The Responses of Single Units to Simple and Complex Sounds from the Superior Olivary Complex of the Guinea Pig

Summary We have recorded the responses of single units in the superior olivary complex (SOC) of the guinea pig to simple and complex stimuli. We can readily identify the responses of the four principal ascending nuclei as described in other species and we are therefore conﬁdent that the guinea pig is a suitable model for studying this region. We found single units in the medial superior olive that represent both diotic iterated rippled noise and dichotic repetition pitch in their temporal discharge patterns. Our data demonstrate a use of the SOC beyond sound-source localisation.


Introduction
The superior olivary complexconsists of agroup of auditory nuclei in the mammalian brainstem.Twoofthe largest nuclei, the medial superior olive (MSO)a nd lateral superior olive (LSO), receive bilateral, direct and/or indirect inputs, from the bushycells of the ventral cochlear nucleus on both sides and are thus major sites of binaural integration.Not surprisingly,previous physiological studies have examined the role of the MSO and LSO in sound localisation.Here we examine, for the first time, the responses of single units in the guinea pig SOC to simple and complex sounds, including stimuli that can evokeab inaural pitch in humans.
Human psychoacoustic studies have demonstrated striking "binaural pitch" phenomena.These pitch percepts are produced by sounds that, presented to either ear on its own, evoked no pitch [1,2].One such example is dichotic repetition pitch (DRP), consisting of Gaussian noise with a small inter-aural time difference (ITD) [ 1].Neural processing to combine information from the twoears is essential for this aspect of pitch perception.Here, we demonstrate neural correlates of binaural pitch in the temporal discharge patterns of single units in the MSO.The MSO is well known to perform inter-aural computations that represent inter-aural time differences.Analysis of the temporal discharge patterns of MSO units to diotic iterated rippled noise, and dichotic repetition pitch reveal an eural correlate of the pitch of these stimuli.These findings suggest a role for the MSO in the neural machinery underlying binaural pitch.

Methods
Our methods are described in detail elsewhere [3], so only abrief description is givenhere.Data were obtained from normal-hearing (based on round-windowC AP thresholds), anaesthetised and normothermic, pigmented guineapigs (Cavia porcellus).Experiments were performed in accordance with the Animals (ScientificP rocedures)A ct 1986 (Amendment Regulations 2012)f ollowing ethical reviewb yt he University of Cambridge Animal Welfare and Ethical ReviewBody.
An estimate of inter-aural crosstalk attenuation wasobtained from single ventral cochlear nucleus units by measuring the difference between monaurally-evokedspiking responses to BF tones presented ipsi-laterally and contralaterally.Between 0.10-and 2.00-kHz, the attenuation was greater than 50 dB.This, combined with our relatively low signal levels, means that the results are unlikely to be contaminated by inter-aural acoustic crosstalk.
Glass-insulated tungsten microelectrodes were positioned at the surface of the dorsal cochlear nucleus (DCN) under operating-microscope control, and advanced parasagittally through the brainstem at 45 • to the horizontal plane using ah ydraulic microdrive.S ignals from the microelectrode were amplified (×1000)and bandpass filtered (0.3-10 kHz)b efore online spiked iscrimination and timing.Spiketimes were stored for offline analysis (10µsresolution).Superior olivary complexu nits were defined as those found at depths greater than 3.3 mm from the surface of the DCN.The presence of aneurophonic in response to monaural or binaural stimulation with low-frequencypure tones gave an initial indication that the electrode wasclose to the MSO or the lowfrequencylimb of the LSO.Receptive fields were measured using 50-ms pure tones, gated with 5-ms cos2 ramps.Forneurons with BFs <5kHz, the frequencyrange was3octavesbelowunit BF to 2-octaves above,f or units with BFs 5kHz or above,t one frequencies ranged from 2o ctavesb elowB Ft o1o ctave above.Forb inaural neurons, receptive fields were measured to ipsilateral-only,contralateral-only and diotic tone presentations.
We collected spiket imes in response to several pitch stimuli.These included: diotic iterated rippled noise (Gaussian noise delayed and added back to itself [comb filtering], identical in both ears), and dichotic repetition pitch (Gaussian noise with as mall ITD between the two ears).

