Empiricist Epistemology: Hume and Positivism

David Hume (1711-1776) extends the empiricist project by insisting that our knowledge of "facts" about the world is based ultimately on experience.  Such claims about the world are what he calls "matters of fact."  By contrast, judgments, statements, or propositions that are true or false by definition are not about facts: they are "relations of ideas."

According to Hume, propositions that rationalists use as models for knowledge--such as "a whole is always greater than any one of its parts" and "triangles have three sides"--are simply matters of definition.  They provide us with no real knowledge other than regarding what terms mean.  As long as we understand the meaning of a certain term, we can analyze it (that is, unpack what is implicit in the term) without learning anything about whether there is anything in the world that the term describes.  Relations-of-ideas statements that identify characteristics that are already implicit in the meaning of a concept or object (such as "all bodies take up some space," "bachelors are human beings") are thus called analytic statements. True analytic judgments are those in which the predicate (that is, what you say about something--for example, "are unmarried") is contained within the subject (e.g., bachelors).  If you try to deny such a proposition (such as saying, "It is not the case that bachelors are unmarried"), you contradict yourself.

In contrast to analytic propositions, propositions in which the predicate is not part of the meaning or definition of the thing. To say, for example, that some birds are yellow is to say something about birds which is not contained within the definition of what it is that makes a bird a bird. After all, not all birds have to be yellow. By saying that a particular bird is yellow, we add a new bit of information which we could not know by simply knowing that something is a bird: that is, we appeal to experience.  In this sense, we combine or synthesize two ideas, one of which (yellow) is not already implicit within the meaning of the other (bird). A proposition such as "Some birds are yellow" is therefore called a synthetic proposition.  Synthetic propositions are statements in which the predicate is not contained within the subject; and if we deny such a "matter of fact" proposition (such as saying, "It is not the case that some birds are yellow"), we do not necessarily contradict ourselves.

The distinction between analytic and synthetic propositions is different from the distinction between a priori and a posteriori propositions. The distinction between a priori and a posteriori judgments deals with whether you have to rely on experience to determine whether the proposition is true or false. The distinction between analytic and synthetic judgments deals with whether what you say about a thing is already contained in the meaning of the thing.  Since analytic judgments can be made in most cases without having to appeal to experience, they happen to be a priori judgments as well; just as in most cases, synthetic judgments happen to be a posteriori judgments.

Hume thinks that all analytic judgments are a priori and all synthetic judgments are a posteriori.  That is, every meaningful statement is known as true or false either by definition--in which case, it tells us nothing about the world--or by experience.  Propositions such as "there is a God" or "there is a spiritual self" are not true by definition nor are they based on sense experience; therefore they are meaningless statements.

Berkeley had claimed that there must be a cause of our ideas which is not a material substance (as Locke maintained) but is rather God.  Hume replies that, since we can say "there is no God" without contradicting ourselves, and since the notion of God is not based on any sense experience, we cannot say that God is the cause of anything.

But Hume does not stop there.  He asks what it means to say that we know that events have causes.  When we try to trace such knowledge back to experience, we discover that all we mean by saying that A causes B is that A occurs before B, A seems to be near B in space and time, and that in our experience events like A seem to be followed with some regularity by events like B. The problem with this, Hume notes, is that we do not experience a necessary connection between A and B.  We have a natural inclination to assume that there is a connection (a "constant conjunction") between an event and its supposed cause, but there is no empirical justification for thinking every event has a cause. Even if every event in the past has had a cause, that is no justification for thinking that future events will have causes as well, since we do not know that future events will resemble past events. So even the assumption that there is a high probability that things in the furture will resemble the past is empirically groundless and philosophically suspect.

Similarly, we assume that our present experiences are linked to our past experiences by means of some "self." But we have no sense datum ("given") or experience of a continuous self.  If we do not know that we have a continuous self, we cannot depend on our memories.  And that means that all we know is what we are immediately experiencing.  Even that is doubtful, because it all might be illusion.  That is, in the end we must doubt whether any knowledge is possible (skepticism).

Objections to Hume:

• Even if our natural inclinations (e.g., about cause, self) cannot be empirically justified, that does not mean they lack philosophical respectability.
• Even if the self and cause cannot be known, they must be presumed to make sense of experience.

Logical Positivism

According to Logical Positivists, a group of philosophers who developed their ideas beginning in the 1920s, to say that a proposition is meaningful means that it is based on sense data (experience) or is a mere linguistic convention (a truth that we simply decide to accept by definition, a tautology).  The meaning of a statement about the world is a prediction of what sense data you would experience if you were to be in a particular situation.  Learning is the result simply of behavioral conditioning (behaviorism).

Sense data are themselves neither mental nor physical.  The mental-physical distinction consists simply in our organization or arrangement of predicates (e.g., size, shape).  Positivism asserts that other minds (and even one's own) are logical constructs, since they are not sensed nor are they composed of sense data.  The self is simply a bundle of perceptions, and there is no continuous substantial self.  Theoretical entities (e.g., atoms, average housewife) exist solely in terms of theories for explanatory and predictive purposes.

[Ockham's razor: Locke drops innate ideas; Berkeley drops Descartes/Locke's material substance; Hume drops mental substance and causality; the positivists drop Hume's belief in things in favor of sense data ("phenomena").]

• Psychological atomism (the view that reality as composed simply of atoms of sensation such as yellow, sharp, hot) is wrong because knowledge of the world is not built up from discrete sensorial impressions.  Rather, perception occurs in a field of meaning (a "gestalt"), in which a sensation becomes a perception by being distinguished from its background.  The social field is a structure that sets the expectation for what we can perceive.  Perception is not simply an objective given.
• The ability to engage in linguistic novelty (that is, to understand combinations of words we have not heard before and to form expressions that parallel others) proves that operant conditioning (behaviorism) cannot explain linguistic ability.  Noam Chomsky claims that this capacity for "deep grammar" in terms of which we describe the world is innate, a result of the structure of the brain.