REVIEWS MULTIPARAMETRIC CORTICOFUGAL MODULATION AND PLASTICITY IN THE AUDITORY SYSTEM. Nobuo Suga and Xiaofeng Ma

Transcription

1 MULTIPARAMETRIC CORTICOFUGAL MODULATION AND PLASTICITY IN THE AUDITORY SYSTEM Nobuo Suga and Xiaofeng Ma The auditory systems of adult animals can be reorganized by auditory experience. The auditory cortex, the corticofugal system and the cholinergic basal forebrain are crucial for this reorganization. The auditory system can undergo two different forms of reorganization expansion and compression. Whereas expanded reorganization has been found in different species and different sensory systems, compressed reorganization has only been found in the auditory system of the moustached bat, which is highly specialized for echolocation. Here, we review recent progress in our understanding of the corticofugal system and the reorganization of the auditory system. Department of Biology, Washington University, One Brookings Drive, St Louis, Missouri 6313, USA. doi:1.138/nrn1222 The auditory system consists of ascending and descending (corticofugal) pathways. Auditory signals received by the cochlea are sent to the brain through the auditory nerve. Before reaching the auditory cortex (AC), auditory stimuli travel through the cochlear nucleus and superior olivary complex in the brainstem, the lateral leminiscal nuclei and inferior colliculus (IC) in the midbrain, and the medial geniculate body (MGB) in the thalamus (BOX 1). The ascending auditory system has an extra level of complexity in that divergent and convergent projections that are relevant to parallel and hierarchical processing occur at several levels. As a result of these complex projections, there are many physiologically distinct types of neuron in the subcortical auditory nuclei and in the AC. In contrast to the physiology of the ascending auditory pathways, the contribution of the corticofugal system to auditory processing has received little attention 1 3.Here, we review the recent data on the function of the corticofugal system in auditory signal processing and on the plasticity of the auditory system. Corticofugal projections Neurons in the deep layers of the AC project back to the MGB, IC and subcollicular auditory nuclei (BOX 1). Corticothalamic fibres project only to the ipsilateral MGB and thalamic reticular nucleus. However, corticocollicular fibres project bilaterally to the IC. The ipsilateral projection is much more extensive and more topographically organized than the contralateral projection 4. Therefore, the ipsilateral modulation by corticofugal fibres is greater than the contralateral modulation in the IC and MGB 5.The corticofugal projections are bilateral to the subcollicular nuclei 6.Corticofugal modulation takes place even in the cochlea, through olivocochlear neurons in the superior olivary complex 7.The central nucleus of the IC projects not only to the MGB and the superior colliculus, but also to medial olivocochlear neurons, which mostly project to contralateral cochlear outer hair cells. In general, olivocochlear neurons project bilaterally to the cochlea, although there are some differences in olivocochlear projections between species. The corticothalamic projection forms the shortest auditory feedback loop, whereas the projection to cochlear hair cells through olivocochlear fibres forms the longest auditory feedback loop 4,8 1. How does the corticofugal auditory system modulate signal processing, and how do the response properties of auditory neurons and the functional map of the auditory system change with experience? Since the late 195s, many studies on corticofugal modulation of thalamic and collicular neurons have been undertaken in anaesthetized animals. In these studies, strong activation or inactivation of the primary AC evoked NATURE REVIEWS NEUROSCIENCE VOLUME 4 OCTOBER

2 Box 1 The brain and the auditory cortex of the moustached bat Dorsal Ventral Within fossa FM FM CF/CF AIa 91 VF khz FM 1 FM n H1 H 2 VA Anterior CER DIF FM 1 FM 4 FM 1 FM 2 CF 1 /CF 2 CF 1 /CF 3 * AC FM 1 FM 3 ( ) VM SC MGB SOC DF FM 1 FM 2 Posterior The red and blue arrows in the upper part of the figure indicate the ascending and descending (corticofugal) systems, respectively. The lower part of the figure represents a physiological map of the auditory cortex (AC). The numbers and lines in the anterior (AIa) and posterior (AIp) divisions of the AC, and in the Doppler-shifted constant frequency (DSCF) area, indicate iso-best-frequency lines. The area of the AC that is sensitive to combinations of constant-frequency signals (CF/CF) consists of two subdivisions that contain a Doppler-shift (velocity) axis. The dorsal fringe (DF) and the FM FM areas that are sensitive to combinations of frequency-modulated signals (FM FM) each consist of three subdivisions. These areas contain an echo-delay (range) axis. CBL, cerebellum; CER, cerebrum; CN, cochlear nucleus; DIF, dorsal intrafossa; DM, dorsomedial area; DP, dorsoposterior area; H 1 H 2, first and second harmonic combination-sensitive area; IC, inferior colliculus; MGB, medial geniculate body; NLL, nucleus of the lateral lemniscus; SC, superior colliculus; SOC, superior olivary complex; VA, ventroanterior area; VF, ventral fringe area; VM, ventromedial area; VP, ventroposterior area. Reproduced, with permission, from REF. 48 (1994) MIT Press. IC NLL FM 1 FM DSCF 62.3 CN FM 1 FM 3 DM CBL 1. mm khz L Alp (2) (15) (1) 1. mm excitation and/or inhibition of these subcortical neurons. These physiological data were contradictory: some authors found only, or predominantly, inhibitory activity ; others found only, or predominantly, excitatory or facilitatory activity ; and a third set of studies showed roughly equal levels of excitatory and inhibitory activity 18 2.As we discuss here, the apparent contradiction VP DP A between studies might be resolved if the frequency dependence of excitation and inhibition, and the relationship in tuning between stimulated and recorded neurons, are considered. To try to resolve these contradictions and make progress in our understanding of the corticofugal modulation of frequency-tuning curves, we have used awake (rather than anaesthetized) animals. In our experimental design, we first obtain tuning curves of both the cortical neurons to be stimulated and the subcortical neurons that respond to this stimulation. With this initial characterization in hand, we then use electrical or pharmacological means to stimulate or inactivate the characterized cortical neurons, and evaluate the effects of this stimulation or drug on the subcortical neurons. We focus on the relationship in tuning between them, and on the frequencydependence of facilitation and inhibition. Using this experimental system, we have found that corticofugal modulation occurs in a specific and systematic way for the adjustment of auditory signal processing in the frequency and time domains. Corticofugal modulation in the frequency domain Research on auditory neurophysiology has mostly focused on auditory processing in the frequency domain; that is, frequency tuning and frequency mapping in the central nervous system. Neuroethological research has focused on the responses of neurons to acoustic stimuli that mimic species-specific sounds. Such research indicates that the auditory system has different types of subcortical neuron that are tuned to the values of the parameters that characterize species-specific sounds 1 3,21. These subcortical neurons are subject to corticofugal modulation in the frequency, amplitude and time domains. We will begin by reviewing corticofugal modulation in the frequency domain, before considering it in the time and amplitude domains. Frequency-dependent facilitation and inhibition. The response of a neuron is usually maximal in magnitude and lowest in threshold at a certain frequency the best frequency (BF). Electrical stimulation of cortical auditory neurons evokes both facilitation and inhibition of the auditory responses of subcortical neurons. The amount of facilitation and inhibition varies depending on the frequency of a tone to which the subcortical neurons respond, so that the frequency-tuning curves of these neurons shift along the frequency axis (FIG. 1a c).the amount of facilitation and inhibition also varies as a function of the relationship in frequency tuning between the stimulated and recorded neurons (FIG. 2). When a recorded neuron is matched in BF to the stimulated cortical neurons, the response of the recorded neuron is augmented at its BF and is inhibited at frequencies that are lower and higher than the BF. As a result, the response of the recorded neuron is often sharpened in frequency tuning. When unmatched, the response of the recorded neuron is inhibited at its BF and is facilitated at other frequencies. As a result, the frequency tuning of the unmatched neuron undergoes what is known as a BF 784 OCTOBER 23 VOLUME 4

