Coding of electric pulse trains presented through cochlear implants in the auditory midbrain of awake rabbit: comparison with anesthetized preparations

Abstract

Cochlear implant (CI) listeners show limits at high frequencies in tasks involving temporal processing such as rate pitch and interaural time difference discrimination. Similar limits have been observed in neural responses to electric stimulation in animals with CI; however, the upper limit of temporal coding of electric pulse train stimuli in the inferior colliculus (IC) of anesthetized animals is lower than the perceptual limit. We hypothesize that the upper limit of temporal neural coding has been underestimated in previous studies due to the confound of anesthesia. To test this hypothesis, we developed a chronic, awake rabbit preparation for single-unit studies of IC neurons with electric stimulation through CI. Stimuli were periodic trains of biphasic pulses with rates varying from 20 to 1280 pulses per second. We found that IC neurons in awake rabbits showed higher spontaneous activity and greater sustained responses, both excitatory and suppressive, at high pulse rates. Maximum pulse rates that elicited synchronized responses were approximately two times higher in awake rabbits than in earlier studies with anesthetized animals. Here, we demonstrate directly that anesthesia is a major factor underlying these differences by monitoring the responses of single units in one rabbit before and after injection of an ultra-short-acting barbiturate. In general, the physiological rate limits of IC neurons in the awake rabbit are more consistent with the psychophysical limits in human CI subjects compared with limits from anesthetized animals.

Keywords: anesthesia; cochlear implant; inferior colliculus; temporal coding.

Figure 1.

Temporal response pattern (left), average firing rates (middle), and spike waveforms (right) in response to trains of biphasic pulses of different rates for three example IC neurons in awake rabbit (A–C) and one neuron from anesthetized cat (D). Left: Dot-rasters in which each dot represents a spike and alternating shades of gray distinguish blocks of stimulus trials at different pulse rates. Middle: Mean firing rates during the on period (excluding the first ∼30 ms after stimulus onset) and off period (excluding the first 100 ms after stimulus offset) versus pulse rate. Right: Superimposed spike waveforms from each neuron (145, 314, 227, and 95 spikes, respectively).

Figure 2.

Fraction of IC units showing excitatory (positive ordinates) and suppressive (negative ordinates) responses to pulse-train stimulation as a function of pulse rate in unilaterally implanted and bilaterally implanted rabbits.

Figure 3.

A, B, Cross-correlograms between stimulus pulse trains and neural spike trains for the same two neurons as in Figure 1, C and B, respectively. Each trace shows the cross-correlogram for one pulse rate. Gray shading indicates the 99.5% upper confidence bound for a random spike train; correlation peaks exceeding the confidence bound are filled in white. C, D, Normalized height of the main correlogram peak as a function of pulse rate. The peaks are normalized to the 99.5% confidence bound (dashed line). Filled marker indicates that the peak height exceeds the confidence bound. The pulse-locking limit is defined as the rate where the peak height intercepts the 99.5% upper confidence bound. E, F, Vector strength and 99.5% upper confidence bound for a random spike train as a function of pulse rate. The limit is defined as the rate where the vector strength intercepts the 99.5% upper confidence bound. G, H, Comparison of firing rates computed using all spike to firing rates computed using only pulse-locked spikes.

Figure 4.

A, Histogram of the distribution of pulse-locking limits based on cross-correlation and vector strength across the population of IC neurons in awake rabbit. B, Distribution of spike latencies to low-rate pulse trains in awake rabbit. C, Scatter plot of pulse-locking limit against spike latency in awake rabbit and anesthetized cat.

Figure 5.

Effect of barbiturate on responses to pulse trains in a single unit from rabbit IC. A–C, Temporal response patterns (dot rasters) just before injection (A), 1 min after injection (B), and 10 min after injection (C). D–E, Absolute (D) and normalized (E) firing rate during the on period against pulse rate for the three time periods in A–C.

Figure 6.

Effect of barbiturate in two single units from rabbit IC. A–B, Unit with high spontaneous rate and suppressive responses to pulse trains. C–D, Unit with multiple spike responses to each pulse at low rates and rebound responses at high pulse rates. For each unit and condition, temporal discharge patterns are shown on the top and mean firing rates during on and off periods versus pulse rate on the bottom.

Figure 7.

A, Scatter plot of background firing rate before and 1–2 min after injection in the 13 units studied with barbiturate injections. B, Scatter plots of half-maximum pulse rate and pulse-locking limit before and 1–2 min after injection.

Figure 8.

Comparison between response properties of sample IC neurons from awake rabbit and anesthetized cat. A, Distribution of SRs. B, Fraction of IC units showing sustained excitatory (positive ordinates) and suppressive (negative ordinates) responses to pulse-train stimulation as a function of pulse rate. C, Cumulative distribution of pulse-locking limits using cross-correlation method excluding units that do not pulse lock at any rate.

Figure 9.

Neurons clusters based on on period and off period responses to pulse train stimuli. A–D, Centroids of the four neuron clusters generated by k-means clustering. Each curve shows the normalized mean firing rate during the on and off periods as a function of pulse rate. A, “Suppressive” cluster. B, “All-pass” cluster. C, “Band-pass” cluster. D, “Low-pass” cluster. E, Relative incidence of each cluster differs between the two preparations.

Figure 10.

Mean pulse-locked firing rate and overall firing rate as a function of pulse rate for each of the four neuron clusters defined from rate responses to pulse trains. A–D, Same as in Figure 9.