eptesicus, however, has an expanded representation of frequencies between about 20–30 kHz, a range that corresponds to the quasi-CF calls used by this species when searching for prey (Casseday and Covey, 1992). The IC of the pallid bat, Antrozous, appears to have lateral and medial functional subdivisions. The medial subdivision includes the higher frequency range used in echolocation while the lateral subdivision contains low frequencies that are used in passive listening when the bat gleans moving prey from the ground and other surfaces (Fuzessery and Hall, 1999). The mustached bat, Pteronotus parnellii, is one of the species in which cochlear specializations produce an acoustic fovea around 60–63 kHz. The isofrequency lamina devoted to this narrow frequency range in the IC is so greatly expanded that it has “mushroomed” out of its normal position to produce a convoluted tonotopic sequence in which it forms a distinct subdivision on the dorsal part of the IC and completely covers the area representing the lower frequencies (e.g., Pollak and Casseday, 1989).
The IC is the target of numerous pathways that originate in the lower brainstem, the auditory cortex, and nonauditory structures (for reviews, see Covey and Casseday, 1995; Casseday and Covey, 1996; Casseday et al., 2002; Covey, 2005). In addition, the IC contains an extensive network of intrinsic connections, which appear to form a cascaded feedforward projection system that could produce a wide range of synaptic delays (Miller et al., 2005). Integration of excitatory and inhibitory inputs with different strengths, latencies, temporal patterns, and response functions results in IC neurons being selective for stimulus features that are important for echolocation (e.g., Covey et al., 1996). These include the delay between two sounds (e.g., Mittmann and Wenstrup, 1995; Yan and Suga, 1996; Portfors and Wenstrup, 1999), sound duration (e.g., Casseday et al., 1994, 2000; Ehrlich et al., 1997; Fremouw et al., 2005), the direction of an FM sweep (e.g., Fuzessery, 1994), and sinusoidal frequency modulation rate (e.g., Casseday et al., 1997; for reviews of integrative mechanisms in the IC, see Covey and Casseday, 1999; Covey, 2004; Covey and Faure, 2005).
Thus, although the IC of bats is not fundamentally different from that of other mammals in its structure and connections, slight adjustments in the relative strength and/or timing of inputs appear to have altered the responses of IC neurons to optimize their processing of the signals that are most relevant for echolocation in each species of bat.
Although the medial geniculate nucleus of bats tends to be relatively large, it is structurally similar to that of other mammals (e.g., Winer and Wenstrup, 1994; for review, see Wenstrup, 1995). Compared to other parts of the auditory system, there have been relatively few electrophysiological studies of the medial geniculate in bats, and those that exist have focused on the mustached bat, Pteronotus. Functionally, neurons in the thalamocortical system of bats possess the same sorts of specialized response properties as do IC neurons, including frequency combination sensitivity, delay sensitivity, and selectivity for FM sweep direction (e.g., Olsen and Suga, 1991a, 1991b; Wenstrup, 1999; O'Neill and Brimijoin, 2002). The suprageniculate nucleus (SG) is prominent in Pteronotus and appears to be part of an extralemniscal pathway from the periphery to the cortex. The extralemniscal input is from the nucleus of the central acoustic tract, a cell group in the lower brainstem that receives direct input from the cochlear nucleus. This input then appears to be integrated with input from the IC and the superior colliculus (Casseday et al., 1989). The SG is the source of a diffuse projection to the auditory cortex and to a region of the frontal cortex that, in turn, projects back to the SC (Kobler et al., 1987). The connections of the SG in Pteronotus are summarized in Figure 12. Neurons in the frontal cortex of two species of bats, Pteronotus parnelliis and Carollia perspicillata, are known to respond to sound (Kobler et al., 1987; Eiermann and Esser, 2000; Kanwal et al., 2000). It has been suggested that the SG sends a fast input to the auditory cortex that then results in rapid feedback to motor circuits accessed by the SC (Kobler et al., 1987; Casseday et al., 1989).
The auditory cortex of all bats is large compared to total cortical area, but unremarkable in its structure, resembling that of primitive insectivores (for review, see O'Neill, 1995). Thus, the auditory cortex, like the IC and MG, exhibits no clear anatomical specializations comparable to those seen in the SOC and VNLL of bats. There have been numerous reviews of the functional organization of auditory cortex in the mustached bat (e.g., Suga, 1990a, 1990b, 1997), but considerably less information is available on the organization of auditory cortex in other bat species. However, enough is known to conclude that there are considerable differences among species. Figure 13 compares the highly specialized functional organization of auditory cortex in Pteronotus with that found in a less specialized species, the big brown bat, Eptesicus fuscus.
The auditory cortex of Eptesicus contains a large primary area with a straightforward caudal-to-rostral tonotopic progression. Rostral to the primary tonotopic progression is an area that is highly variable across individuals, followed by another small tonotopically organized region that represents a reversal of the primary auditory cortical frequency map. The organization in Pteronotus appears to be much more complicated. Like auditory cortex of other mammals, it contains a primary tonotopic area with a caudal-to-rostral frequency progression. The auditory foveal frequency region occupies its logical place in the progression, but is greatly expanded to form what is referred to as the Doppler-shifted CF region (DSCF) area. The primary area is surrounded by a number of other regions in which neurons respond best to specific combinations of CF or FM harmonic frequencies and/or the time relation between two sounds (for review, see O'Neill, 1995). It remains to be determined whether Pteronotus has evolved unique and highly specialized areas to process CF-FM calls, or whether Pteronotus has simply been studied in greater detail than other species of bats, resulting in more nonprimary areas having been identified. For this reason, comparative studies of the organization of auditory cortex across bat species would be a potentially fruitful area for future work.