3 a Number of impulses / 5 stimuli BF C 31.5 khz 4 1 Control d BF shift (khz) 2 BF S 29. khz 3 Time (ms) min after 18 min after Big brown bat 3 d c Conditioning Number of impulses / 5 stimuli Number of impulses / 5 stimuli a b Centripetal shift BF S Control Recovery Frequency of tone burst (khz) c Centrifugal shift 16 8 Big brown bat Moustached bat BF S Conditioning BF C BF C Frequency of tone burst (khz) b Time (hours) Figure 1 Corticofugal modulation of the auditory responses, frequency-response curves of collicular neurons, and time courses of collicular and cortical best-frequency shifts. a The peri-stimulus-time (PST) histograms show inhibition (left column) or facilitation (right column) of the responses of a single collicular neuron at the best frequencies in the control (BF C ) and shifted (BF S ) conditions. The PST histograms showing the responses were obtained before (1), immediately after (2), 9 min after (3) and 18 min after (4) electrical stimulation of cortical neurons ( ) tuned to 26 khz. Note the changes in response in a 2 compared with a 1. The horizontal bars at the bottom represent sounds of 2 ms duration. Data reproduced, with permission, from REF. 5 (21) The American Physiological Society. b,c Centripetal (b) and centrifugal (c) BF shifts evoked by electrical stimulation of cortical neurons tuned to 26 khz (arrow in b) or 61.5 khz (arrow in c). The green and blue circles and the broken lines indicate the frequency-response curves at 1 db above minimum threshold in the control, shifted and recovery conditions, respectively. Note the shift in BF from BF C to BF S, and the difference in the direction of the BF shift between b and c. b reproduced, with permission, from REF. 5 (21) The American Physiological Society; c reproduced, with permission, from REF. 24 (2) The American Physiological Society. d Time course of the BF shifts of collicular (curves a and c) and cortical (b and d) neurons of the big brown bat evoked by cortical electrical stimulation (a and b) or auditory fear conditioning (c and d). A second conditioning session 3.5 h after the first also evoked collicular (e) and cortical (f) BF shifts. Note that the time course of the collicular BF shift evoked by electrical stimulation is identical to that caused by conditioning, whereas the time course of the cortical BF shift evoked by electrical stimulation is different. The horizontal bars indicate electrical stimulation or conditioning of 3 min duration. Data reproduced, with permission, from REFS 5, 25 and 31. f e shift (FIG. 1b,c). The corticofugal effects are clearly different between matched and unmatched neurons. These frequency-dependent effects improve the input to the stimulated cortical neurons and the subcortical representations of the stimulus parameters to which the cortical neurons are tuned 5,22 27.This function of the corticofugal system has been termed egocentric selection 28.In the visual system, orientation-sensitive cortical neurons reorganize their own input through the corticofugal system to improve their orientation sensitivity 29.So, egocentric selection occurs in both the auditory and visual systems. Focal electrical stimulation of the AC for 3 min leads to the simultaneous development of collicular and cortical BF shifts, which reach a peak at the end of the stimulation. These BF shifts disappear h after the stimulation. The cortical BF shifts tend to last slightly longer than the collicular shifts 5,25,3 (FIG. 1d).It is interesting to compare these dynamic properties to what is observed after a 3-min-long training session on an auditory FEAR-CONDITIONING task. In this case, the collicular BF shift develops and recovers similarly to that evoked by electrical stimulation, whereas the cortical BF shift develops slowly and shows no recovery. That is, the collicular BF shift is a short-term change, but the cortical BF shift is long term. The long-term cortical BF shift does not represent a saturation of BF shift, because a second conditioning session 3.5 h after the first elicits another short-term collicular BF shift and a further increase in the long-term cortical BF shift 31 (FIG. 1d).The short-term collicular BF shift contributes to evoking the long-term cortical BF shift 32.Here,we operationally define short-term plasticity as changes that recover within 3.5 h of cortical electrical stimulation or conditioning, whereas long-term plasticity refers to changes that do not recover by more than 1% after 3 h. Expanded and compressed reorganizations of frequency maps. BFs vary systematically across the AC and the subcortical auditory nuclei, so that these structures have a frequency axis. BF shifts that are evoked by focal electrical stimulation of the AC affect the frequency axis. There are two types of BF shift of unmatched neurons centripetal and centrifugal. Centripetal BF shifts are shifts towards the BF of the stimulated cortical neurons (FIG. 2b),whereas centrifugal BF shifts are shifts away from the cortical BF (FIG. 2c).These two types of shift occur in a specific spatial pattern along the frequency axes of the IC and AC, and the shift is essentially the same in both structures 5,23,25,33 35.In the AC of the big brown bat 5,25 (FIGS 3a and 4c) and Mongolian gerbil 26 (FIGS 3d and 4a), and in the IC of these two species and the house mouse 36 (FIG. 3e), centripetal BF shifts occur in a large area around the matched neurons, and small centrifugal BF shifts occur in the narrow zone that surrounds the centripetal area. The main reorganization in the AC of these species is due to centripetal BF shifts. But in a specialized area of the AC of the moustached bat the DOPPLER-SHIFTED constant frequency (DSCF) area centrifugal BF shifts occur in a large area that surrounds the matched neurons (FIGS 3c and 4b). In a non-specialized area of the AC of the moustached bat, centripetal BF shifts occur in a large area around the matched neurons (FIGS 3b and 4b). A BF-shift difference curve can differ between species and between cortical areas of the same species 26,37 (FIG. 3). The difference in BF shift is related to the difference in NATURE REVIEWS NEUROSCIENCE VOLUME 4 OCTOBER

4 FEAR CONDITIONING A form of Pavlovian (classical) conditioning in which an animal learns that an innocuous stimulus (for example, an auditory tone the conditioned stimulus, CS) comes to reliably predict the occurrence of a noxious stimulus (for example, foot shock the unconditioned stimulus, US) following their repeated paired presentation. As a result of this procedure, presentation of the CS alone elicits conditioned fear responses previously associated with the US only. DOPPLER SHIFT The Austrian physicist C. J. Doppler discovered that a light or sound wave measured by a moving observer will be shifted as a direct function of the speed of the observer and as an inverse function of the speed of the wave. A prototypical example of the Doppler shift is the sound of a train s horn as we stand at a station; the sound is shifted to a higher pitch as the train approaches, and then abruptly to a lower pitch as it passes by. a AC MGB or IC b Centripetal BF shifts evoked by focal activation of AC Response Matched neurons Frequency Matched neurons Frequency (big brown bat) c Centrifugal BF shifts evoked by focal activation of AC Response Unmatched neurons Unmatched neurons Electrical stimulation Unmatched neurons Unmatched neurons (moustached bat) Figure 2 Facilitation of matched neurons, and centripetal or centrifugal best-frequency shifts of unmatched neurons, evoked by electrical stimulation of cortical neurons. a In the auditory cortex (AC) and the subcortical auditory nuclei (IC, inferior colliculus; MGB, medial geniculate body), there are arrays of neurons that are tuned to different frequencies. Electrical stimulation of cortical neurons (filled red circle) evokes changes that differ between matched (filled blue circle) and unmatched (open circles) neurons. Electrical stimulation of cortical neurons evokes facilitation, inhibition and best-frequency (BF) shifts within the AC and subcortical auditory nuclei. There are two types of BF shift centripetal (b) and centrifugal (c). The unbroken and broken triangular curves represent the frequency-response curves in the control and shifted conditions, respectively. Data in b is from REFS 5 and 25; data in c is from REF. 34. frequency axis. BF shifts are smaller in an expanded portion of the frequency axis than in other portions. An expanded portion contains neurons that are more sharply tuned than others, such that BF shifts are smaller for sharply tuned neurons in the expanded portion. The best example of this is the DSCF area of the AC of the moustached bat (BOX 1 and FIG. 3c). BF shifts are largest for neurons along the frequency axis that crosses stimulated cortical neurons 27. In the big brown bat, the BF shifts elicited by electrical stimulation are essentially the same as those produced by auditory fear conditioning 33. Centripetal BF shifts result in an increase in the number of neurons that respond to a frequency that is equal to the best frequency of the stimulated cortical neuron 5,23,31,33.That is, they result in an expanded reorganization. By contrast, centrifugal BF shifts result in a reduced representation that is associated with the augmentation of responses and the sharpening of tuning curves of matched neurons 34.Such reorganization is referred to as compressed 38.Compressed reorganization increases contrast in the neural representation of an auditory signal; it is therefore presumably more suited than expanded reorganization to improving the discrimination of acoustic signals. Electrical stimulation