Some Terminological Distinctions Pertinent to Carnap's Development:

Phenomenalism: The metaphysical view that reality consists of nothing but
"phenomena."

The term "phenomenon" was introduced by Kant to mean the object as it appears to an
experiencing subject. Since to speak of an "appearance" implies the existence of
something which "appears." Kant held that if one regards the phenomena as distinct from
an object "in-itself" then the mind must posit that there is an unexperienced object
standing "behind" the phenomenon which "appears" to the experiencing subject. Kant
called this object posited by the mind ("nous" in Greek) the "noumenon." Unfortunately
Kant's epistemology makes this noumenon unknowable. Thus post-Kantian philosophers
have sought to do away with this conception of something "behind" the appearance of
things to the experiencing subject. One way of doing this is to assert that there is nothing
behind the phenomenon at all. All that is are phenomena. Thus was born the
metaphysics of "phenomenalism."

"Sense data" epistemology: a theory of knowledge which holds that what we directly
experience are bits (atoms) of sensory data, and that all empirical knowledge rests on this
direct acquaintance with such sensory bits. Presumably simple referential terms like
"red," "hot," "soft," etc. would refer to such sense data.

While sense data theory was developed from empiricist epistemology and phenomenalism
is a metaphysical view, the two positions are often found together and tend to reinforce
each other. In particular in introducing the term "sense data" Russell sought to avoid the
subjectivist connotation of words used by earlier empiricists such as "sense perceptions"
(Berkeley) or "sense impressions" (Hume) that were imagined to refer to what was in the
experiencing subject's "understanding" or "mind." Instead, Russell regarded "sense data"
as objectively real, existing in the "real world" independently of any experiencing subject.
Russell tried (rather unsuccessfully) to show how the ordinary physical world of
experience, as well as one's mental life, is a construction out of such sense data (neutral
monism). Sense data empistemology is an attempt to establish certain propositions as
"incorrigible," or known directly, to serve as the "foundation" for any other empirical
propositions which they serve to justify. Thus this is a form of "foundationalistic
epistemology."

Reductionism: the position that all propositions about the ordinary physical objects of
the world can in principle be "reduced" to large sets of propositions which refer
exclusively to experienced phenomena or sense data.

Originally positivists embraced a reductionist position that held that the scientists'
description of the world, which constitutes all meaningful synthetic empirical knowledge
could (in principle) be reduced to propositions which referred exclusively to sense data.
Carnap attempted to carry out such a reduction in his Logical Structure of the World
(popularly known as the Aufbau program), but he eventually realized such a reduction
could not be carried out even in principle.

Physicalism: the position that propositions which are taken as directly known are not
those which refer only to reduced to sense data but include those which refer to the
ordinary objects of the scientists' world as are actually reported in the observational
language used by science. Thus a physicalist could give observational reports about, for
example, such things as microscopes or voltmeters; he could talk about observing
animals or people without having any obligation to show that such terms could (in
principle) be reduced to talk about sense data.

After abandoning sense data theory, with its commitment to phenomenalism, Carnap
introduced the notion of physicalism as defined above. He spoke of such reports as
expressed in what he called the "thing language." Obviously what one regards as things
is relative to one's language and its conceptual vocabulary. Since the acceptance of a
particular language is a "convention" adopted because of the usefulness of that language
in communicating the sorts of observation reports which are pertinent to different
theories, the move from phenomenalism to physicalism constitutes a move from a
"foundationalistic" to a "conventioanlistic" justification of the knowledge claims made by
synthetic empirical propositions. Carnap explicitly admits that the decision to adopt one
vocabulary or another, i.e. which language to employ, is an arbitrary convention, which
may be justified pragmatically (i.e. by its usefulness), but that no choice of language can
be shown to be the "true" or "correct" language for describing experience.