Although it was originally thought that each area in the mustached bat auditory cortex was dedicated to processing a particular feature of echolocation calls, it now appears that these same areas also process communication sounds (e.g., Ohlemiller et al., 1996; Esser et al., 1997; Portfors, 2004). In the pallid bat, Antrozous pallidus, the same neurons process echolocation sounds and other environmental sounds (Razak et al., 1999). Therefore, the basic mechanisms involved in information processing in bat auditory cortex may not differ significantly from those in nonecholocating mammals.
Not only does there appear to be considerable variability in auditory cortical organization among bat species, there is also evidence for sexual dimorphism and a high degree of variability among individuals of the same species (Suga et al., 1987; Dear et al., 1993). However, there does not appear to be any hemispheric lateralization, at least in Pteronotus (Sherwood et al., 2005).
In all mammalian species, the auditory cortex provides descending feedback to the thalamus and brainstem (for review, see Huffman and Henson, 1990). The role of corticofugal descending projections in shaping the responsiveness of neurons in the thalamus and midbrain of bats has been extensively studied and reviewed (e.g., Suga and Ma, 2003), but it appears that these mechanisms in bats are not fundamentally different from those in nonecholocating mammals (for review, see Weinberger, 2004).
AUDITORY-MOTOR PATHWAY SPECIALIZATIONS
Not surprisingly for an animal that directs much of its behavior on the basis of hearing, there are prominent pathways from the IC to motor-related areas, including the deep superior colliculus (SC) and the pontine-gray-cerebellar pathway. In at least some species of bats, these auditory-motor pathways appear to have largely taken over the function of the corresponding visual pathways.
The SC is a center for orientation reflexes and consists of several layers, each of which has a characteristic pattern of connections. The superficial layers receive direct input from the eye and in most mammals are large and well developed. In the mustached bat, however, the superficial layers are almost vestigial, occupying a layer no more than 50 μm in thickness. Moreover, this species does not appear to possess any structure that resembles the dorsal lateral geniculate nucleus (LGN) (Covey et al., 1987). The main target of the optic nerve in Pteronotus and other microchiroptera is the suprachiasmatic nuclei (data not shown) with a lesser projection to the ventral LGN. This pattern of projections is consistent with the idea that the bat does not rely on vision to direct its behavior, but that vision is important in regulating circadian rhythms.
The intermediate and deep layers of the SC are relatively enlarged in bats and receive input from several structures in the auditory system, including the IC, the dorsal nucleus of the lateral lemniscus, and the nucleus of the central acoustic tract, a cell group that provides an ascending auditory pathway that bypasses the IC to terminate in the SC and thalamus (Covey et al., 1987; Casseday et al., 1989). Some neurons in the deep layers of the SC of all mammals have sensory responses to sound (for reviews, see Sparks and Hartwich-Young, 1989; Stein, 1998). Auditory neurons in the deep layers are a prominent feature of the SC in bats (e.g., Reimer, 1991; Jen et al., 1993; Valentine and Moss, 1997). Electrical microstimulation in the intermediate and deep layers of the SC has been shown to elicit vocalizations and/or orientation responses in several species of bats (e.g., Schuller et al., 1990; Valentine et al., 2002), confirming the link between auditory sensory input and sound-directed patterns of motor activity.
The other major system for motor control that has subcortical connections with the auditory system in bats is the pontine gray, which projects to the cerebellum. It has been shown that neurons in the pontine gray of the horseshoe bat, Rhinolophus rouxi, have robust sensory responses to auditory stimuli and receive a massive projection from the IC (Schuller et al., 1991). A similar projection to the pontine gray has been seen in the mustached bat (Wenstrup et al., 1994) and in the big brown bat (data not shown). In nonecholocating mammals, projections from the IC to the pontine gray are a minor pathway, if they exist at all (Schuller et al., 1991), so this connection between the IC and cerebellum may be a specialization for motor control in a mammal that performs rapid navigational maneuvers in three dimensions using its sense of hearing as a guide.
It seems likely that some of the most striking specializations for echolocation may be found in the auditory-motor connections, rather than the sensory processing side, so this area would seem to be a particularly interesting focus for future investigation.
The structural features of the central nervous system of echolocating microchiropteran bats are basically the same as those of more generalized mammals. However, certain pathways, mainly those having to do with accurate processing of temporal information and auditory control of motor activity, are hypertrophied and/or organized somewhat differently from those same pathways in nonecholocating species. Through the resulting changes in strengths and timing of synaptic inputs to neurons in these pathways, bats have optimized the mechanisms for analysis of complex sound patterns to derive accurate information about objects in their environment and direct behavior toward those objects.
The author thanks Kimberly Miller for her help in preparing the photomicrographs.