Results
Recordings were made from all main divisions of the SOC: the superior paraolivary nucleus (SPN), the medial nucleus of the trapezoid body (MNTB), the lateral superior olive (LSO)a nd the medial superior olive (MSO).Physiological response properties of the principal cells of these nuclei have been characterised in other species.However, to our knowledge, responses from such units in the guinea pig SOC have not been previously reported.Responses to pure tones at unit BF,p lotted as post-stimulus time histograms (PSTHs), were used to form an initial classification.MNTB units showed a"primary-like" (PL) PSTHs hape or a" primary-likew ith notch" (PN) shape, and the characteristic three-component spike-waveform shape reported in other species (Figure 1A).SPN units showed chopper-liker esponses at stimulus offset (Figure 1B), and sensitivity to gaps between tones.We also confirmed interaural phase difference (IPD)-sensitivity in low-BF binaurally-responsive cells of the LSO and MSO, using binaural-beat stimuli (Figures 1C and 1D).MSO units responded weakly to monaural stimulation butmore strongly to diotic stimulation, with manyu nits showing binaural facilitation.LSO units were sensitive to interaural leveldifferences of BF tones (Figure 1F).
The microelectrode recording of an MSO unit to stimulation with binaural beats is shown in Figure 2A.This shows three clusters of spikes during the stimulus presentation, corresponding to the best IPD in each 1s econd beat period.Figure 2B shows an expanded segment of the same recording to showt he presence of the driven neurophonic, readily identifiable as the approximately sinusoidal oscillation of the background noise.The oscil- lation frequencyw as quantified by an FFT of the electrode trace for twotime-windows.The first wasinthe presence of the drivenneurophonic and the second wasinthe later 3s econds of the recording, when the stimulus was switched off (the spontaneous neurophonic -Figure 2C). Figure 2D shows aclear peak in the spectrum of the waveform around the stimulus frequencyo ft he binaural beat stimulus 333/334 Hz in the drivenn europhonic condition The spectrum shown by the black line is the electrode noise recorded when the stimulus is off (see scale in A).Facilitation in response to diotic tone bursts for twoMSO units is shown in the lower row.In (E) the monaural receptive fields are non-existent while the diotic receptive field (BF = 0.46 kHz)shows conventional tuning.Forcomparison, the continuous blue line is the frequency-threshold curve of asimilar BF filter recorded from asingle unit in the ventral cochlear nucleus of the guinea pig.In (F)w eshowaunit that showed weak excitatory responses in the monaural receptive fields.BF wasestimated as 0.74 kHz.
The facilitation to diotic tones is illustrated in Figures 2E and Fwhich showipsi-lateral, contralateral and diotic receptive fields for twoM SO units.In the first unit (Fig- ure 2E)t he monaural response, both ipsilateral and contralateral, is very weak or non-existent [0, 0] while the diotic response shows aconventional "V" shaped response.This type of response wasobserved for 29% of our MSO unit population.In contrast, the second unit (Figure 2F) shows atuned response to monaural stimulation of either ear [E, E], with as tronger response to diotic stimulation.This [E, E] pattern wasfound for 41% of our MSO units.This percentage is lower than seen in the dog (65%, [3]), cat (58%, [4])and the gerbil (67% [5]).
Figure 3s hows the responses of as ingle MSO unit to diotic iterated rippled noise (IRN)a nd dichotic repetition pitch (DRP).The IRN wasgenerated for one iteration with positive gain at three delays, 8, 16 and 32 ms.There is a clear peak in the all-order inter-spikei nterval histogram for each of the delays.It is important to note that the IRN wasp resented diotically,a nd the neural representation of the corresponding pitch is available in monaural neuralinputs.In contrast, DRP is generated by presenting broadband noise to one ear and the same noise, delayed by τ ms, to the other ear.DRP is adichotic pitch, requiring binaural interaction.
Humans do not report the pitch corresponding to 1/τ to be lateralized; it is heard in the centre of the head.If both the delayed and un-delayed noise waveforms are in phase (DRP+), the pitch is unambiguous at 1/τ.W hen a broadband phase shift of τ is applied to the delayed noise (DRP-),t he DRP is ambiguous, being either ∼ 0.9/τ or ∼ 1.1/τ [6].
This result is very similar to that of the more familiar monaural repetition pitch, that is "[iterated] rippled noise" [6].In Figure 3B, aclear peak can be seen (red asterisks) in the all-order inter-spikeinterval histograms in response to the same three delays shown in Figure 3A.
It should be noted that only 3o f1 1M SO units in our dataset showed aclear correlate of DRP in their inter-spike interval distributions.In contrast, all of our MSO units showed aclear neural correlate of the pitch of diotic IRN. Figure 4s hows an example of an MSO unit with ac lear temporal representation of the pitch of IRN with 16 iterations, buto nly aw eak representation of the pitch of IRN with 1iteration.This could be interpreted as reflecting the salience of IRN pitch.However, this same MSO unit fails to showresponses at the delay in the DRP stimulus (4C).