of cortical neurons evokes centripetal BF shifts in the IC and AC of the big brown bat 5,23,25, moustached bat 26,Mongolian gerbil 26,27 and house mouse 36.Acoustic tone bursts paired with electrical stimulation of the basal forebrain evoke centripetal BF shifts in the rat AC 39.Auditory fear conditioning causes centripetal BF shifts in the AC of the guinea-pig 4, and in the IC and AC of the big brown bat Shifts in receptive fields of somatosensory cortex neurons are centripetal in different species of mammals 41,42, and shifts in orientation selectivity of neurons in the cat visual cortex are also centripetal 43.So, expanded reorganization is common to many mammalian sensory systems. In the visual system, corticofugal modulation of thalamic visual neurons has been extensively studied 44,45. In the somatosensory system, the reorganization of the thalamic somatosensory nucleus partly depends on the corticofugal system 46,47.However, it is not yet known whether focal activation of the visual or somatosensory cortices evokes a shift of the receptive fields of thalamic visual or somatosensory neurons, respectively. The role of facilitation and inhibition in reorganization. Do the two types of reorganization depend on different mechanisms? A BF shift depends on inhibition of responses at certain frequencies and facilitation of responses at other frequencies (FIG. 1a c).iffacilitation of neighbouring unmatched neurons is strong and widespread and inhibition is weak, then the AC and the corticofugal system might evoke centripetal BF shifts in subcortical and cortical neurons. By contrast, if facilitation is highly focused to matched neurons and inhibition is strong in neighbouring unmatched neurons, then the AC might evoke centrifugal BF shifts 37. This hypothesis was tested in the moustached bat. In this species, electrical stimulation of cortical DSCF neurons evokes centrifugal BF shifts of collicular and cortical DSCF neurons (FIGS 3c and 4b). However, application of the GABA (γ-aminobutyric acid)-receptor antagonist bicuculline to the stimulation site changes the centrifugal BF shifts into centripetal shifts. As bicuculline suppresses inhibition, leading to the excitation of cortical neurons, the drug alone also evokes centripetal BF shifts of collicular and cortical neurons. These observations indicate that compressed reorganization changes into expanded reorganization when cortical inhibition is removed. Electrical stimulation of the AC stimulates both excitatory and inhibitory neurons. In the posterior division of the AC (AIp), facilitation is probably stronger than inhibition. So, electrical stimulation of cortical AIp neurons evokes centripetal BF 786 OCTOBER 23 VOLUME 4

5 a Big brown bat b Moustached bat (Alp) c Moustached bat (DSCF) min.2 7 min 3..2 ms, 1 na n = 98 or min n = 76.3 n = 38 or BF difference (AC R or IC R AC S ) (khz) BF difference (AC R AC S ) (khz) BF difference (AC R or IC R AC S ) (khz) BF shift (khz) d Mongolian gerbil BF shift (khz) min BF shift (khz) e House mouse BF shift (khz) min n = ms, 5 na n = BF difference (AC R AC S ) (khz) BF difference (IC R AC S ) (khz) BF shift (khz) Distance along the AC (mm) Figure 3 Shifts in best frequency evoked by electrical stimulation of cortical auditory neurons. The best-frequency (BF) shift changes as a function of the difference in BF between recorded collicular (IC R, blue) or cortical (AC R, green) neurons and electrically stimulated cortical neurons (AC S ). Each BF-shift difference curve encompasses a scatter plot of BF shifts of many neurons studied (n). Note the differences in curves between species and between different areas of the same species. a,b,d,e Centripetal BF shifts, except where indicated by arrows. A prominent centripetal BF shift occurs at ~5 khz higher than the stimulated cortical BF in the big brown bat (a) and at ~1 khz higher than that in the Mongolian gerbil (d). By contrast, the centripetal BF shift occurs at ~1 khz lower than the stimulated cortical BF in the posterior division of the primary AC (AIp) of the moustached bat (b). In the house mouse, prominent centripetal BF shifts occur at ~9 khz higher and lower than the stimulated cortical BF (e). c Centrifugal BF shifts in the Doppler-shifted constant frequency (DSCF) area of the moustached bat. Prominent centrifugal BF shifts occur at ~.5 khz higher and lower than the stimulated cortical BF. The shape of these BF-shift difference curves might change with the mean BF of stimulated cortical neurons (AC S ). The characteristics of the electrical stimulation ( ) were.2 ms, 1 na pulses for a d and 1 ms, 5 na pulses for e. More intense electrical stimulation presumably increases both BF shifts and the frequency range at which BF shifts occur. Data in a is from REFS 5 and 25; data in b is from REF. 26; data in c is from REF. 35; data in d is from REF. 27; data in e is from REF. 36. AC R IC R shifts of neighbouring cortical neurons (FIGS 3b and 4b). Bicuculline applied to the stimulation site augments these centripetal shifts 35.This observation can be interpreted as evidence that elimination of inhibition led to the augmentation of centripetal BF shifts. The DSCF area is specialized for the representation of biosonar information and is large relative to the rest of the AC 1,48,whereas AIp is not specialized for fine frequency analysis. So, depending on auditory experience, cortical and subcortical reorganization can be quite different between specialized and non-specialized areas of a single species. Jen and co-workers have studied corticofugal modulation of collicular neurons in the big brown bat. They found two groups of collicular neurons that were modulated by corticofugal projections facilitated (26%) and inhibited (74%) neurons 2.In the case of facilitated neurons, cortical stimulation augments their responses and broadens their frequency-tuning curves. In the case of inhibited neurons, electrical stimulation inhibits their responses and sharpens their frequencytuning curves. Such corticofugal modulation occurs even in collicular neurons with BFs that differ by as much as 5 khz from the BF of stimulated cortical neurons. So, perhaps corticofugal modulation occurs over the entire central nucleus of the IC. The strong electrical stimulation that the authors used in this study 2 presumably stimulated a large portion of the bat AC, which contained many cortical neurons that were matched and unmatched in BF to the recorded collicular neuron. Therefore, it is quite difficult to refer to the difference in BF between the stimulated and recorded neurons for the further evaluation of corticofugal modulation. Zhou and Jen 49 also found two groups of collicular neurons that were modulated by corticofugal activity plastic neurons (49%), for which corticofugal effects last 5 35 min, and non-plastic neurons (51%), for which the effects last only up to 1 s. BF shifts evoked by cortical stimulation depend on the relationship in tuning between cortical and collicular neurons 5,23,24,3, and on the relationship in relative location along an iso-bf line between them 27.It is important to take these facts into account when calculating the percentages of plastic and non-plastic neurons. The percentage of collicular neurons that exhibit BF shifts (that is, plastic changes) is higher (84%) when the BFs of the stimulated and recorded neurons are in an appropriate relation for evoking BF shifts. It increases to 92% when the auditory cortical electrical stimulation is delivered during electrical stimulation of the basal forebrain 3.So, it is probable that almost all collicular neurons are potentially plastic. NATURE REVIEWS NEUROSCIENCE VOLUME 4 OCTOBER

6 a Mongolian gerbil AI b Moustached bat MCA 1 Ala FM FM DSCF khz Alp khz 1 c Big brown bat MCA Anterior Dorsal 1 khz AI.5 mm Centripetal BF shift Centrifugal BF shift Figure 4 Expanded and compressed reorganizations in the auditory cortices of three species of mammal. In each panel, crosses mark the location of cortical neurons that are electrically stimulated, and the red lines intersecting these indicate the best frequencies (BFs) or best delays (BDes) of matched neurons that do not show BF or BDe shifts. The broken lines represent iso-bf lines. In the Mongolian gerbil (a) and big brown bat (c), the large area for centripetal BF shifts (expanded reorganization) is surrounded by a narrow zone for small centrifugal BF shifts. In the moustached bat (b), the Doppler-shifted constant frequency (DSCF) and FM FM areas show centrifugal BF shifts, whereas the posterior division of the primary auditory cortex (AIp) shows centripetal BF shifts. Small centrifugal BF shifts are observed anteriorly to the large area for centripetal BF shifts. It is probable that in the AIp, the zone for centrifugal BF shifts surrounds the area for centripetal BF shifts. MCA., middle cerebral artery. a reproduced, with permission, from REF. 27 (22) National Academy of Sciences; data in b is from REFS 24, 26 and 28 ; data in c is unpublished (N. S. and X. M.), and from REFS 5 and 25. Corticofugal modulation of cochlear hair cells. In the moustached bat, the cochlear MICROPHONIC RESPONSE is sharply tuned to ~61 khz. Electrical stimulation of cortical DSCF neurons at a low frequency (5 Hz) evokes centrifugal BF shifts