Modulation of auditory processing by cortico-cortical feed-forward and feedback projections

Abstract

The auditory center in the cerebrum, the auditory cortex, consists of multiple interconnected areas. The functional role of these interconnections is poorly understood. The auditory cortex of the mustached bat consists of at least nine areas, including the frequency modulation–frequency modulation (FF) and dorsal fringe (DF) areas. The FF and DF areas consist of neurons tuned to specific echo delays carrying target-distance information. The DF area is hierarchically at a higher level than the FF area. Here, we show that the feedback projection from the DF area to the FF area shifts the delay-tuning of FF neurons toward that of the stimulated DF neurons. In contrast, the feed-forward projection from the FF area to the DF area shifts the delay-tuning of DF neurons away from that of the stimulated FF neurons. The lateral projection within the DF area shifts the delay-tuning of DF neurons toward that of the stimulated DF neurons. In contrast, the lateral projection within the FF area shifts the delay-tuning of FF neurons away from that of the stimulated FF neurons. The delay-tuning shift evoked by the DF stimulation was 2.5 times larger than that evoked by the FF stimulation. Our data indicate that the FF–DF feed-forward and FF–FF lateral projections shape the highly selective neural representation of the tuning of the excited DF neurons, whereas the DF–FF feedback and DF–DF lateral projections enhance the representation of the selected tuning, perhaps, for focal processing of information carried by the excited FF neurons.

The auditory center in the cerebrum, the auditory cortex, consists of multiple, anatomically distinct areas, as do other sensory cortices. Its functional organization beyond tonotopy, however, has hardly been explored, except for the auditory cortex of the mustached bat, Pteronotus parnellii. Tonotopically organized cortical auditory areas are commonly interconnected (1). The cortico–cortical interaction within the primary auditory cortex has been studied (2–6). However, the cortico–cortical interaction between different cortical auditory areas has not yet been studied. The interaction between different cortical areas in a sensory system has thus far been studied only in the cat visual cortex by Galuske et al. (7). They found that inactivation of the visuoparietal cortex decreases the orientation and direction sensitivities of neurons in area 18. That is, they studied the effect of the feedback projection (the projection from a higher to a lower area in a signal processing hierarchy), but not that of the feed-forward projection (the projection from a lower to a higher area).

A microchiropteran bat emits orientation sounds (biosonar pulses or, simply, pulses) and listens to their echoes for echolocation. The delay of an echo from the emitted pulse carries target-distance information. In the mustached bat, the biosonar pulse consists of a long constant frequency (CF) and a short frequency-modulated (FM) component, and these components consist of four harmonics: CF_1–4 and FM_1–4. The CF and FM components, respectively, are suited for bearing velocity information carried by Doppler shifts and distance information carried by echo delays from the emitted pulse. In the auditory cortex, there are three major types of FM–FM combination-sensitive neurons that are tuned to specific delays of echo FM_2–4 from a pulse FM₁. They are called FM₁–FM₂, FM₁–FM₃, and FM₁–FM₄. These FM–FM neurons systematically represent the target-distance information carried by an echo delay in two cortical areas: the frequency modulation–frequency modulation (FF)^† and dorsal fringe (DF) areas. Namely, in these areas, the best delays (BDs) for their excitation systematically vary along the cortical surface and form an echo delay axis that is up to ≈18 ms in the FF area and ≈8 ms in the DF area (Fig. 1) (8–10). Focal electric stimulation of the FF area augments surrounding FF neurons that match the stimulated neurons in BD and sharpen their delay tuning, whereas it slightly inhibits other surrounding neurons that are unmatched in BD and shifts their delay tuning and BDs (11). This BD shift occurs away from the BD of the stimulated FF neurons. Such a shift in tuning is called a “centrifugal” shift. On the other hand, a shift in tuning toward the tuning of stimulated neurons is called a “centripetal” shift (12). The DF area is at a higher hierarchical processing level than the FF area. The size and the echo-delay axis of the DF area are smaller and shorter than those of the FF area. These two delay-tuned areas are strongly interconnected (13). The role of these cortico–cortical interconnections in processing distance information has not yet been studied. Our current study explores how focal electric stimulation of the DF area modulated neurons in the FF area through the feedback projection, how stimulation of the FF area modulated neurons in the DF area through the feed-forward projection, and how stimulation of DF neurons modulated nearby DF neurons through the lateral projection. We found that focal electric stimulation of the DF area evoked centripetal BD shifts of “BD-unmatched” FF and DF neurons, whereas focal electric stimulation of the FF area evoked centrifugal BD shifts of BD-unmatched DF neurons. “BD-matched” FF neurons were facilitated, without a BD shift, by electric stimulation of the DF neurons.

