The field of science studies is the site of an explicit reflection on the o
ntological premises of sociology, with rival approaches defined by distinct
ive ways of specifying the basic constituents of reality. This article take
s advantage of this debate to compare three types of ontological schemes in
terms of their internal coherence and their consequences for sociology. So
ciological in terms of their internal coherence and their consequences for
sociology. Sociological humanism-represented by proponents of the sociology
of scientific knowledge (SSK)-distinguishes between an immanent domain of
social relations, a transcendent and meaningless material reality, and an i
ntermediate, socially constructed level of knowledge, meaning and culture.
Symmetrical humanism-as found in the recent writings of Andrew Pickering-in
sists that society too should be placed among the constructions, thereby di
squalifying it as a source of explanations of human agency and leaving a de
tached and self-moving human agent. The relational ontology-exemplified by
the "actor-network" approach of Bruno Latour adn others-make no a priori di
stinctions between humans and others, or between trandscendent reality and
construction, treating these properties as outcomes. The two humanist appro
aches are found to be incoherent as ontological schemes and also, contrary
to the antisociological stance of the actor-network approach, it is found t
hat the relational ontology provides a consistent basis for sociological ex
planations of human practices.