in the IC and MGB 24.However,such stimulation does not change the cochlear microphonic response. On the other hand, cortical stimulation at a higher frequency (33 Hz) evokes a short-term centrifugal BF shift of the contralateral microphonic response, but a centrifugal or centripetal BF shift of the ipsilateral microphonic response. The BF of the cochlear microphonic response systematically shifts by as much as.25 khz, depending on the BF and the location of the stimulated cortical DSCF neurons 7. During the emission of biosonar signals and in the absence of electrical stimulation, the BF of the cochlear microphonic response changes by as much as.15 khz 53.Such a change might be evoked by the corticofugal system and be related to auditory attention to echoes. Multiparametric corticofugal modulation Biologically relevant sounds are characterized by several parameters such as frequency, amplitude and duration. The auditory system has various types of neuron that are tuned to acoustic parameters other than frequency 1,3,48,54. For processing in the frequency domain and other acoustic parameters, corticofugal modulation seems to occur in a specific and systematic way according to the relationship in tuning between activated cortical and recorded subcortical neurons. We now turn our attention to corticofugal modulation of duration tuning in the big brown bat 55,delay tuning in the moustached bat 28, minimum threshold in the house mouse 36 and spatial tuning in the big brown bat 2. MICROPHONIC RESPONSE The summated auditory response of the cochlear hair cells. Nwabueze-Ogbo et al. 5 found that inactivation of the entire AC with tetrodotoxin evokes no BF shifts of collicular neurons in the rat, although their responses are either increased or decreased. These authors concluded that corticofugal modulation differs between rats and bats. In the moustached bat, focal inactivation of the AC evokes BF shifts of subcortical neurons, whereas nonfocal cortical inactivation does not, even though it evokes a large decrease in their auditory responses 34,51.This finding indicates that widespread activation of the AC by strong electrical stimulation and widespread inactivation of the AC by a drug might be ineffective in evoking BF shifts, although these stimuli evoke increases or decreases of subcortical auditory responses. So, there might not be a clear difference in corticofugal modulation between the rat and the big brown bat. Furthermore, the data obtained in the Mongolian gerbil and the big brown bat (FIGS 3 and 4) indicate that corticofugal modulation is essentially the same in these two species. It should be noted that Jen et al. 52 found that neurons in the external nucleus of the IC are inhibitory and broadly tuned in frequency. They are excited by corticocollicular fibres and inhibit neurons in the central nucleus of the IC, which are facilitated by corticofugal input. This inhibitory pathway is comparable to the pathway from the AC to the MGB through the thalamic reticular nucleus. Modulation of duration tuning in the big brown bat. In the big and little brown bats, duration-tuned neurons have been found in the IC and the AC 59.They are sensitive to a particular sound duration, as well as to the frequency of the sound. The response of a neuron is maximal for a specific duration the best duration (BDu). When cortical duration-tuned neurons are electrically stimulated, the response of a duration-tuned collicular neuron with the same BDu and BF is augmented, and its duration tuning is sharpened (FIG. 5a). By contrast, the BDu of an unmatched duration-tuned collicular neuron is either shifted (FIG. 5b) or broadened. All of these changes occur when the BDu and BF differences between the recorded collicular and stimulated cortical neurons are less than 4 ms and 6 khz, respectively (FIG. 5c,d). BDu shifts occur towards the BDu of the stimulated cortical neurons the larger the BDu difference, the larger the BDu shift (FIG. 5e). The broadening of duration tuning mostly occurs towards the BDu of stimulated cortical neurons the larger the BDu difference, the greater the broadening (FIG. 5f). So, these changes are centripetal. Corticofugal modulation of BDu, as for that of BF, depends specifically and systematically on the relationship in BDu between recorded and stimulated neurons OCTOBER 23 VOLUME 4

7 a Impulses/1 stimuli c BF difference (IC R AC S ) (khz) e BDu shift (ms) AC S Duration of tone burst (ms) Duration of tone burst (ms) d BDu difference (IC R AC S ) (ms) BDu: 5.5 ± 1.8 ms.8 Control Recovery Impulses/1 stimuli BF difference (IC R AC S ) (khz) Duration width change (%) f BDu difference (IC R AC S ) (ms) BDu: 5.6 ± 2.4 ms BDu difference (IC R AC S ) (ms) BDu difference (IC R AC S ) (ms) Figure 5 Corticofugal modulation of duration-tuned collicular neurons evoked by electrical stimulation of cortical duration-tuned neurons. a,b The stimulated cortical (AC S ) and recorded collicular (IC R ) neurons were matched (a) or unmatched (b) in terms of both best frequency (BF) and best duration (BDu). The blue and black arrows indicate BDu of IC R and AC S neurons, respectively. Cortical stimulation sharpened (a) or shifted (b) duration-response curves. c,d Distributions of three types of change in duration-response curves: BDu shifts (triangles), sharpening (blue circles) and broadening (green circles). The abscissa and ordinate, respectively, represent BDu and BF differences between IC R and AC S neurons. Each triangle in c represents a BDu and a BF difference between paired AC S and BDu-shifted neurons. Each blue or green circle in d, respectively, represents a BDu and a BF difference of paired AC S and IC R neurons that showed sharpening (blue circles) or broadening (green circles) of the duration-response curve. Crosses in c,e,f mark neurons that showed no changes in BDu and width of the duration-tuning curve. Note that changes in duration tuning occur only when the BF and BDu differences are less than 6 khz and 4 ms, respectively. e,f Distributions of the BDu shifts (e) and width changes (f) in durationresponse curves. The extent of change depends on BDu differences between IC R and AC S neurons. Note that the larger the BDu difference, the larger the change. The correlation coefficient is shown for each regression line. The mean BDu of AC S neurons is also shown in e and f. Reproduced, with permission, from REF. 55 (21) National Academy of Sciences. b AC S Duration-tuned neurons are tuned in both duration and frequency. For cortical electrical stimulation, a centripetal BF shift occurs when the difference in BF between recorded collicular and stimulated cortical neurons is less than 15 khz (FIG. 3a).So, the auditory responses of duration-tuned neurons are modulated as a function of both the BF and BDu differences between recorded and stimulated neurons 55.Centripetal BF and BDu shifts are expected to evoke an expanded representation of a particular sound in the IC and AC. It has been shown that the IC of the big brown bat has an array of neurons that are tuned to different duration values 58.However, no anatomical map for the systematic representation of durations has been found 56,58,59. An intriguing problem is how the corticofugal system can modulate subcortical duration-tuned neurons in a highly specific and systematic way without a duration map. Modulation of delay tuning in the moustached bat. In echolocation, the delay of an echo from the sound emitted by a bat is directly related to the target distance. In the moustached bat, delay-tuned neurons that are sensitive to certain echo delays are found in the AC, MGB and IC. The AC and MGB have a delay map for the systematic representation of echo delays 1,48.It is not yet clear how delay-tuned neurons are arranged in the IC. Electrical stimulation of delay-tuned neurons located in a cortical region called the FM FM area (FM stands for frequency modulation ) augments the response at the best delay (BDe) of a subcortical delaytuned neuron that is matched in BDe to the stimulated cortical neurons. This stimulation also sharpens the delay tuning of the subcortical neuron without shifting its BDe. It simultaneously suppresses the responses at the BDe of unmatched subcortical delay-tuned neurons, and shifts their BDe away from that of the stimulated cortical neurons (centrifugal BDe shifts). The values of BDe shifts are proportional to the differences in BDe between stimulated and recorded delay-tuned neurons the larger the difference, the larger the shift (FIG. 6a). Focal inactivation of cortical delay-tuned neurons with a local anaesthetic evokes changes of subcortical delaytuned neurons that are exactly opposite to those evoked by cortical activation 28.The inactivation experiment indicates that auditory responses of delay-tuned neurons and the delay map in the normal condition are maintained by the corticofugal system. Specific and systematic corticofugal modulation for signal processing (that is, egocentric selection) occurs in the time domain as in the frequency domain. Centripetal and centrifugal shifts occur in the time domain as in the frequency domain centripetal shifts for duration tuning and centrifugal shifts for delay tuning. Modulation of minimum threshold in the house mouse. A minimum threshold (MT) is the threshold of an auditory response at the BF of a neuron. In the auditory system, MT and BF differ from neuron to neuron. In the house mouse, cortical electrical stimulation evokes shifts in both the MT and BF of collicular neurons. Collicular neurons matched to the stimulated cortical neurons in BF, but not in MT, show no BF shift, but do show an MT shift towards the MT of stimulated neurons (centripetal MT shift) the larger the MT difference, the larger the MT shift (FIG. 6b). When the MT of a recorded neuron is higher than that of the stimulated neurons, the former decreases towards the latter. However, when the MT of a recorded neuron is lower than that of the stimulated neurons, the former increases towards the latter. On the other hand, collicular neurons that are unmatched in both BF and MT show both BF and MT shifts in response to cortical electrical stimulation. Almost all of these neurons increase their MT regardless of whether MT or BF differences are positive or negative. NATURE REVIEWS NEUROSCIENCE VOLUME 4 OCTOBER