Fig. 1.

Cortical auditory areas and the delay-tuned DF and FF areas. (A) Dorsolateral view of the left cerebral cortex of the mustached bat. 1, Doppler-shifted constant frequency (DSCF). 2, Ala (anterior division of the primary auditory cortex, AI). 3, Alp (posterior division of AI). 4, FF area. 5, CF/CF area. 6, DF area. The FF area was previously called the FM–FM area. (B) The FF and DF areas each consist of three subdivisions in terms of combination sensitivity: FM₁–FM₂, FM₁–FM₃, and FM₁–FM₄. The subscripts 1–4 indicate the first through fourth harmonics of FM components of biosonar pulses and echoes. The arrows indicate the delay axes, from short to long.

Results

Feedback Modulation of FF Neurons by DF Neurons.

In five unanesthetized bats, the delay-response curves of 23 cortical FF neurons were studied before and after the electric stimulation of ipsilateral cortical DF neurons. The combination sensitivity of the recorded FF neurons was either FM₁–FM₂ (19 neurons) or FM₁–FM₃ (4 neurons). Their BDs ranged from 2.0 to 14 ms (mean ± SE, 7.3 ± 0.7 ms, n = 23). The combination sensitivity of the stimulated DF neurons was either FM₁–FM₂ (16 neurons) or FM₁–FM₃ (7 neurons) (see Table 1). Their BDs ranged from 1 to 6 ms (4.3 ± 1.4 ms, n = 23). Of the 23 FF neurons, 22 had a BD that was different by >1.0 ms from that of the stimulated DF neurons. They were BD-unmatched neurons. The remaining FF neuron had a BD that was the same as that of the stimulated DF neurons. So it is referred to as a BD-matched neuron.

Table 1.

The combinations of recorded FF and stimulated DF neurons

Recorded FF neurons	Stimulated DF neurons
	FM1–FM2	FM1–FM3	FM1–FM4	Total
FM1–FM2	14	5	0	19
FM1–FM3	2	2	0	4
FM1–FM4	0	0	0	0
Total	16	7	0	23

Each number indicates the number of neurons studied, whose combination sensitivities are indicated by the recorded FM₁–FM_n and stimulated FM₁–FM_n neurons. n = 2–4.

The effects of the electric stimulation of the DF neurons on the FF neurons were different depending on the difference in BD between the recorded FF and stimulated DF neurons, not on the difference in FM–FM combination sensitivity between them (Table 1). Electric stimulation of DF neurons augmented the responses of a BD-matched FF neuron to paired FM sounds and did not shift its BD. In Fig. 2A, a FF neuron tuned to a 6.0-ms delay responded only slightly to a pulse FM₁ alone or an echo FM₂ alone, but strongly responded to a FM₁–FM₂ pair corresponding to the essential components of a pulse–echo pair. The responses to the FM₁–FM₂ pairs were much larger than the sum of the responses to FM₁ alone and FM₂ alone when the FM₂ in the pair delayed from the FM₁ by 3.0–9.0 ms. Therefore, these responses were facilitatory. The BD for the facilitatory response of the neuron was 6.0 ms (Fig. 2A1). The number of spikes in each poststimulus time (PST) histogram was plotted to obtain a delay-response curve (Fig. 2B, open circles). Electrical stimulation of DF neurons tuned to a 6.0-ms delay increased the response of the FF neuron at its BD by 74% (Fig. 2A2 and filled circles in Fig. 2B). The effect of the electric stimulation reverted back to the control condition ≈122 min after the onset of the electric stimulation (Fig. 2 A3 and open triangles in Fig. 2B).

Fig. 2.

