Consider the empiricist idea that knowledge is justified by sense experience. We have basement terms for what we sense: colors, shapes, textures, smells, tastes, sounds, etc. And we have terms for macro-objects such as tables and chairs that have these basement properties. One might think that all language is built out of the basement terms that describe what we sense. If words were only related to other words without appeal to experience, we would spiral into a regress of definitions, and we could never learn the meanings of terms. We can produce an infinite number of sentences, but with a finite mind, it seems we must have some basement terms that we use to construct the infinite possibilities. Considerations like this are traced to empiricists like Berkeley and Hume. Logical empiricists wanted to take the next step and reconstruct science in empiricist terms.
The logical empiricists were critical of terms like Marx’s ‘surplus value’, Freud’s ‘oedipal complex’, or ‘vital force’, because they were not translatable into experiential terms. Science uses theoretical terms like ‘gravity’ and ‘atom’, which need to be accounted for in experiential terms. This led to the project of taking theoretical terms in science and reducing them to observable terms.
Consider a translation of theoretical term ‘temperature’ into observational terms. Temperature change could be defined by the change in mercury level in a tube, or change in resistance of an ohm-meter, or an infinite number of other ways. There doesn’t seem to be a way to reduce temperature to a finite list of observation terms. Moreover, we think there are small temperature changes even if we cannot observe small changes in mercury level with the human eye.
There is also a problem with reference to unobservables. Most terms describing unobservable properties, processes, states and events are not translatable into entirely observational terms. The theoretical term ‘acid’ is defined as a ‘proton donor’. Protons are unobservable; we can only see the effects of protons. This has led some empiricists to deny that ‘proton’ has meaning. Instead, say we define ‘acid’ as ‘whatever turns red litmus paper blue’; then, we can’t explain why some liquids turn red litmus paper blue and some don’t: we lose the explanatory power of protons. Issues like this illustrate the tension between empiricist principles about staying true to observation, and the need for science to predict and explain.
Since we cannot define theoretical terms in terms of all possible observation, theoretical terms were given “partial interpretation.” Interpretation is partial because the term is only applied to objects that meet certain test conditions. Recall the temperature example: temperature can be measured by mercury in tubes, or in an infinite number of ways. And having a complete interpretation of temperature wouldn’t help anyway since temperature would be turned into an abbreviation of observations (i.e., a summary of experience) instead of it being used as an explanation. So, it turns out partial definitions are what allows us to predict and explain. In practice, while some partial definitions are connected to experience, most partial definitions will be connected to other partial definitions with the upshot being that most theoretical terms are very far removed from observation. So we have this tension between the need to predict and explain and the need to stay true to empiricist principles about connecting to observation.
But a problem arises with partial definitions: it is easy to make partial definitions like “A sample of water is ‘holy’ if it is found in church.” The classical response is that this new term doesn’t allow us to make new predictions. But it turns out there are many terms in science that are used without making new predictions, like the classical ‘gene’ before DNA was discovered. So it seems this restriction may keep out terms we want. On the other hand, it won’t keep out terms we don’t want, because we can fabricate terms like ‘vital forces’, which are needed to derive generalizations about what we observe.
One remedy for the logical empiricists was to become instrumentalists and say theories don’t describe reality but are used to make predictions about experience. Today, the popular, empiricist-unfriendly position is to be a scientific realist and take theoretical terms to be real and not partial interpretations or useful fictions.
Rosenberg, Alex. Philosophy of Science.