5 General discussion and outlook
5.2 Processing of concurrent sounds by neurons sensitive to interaural time
5.2.1 Localization of concurrent sounds
Psychophysical studies, investigating the localization of tones in background noise, showed a decline of accurate discrimination with increasing noise levels (Stern et al. 1983; Good and Gilkey 1996). These findings are consistent with the single cell responses shown in the study described in Chapter 3. Here I could show that DNLL neurons, which were sensitive to ITDs of tones, lost this sensitivity if white noise as concurrent background was presented. This decrease of the ITD sensitivity mostly resulted from a decrease of the response to tones at favorable ITDs. Interestingly, the concurrent noise only marginally increased the response to tones at unfavorable ITDs. Thus, minimal or no response to tones at unfavorable ITD was achieved, even if concurrent noise alone elicited responses. The tone at unfavorable ITDs suppressed the response to the noise and thereby assured the robustness of ITD sensitivity to tones in background noise (see chapter 3). This finding is of particular interest, since most studies about ITD coding focused so far on the maximal response at favorable ITD. Based on these results, I suggest that neuronal ITD sensitivity is not exclusively encoded by the maximal response to favorable ITD but also by the minimal response to unfavorable ITDs. This hypothesis is in contrast to the place-code model, which suggests, that ITDs are exclusively encoded by a maximal response to a certain ITD
The presented data are insufficient to draw conclusions about the neuronal mechanism. However, they suggest that the suppressive effects of tones at unfavorable ITDs occur at the level of the coincidence-detector. Such suppressive effects of tones at unfavorable ITDs, albeit less pronounced, had already been observed in the MSO years ago by Goldberg and Brown (1969). They found that the binaural responses to tones at unfavorable ITDs could suppress the responses to monaural stimuli. Two different hypotheses exist to explain this suppression. The first hypothesis formulated by Brand et al. (2002) suggested that inhibitory inputs suppress the response at unfavorable ITDs. Inhibition and excitation arriving at the coincidence-detector are phase-locked to the stimulus waveform. At a particular unfavorable ITD, inhibitory and excitatory inputs are
109
on top of each other and the inhibition suppresses the response (Brand et al. 2002; see chapter 1.3). In contrast, a second hypothesis postulated that a simple coincidence model without inhibition is adequate to describe suppressive effects of tones at unfavorable ITDs (Colburn et al. 1990). Using this model with two excitatory inputs, Colburn et al. explained the suppressive effects with the following three assumptions: (i) coincidence- detector neurons are spontaneously innervated; (ii) acoustic stimulation induces synchronization of the response and (iii) an absolute neuronal refractory time follows after a response. They found that the synchronized response led to a synchronized “no- response”, which was below the spontaneous activity of the neuron. This resulted in a lack of coincidences, or suppressive effects of the binaural response, when the synchronized excitatory inputs were out of phase and the synchronized “no-responses” were in phase (Colburn et al. 1990). Additional evidence comes from in vitro patch clamp studies in the MSO equivalent of birds, the nucleus laminaris, by Reyes et al. (1996), showing that two out of phase stimulations evoked less neuronal firing than a single stimulation. The authors suggested that the modulation of the discharge rate, at least in birds, did not critically depend on inhibitory inputs but rather on membrane properties of the coincidence-detector neurons. Subsequent experiments showed that these membrane properties of neurons in the nucleus laminaris are strongly modulated by inhibition (Brückner and Hyson 1998; Monsivais et al. 2000). However, several studies showed that mammals and birds probably developed different mechanisms to encode ITDs (for review: Grothe 2003). Now, both mechanisms described above might contribute to the suppressive effects of tones at unfavorable ITD. On the basis of the experiments presented here, neither hypothesis can be favored. Further studies, e.g. using patch clamp experiments in mammals, are necessary to clarify this issue.
While extra-cellular single cell recordings measure the result of the neuronal processing, the action potentials of a neuron, the underlying cellular mechanisms can only be hypothesized. In contrast to this, patch clamp studies allow a detailed study of the cellular mechanisms. However, most patch clamp studies are performed in vitro in acute brain slices or cell cultures and are restricted to “unphysiological” stimulations of the neuronal circuitry within this brain slice. Furthermore, spontaneous activity is mostly absent in neurons in brain slices. This absence of spontaneous activity may influence the synaptic transmission of auditory brain neurons as recently shown by Hermann et al. (2007). They
110
reintroduced in vivo-like spontaneous activity to in vitro brainstem synapses and found that the steady-state amplitude of the excitatory postsynaptic currents and, accordingly, the excitability of the neurons were reduced. As spontaneous active inputs are also one of the important feature of the coincidence-detector model described by Colburn et al. (1990), a combined approach such as in vivo patch clamp studies would be useful as have been performed for example more than 10 years ago by Casseday et al. (1994). This method combines the advantages of both methods and permits in vivo investigation of cellular mechanism, thus enabling a test of the hypothesized sub-threshold excitatory or inhibitory currents.
Investigating the general role of coincidence-detector neurons in the processing of concurrent sounds by using tones and background noise was experimentally challenging. First, the noise activated strong inhibitory interactions across frequency, which most likely did not involve the coincidence-detection. Second, the effective intensities of these two sound sources were not easily comparable (see chapter 3). Based on these two findings, two pure tones instead of noise and tone as concurrent sound sources were used in subsequent experiments. The rational of such an approach is the possibility to compare the effective levels of the two sound sources and to diminish cross-frequency interactions. Preliminary data using two tones with different frequencies showed that neurons could be sensitive to ITDs of both tones if the difference between the effective levels of the tones was below 20 dB. If this difference is increased, the neurons were only sensitive to ITDs of the tone with the higher intensity. This simple finding opens the field for many further investigations. In the next part I will focus on one possible experiment.
ITDs are encoded by a population of neurons (Jeffress 1948; Fitzpatrick et al. 1997; Hancock and Delgutte 2004; Harper and McAlpine 2004; Stecker et al. 2005). Therefore, an increase in the number of ITD-sensitive neurons could increase the precision of sound localization. How many ITD-sensitive neurons are necessary to achieve the psychophysical precision in detecting just noticeable differences in ITD? As mentioned in the introduction, Skottun and colleagues hypothesized that this precise detection of ITDs could be accomplished by single ITD-sensitive neurons (Skottun et al. 2001; Shackleton et al. 2003). A combined electrophysiological and psychophysical study analyzing the ITD sensitivity to tones with an increasing masking level of another concurrent tone
111
could test this hypothesis. This experimental approach would take advantage of the finding that the psychophysical accuracy in detecting just noticeable differences in ITD and the ITD sensitivity in the DNLL decrease with masking sounds. Thus, comparing the psychophysical accuracy with the ITD sensitivity in the DNLL could give an estimate how many DNLL neurons are needed for the precise encoding of ITDs.