Modulation of the delay-response curves of FF neurons by DF neurons. (A) Augmentation of the delay-dependent responses of a FF neuron by electric stimulation of BD-matched DF neurons. The arrays of PST histograms display the responses of the FF neuron to paired stimuli obtained before (A1), 62 min after (A2), and 122 min after (A3) electric stimulation (ES) of the DF neurons. The sound stimuli were pulse FM₁ (P) alone, echo FM₂ (E) alone, and P–E pairs with different echo (E) delays ranging from 0.0 to 19 ms. (B) The delay-response curves of the FF neuron obtained before (○, control), 62 min after (●), and 122 min after (▵, recovery) ES of the BD-matched DF neurons. Each data point indicates a mean and a standard error. FM₁: 28.5- to 22.5-kHz sweep at 40 dB sound pressure level (SPL). FM₂: 58.0- to 46.0-kHz sweep at 20 dB SPL. (C) The arrays of PST histograms display the responses of a FF neuron to paired stimuli obtained before (C1), 55 min after (C2), and 94 min after (C3) ES of the BD-unmatched DF neurons. (D) The arrays of PST histograms display the responses of a FF neuron to paired stimuli obtained before (D1), 91 min after (D2), and 134 min after (D3) ES of the BD-unmatched DF neurons. The combination sensitivities of the FF and DF neurons are shown in each graph. The BDs in the control condition are indicated by the open circles for the recorded neurons and by the arrows for the stimulated neurons.

The responses of the BD-unmatched FF neurons to pulse–echo pairs were inhibited or facilitated by the electric stimulation of the DF neurons depending on the echo delays. These delay-dependent changes in the responses caused the BD shifts of the FF neurons. For example, electric stimulation of 4.0-ms delay-tuned DF neurons decreased the response of a 10-ms delay-tuned FF neuron at a 10-ms delay and increased its response at a 6.0-ms delay (Fig. 2C2). Because of this delay-dependent inhibition and facilitation, the BD of the neuron shifted from 10 to 6.0 ms. That is, the BD shifted toward that of the stimulated DF neurons (a centripetal BD shift). These delay-dependent changes reverted back to the control condition 94 min after the electric stimulation (Fig. 2C3). When the BD of a FF neuron was 14 ms (Fig. 2D1) and the BD of stimulated DF neurons was 6.0 ms, the BD difference between them was large, 8.0 ms. Electrical stimulation of the DF neurons reduced the response of the FF neuron at a 14-ms delay, but increased it at a 6.0-ms delay, and the delay-response curve of the neuron became double-peaked (Fig. 2D2). Because the magnitude of the response at 14 ms was statistically insignificant from that at 6.0 ms, the neuron had two BDs. These changes disappeared ≈130 min after the electric stimulation and the delay-response curve recovered to the control (Fig. 2D3). The change in the response of this FF neuron is further explained later (Fig. 3D). BD representation covers values from 0.4 to ≈8 ms in the DF area (10) and from 0.4 to ≈18 ms in the FF area (8), so Fig. 2 C and D clearly shows that DF neurons modulated the delay tuning of FF neurons, although their long BDs were not represented in the DF area.

Fig. 3.

The BD shifts of four FF neurons evoked by electric stimulation of BD-unmatched DF neurons. Each graph shows three frequency-response curves: before (control, ○), 55–91 min after (●), and 94–134 min after (recovery, ▵) the electric stimulation. D Inset shows many action potentials that were superimposed.