Discussion
We have recorded the responses of single units in the superior olivary complex( SOC)o ft he guinea pig to simple and complexs timuli.We can readily identify the responses of the four principal ascending nuclei as described in other species and we are therefore confident that the guinea pig is as uitable model for studying this region.We also demonstrated that single units in the MSO can respond to the pitch of diotic IRN, even with as ingle iteration.Humans only hear aw eak pitch with one iteration.As mall subset (3/11)o ft he same units showed a neural correlate to the pitch of DRP.A lthough DRP produces arather weak pitch percept, it is currently unknown whyo nly as ubset of MSO units responded to DRP.O ne possibility is am ismatch in best frequencyo ft he ipsilateral and contralateral input filters to the MSO unit.Future studies should consider measuring the responses of single SOC units to sounds which are capable of producing much stronger binaural pitch percepts, such as dichotic complex tones or Huggins pitch.

Figure 1 .
Figure 1.Physiological signatures of the main cell types in the principal nuclei of the superior olivary complex.A&B ) post-stimulus time histograms in response to supra-threshold BF tone bursts from the MNTB (BF = 12.5 kHz)a nd SPN( BF = 15.64 kHz), respectively.B elowe ach histogram is the corresponding dot-raster plot.The horizontal bar above each histogram represents the stimulus duration.C) Response of an MSO unit (BF = 0.13 kHz)t ob inaural beats at BF and E) an exemplar composite delay curve.D) Responses of an LSO unit (BF = 0.39 kHz)t ob inaural beats at BF and F) an exemplar ILD function.Average spikewaveform shapes are shown as insets for plots A-D.Note the signature three-component shape for MNTB units (inset in A).

Figure 2 .
Figure 2. (A) Am icroelectrode trace showing as ingle MSO unit responding with ab urst of action potentials at the 1s period of a binaural beat stimulus.The duration of the beat stimulus (3 seconds)isshown by the black bar.The drivenneurophonic is clearly seen as an amplitude modulation in the second trace [B] which is an expanded section of the waveform in A. The spontaneous neurophonic (inthe absence of controlled acoustic stimulation)i sshown in C. The frequencyofthe neurophonic potential is shown in [D] by the peak just below350 Hz in the spectra of the traces shown in Band C. The carrier frequencies of the binaural beat were 333 and 334 Hz.The spectrum shown by the black line is the electrode noise recorded when the stimulus is off (see scale in A).Facilitation in response to diotic tone bursts for twoMSO units is shown in the lower row.In (E) the monaural receptive fields are non-existent while the diotic receptive field (BF = 0.46 kHz)shows conventional tuning.Forcomparison, the continuous blue line is the frequency-threshold curve of asimilar BF filter recorded from asingle unit in the ventral cochlear nucleus of the guinea pig.In (F)w eshowaunit that showed weak excitatory responses in the monaural receptive fields.BF wasestimated as 0.74 kHz.

Figure 3 .
Figure 3. (A) The responses of as ingle MSO unit (BF = 0.25 kHz)t oI RN with 1i teration presented diotically for three different IRN delays (but zero ITD).In (B),wesee the responses of the same unit to DRP with the same three delays presented as ITDs.The all-order interspikeinterval histograms are plotted with abinwidth of 0.2 ms.The asterisks mark the stimulus delay.

Figure 4 .
Figure 4. Responses of as ingle MSO unit (BF = 0.13 kHz)t o 1i teration of diotic iterated ripple noise at ad elay of 8ms and either 1or16iterations for both positive and negative gain.The lack of response by the same cell to positive and negative DRP is also shown (C).The all-order interspikeinterval histograms are plotted with abinwidth of 0.2 ms.