Epistemology of screening for behavioral toxicity.

A method is described for the assay of the behavioral effects of volatile solvents on mice and illustrated with pilot results on trichlorethylene. A dose-effect curve has been determined for the effects on schedule controlled responding and compared with the dose-lethality curve and the TLV for man. The OR50 for behavioral effects was 1/5 of the LD50 and 50 times the TLV for long-term exposure of man. An analysis of the errors involved in determination of effects on whole animals leads to the conclusion that subtle effects, representing a few per cent change, will not be detectable in routine screening. It is suggested nevertheless that information on the midrange, knowable, part of the dose-effect curve may prove useful in predicting safe levels for man.