8 a Moustached bat BDe shift (ms) n = BDe difference (IC R AC S ) (ms) b House mouse MT shift (db) 2 2 n = MT difference (IC R AC S ) (db) Figure 6 Corticofugal modulation of best delays and minimum thresholds. Centrifugal best-delay (BDe) shifts in the moustached bat and centripetal minimum-threshold (MT) shifts in the house mouse. The extent of BDe and MT shifts is linearly related to the difference in BDe and MT between recorded collicular (IC R ) and electrically stimulated cortical (AC S ) neurons, respectively. The correlation coefficient is shown for each regression line. n, number of neurons studied. a reproduced, with permission, from REF. 28 (1996) American Association for the Advancement of Science; b reproduced, with permission, from REF. 36 (22) Blackwell Publishing. An MT increase is accompanied by a large decrease in response. So, corticofugal modulation enhances the contrast in neural representation of an auditory signal by the centripetal MT shifts of BF-matched neurons and by the suppression of BF-unmatched neurons 36.In contrast to the house mouse, a large MT shift has not been observed in the big brown bat, and centripetal BF shifts of unmatched neurons are associated with a facilitation of the response at a shifted BF (FIG. 1b).So,there is a difference in corticofugal modulation between the house mouse and the big brown bat. In the house mouse 36 and the big brown bat 49, the dynamic range of a collicular neuron for coding stimulus amplitude or intensity decreases when the MT of the neuron becomes higher after cortical electrical stimulation. In these studies, the decrease in dynamic range was studied only at the BF of the control condition; that is, at the BF measured without electrical stimulation. The dynamic ranges at the control and shifted BFs of a given collicular neuron should be compared with each other for the evaluation of corticofugal modulation. If cortical electrical stimulation evokes BF shifts of collicular neurons (FIG. 1b), the dynamic range must increase at the shifted BF and decrease at the control BF. Modulation of spatial tuning in the big brown bat. The spatial tuning (direction sensitivity) of the ear varies with the frequency of an auditory stimulus. In the auditory system, binaural interactions take place and give rise to neurons with spatial tuning properties that are different from those determined by the ear. In the big brown bat, cortical electrical stimulation evokes inhibition or facilitation of collicular neurons, and changes their spatial-tuning curves spatial-tuning curves of neurons that are inhibited by corticofugal projections are sharpened, and those of neurons that are facilitated by these projections broaden 2.Corticofugal modulation presumably depends on the relationship in spatial tuning between stimulated and recorded neurons. It remains to be established whether the spatial-tuning curves of collicular neurons are shifted by cortical electrical stimulation according to the relationship in spatial tuning between stimulated and recorded neurons, and whether the changes in spatial tuning are independent of facilitation or inhibition in the frequency domain. Despite this lack of information, it is clear that corticofugal modulation takes place in the spatial domain. Plasticity elicited by auditory fear conditioning Does acoustic stimulation evoke corticofugal modulation in the same way as electrical stimulation does? In the big brown bat, BF shifts of collicular and cortical neurons have been studied with three types of stimulus a tone repetitively delivered to the animal, focal electrical stimulation of the AC 5,23,25,33 and auditory fear conditioning These three types of stimulus evoked basically the same BF shifts in the IC, although the tone was much less effective than the other stimuli. Several experiments 37,38 indicate that fear conditioning causes collicular changes through the corticofugal system, as does electrical stimulation (FIG. 1d).So, the collicular changes that are evoked by electrical stimulation are not epiphenomena, and cortical electrical stimulation is an adequate method for exploring the function of the corticofugal system. Weinberger and his co-workers 6,61 proposed the following model to explain BF shifts caused by auditory fear conditioning (FIG. 7a). An auditory signal (the conditioned stimulus) reaches the AC through the ventral division of the MGB (MGBv), which is thought not to be plastic. It also reaches the AC through the magnocellular division of the MGB (MGBm), which is plastic, and through the posterior intralaminar nucleus in the thalamus. The somatosensory signal elicited by the foot shock (the unconditioned stimulus) also reaches the MGBm and the posterior intralaminar nucleus. These structures are therefore the sites where the auditory and somatosensory signals first converge for associative learning. The associated signal is then sent to the AC to strengthen the effect of MGBv neurons which were excited by the conditioned stimulus on cortical neurons. This associated signal is also sent to the amygdala, which in turn projects to the cholinergic basal forebrain (the nucleus basalis). The nucleus basalis increases cortical acetylcholine concentrations and further amplifies the effect of the MGBm neurons on the cortical neurons. Several studies indicate that the cholinergic basal forebrain is crucial for cortical plasticity. For example, electrical stimulation of the basal forebrain paired with a tone evokes centripetal BF shifts for the expanded representation of the tone frequency 3,39,62,63. The MGBv exhibits short-term, frequency-specific plasticity for fear conditioning 64.However, it is not plastic in the aforementioned model, whereas the MGBm is 6,61. MGBm neurons often have a broad or multipeaked frequency-tuning curve, and habituate after several stimulus presentations 65,66.So, the MGBm might not be suited to fine adjustment of the auditory system for signal processing. Instead, other parts of the auditory system, where the fine analysis of auditory signals takes place and where the corticofugal system evokes subcortical changes 79 OCTOBER 23 VOLUME 4

9 a Auditory cortex b Auditory system (egocentric selection in FATS domains) Somatosensory systems Cholinergic system I II Plasticity (augmentation according to behavioural relevance) Auditory cortex Somatosensory cortex Cholinergic basal forebrain III IV ACh TRN Associated cortex Medial geniculate body Vent. post. LTN MGBv MGBm PIN V VI Nucleus basalis Inferior colliculus Lat. LN DCN Amygdala (conditioned behavioural response) SOC Cholinergic brainstem nuclei IC Amygdala CS (tone) US (shock) Conditioned response CN Cochlea Leg Conditioning Modulatory systems 1. Cholinergic basal forebrain 2. Dopaminergic ventral tegmental area 3. Serotonergic raphe nuclei 4. Noradrenergic locus coeruleus Figure 7 Working models for the changes in auditory processing during auditory fear conditioning (associative learning). a Weinberger and colleagues model 6,61. ACh, acetylcholine; CS, conditioned stimulus (tone); IC, inferior colliculus; MGBm, magnocellular division of the medial geniculate body; MGBv, ventral division of the medial geniculate body; PIN, posterior intralaminar nucleus in the thalamus; US, unconditioned stimulus (shock). b Gao and Suga s model 31,33. See text for details on both models. CN, cochlear nucleus; DCN, dorsal column nuclei; FATS, frequency, amplitude, time and spatial domains; Lat. LN, lateral leminiscal nuclei; SOC, superior olivary complex; TRN, thalamic reticular nucleus; Vent. post. LTN, ventro-postero-lateral thalamic nucleus. that are highly specific to acoustic stimuli 5,23,28,33,34, might be involved. The IC shows plastic changes in relation to associative learning 33, as does the AC 31. So, the plastic changes of the MGBm might be due to those in the IC. It should be noted that the BF shifts that are observed in the IC and the AC also take place in the MGBv 28,34. The corticofugal auditory system and the somatosensory cortex have important roles in the plasticity of the IC and AC that is elicited by auditory fear conditioning. So, Gao and Suga 31,33 proposed an alternative working model (FIG. 7b). In response to a behaviourally insignificant sound that