Fig. 3 further demonstrates changes in the delay-response curves of four BD-unmatched FF neurons evoked by the electric stimulation of DF neurons: the BDs of two FF neurons were represented in the DF area, whereas those of the other two were not. Fig. 3A shows the delay-response curve of a FF neuron tuned to a 3.0-ms delay (open circles). Electric stimulation of the DF neurons tuned to a 5.0-ms delay decreased the response of the FF neuron at a 3.0-ms delay and increased it at a 5.0-ms delay. Accordingly, the BD of the FF neuron increased by 2.0 ms (Fig. 3A, filled circles). These changes disappeared ≈108 min after the electric stimulation (Fig. 3A, open triangles). Fig. 3B shows the delay-response curve of a FF neuron tuned to a 6.0-ms delay (open circles). Electric stimulation of the DF neurons tuned to a 3.0-ms delay decreased the response of the FF neuron at a 6.0-ms delay and increased it at a 3.0-ms delay. Accordingly, the BD of the FF neuron decreased by 3.0 ms (Fig. 3B, filled circles). These changes disappeared ≈120 min after the electric stimulation. The BD shift was frequently associated with a decrease in overall response, which is further documented later (Fig. 4B). Fig. 3C shows the delay-response curve of a FF neuron tuned to a 10-ms delay, a value not represented in the DF area (open circles). Electric stimulation of DF neurons tuned to a 4.0-ms delay shortened the BD of the FF neuron from 10 to 6.0 ms (Fig. 3C, filled circles). As shown in Fig. 3 A–C, the BD shift became either longer or shorter depending on whether the BDs of the recorded FF neurons were shorter or longer than the BDs of the stimulated DF neurons. That is, the BD shifts of the FF neurons were toward the BDs of the stimulated DF neurons (centripetal BD shifts). In 21 of the 22 BD-unmatched FF neurons studied, the BD shifts were centripetal. The change in the delay-response curve of the remaining FF neuron is described below (Fig. 3D). It was clear that focal electric stimulation of the DF area evoked delay-dependent inhibition and facilitation of the FF neurons even though the BDs of the FF neurons were much longer than the BDs of the stimulated DF neurons and were not represented in the DF area.

Fig. 4.

Changes in the BDs and responses of the FF neurons as a function of BD differences between the recorded FF and stimulated DF neurons or as a function of BD shifts. (A) BD shifts as a function of BD differences between the recorded FF and stimulated DF neurons. r, correlation coefficient. (B and C) Changes in the response magnitude as a function of BD differences between the recorded FF and stimulated DF neurons (B) and as a function of BD shifts (C). Filled circles show changes in the response magnitudes of 22 BD-unmatched FF neurons at their BDs in the control condition (BD_c) evoked by electric stimulation of DF neurons. Open circles show changes in their response magnitudes between the responses at the new, i.e., shifted BD (BD_s) and at the original, i.e., control BD (BD_c). The open triangles show the change in the response of the BD-matched FF neuron. The changes are expressed in percentage of the response at the BD_c before electric stimulation. Each data point indicates a mean and a standard error.

Fig. 3D shows the delay-response curve of a FF neuron tuned to a 14-ms delay (open circles). Electrical stimulation of DF neurons tuned to a 6.0-ms delay decreased the response of this FF neuron at a 14-ms delay, but increased it at a 6-ms delay. As a result, the delay-response curve of the FF neuron became double-peaked (Fig. 3D, filled circles; also see Fig. 2D). The response magnitude at 14 ms was the same as that at 6 ms. These changes disappeared ≈130 min after the electric stimulation (Fig. 3D, open triangles). One might consider that the recording of action potentials was not from a single neuron, but from two neurons. This possibility, however, was unlikely, because the waveforms of action potentials were consistently very similar to each other throughout the recording period (Fig. 3D Inset) and the delay-response curve returned to that in the control condition. Therefore, we concluded that the delay tuning of the stimulated DF neurons was added to that of the recorded FF neuron. Because this FF neuron became slightly insensitive to a delay >10.0 ms and became sensitive to the BD of the stimulated DF neurons, we may also conclude that the response properties of this FF neuron underwent a centripetal shift.

The amount of the BD shift varied as a function of the difference in the BD between the recorded FF and stimulated DF neurons (Fig. 4A). The BD shift became larger with a larger BD difference and reached its maximum (5.0 ms) at a 5.0-ms BD difference. The slope of the BD shift-difference curve between −2.0-ms and +5.0-ms BD differences was −0.87 (r = 0.98, n = 17). Therefore, the BDs of the BD-unmatched FF neurons became almost the same as those of the electrically stimulated DF neurons. The BD shift became smaller with an increase in the BD difference beyond 5.0 ms and was zero at an 8.0-ms BD difference. The slope of the BD shift-difference curve between 5.0- and 8.0-ms BD differences was 1.48 (r = 0.86, n = 7).

