The influence of noise on quantal EPSP size obtained by deconvolution in spinal motoneurons in the cat.

1. The amplitudes of quantal components that make up single-fiber excitatory postsynaptic potentials (EPSPS) were determined by a deconvolution technique and by simulation studies and were compared with the background noise. 2. A strong correlation was found between the sizes of EPSP quantal components and the standard deviation of the noise from which the data were extracted by deconvolution. A similar correlation was then shown in published data from several other laboratories. 3. EPSPS having amplitudes less than 100 microV were recorded that had little or no variance in their amplitudes. Most of these EPSPS showed a much smaller peak variance than would be expected if they fluctuated among amplitudes in steps of approximately 100 microV--the proposed mean value for the amplitude of the quantal EPSP. 4. Deconvolution of simulated data with the maximum likelihood algorithm resulted in the suppression of components less than 1.5 SD of the background noise. The remaining components were approximately equally spaced. No way was found to detect this error, and rejection of deconvolved data with components less than 1.5 noise SD did not eliminate it. The resulting erroneous data showed a strong correlation between the amplitudes of the components obtained and the noise standard deviation. 5. It is concluded that at least some EPSPS generated by single Ia-afferents on motoneurons are composed of quantal components significantly less than 100 microV and that deconvolution procedures are not capable of detecting such small components.