is repetitively delivered to an animal, cortical auditory neurons change their response properties and, through the corticofugal system, evoke changes in the response properties of subcortical auditory neurons. These changes, which result from egocentric selection, are minor, short-term and highly specific to the values of the parameters that characterize the sound. When the sound is paired with a foot shock, the auditory and somatosensory signals ascend from the periphery to the auditory and somatosensory cortices, respectively, and then presumably to the amygdala through association cortices. These signals might be associated in these cortical regions and/or the amygdala, so that the sound becomes behaviourally relevant to the animal 31,33. The amygdala is essential for evoking a conditioned response 61,67. It sends the associated auditory somatosensory signal to the cholinergic basal forebrain, which in turn increases the cortical acetylcholine concentration, as hypothesized by Weinberger et al. 6. This modulatory system alone has presumably no specific information about (that is, no neurons tuned to) the multipleparameters that characterize sounds, but augments the response of cortical neurons to the conditioned sound. Accordingly, the activity of the corticofugal system is augmented, and the subcortical changes that result from egocentric selection become larger, but are short-lived. The cortical changes also become larger but, in contrast to the collicular changes, they become longlived because of increased acetylcholine concentration in the AC 32.This cortico-colliculo-cortical feedback loop is presumably controlled by inhibition from the thalamic NATURE REVIEWS NEUROSCIENCE VOLUME 4 OCTOBER

10 reticular nucleus 37,which receives cholinergic input from the basal forebrain 68,69.As corticofugal modulation of collicular neurons has been best characterized, the previous discussion centres on the cortico-collicular feedback. But there are several corticofugal feedback loops, such as the cortico-thalamo-cortical loop, which might also contribute to cortical plasticity. The augmentation of auditory cortical plasticity in the frequency domain (BF shifts) by the cholinergic basal forebrain 3,39,62 and the dopaminergic ventral tegmental area 7 has been well characterized in different species. In rats, the responses of cortical neurons to amplitude-modulated sounds 71 or to a combination of sounds 72 are augmented by electrical stimulation of the basal forebrain. FM FM neurons of the moustached bat are delay-tuned, combination-sensitive neurons that are tuned to particular spectrotemporal sound patterns 73,74. As mentioned above, activation or inactivation of cortical FM FM neurons modulate subcortical FM FM neurons 28 (FIG. 6a). Combination sensitivity (that is, facilitatory responses of cortical FM FM neurons to particular paired sounds) depends to a large extent on corticofugal feedback 51.So, we speculate that corticofugal feedback also has a role in the augmentation of the responses of rat cortical neurons to amplitude-modulated sounds or to a combination of sounds by the basal forebrain. In general, we believe that the AC and the corticofugal system evoke plastic changes that are specific and systematic to the values of the multiple parameters that characterize an acoustic stimulus, so that the augmentation of these changes by cholinergic and dopaminergic systems is also multiparametric. In two species of bats, the basal nucleus of the amygdala projects bilaterally to the IC, including to its central nucleus 75.This projection is probably glutamatergic 76.Inactivation of the AC or the somatosensory cortex abolishes the development of the collicular BF shift caused by conditioning 31,33.Application of the cholinergic antagonist atropine to the IC abolishes the development of the collicular BF shift and reduces the cortical BF shift caused by conditioning 32. The IC presumably receives a projection from the cholinergic brainstem nuclei 77.So, the collicular BF shift is probably not evoked by the amygdalo-collicular projection. The functional role of this projection has yet to be explored. Multiple functions of the corticofugal system Egocentric selection. Acoustic stimulation, electrical stimulation or auditory fear conditioning can evoke egocentric selection, which modifies the input of the activated cortical neurons through the modifications of subcortical auditory processing. Egocentric selection is based on focused positive feedback that is associated with lateral inhibition in the AC and the corticofugal feedback loops 28,34.Egocentric selection can sharpen the filter properties of neurons in the frequency 25,34, amplitude, time 28,55 and spatial domains 2, and augments sensitivity to certain combinations of sounds 51. Egocentric selection occurs even in the cochlea of the moustached bat 7 and probably the human 78.In the guinea-pig and cat, the responses of auditory nerve fibres to tones masked by homolaterally delivered noise increase when another noise is continuously delivered to the contralateral ear to activate the crossed olivocochlear bundle (COCB). That is, the COCB increases the signalto-noise ratio In monkeys, the COCB improves discrimination of complex sounds in noise 82.Such improvement might be partly due to egocentric selection mediated by the corticofugal system, assuming that the COCBs were modulated by the corticofugal system. This needs to be explored further. In the cat visual system, cortical orientation-sensitive neurons perform egocentric selection 29,83,and corticofugal feedback changes the response properties of thalamic neurons 84.The feedback projection from motionsensitive neurons to area 18 in the cortex augments visual responses of neurons in area 18 and contributes to the emergence of direction sensitivity in area 18 (REF. 85). These findings indicate that the corticothalamic projection and cortical feedback within the cortex might perform egocentric selection. For acoustic stimuli alone, egocentric selection evokes minor, short-term plasticity at the IC and AC. For auditory fear conditioning, however, it evokes large, short-term collicular plasticity, and large, long-term cortical plasticity. The short-term collicular plasticity participates in evoking the long-term cortical plasticity. Large collicular and cortical plasticity are caused by the augmentation of the cortical and corticofugal modulation in the auditory system, which involves other systems such as the somatosensory cortex and the basal forebrain In the somatosensory systems of monkeys 46 and rats 47, the corticofugal system is involved in the plasticity of the thalamic somatosensory nucleus. Attentional modulation. The corticofugal system might mediate attentional modulation of auditory processing 7. In cats, visual attention to a mouse reduces auditory responses of the dorsal cochlear nucleus 86 and a visual discrimination task reduces auditory nerve responses to clicks 87,88.In humans, visual attention reduces auditory nerve responses 89 and sound emissions by the cochlea evoked by a click 9.In the moustached bat, the cochlear BF fluctuates during the emission of biosonar pulses 53, but it systematically varies with the location of focal cortical activation 7. Gain control. The facilitatory and inhibitory corticofugal modulation in cats 11 13,15 18,rats 19 and bats 14,2,34,51 could be interpreted as gain control for auditory signal processing. The COCB changes the MT of cochlear hair cells in the guinea-pig 91 and of cochlear nerve fibres in the cat 92.Egocentric selection based on positive feedback and lateral inhibition might be viewed as selective gain control. In cats, the COCB changes the dynamic range of intensity coding by auditory nerve fibres 93. Corticofugal modulation of a dynamic range in intensity coding also occurs in the IC of mice 36 and bats 49. Modulation of a dynamic range in intensity coding is a consequence of gain control. 792 OCTOBER 23 VOLUME 4

