blog.kennypearce.net

Subjunctive Phenomenalism and Logical Construction Idealism

Within the last week, I have seen the same mistake in two different recent books on the philosophy of perception:

According to phenomenalism, objects are (in John Stuart Mill's excellent phrase) "permanent possibilities of sensation"; they are, in a more recent idiom, "logical constructions" of sense data. (Alva Noë, Action in Perception, 79)

Berkeley observed that the philosophical conception that the objects of direct awareness are sense-data (or, in Berkeley's terminology, "ideas") is perfectly compatible with the commonsense conception that the objects of direct awareness are ordinary things (e.g., tomatoes). We can accept both, Berkeley argued, if we recognize the truth of idealism: that ordinary things are sense-data. (In Berkeley's terminology, "things are ideas.") This move allows us to reject as illusory the gulf between subjective sense-data and the objective world external to us. Russell sharpened Berkeley's dictum to say that things are "logical constructions" out of sense-data. The point here is that, while it may be impossible to identify a thing (e.g., a tomato) with a particular sense-datum (or a collection of sense-data), the content of any claim about external objects can be specified entirely in terms of sense-data. This is the thesis of phenomenalism ... (Anil Gupta, Empiricism and Experience, 43)

First, let's get some terminology straight here. Berkeley's name for his view is immaterialism. Immaterialism is the denial that material substance is ontologically fundamental. Idealism is (to a first approximation) the view that all of the ontologically fundamental entities are on the 'mind' side of the mind-body dichotomy. Finally, phenomenalism is the view that true statements about familiar objects are true in virtue of some facts about the experiences of minds.

Having cleared that up, there is a more specific confusion in the above two quotations: they both confuse subjunctive phenomenalism with logical construction idealism. (Lest you think Gupta is not talking about subjunctive phenomenalism, he cites C. I. Lewis as a paradigm phenomenalist later on the page.) These two views are importantly different. Subjunctive phenomenalism holds that all familiar object talk can be translated into subjunctive conditionals about sense experience. (I discussed this view in a recent post on familiar object talk.) Logical construction idealism posits the existence of certain entities called 'ideas' (Berkeley) or 'sense data' (Russell) and holds that familiar objects are made out of these. By calling them 'logical constructions' the view further claims that these are merely nominal entities, having no metaphysical unity. We put them together out of the sense data. If we deny the existence of unsensed sense data (which seems like a reasonable thing to do), then logical construction idealism will be a form of phenomenalism, but it will still not be the same as subjunctive phenomenalism. The most important difference is that logical construction idealism is committed to some entities, sense data, to which subjunctive phenomenalism is not necessarily committed. For subjunctive phenomalism, facts about events (actual and counterfactual) ground the truth or falsity of familiar object statements, whereas for logical construction idealism familiar object statements are actually about some entities (albeit mere logical constructions of sense data).

Which of these views Berkeley holds is a disputed matter (which renders Gupta's confusion between the two the more problematic, since he is discussing Berkeley here). On this dispute, see Kenneth P. Winkler, Berkeley: An Interpretation, sect. 6.8. Personally, I think subjunctive phenomenalism is closer to Berkeley's view. However, because of Berkeley's distinctive theory of language (which is too rarely deployed by interpreters in these contexts, despite Berkeley's own repeated insistence on its relevance) his view is actually more plausible and sophisticated than the views of explicit subjunctive phenomenalists like C. I. Lewis. More on that another day.