Alan R. Palmer
MRC Institute of Hearing Research
Nottingham NG7 2RD
Joint work with D. McAlpine of the Department Biomedical Science, University of Sheffield, S10 2TN.
At frequencies below 1500 Hz, the difference in the phase of the sound waveform arriving at the ears (the interaural phase difference: IPD) provides a cue for localisation. This cue is transformed into changes in neural firing rate by cells in the medial superior olive (MSO; Yin, T.C.T and Chan, J.K. J. Neurophysiol., 64, 465-488., 1990). MSO cells fire when the inputs from the two ears arrive coincidentally and the IPD to which they are most sensitive is determined by the difference in the length of the transmission path of the neural activity from each ear. Cells in the MSO target nucleus, the inferior colliculus (IC), largely reflect the IPD sensitivities of the MSO, but with subtle variations most likely as a result of convergent input (McAlpine et al, J. Neurosci., 18, 6026-6039, 1998). When sound sources move they generate dynamically-changing IPDs that convey both the direction and velocity of the motion. MSO cells appear to be insensitive to either the direction or velocity of moving sources (Spitzer, M.W. and Semple, M.N. J. Neurophysiol., 80, 3062-3076, 1998), whereas cells in the IC can show very different firing rates to the same IPD depending on the direction and past history of the stimulation (Spitzer and Semple, J. Neurophysiol., 69, 1245-1263, 1993).
Two hypotheses for motion sensitivity are: (i) separate detectors tuned to respond selectively to moving stimuli and (ii) sensitivity to motion as a result of sequential responses to instantaneous IPDs across a population of IPD sensitive cells (the snapshot hypothesis). To investigate these hypotheses and the mechanisms underlying the emergent sensitivity to virtual sound motion, we have made extracellular recordings from cells in the IC of anaesthetised guinea pigs, using closed-field sounds which human listeners perceive as moving.
Best-frequency tones were presented to one ear while the phase of a tone, at the same frequency, in the other ear was sinusoidally modulated (interaural phase modulation: IPM). At low IPM rates only a sub-set of cells were sensitive to the modulation direction and for these we have independent evidence of convergence from lower brainstem levels. At constant IPM depth, as the rate increased, the IPDs to which a cell best responded in the forward and reverse directions, progressively separated. A plot of the IPD evoking maximum response against the modulation rate showed that the phase delay was close to the neural first-spike latency. Thus directional sensitivity in the measured responses at higher IPM rates is explicable in terms of the transmission time to the IC cell (Palmer et al. In: Psychophysical and physiological advances in hearing. Eds. Palmer et al., 368-375, 1998) and does not represent any specialisation for detecting moving stimuli. Sinusoidal IPM produces a non-constant motion velocity. To separate the effects of motion extent and velocity we have used linear IPM at a variety of depths and centres to probe the mechanisms that contribute to the motion sensitivity. With linear IPM, the response magnitudes to each direction of motion depended on the IPD around which the phase was modulated, and also decreased with decreasing depth of modulation. Cells that showed the greatest decrease in discharge rate as depth was decreased around the most effective IPD also showed greatest variation in response as the centre of the motion was progressively changed. Those conditions that resulted in more prolonged stimulation with the most effective IPD caused the largest firing rate reductions. This suggests that adaptation of excitation and hence the available recovery time may account for the directional sensitivity.
Recently we have obtained preliminary data measuring responses to IPM in the IC while administering antagonists to the inhibitory neurotransmitter GABAA. The release from tonic inhibition increased the firing rate non-specifically, thereby increasing the adaptation state of the cell and exacerbating the directional dependence of the response to IPM.
These results suggest that adaptation of excitation, convergence and the neural transmission delay all contribute to the sensitivity of IC cells to virtual motion cues. Other data (Semple, 1998, ARO, 21st Midwinter Meeting, 426), however, indicate that adaptation of excitation alone is insufficient: adaptation of inhibitory inputs to the IC is also required.
The responses to moving stimuli in the IC are consistent with sensitivity to instantaneous IPDs followed by adaptation (the snapshot hypothesis) rather than with specialised motion-selectivity.