SYNCHRONY BETWEEN SINGLE-UNIT ACTIVITY AND LOCAL-FIELD POTENTIALS IN RELATION TO PERIODICITY CODING IN PRIMARY AUDITORY-CORTEX

1. We recorded responses from 136 single units and the corresponding l ocal field potentials (LFPs) from the same electrode al 44 positions i n the primary auditory cortex of 25 juvenile, ketamine-anesthetized ca ts in response to periodic click trains with click repetition rates be tween 1 and 32 Hz; to Poisson-distributed click trains with an average click rate of 4 Hz; and under spontaneous conditions. The aim of the study is to evaluate the synchrony between LFPs and single-unit respon ses, to compare their coding of periodic stimuli, and to elucidate mec hanisms-that limit this periodicity coding in primary auditory cortex. 2. We obtained averaged LFPs either as click-triggered averages, the classical evoked potentials, or as spike-triggered averages. We quanti fied LFPs by initial negative peak-to-positive peak amplitude. In addi tion, we obtained trigger events from negativegoing level crossings (a t similar to 2 SD below the mean) of the 100-Hz low-pass electrode sig nal. We analyzed these LFP triggers similarly to single-unit spikes. 3 . The average ratio of the LFP amplitude in response to the second cli ck in a train and the LFP amplitude to the first click as a function o f click rate was low-pass with a slight resonance at similar to 10 Hz, and, above that frequency, decreasing with a slope of similar to 24 d B/octave. We found the 50% point at similar to 16 Hz. In contrast, the LFP amplitude averaged over entire click trains was low-pass with a s imilar resonance but a high-frequency slope of 12 dB/ octave and a 50% point at similar to 12 Hz. 4. The LFP amplitude for click repetition rates between 5 and 11 Hz often showed augmentation, i.e., the amplitu de increased in response to the first few clicks in the train and ther eafter decreased. This augmentation was paralleled by an increase in t he probability of firing in single units simultaneously recorded on th e same electrode. 5. We calculated temporal modulation transfer functi ons (tMTFs) for single-unit spikes and for LFP triggers. They were typ ically bandpass with a best modulating frequency of 10 Hz and similar shape for both single-unit spikes and LFP triggers. The tMTF per click , obtained by dividing the tMTF by the number of clicks in the train, was low-pass with a 50% cutoff frequency at similar to 12 Hz, similar to that for the average LFP amplitude. 6. The close similarity of the tMTFs for single-unit spikes and LFP triggers suggests that single-uni t tMTFs can be predicted from LFP level crossings. The average ratio o f the tMTF for single-unit spikes and LFP triggers was 0.38 +/- 0.25 ( SD). Cross-correlations between single-unit spikes and LFP triggers sh owed a mean peak at a lag of 1.6 +/- 3.2 (SD) ms, a mean peak width of 13.9 +/- 7.7 (SD) ms, and a mean efficacy of 0.21 +/- 0.28 (SD). 7. W e calculated first-order Poisson kernels between clicks and single-uni t spikes, and LFP triggers and LFPs. Kernels were generally similar in shape for single-unit spikes and LFP triggers. The sequence of activa tion and suppression around the mean firing rate could generally be pr edicted from the waveform of the integrated click-evoked LFP. Poisson kernels accurately predicted the presence and salience of rebounds in single-unit responses to click train stimuli. 8. Spike-triggered avera ge LFPs were similar under Poisson click stimulation and spontaneous c onditions. Their integrated waveforms, called average compound excitat ion potentials, lacked the strong excitatory rebounds of the click-evo ked LFPs and the single-unit firing patterns. 9. Double click transfer functions provided insufficient information to predict the response t o successive clicks in periodic click trains for repetition frequencie s in the range of 5-11 Hz, i.e., in the augmentation region. In contra st, the prediction for repetition rates <5 Hz and >12 Hz was quite acc urate. 10. Predicting the double-click modulation transfer function on the basis of first- and second-order Poisson kernels calculated from the response to Poisson-distributed click trains was impossible. A goo d prediction was, however, obtained for the average LFP amplitude modu lation transfer function up to repetition rates of similar to 16 Hz. T he discrepancy for repetition rates >16 Hz suggests that the system ha s a much higher nonlinearity than second order and/or that the paramet ers of the system change during prolonged stimulation.