11 Low-frequency modulation of brain rhythms. The corticofugal visual system transmits slow oscillatory changes in cortical activity to the thalamus. This slow oscillation interacts with spindles generated in the thalamus, modulates neural excitability and produces different brain rhythms that characterize various behavioural states 94.The corticofugal auditory system might also transmit slow oscillatory changes in cortical activity to the thalamic auditory nucleus. It remains to be established whether these possible functions of the corticofugal system cause behavioural changes. For example, it remains to be seen whether the compressed reorganization in the frequency domain improves the detection and discrimination ability of the bat. Similarly, we do not know whether the expanded reorganization in the frequency domain increases the sensitivity to a sound at a particular frequency. Epilogue Are the mechanisms that have been found in an organism such as the bat relevant to other species? The corticofugal visual system 29,83 and feedback projections within the visual cortex of cats 85 perform egocentric selection. The corticofugal somatosensory systems of monkeys and rats participate in the plasticity of the thalamic somatosensory nucleus 46,47. Electrical stimulation of the AC evokes centripetal BF shifts in the AC of the Mongolian gerbil 26,27.In humans, corticofugal effects on cochlear hair cells 78 are comparable to those found in the moustached bat 7.So, corticofugal functions that have been found in bats, such as egocentric selection and subcortical plasticity, are apparently shared by other sensory modalities in other species. On the other hand, compressed reorganization has only been found so far in the specialized auditory subsystems of the moustached bat 7,28,34,35.It is not clear whether it is specific to the moustached bat, because no comparable studies have been performed on other species. We hope that this article stimulates research on the corticofugal system, and that future research on different species will reveal which corticofugal functions are species-specific and which are more general. 1. Suga, N. in Dynamic Aspects of Neocortical Function (eds Edelman, G. M., Gall, W. E. & Cowan, W. M.) (John Wiley and Sons, New York, 1984). 2. Suga, N. in Auditory Function (eds Edelman, G. M., Gall, W. E. & Cowan, W. M.) (John Wiley & Sons, New York, 1988). 3. Covey, E. & Casseday, J. H. Timing in the auditory system of the bat. Annu. Rev. Physiol. 61, (1999). 4. Saldana, E., Feliciano, M. & Mugnaini, E. Distribution of descending projections from primary auditory neocortex to inferior colliculus mimics the topography of intracollicular projections. J. Comp. Neurol. 371, 15 4 (1996). 5. Ma, X. & Suga, N. Plasticity of bat s central auditory system evoked by focal electric stimulation of auditory and/or somatosensory cortices. J. Neurophysiol. 85, (21). 6. Feliciano, M., Saldana, E. & Mugnaini, E. Direct projections from the rat primary auditory neocortex to nucleus sagulum, paralemniscal regions, superior olivary complex and cochlear nuclei. Aud. Neurosci. 1, (1995). 7. Xiao, Z. & Suga, N. Modulation of cochlear hair cells by the auditory cortex in the mustached bat. Nature Neurosci. 5, (22). Experimental evidence for systematic corticofugal modulation of the frequency tuning of cochlear hair cells. 8. Kelly, J. P. & Wong, D. Laminar connections of the cat s auditory cortex. Brain Res. 212, 1 15 (1981). 9. Ojima, H. Terminal morphology and distribution of corticothalamic fibers originating from layers 5 and 6 of cat primary auditory cortex. Cereb. Cortex 6, (1994). 1. Huffman, R. F. & Henson, O. W. Jr. The descending auditory pathway and acousticomotor systems: connections with the inferior colliculus. Brain Res. Brain Res. Rev. 15, (199). A comprehensive review of the anatomy and physiology of the corticofugal auditory system. Our current article only reviews the recent progress in physiological research of corticofugal modulation and plasticity. Reference 1 complements our review because it describes what was known about the corticofugal system before Massopust, L. C. Jr & Ordy, J. M. Auditory organization of the inferior colliculus in the cat. Exp. Neurol. 6, (1962). 12. Watanabe, T., Yanagisawa, K., Kanzaki, J. & Katsuki, Y. Cortical efferent flow influencing unit responses of medial geniculate body to sound stimulation. Exp. Brain Res. 2, (1966). 13. Amato, G., La Grutta, V. & Enia, F. The control exerted by the auditory cortex on the activity of the medial geniculate body and inferior colliculus. Arch. Sci. Biol. (Bologna) 53, (1969). 14. Sun, X., Chen, Q. C. & Jen, P. H. Corticofugal control of central auditory sensitivity in the big brown bat, Eptesicus fuscus. Neurosci. Lett. 212, (1996). 15. Andersen, P., Junge, K. & Sveen, O. Cortico-fugal facilitation of thalamic transmission. Brain Behav. Evol. 6, (1972). 16. Orman, S. S. & Humphrey, G. L. Effects of changes in cortical arousal and of auditory cortex cooling on neuronal activity in the medial geniculate body. Exp. Brain Res. 42, (1981). 17. Villa, A. E. et al. Corticofugal modulation of the information processing in the auditory thalamus of the cat. Exp. Brain Res. 86, (1991). 18. Ryugo, D. K. & Weinberger, N. M. Corticofugal modulation of the medial geniculate body. Exp. Neurol. 51, (1976). 19. Syka, J. & Popelar, J. Inferior colliculus in the rat: neuronal responses to stimulation of the auditory cortex. Neurosci. Lett. 51, (1984). 2. Jen, P. H., Chen, Q. C. & Sun, X. D. Corticofugal regulation of auditory sensitivity in the bat inferior colliculus. J. Comp. Physiol. A 183, (1998). 21. Suga, N. Biosonar and neural computation in bats. Sci. Am. 262, 6 68 (199). 22. Zhang, Y. & Suga, N. Corticofugal amplification of subcortical responses to single tone stimuli in the mustached bat. J. Neurophysiol. 78, (1997). 23. Yan, W. & Suga, N. Corticofugal modulation of the midbrain frequency map in the bat auditory system. Nature Neurosci. 1, (1998). 24. Zhang, Y. & Suga, N. Modulation of responses and frequency tuning of thalamic and collicular neurons by cortical activation in mustached bat. J. Neurophysiol. 84, (2). 25. Chowdhury, S. A. & Suga, N. Reorganization of the frequency map of the auditory cortex evoked by cortical electrical stimulation in the big brown bat. J. Neurophysiol. 83, (2). 26. Sakai, M. & Suga, N. Plasticity of the cochleotopic (frequency) map in specialized and nonspecialized auditory cortices. Proc. Natl Acad. Sci. USA 98, (21). 27. Sakai, M. & Suga, N. Centripetal and centrifugal reorganizations of frequency map of the auditory cortex in gerbils. Proc. Natl Acad. Sci. USA 99, (22). This paper reports that the reorganization of the frequency map evoked by focal cortical electrical stimulation consists of a central area for centripetal shifts and a surrounding zone for centrifugal shifts. 28. Yan, J. & Suga, N. Corticofugal modulation of time-domain processing of biosonar information in bats. Science 273, (1996). This paper reports that specific and systematic corticofugal modulation evokes centrifugal shifts of the delay-tuning curves of thalamic and collicular neurons. 29. Murphy, P. C., Duckett, S. G. & Sillito, A. M. Feedback connections to the lateral geniculate nucleus and cortical response properties. Science 286, (1999). 3. Ma, X. & Suga, N. Augmentation of plasticity of the central auditory system by the basal forebrain and/or somatosensory cortex. J. Neurophysiol. 89, 9 13 (23). 31. Gao, E. & Suga, N. Experience-dependent plasticity in the auditory cortex and the inferior colliculus of bats: role of the corticofugal system. Proc. Natl Acad. Sci. USA 97, (2). Experimental evidence that the collicular plasticity caused by auditory fear conditioning is evoked by corticofugal feedback, that the somatosensory cortex has an important role in the development of collicular plasticity, and that collicular plasticity is evoked not only by conditioning, but also by repetitive acoustic stimuli. 32. Ji, W., Gao E. & Suga, N. Effects of acetylcholine and atropine on plasticity of central auditory neurons caused by conditioning in bats. J. Neurophysiol. 86, (21). Together with reference 31, this paper provides experimental evidence that the short-term collicular plasticity evoked by corticofugal feedback contributes to the production of the long-term cortical plasticity caused by conditioning. 33. Gao, E. & Suga, N. Plasticity of midbrain auditory frequency map mediated by the corticofugal system in bat. Proc. Natl Acad. Sci. USA 95, (1998). This paper reports the time courses of collicular and cortical plasticity caused by conditioning or cortical electrical stimulation. It also reports that inactivation of the somatosensory cortex abolishes both collicular and cortical plasticity caused by conditioning. 34. Zhang, Y., Suga, N. & Yan, J. Corticofugal modulation of frequency processing in bat auditory system. Nature 387, 9 93 (1997). This paper reports that specific and systematic corticofugal modulation evokes centrifugal shifts of frequency-tuning curves of thalamic and collicular neurons in a part of the auditory system specialized for fine frequency analysis. 35. Xiao, Z. & Suga, N. Reorganization of the cochleotopic map in the bat s auditory system by inhibition. Proc. Natl Acad. Sci. USA 99, (22). 36. Yan, J. & Ehret, G. Corticofugal modulation of midbrain sound processing in the house mouse. Eur. J. Neurosci. 16, 1 11 (22). 37. Suga, N., Gao, E., Zhang, Y., Ma, X. & Olsen, J. F. The corticofugal system for hearing: recent progress. Proc. Natl Acad. Sci. USA 97, (2). 38. Suga, N, Xiao, Z., Ma, X. & Ji, W. Plasticity and corticofugal modulation for hearing in adult animals. Neuron 36, 9 18 (22). NATURE REVIEWS NEUROSCIENCE VOLUME 4 OCTOBER

