1. You can't get correspondence rules.
   • Examples

     solubility (dispositional terms—Carnap)

     What observation is connected with the solubility of salt? Solubility "means" if some salt were to be placed in water, then it would dissolve. That is far from observable. Of course, "dissolved" is easy: the salt was placed in water, and it dissolved. But that is not at all the same.

     thermometer (multiple realizability—Fodor)

     There are many different kinds of thermometers (mercury, infrared, resistive, just to name ones available at drug stores for under $20). There is no reason to think that they all have some observational properties in common. What they clearly have in common is their function, and functions are theoretical, not observational.

2. You can produce theoretical sentences using only observational terms: People too little to see. (Hilary Putnam's example.)
3. Logic is not conventional.
4. There is no clear distinction between analytic and synthetic. (Quine)

• Is "Everything that is green is extended" analytic or synthetic? (Quine's example.)
• In scientific discourse, we often define various technical terms in terms of others.

Example (based on one of Field's)

Consider Ohm's law:

voltage = current X resistance.

• We could take voltage and current to be defined in terms of correspondence rules, in which case we could use the law to define resistance as a theoretical term. In that case, the "law" is an analytic truth, and it is a synthetic truth that whatever observational techniques we use to measure resistance do, in fact, measure resistance.
• We could take voltage, current, and resistance to be defined in terms of correspondence rules (using whatever observational techniques we use to measure resistance to give the correspondence rule for resistance). In that case, Ohm's law is a synthetic truth.

• So, is Ohm's law analytic or synthetic? It depends on which of the many things we believe we take to be definitions, and which facts. That is an arbitrary choice, and so it is a matter of arbitrary choice whether it is analytic or synthetic. That shows that there is no clear distinction.

The fallback position is logical empiricism, which

1. gives up on correspondence rules. Instead there are just some conventions.
   *Aside:* That makes it possible to handle common types of dispositional terms in the manner Carnap proposed. We don't get an equivalence, a correspondence rule, between soluble and something observational, but we get the convention that anything that has been placed in water is soluble if and only if it dissolves.
2. It gives up on observational terms in favor of observational sentences.
3. Instead of looking at some absolute system of conventions, one adopts frameworks, vocabularies that include conventions established for some purpose or other, rather than once and for all.

According to logical empiricism we adopt frameworks of rules that include logic and mathematics and rules for the use of lots of words. Some sentences are observational, and theoretical ones had better have, according to the framework, enough connections with the observational ones so that they can be tested, at least enough of the time.

-- ShaughanLavine - 12 Sep 2005 - 05 Sep 2007 ()