Four decades of work and no characterised binding site. What exists instead is a set of constraints: acetylcholine preserved under conditions that deplete it, protein kinase C translocated only in learning-impaired animals, and a release effect that disappears when slices are replaced by synaptosomes.