12 39. Kilgard, M. P. & Merzenich, M. M. Cortical map reorganization enabled by nucleus basalis activity. Science 279, (1998). This paper reports the massive reorganization of the auditory cortex that is evoked by acoustic and electrical stimulation of the nucleus basalis. 4. Weinberger, N. M, & Bakin, J. S. Learning-induced physiological memory in adult primary auditory cortex: receptive fields plasticity, model, and mechanisms. Audiol. Neurootol. 3, (1998). 41. Buonomano, D. V. & Merzenich, M. M. Cortial plasticity; from synapses to maps. Ann. Rev. Neurosci. 21, (1998). This is a comprehensive review of cortical plasticity. This reference is complementary to our current article. 42. Rasmusson, D. D. The role of acetylcholine in cortical synaptic plasticity. Behav. Brain Res. 115, (2). 43. Godde, B., Leonhardt, R., Cords, S. M. & Dinse, H. R. Plasticity of orientation preference maps in the visual cortex of adult cats. Proc. Natl Acad. Sci. USA 99, (22). 44. Sillito, A. M., Jone, H. E., Gerstein, G. L. & West, D. C. Feature-linked synchronization of thalamic relay cell firing induced by feedback from the visual cortex. Nature 2, 1 2 (1994). 45. Singer, W. Synchronization of cortical activity and its putative role in information processing and learning. Annu. Rev. Physiol. 55, (1993). 46. Ergenzinger, E. R., Glasier, M. M., Hahm, J. O. & Pons, T. P. Cortically induced thalamic plasticity in the primate somatosensory system. Nature Neurosci. 1, (1998). 47. Krupa, D. J., Ghazanfar, A. A. & Nicolelis, M. A. Immediate thalamic sensory plasticity depends on corticothalamic feedback. Proc. Natl Acad. Sci. USA 96, (1999). 48. Suga, N. in The Cognitive Neurosciences (ed. Gazzaniga, M. S.) (MIT Press, Cambridge, Massachusetts, 1994). 49. Zhou, X. & Jen, P. H. Brief and short-term corticofugal modulation of subcortical auditory responses in the big brown bat, Eptesicus fuscus. J. Neurophysiol. 84, (2). 5. Nwabueze-Ogbo, F. C., Popelar, J. & Syka, J. Changes in the acoustically evoked activity in the inferior colliculus of the rat after functional ablation of the auditory cortex. Physiol. Res. 51, S95 S14 (22). 51. Yan, J. & Suga, N. Corticofugal amplification of facilitative auditory responses of subcortical combination-sensitive neurons in the mustached bat. J. Neurophysiol. 81, (1999). 52. Jen, P. H., Zhou, X. & Wu, C. H. Temporally patterned sound pulse trains affect intensity and frequency sensitivity of inferior collicular neurons of the big brown bat, Eptesicus fuscus. J. Comp. Physiol. A 187, (21). 53. Goldberg, R. L. & Henson, O. W. Jr. Changes in cochlear mechanics during vocalization: evidence for a phasic medial efferent effect. Hear. Res. 122, (1998). 54. Rauschecker, J. P. & Tian, B. Mechanisms and streams for processing of what and where in auditory cortex. Proc. Natl Acad. Sci. USA 97, (2). 55. Ma, X. & Suga, N. Corticofugal modulation of duration-tuned neurons in the midbrain auditory nucleus in bats. Proc. Natl Acad. Sci. USA 98, (21). The best example of multiparametic corticofugal modulation for hearing. 56. Pinheiro, A. D., Wu, M. & Jen, P. H. Encoding repetition rate and duration in the inferior colliculus of the big brown bat, Eptesicus fuscus. J. Comp. Physiol. A 169, (1991). 57. Casseday, J. H., Ehrlich, D. & Covey, E. Neural tuning for sound duration: role of inhibitory mechanisms in the inferior colliculus. Science 264, (1994). 58. Ehrlich, D., Casseday, J. H. & Covey, E. Neural tuning to sound duration in the inferior colliculus of the big brown bat, Eptesicus fuscus. J. Neurophysiol. 77, (1997). 59. Galazyuk, A. V. & Feng, A. S. Encoding of sound duration by neurons in the auditory cortex of the little brown bat, Myotis lucifugus. J. Comp. Physiol. A 18, (1997). 6. Weinberger, N. M., Ashe, J. H., Metherate, R., Diamond, D. M. & Bakin, J. Retuning auditory cortex by learning: a preliminary model of receptive field plasticity. Concepts Neurosci. 1, (199). 61. Weinberger, N. M. Physiological memory in primary auditory cortex: characteristics and mechanisms. Neurobiol. Learn. Mem. 7, (1998). Together with reference 4, this is a comprehensive review of plasticity of the auditory cortex caused by auditory fear conditioning. As our review discusses Weinberger s model only in relation to corticofugal modulation, this reference is complementary to our current article. 62. Bakin, J. S. & Weinberger, N. M. Induction of physiological memory in the cerebral cortex by stimulation of the nucleus basalis. Proc. Natl Acad. Sci. USA 93, (1996). 63. Bjordahl, T. S., Dimyan, M. A. & Weinberger, N. M. Induction of long-term receptive field plasticity in the auditory cortex of the waking guinea pig by stimulation of the nucleus basalis. Behav. Neurosci. 112, (1998). 64. Edeline, J. M. & Weinberger, N. M. Thalamic short-term plasticity in the auditory system: associative returning of receptive fields in the ventral medial geniculate body. Behav. Neurosci. 15, (1991). 65. Aitkin, L. M. Medial geniculate body of the cat: responses to tonal stimuli of neurons in medial division. J. Neurophysiol. 36, (1973). 66. Calford, M. B. The parcellation of the medial geniculate body of the cat defined by the auditory response properties of single units. J. Neurosci. 3, (1983). 67. LeDoux, J. E. Emotion circuits in the brain. Annu. Rev. Neurosci. 23, (2). 68. Levey, A. I., Hallanger, A. E. & Wainer, B. H. Cholinergic nucleus basalis neurons may influence the cortex via the thalamus. Neurosci. Lett. 74, 7 13 (1987). 69. Steriade, M., Parent, A., Pare, D. & Smith, Y. Cholinergic and non-cholinergic neurons of cat basal forebrain project to reticular and mediodorsal thalamic nuclei. Brain Res. 48, (1987). 7. Bao, S., Chan, V. T. & Merzenich, M. M. Cortical remodeling induced by activity of ventral tegmental dopamine neurons. Nature 412, (21). This paper reports reorganization of the auditory cortex evoked by auditory and electrical stimulation of dopaminergic neurons. Differences between cholinergic and dopaminergic modulation are discussed. 71. Kilgard, M. P. & Merzenich, M. M. Plasticity of temporal information processing in the primary auditory cortex. Nature Neurosci. 1, (1998). 72. Kilgard, M. P. & Merzenich, M. M. Order-sensitive plasticity in adult primary auditory cortex. Proc. Natl Acad. Sci. USA 99, (22). 73. O Neill, W. E. & Suga, N. Encoding of target range and its representation in the auditory cortex of the mustached bat. J. Neurosci. 2, (1982). 74. Suga, N., O Neill, W. E., Kujirai, K. & Manabe, T. Specificity of combination-sensitive neurons for processing of complex biosonar signals in auditory cortex of the mustached bat. J. Neurophysiol. 49, (1983). 75. Marsh, R. A., Fuzessery, Z. M., Grose, C. D. & Wenstrup, J. J. Projection to the inferior colliculus from the basal nucleus of the amygdala. J. Neurosci. 22, (22). 76. McDonald, A. J. Glutamate and aspartate immunoreactive neurons of the rat basolateral amygdala: colocalization of excitatory amino acids and projections to the limbic circuit. J. Comp. Neurol. 365, (1996). 77. Aigner, T. G. Pharmacology of memory: cholinergic glutamatergic interactions. Curr. Opin. Neurobiol. 5, (1995). 78. Khalfa, S. et al. Evidence of peripheral auditory activity modulation by the auditory cortex in humans. Neuroscience 14, (21). 79. Nieder, P. & Nieder, I. Antimasking effect of crossed olivocochlear bundle stimulation with loud clicks in guinea pig. Exp. Neurol. 28, (197). 8. Dolan, D. F. & Nuttall, A. L. Inner hair cell responses to tonal stimulation in the presence of broadband noise. J. Acoust. Soc. Am. 86, (1989). 81. Kawase, T., Delgutte, B. & Liberman, M. C. Antimasking effects of the olivocochlear reflex. II. Enhancement of auditory-nerve response to masked tones. J. Neurophysiol. 7, (1993). 82. Dewson, J. H. Efferent olivocochlear bundle: some relationships to stimulus discrimination in noise. J. Neurophysiol. 31, (1968). 83. Tsumoto, T., Creutzfeldt, O. D. & Legendy, C. R. Functional organization of the corticofugal system from visual cortex to lateral geniculate nucleus in the cat (with an appendix on geniculo-cortical mono-synaptic connections). Exp. Brain Res. 32, (1978). 84. Sillito, A. M., Cudeiro, J. & Murphy, P. C. Orientation sensitive elements in the corticofugal influence on centresurround interactions in the dorsal lateral geniculate nucleus. Exp. Brain Res. 93, 6 16 (1993). 85. Galuske, R. A., Schmidt, K. E., Goebel, R., Lomber, S. G. & Payne, B. R. The role of feedback in shaping neural representations in cat visual cortex. Proc. Natl Acad. Sci. USA 99, (22). 86. Hernandez-Peon, R., Scherrer, H. & Jouvet, M. Modification of electric activity in cochlear nucleus during attention in unanesthetized cats. Science 123, (1956). 87. Oatman, L. C. Role of visual attention on auditory evoked potentials in unanesthetized cats. Exp. Neurol. 32, (1971). 88. Oatman, L. C. & Anderson, B. W. Effects of visual attention on tone burst evoked auditory potentials. Exp. Neurol. 57, (1977). 89. Lukas, J. H. Human auditory attention: the olivocochlear bundle may function as a peripheral filter. Psychophysiology 17, (198). 9. Puel, J. L., Bonfils, P. & Pujol, R. Selective attention modifies the active micromechanical properties of the cochlea. Brain Res. 447, (1988). 91. Brown, M. C. & Nuttall, A. L. Efferent control of cochlear inner hair cell responses in the guinea-pig. J. Physiol. (Lond.) 354, (1984). 92. Wiederhold, M. L. Variations in the effects of electric stimulation of the crossed olivocochlear bundle on cat single auditory-nerve-fiber responses to tone bursts. J. Acoust. Soc. Am. 48, (197). 93. Geisler, C. D. Model of crossed olivocochlear bundle effects. J. Acoust. Soc. Am. 56, (1974). 94. Steriade, M. Coherent oscillations and short-term plasticity in corticothalamic networks. Trends Neurosci. 22, (1999). Acknowledgements Our work has been supported by a research grant from the National Institute on Deafness and Other Communication Disorders. Online links FURTHER INFORMATION MIT Encyclopedia of Cognitive Science: auditory plasticity echolocation Access to this interactive links box is free online. 794 OCTOBER 23 VOLUME 4

13