The centripetal BD shifts of the BD-unmatched FF neurons were always associated with changes in response magnitude as shown in Figs. 2 and 3. The responses at their BDs in the control condition decreased after the electric stimulation of DF neurons, as a function of the BD difference between the recorded FF and stimulated DF neurons (Fig. 4B, filled circles) and also as a function of the magnitude of the BD shift (Fig. 4C, filled circles). The larger the BD difference and the larger the BD shift, the larger the decrease in response. The responses at the shifted (i.e., new) BDs increased 138% on the average. The amount of the increase also depended on the amount of the BD difference and BD shift. The physiologically meaningful question was whether the increased responses at the shifted BDs were larger than the responses at the control (i.e., original) BDs. Therefore, in the 21 neurons showing a centripetal BD shift, the response at the shifted BD was compared with the response at the control BD. In addition, the change in the response at the control BD evoked by the electric stimulation was calculated for the BD-matched FF neuron that showed no BD shift. The curve in Fig. 4B (open circles) shows these percent changes in response as a function of BD differences. On the average, the responses at the shifted BDs were 10.3 ± 5.9% smaller than the responses at the control BDs, including the 74% increase at the zero BD difference.

Feed-Forward Modulation of DF Neurons by FF Neurons.

In four unanesthetized bats, the delay-response curves of 12 cortical DF neurons were studied before and after the electric stimulation of homolateral cortical FF neurons. The combination sensitivity of the recorded DF neurons was either FM₁–FM₂ (11 neurons) or FM₁–FM₄ (one neuron). Their BDs ranged from 1.0 to 6.0 ms (mean ± SE, 3.6 ± 0.4 ms, n = 12). The combination sensitivity of the stimulated FF neurons was either FM₁–FM₂ (six neurons) or FM₁–FM₃ (six neurons) (Table 2). Their BDs ranged from 2.0 to 8.0 ms (4.8 ± 0.5 ms, n = 12). All 12 DF neurons studied had a BD that was different from that of the stimulated FF neurons by >0.5 ms, i.e., the neurons were BD-unmatched.

Table 2.

The combinations of recorded DF and stimulated FF neurons

Recorded DF neurons	Stimulated FF neurons
	FM1–FM2	FM1–FM3	FM1–FM4	Total
FM1–FM2	6	5	0	11
FM1–FM3	0	0	0	0
FM1-FM4	0	1	0	1
Total	6	6	0	12

Each number indicates the number of neurons studied, whose combination sensitivities are indicated by the recorded FM₁-FM_n and stimulated FM₁-FM_n neurons. n = 2–4.

Electric stimulation of FF neurons inhibited or facilitated the responses of the BD-unmatched DF neurons to pulse–echo pairs depending on the echo delays and the relationship in the BD between the recorded DF and stimulated FF neurons. These delay-dependent response changes caused the BDs of the DF neurons to shift away from the BDs of the stimulated FF neurons. That is, the BD shifts were centrifugal. For example, electric stimulation of the FF neurons tuned to a 7.0-ms delay decreased the response of a DF neuron tuned to a 4.0-ms delay at its BD and increased it at a 2.0-ms delay (Fig. 5A2). Because of this delay-dependent inhibition and facilitation, the BD of the DF neuron shifted from 4.0 to 2.0 ms. These delay-dependent changes disappeared 107 min after the electric stimulation (Fig. 5A3). The mean recovery time of the BD shift was 107 ± 6.6 min (n = 12) for the DF neurons, which was similar to 114 ± 6.5 min (n = 22) for the FF neurons evoked by the DF stimulation (P = 0.48). Fig. 5B shows the change in the delay-response curve of the DF neuron evoked by electric stimulation of the FF neurons.

Fig. 5.

BD shift of a DF neuron evoked by electric stimulation of FF neurons. (A) PST histograms display the responses at the best delays of the DF neuron in the control (BD_c) and shifted (BD_s) conditions. (A1) The responses recorded before (control) the electric stimulation. (A2) The responses recorded 55 min after the electric stimulation. (A3) The responses recorded 107 min after the electric stimulation. (B) Delay-response curves of the DF neuron obtained in the above three conditions. See Fig. 2B legend for symbols.

The BD shifts of the 12 DF neurons are plotted as a function of the BD differences between the recorded DF and stimulated FF neurons (Fig. 6, open circles). These BD shifts were dramatically different from those of the FF neurons evoked by the electric stimulation of the DF neurons shown in Fig. 4A (see also Fig. 6, filled circles). That is, the direction and amount of the BD shifts are centrifugal and small for the DF neurons, but centripetal and large for the FF neurons. Electric stimulation of the FF neurons evokes the centrifugal BD shifts of FF neurons surrounding the stimulated neurons (11). The BD shifts of those FF neurons, shown by the BD shift-difference curve in Fig. 6, were identical in amount and direction to the BD shifts of DF neurons evoked by electric stimulation of FF neurons.

