The following is a re-post from an earlier incarnation of the site; it was originally published 13 January 2005.

Ira M. Schnall’s “Constancy, Coherence, and Causality” 1 is a welcome addition to a new trend in the study of Hume’s theory of the external world. Traditionally this area of Hume scholarship has been heavily dominated by H. H. Price’s landmark work, Hume’s Theory of the External World2. There has recently been a tendency in the scholarship to pull away from Price’s interpretation, a tendency which, I think, is quite right. Schnall does an excellent job of presenting the main problems with Pricean interpretations in general, and sets the reader on the right track.

In Treatise 1.4.2, Hume attributes our believe in the existence of external and independent bodies to the constancy and coherence of our perceptions.3 For instance, when I shut my eyes or turn my head, I interrupt my perceptions of a given object (e.g., this computer). When I re-open my eyes or turn my head back, however, the computer presents itself “in the same uniform manner”; it is unchanged (T.1.4.2.18, SBN 194-195). There is a constancy in my impression of the computer. On the other hand, there are many cases in which the constancy of our perceptions admits of rather considerable exceptions. As Hume says, “Bodies often change their position and qualities, and after a little absence or interruption may hardly become knowable” (T.1.4.2.19, SBN 195). If we leave a fire for an hour, it is never in exactly the same state as it was; but this change is much like any number of changes with which I am previously familiar. There is, then, a coherence in our perceptions as well, a regularity even where there is something less than perfect constancy.

Price interprets this distinction in the following way. Suppose you have several series of impressions:

(1) A B C D E(2) A - C D E(3) A B - - E

And so forth. (2), (3) and the like partially resemble (1); (2) and (3), in fact, are like (1) except with gaps. This is how Price understands coherence. When we experience a gappy series like (2) or (3), we assimilate it to the more complete series (1), and thereby attribute a greater degree of regularity to the world than we find in our actual experience.

The Pricean account of constancy is similar. Constancy involves closely resembling perceptions, e.g.:

(1) A A A A A(2) A - A A A(3) A - - A A

And so forth. Price treats constancy as a special case of coherence. Faced with (2) or (3) we assimilate that to (1) . On this interpretation coherence and constancy exhibit the same basic pattern; Price calls this ‘gap-indifference’.

There is much in a Pricean gap-indifference account of Hume’s theory that sounds plausible. However, there are a number of difficulties with this approach. Schnall has an excellent discussion of the problem cross-series identification poses for the Pricean account, for which I will simply direct my reader to the article with the assertion that this part of the article is well worth reading. I agree with Schnall’s analysis, and note that there are additional problems (related, I think) for a Pricean interpretation as well. For instance, not only does Price’s leave out issues like cross-series identification and inter-series sensibilia, it is difficult to see that how it can properly characterize a fairly obvious aspect of our experience, one which plays a key role in Hume’s account of constancy, namely, the fact that we very often do not seem to be assimilating one series to another at all. Hume’s view that we simply treat as identical closely resembling impressions (T.1.4.2.34, SBN 203-204) easily and plausibly handles such cases without presuming any prior more complete series. When we perceive A (time interval) A, we often process this not as A A A A A but simply and entirely as A, with no consideration of a series at all.

One puzzle I find in the article is Schnall’s treatment of Hume’s claims about the idea of time and unchanging objects. Hume holds that our idea of time is an idea of succession; we therefore have no idea of the temporal duration of unchanging objects. On this Schnall says:

But there is no reason not to say that the more complex idea of an unchanging object plus a succession of perceptions which is contemporaneous with it is a perfectly respectable idea of identity, or of an object that endures through a real interval of time. (p. 45)

However, Hume himself allows that we can apply our idea of time fictitiously to unchanging objects. What makes it fictitious is that there is nothing in the unchanging object that is identifiable as time; it therefore contributes nothing to the complex idea of [unchanging object + succession of perceptions], and receives nothing from it. We fictitiously attribute temporal duration to unchanging objects because we associate them with objects that have temporal duration (T.1.2.5. 29, SBN 65), but the mere fact of association with a succession of perceptions doesn’t (and shouldn’t) tell us anything about the unchanging object itself. Nor does there seem to be anything gained, in a Humean framework at least, by denying that this is a fiction. Fictions, like the standard of perfect equality, indicate cases where we cannot on empiricist principles allow that there is a genuine idea, but in which a particular procedure (e.g., mental impulse in the case of equality, or association with temporal change in the case of time) in a sense takes the place of an idea (to the point that if we are not sufficiently precise in our analysis we might think we really had an idea).

This brings me to what I felt the article most lacked, and it is what I think most discussions of Hume’s theory of the external world most lack. At T.1.4.2.22 (SBN 198), when he is discussing the mental impulse toward greater regularity, Hume takes the trouble explicitly to place a comment and a footnote directing the reader back to the discussion of a perfect standard of equality in T.1.2.4. This is generally ignored in discussions of coherence; yet it seems to me the key to the entire passage. An adequate interpretation of Hume’s discussion of coherence must be able to make clear sense of the role the impulse to greater regularity plays in both these contexts, since it is this that prevents Hume’s mechanism for coherence from being ad hoc. It is also a sign that the discussion of coherence is not nearly so simply as it might look for someone who just looks at T.1.4.2; and any precise characterization of coherence has to be tested against the discussion in T.1.2.4. This will, I think, be the most important part of a successful interpretation of Hume’s theory of the external world.

1. 1. Hume Studies 30 (April 2004) 33-50
2. 2. Oxford: Clarendon Press, 1940
3. 3. Note on references: The format T.1.2.3.4 means Treatise Book One, Part Two, Section Three, Paragraph Four. SBN indicates the page number in David Hume, A Treatise of Human Nature, ed. L. A. Selby-Bigge, 2nd ed., revised by P. H. Nidditch (Oxford: Clarendon Press, 1978).