Bertrand Russell,
Philosophy of Logical Atomism
Some Issues to Discuss

Curtis Brown

1. What is the property of redness? Russell says that you need to experience red in order to know the meaning of "red." The implication appears to be that redness is a property of experience, not a property of things. (Or at least that it is to be analyzed in terms of the experience it produces, not in terms of "objective" properties.) "The color with the longest wavelength" may be a true description of redness, but it's not acceptable as an analysis.

simple complex
atoms molecules
known only by acquaintance can be known by description
end result of analysis (so can't be analyzed further) can be analyzed further
a definition can't give the meaning of a simple a definition can give the meaning of something complex

2. Logical constructions. Russell applies Occam's Razor in a pretty extreme way. A theory is safer the fewer assumptions it makes (less ways to be wrong), so we should minimize assumptions as far as possible. OK so far, in my view. But Russell goes on to suggest that we can cut down on metaphysical assumptions by using logical constructions.

example: defining numbers in terms of sets. Numbers are ontologically puzzling entities: they don't have locations, they don't exist in time, they don't have causes or effects, and yet they seem to be things of some sort. Maybe there are such things as numbers in that sense. But Russell says we don't need to assume that there are in order to get on with mathematics. Instead, we can construct logical objects which can do all the work numbers do. A very simple way:

0  {}
1 {{}}
2  {{{}}}

that is, we can analyze 0 as the empty set, 1 as the set whose only member is the empty set, etc. Then for any number x, the number x + 1 is just {x}. Now we can define addition, multiplication, etc. in such a way that everything comes out the way it should, e.g. 2 + 4 = 6 etc.

addition: x + 0 = x
  x + y' = (x + y)'

Anyway, here's the point: in mathematics, this seems like a way to avoid ontologically dubious entities. We construct logical entities that we can use to accomplish the same purposes, but which do not require the same ontological assumptions, so that we don't need to make the ontological assumptions.

Now, Russell advocates doing the same thing in the case of other entities that he regards as ontologically dubious. For instance, people. Are there really such things as people in the way we ordinarily think of them? (I.e. as subjects of experience that remain the same over time even though their experiences change?) Maybe, maybe not. But we can make logical constructions instead! No one doubts that experiences exist. Take a particular experience -- my experience of redness right now. Now add all my other experiences right now. Those are all ontologically unproblematic. Then so is the set of all of them: {e1, e2, . . . en}. Now consider the set of my experiences at the following moment: {e'1, e'2, . . ., e'n}. Another perfectly good set. Now construct the sequence (ordered set, roughly) of my experiences at consecutive times: <{e1, e2, . . ., en}, {e'1, e'2, . . . , e'n}, . . . >. Russell's suggestion: this is ontologically innocuous, but it can do all the work that the original notion of a person can do.

3. neutral monism. R doesn't quite commit himself to this, but seems to think there's a good chance that it's right. On this view, people are one sort of logical construction out of experiences (see above). But also, physical objects are just another sort of logical construction out of experiences.

4. Lecture 5, p. 199 [open court p 102]: primitive knowledge of general propositions: apparent neglect of any kind of reasoning other than deductive reasoning.

Last update: August 29, 2012
Curtis Brown  |  Philosophy of Language   |  Philosophy Department  |  Trinity University