Fig. 6.

Difference between the BD shifts evoked by the FF–DF feed–forward (○) and DF–FF feedback (●) cortico–cortical projections. The BD shifts are plotted as a function of BD differences between the recorded (DF_r or FF_r) and electrically stimulated neurons (DF_s or FF_s). The BD shifts of DF neurons evoked by electric stimulation of ipsilateral DF neurons, i.e., by the DF–DF lateral projection, are indicated by ▵. The BD shifts of FF neurons evoked by electric stimulation of ipsilateral FF neurons, i.e., by the FF–FF lateral projection, are also shown by the curve that was obtained by Xiao and Suga (11).

Both the DF–FF feedback modulation (Table 1) and the FF–DF feed-forward modulation (Table 2) occurred regardless of the differences in the types of FM–FM combination sensitivity between the recorded and stimulated neurons. That is, the feedback modulation was always centripetal for the FF neurons, whereas the feed-forward modulation was always centrifugal for the DF neurons.

Lateral Modulation of DF Neurons by DF Neurons.

Because electric stimulation of FF neurons evokes the centrifugal BD shifts of the BD-unmatched FF neurons surrounding the stimulated neurons (11), we also examined whether electric stimulation of DF neurons evoked the centrifugal or centripetal BD shifts of the DF neurons surrounding them. All six DF neurons studied in three mustached bats showed a centripetal BD shift. These centripetal BD shifts were the same in amount as those of the FF neurons evoked by electric stimulation of the DF neurons (Fig. 6, open triangles). The slope of the BD shift-difference curve for these six DF neurons was −0.86 (r = 0.98), which was the same as that (−0.87, r = 0.98) for the DF–FF feedback modulation (P > 0.8).

Discussion

Functions of the Feed-Forward and Feedback Projections Between the FF and DF Areas.

The DF area is moderately or strongly connected with the perirhinal and retrosplenial cortices, but the FF area is not or, at best, is weakly connected (13). The response latencies of DF neurons are longer than those of FF neurons (10). These anatomical and neurophysiological data suggest that the DF area is at a higher hierarchical level than the FF area. The delay axis starting from 0.4 ms extends up to ≈18 ms in the FF area and ≈8 ms in the DF area. Short delays, i.e., short target distances, are therefore represented in these multiple areas. The functional role of this multiple representation may be hypothesized as follows: the processing of distance information during target-directed flight becomes more critical at shorter target distances and more neurons with short BDs are recruited across the different areas. The response properties of DF neurons are similar to those of FF neurons, except that they show larger trial-by-trial fluctuation in response magnitude than FF neurons (10). The functional difference between the FF and DF areas remains to be further studied.

Focal electric stimulation of the FF area evokes centrifugal BD shifts of DF neurons (our current data) and nearby FF neurons (11) as well as collicular FM–FM neurons (14), whereas focal electric stimulation of the DF area evokes centripetal BD shifts of FF neurons and nearby DF neurons. The centrifugal BD shifts shape the highly selective neural representation of a specific target distance, whereas the centripetal BD shifts result in the expanded representation of the selected specific distance. The large BD shifts of FF neurons evoked by the DF stimulation and the steep slope of the BD shift-difference curve of FF neurons (Fig. 6) indicate that the excitation of the DF neurons tuned to a given delay can dramatically increase the number of FF neurons tuned to that same delay. Therefore, we hypothesize that the DF–FF feedback projection plays a role in focusing the neural processing on target information at a specific distance.

The BD shifts evoked by cortical electric stimulation were large and long-lasting. In the natural condition, the BD shifts are presumably small and short-lasting. However, it is unlikely that they occur every time when the bat hears an echo after a biosonar pulse emission. An echo delay systematically shortens during target-directed flight, but may not change much during exploration of nearby objects by the bat roosting in a cave. In such a situation, the FF area may play a role in selecting a specific echo delay and the DF area may play a role in focusing on that specific echo delay. The long recovery time of the BF shifts evoked by cortical electric stimulation may suggest that the cortico–cortical modulation is a basic neural mechanism for adaptation of the auditory cortex for optimal processing of auditory signals that frequently stimulate an animal and are behaviorally important to the animal.

