| Name | What are scientific theories about? | What is the status of unobservables? | Response to the underdetermination of theory by experiment | Main benefits | Main limitations | Additional notes |
|---|---|---|---|---|---|---|
| Scientific realism | Approximately true descriptions of the universe that exists independently of our perceptions, including claims about unobservable entities. | The problem does not arise in scientific practice, since we struggle to come up with theories that make empirically adequate predictions (counterexample: quantum mechanics). We have other criteria for evaluating theories such as explanatory power, simplicity, alignment with scientific history, and whether they can be integrated into a larger framework. | Provides a straightforward answer to why science has been successful. | |||
| Positivism | Directly observable procedures of predication (e.g., positions of clocks, behavior of oscilloscopes). | Claims about unobservables are either about directly observable procedures of predication or meaningless discourse—”grammatical illusions.” | There is no problem, since two theories that make the same predictions about unobservables have exactly the same content. | Avoids problems of skepticism—how we know about the unobservable—and underdetermination—distinguishing between different theories about unobservables. | (1) Challenges with translation: There is no way to turn a proposition about the world into a finite set of observables, since making definitions and theories precise would implicate all the facts of the world. (2) No clear distinction between analytic and synthetic statements: The positivist depends on certain statements being true independent of empirical fact, such as the very claim that scientific theories are about observables, but the holist argues that any established truths can be changed to accommodate an unexpected observation. | Verificationist theory of meaning: A word or proposition has a clear meaning if has observable procedures of predication, i.e., a clear account of what kinds of observations one would need to make to decide whether the proposition was true or false. |
| Structuralism | Theories about unobservable entities can only describe their mathematical and structural relations, not their intrinsic nature. | Unlike positivists, structuralists believe that unobservable entities exist, hence the content of scientific theories includes relations between unobservables. | Avoids the translation challenge of the positivist project. | |||
| Constructive empiricism (van Fraassen) | Not truth, but claims that are empirically adequate, i.e., make correct predictions about observables. | Unlike positivists, constructive empiricists accept underdetermination, but do not consider it a problem for science. | ||||
| Instrumentalism | Not propositions about how the world is at all, but practical tools for what will happen in the future based on what happened in the past. | Claims about unobservables are essentially ”pneumonic devices to help you remember how to do a calculation.” | Do not need to precisely distinguish between observables and unobservables. |