Neural nets identify sensory receptors from somal spikes

Retrospective analysis of somal electrophysiology from intracellularly recorded, physiologically identified afferents demonstrates that neural nets can be readily trained to identify the type of peripheral receptor supplied. Specifically, cat spinal ganglion somata could be identified as innervating muscle spindles, hairs or high-threshold mechanoreceptors. Further, both hair afferents and high-threshold mechanoreceptors could be separated into three distinct subclasses. The neural net sorting reported here utilizes only the electrophysiological properties of the somata plus conduction velocity and can with this information alone predict the functional properties of the sensory endings. Interestingly, neural net sorting could also distinguish between different types of hair afferents (or nociceptors), even when conduction velocity information was ignored. It is suggested that neural nets, in combination with computer-controlled data-acquisition systems, could greatly increase investigator efficiency and decrease the number of animals needed to demonstrate specific phenomena, such as drug effects on particular cell types. A double-edged sword of increased investigator efficiency and decreased animal usage may be of particular usefulness in the present socio-political research arena.