Cortico–Cortical Interactions in Cortical Areas Other than the DF and FF Areas.

The cortico–cortical interaction within a specific cortical area has been studied in the primary auditory cortex [rat (15), big brown bat (3), mongolian gerbil (5)], the primary visual cortex [cat (16, 17)], and the primary somatosensory cortex [rat and monkey (18), human (19)]. In all of these cortices and animal species, focal cortical stimulation changes the receptive fields (tuning curves) of neurons surrounding the stimulated neurons toward the receptive fields of the stimulated neurons. Such centripetal tuning shifts are much more common than the centrifugal tuning shifts. Therefore, one may expect that both the feedback and feed-forward connections between cortical auditory areas in non-mustached bat species evoke centripetal tuning shifts for the expanded representation, without highly selective reorganization of the auditory cortex, of the auditory signal that is frequently heard by an animal and is important for the animal.

In the mustached bat, focal electric stimulation of the posterior division of the primary auditory cortex, which is not specialized for processing biosonar signals, evokes centripetal best frequency (BF) shifts of nearby BF-unmatched neurons (4). However, focal electric stimulation of the Doppler-shifted CF (DSCF) area (20) evokes centrifugal BF shifts of nearby BF-unmatched DSCF neurons. The DSCF area is highly specialized for the fine analysis of biosonar signals in the frequency domain (21). Therefore, cortico–cortical interactions in the DSCF area, as those in the FF area, are unique. An antagonist of GABA_A receptors applied to the DSCF area changes centrifugal BF shifts to centripetal BF shifts (20). Likewise, the centrifugal BD shifts of FF neurons change to the centripetal BD shifts when the GABA_A receptor antagonist is applied to the FF area (11). An agonist of GABA_A receptors applied to the auditory cortex of the big brown bat changes centripetal BF shifts to centrifugal BF shifts (22). Therefore, the highly specialized and less specialized cortical auditory areas share basically the same neural circuit, and the uniqueness of the FF and DSCF areas derives from enhanced inhibition within these cortical areas.

BF-matched neurons in the DSCF area and the posterior division of the primary auditory cortex (4) and BD-matched neurons in the FF area (Fig. 2 A and B) are facilitated by electric stimulation of the cortical neurons corresponding to them. Such facilitation is common regardless of whether the cortical auditory areas are specialized or less specialized and whether they are specialized for processing acoustic signals in the frequency or time domain (23).

When a difference in BF between the stimulated and recorded cortical or collicular neurons is large, the recorded neuron shows an additional peak in its frequency-tuning curve at the frequency equal to the BF of the stimulated neurons [rat (15), house mouse (24)]. In our current experiments, when a BD difference was large between the recorded FF and stimulated DF neurons, the FF neurons showed an additional peak in their delay-tuning curve at the delay equal to the BD of the stimulated neurons. Therefore, the same phenomenon was found in both the frequency and time domains and in both the highly specialized and less specialized auditory cortices.

Materials and Methods

General.

Surgery, acoustic and electric stimulation, and recording of neural activity were the same as those described (11, 25). The protocol for the present research was approved by the Animal Studies Committee of Washington University.

Eleven adult mustached bats (P. parnellii rubiginosus) from Trinidad were used. Under neuroleptanalgesia (Innovar 4.08 mg/kg body weight), a 1.5 cm-long metal post was glued on the dorsal surface of the bat's skull. A local anesthetic (Lidocaine HCl) and antibiotic ointment (Furacin) were applied to the surgical wound. Three to 4 days after surgery, the awake animal was placed in a polyethylene-foam body mold, which was hung with an elastic band at the center of a 31°C soundproof room. The metal post glued on the skull was attached to a metal rod with set screws to immobilize the animal's head, which was adjusted directly toward the loudspeakers located 74 cm away. A few holes (50–100 μm in diameter) were made in the skull covering the FF and DF areas of the right or left auditory cortex. A pair of tungsten-wire electrodes (≈7 μm in tip diameter, ≈35 μm apart, one proximal to the other) was orthogonally inserted 500–700 μm deep into the DF area through one of the holes. The responses (action potentials) of DF neurons to paired FM sounds were recorded, and their BDs were measured. Then, this electrode pair was used to electrically stimulate the neurons. A single glass micropipette electrode (≈1 μm in tip diameter) was inserted into the FF area homolateral to the DF area to examine the effect of the electric stimulation of the DF neurons on a FF neuron. To study the effect of the electric stimulation of FF neurons on a DF neuron, the DF neuron was recorded with a glass micropipette electrode, and the FF neurons were electrically stimulated with the paired tungsten-wire electrodes.

Acoustic Stimulation.

The mustached bat emits orientation sounds (biosonar pulses or, simply, pulses). Each pulse consists of CF and FM components, which contain four harmonics. Therefore, each pulse contains eight components: CF_1–4 and FM_1–4. CF₁ was ≈29 kHz. In FM₁, the frequency swept from ≈29 to ≈23 kHz. FF and DF neurons are tuned to a combination of the FM₁ of the pulse stimulus and the FM_n (n = 2, 3, or 4) of an echo stimulus with a specific time delay from the pulse stimulus. In both the FF and DF areas, different types of FM–FM neurons (FM₁–FM₂, FM₁–FM₃, and FM₁–FM₄) are separately clustered (Fig. 1B) (8, 9). Therefore, acoustic stimuli delivered to the animal were FM₁–FM_n pairs. Each FM sound was 3 ms long, including a 0.5-ms rise-delay time.

To generate a FM sound, a voltage-controlled oscillator (Wavetek 134) was frequency-modulated with a linear voltage-ramp generator to mimic the FM components in the species-specific biosonar pulse. The frequency modulation and the amplitude of each FM sound and the time interval (echo delay) between paired FM sounds were first manually varied to identify the best combination of two FM sounds (FM₁–FM_n) and the BD to excite a given neuron. Then, the FM₁ sound (pulse stimulus) was fixed at the frequency sweep and amplitude to evoke the largest facilitation. The FM_n sound (echo stimulus) was fixed at the frequency sweep to evoke the largest facilitation and at 10 dB above the minimum threshold, i.e., the threshold of the response at the best FM_n. Then, the FM_n delay from the FM₁ was varied with a computer and other hardware (Tucker-Davis Technologies) to obtain a delay-response curve. The computer-controlled delay scan consisted of 22 150-ms time blocks of pulse only, echo only, and 20 pulse-echo pairs in which the echo delay randomly varied between 0 and 19 ms in 1.0-ms steps. An identical delay scan was delivered 50 times to obtain an array of PST or PST cumulative histograms displaying the responses of a single neuron.

Electric Stimulation of Cortical FF and DF Neurons.

Electric stimulation was a monophasic electric pulse (0.2 ms long, 100 nA) delivered at a rate of 5/s for 7 min with a constant current stimulator (WPI; modified model A360). Such electric pulses delivered at a low rate evoke changes in the cortical, thalamic, and collicular FM–FM (14, 26) and DSCF neurons (6, 27), but do not evoke any noticeable change in the cochlear microphonic responses (6). These electric pulses were estimated to activate neurons within a 60-μm radius in the plane orthogonal to the cortical columns (14).

Data Acquisition.

Action potentials of a single FF or DF neuron tuned to a pair of FM₁–FM_n sounds with a specific FM_n delay from FM₁ were selected with time amplitude–window discrimination software (Tucker-Davis Technologies). At the beginning of data acquisition, the waveform of an action potential was stored and displayed on the monitor screen. This action potential (i.e., template) was compared with other action potentials obtained during data acquisition. The response of the single cortical neuron to a delay scan delivered 50 times was recorded before and after electric stimulation and was displayed as an array of PST or PST cumulative histograms. The data were stored on the computer hard drive and used for off-line analysis. In a 1-day experiment, one or two neurons were studied for the effect of and recovery from electric stimulation.

Off-Line Data Processing.

The magnitude of auditory responses of a neuron was expressed by the number of spikes per stimulus (50 identical stimuli) and plotted as a function of echo delays. The BD shift evoked by the cortical electric stimulation was considered significant if it reverted back to the BD in the control condition and if the auditory responses changed by the electric stimulation recovered by >85%. A t test was used to examine the significance of the difference between the auditory responses obtained before and after the electric stimulation.