Empirical structure physicalism and realism, Hempel's dilemma, and an optimistic meta-induction
Abstract: Motivated by a generalization of Hempel's dilemma, I introduce a novel notion of empirical structure, as well as theory supervenience as a new reductive relationship between theories. One theory supervenes on another theory if the empirical structure of the latter theory refines the empirical structure of the former theory. I then argue that (1) empirical structure physicalism, the thesis that the current special sciences supervene both on current and on future physics, avoids both horns of Hempel's dilemma; (2) in particular, mental theories remain empirically dispensable in the future; (3) empirical structure realism, the thesis that earlier theories of physics supervene on later theories of physics, is supported by an optimistic meta-induction; (4) this optimistic meta-induction can coexist with the well-known pessimistic meta-induction; (5) empirical structure physicalism is appropriately labeled as a type of physicalism; and (6) empirical structure physicalism is compatible with multiple realization. To illustrate the plausibility of empirical structure physicalism, I also briefly address the so-called knowledge argument.
Paper Prompts
Sign up for free to create and run prompts on this paper using GPT-5.
Top Community Prompts
Explain it Like I'm 14
Overview
This paper is about a new way to connect physics with the “special sciences” (like biology, psychology, and economics) so that we can avoid a famous problem called Hempel’s dilemma. The author proposes “empirical structure physicalism,” a simple, test-friendly idea that says the patterns the special sciences use to tell different situations apart can always be matched—and even sharpened—by physics, both now and in the future.
The main questions
The paper asks:
- What kind of connection between physics and the special sciences can avoid Hempel’s dilemma?
- Can we define a “reduction” (a way the special sciences depend on physics) without getting stuck by the fact that current physics is incomplete and future physics is unknown?
- Do newer physics theories keep and improve the successful parts of older theories?
- If this works, does it still count as “physicalism” (the view that everything is physical or depends on the physical), and can it handle issues like multiple realization and the knowledge argument in philosophy of mind?
How the author approaches the problem
Instead of tying physicalism to a fixed list of “physical stuff” from a specific theory, the paper focuses on what theories can do empirically—how well they sort real-world situations based on what can be observed and measured.
Key ideas in everyday terms
To make this approach clear, the paper defines a few simple ideas:
- History-description: Think of this as the story a theory tells about “what happens” in a particular place and time. For example, Newton’s story of how planets move, or thermodynamics’ story of gas pressure and temperature.
- Space-time region: A chunk of the world across some area and some duration—like “our Solar System over 10 years” or “two rooms for five minutes.”
- Empirical adequacy: A story is empirically adequate to a region if the theory says it’s okay to simplify (idealize), focus on certain features (like positions or temperatures), limit attention to that region, and the story matches measurements closely enough (good approximation).
- Empirical content: For a given theory and region, this is the set of all its stories that fit the measurements in that region well enough. It’s like, “all the possible versions of the theory’s story that would have passed the tests there.”
- Empirical structure: Group all the regions in the world into piles so that each pile looks the same to the theory—meaning the theory has the same empirical content for them. Empirical structure is that grouping. It’s like a filter: which regions the theory can tell apart and which it treats as the same.
- Supervenience between theories: One theory supervenes on another if whenever the first theory can tell two regions apart, the second can too. In other words, the second theory’s filter is at least as sharp. If the second can sometimes tell regions apart that the first can’t, the second “strongly” supervenes on the first.
Helpful examples
- Newton vs. Einstein: Newtonian gravity does well for planet motions in the near term (say, 10 years), but breaks down over very long times or very precise measurements (like Mercury’s orbit over centuries). Einstein’s General Relativity can tell apart regions Newton treats as “the same” (it captures Mercury’s extra shift). So Newton’s theory supervenes on Einstein’s: Einstein’s filter keeps all the distinctions Newton makes and adds more.
- Thermodynamics vs. Statistical Mechanics: Thermodynamics gives big-picture properties (temperature, pressure). Statistical Mechanics looks at the behavior of individual particles. If two rooms of gas have the same thermodynamic values, Thermodynamics can’t tell them apart, but Statistical Mechanics can (the particle-level details differ). So Thermodynamics strongly supervenes on Statistical Mechanics.
Main findings and why they matter
Here are the paper’s key results, explained simply:
- Empirical structure physicalism avoids Hempel’s dilemma. Hempel’s dilemma says: pick current physics and your definition of “physical” is likely wrong (since physics changes); pick future physics and your definition is vague (since we don’t know what future physics says). The new approach avoids this trap by not fixing “physical” to a list. Instead, it focuses on how physics sorts situations by observable patterns—now and in the future.
- Current and future physics can make special sciences “empirically dispensable.” This doesn’t mean biology or psychology are useless. It means: whenever a special science can tell two situations apart using observations, physics can also tell them apart—now or later. Physics’ filter is at least as sharp. So in terms of distinguishing what actually happens, you don’t need the special science to do something physics can’t do.
- Mental theories remain empirically dispensable in the future. Even for mental phenomena (like pain, perception), the claim is that future physics will still be able to distinguish all the situations that mental theories can distinguish, based on observable, measurable patterns.
- Empirical structure realism is supported by an optimistic meta-induction. The pessimistic meta-induction says: “Old theories were false; new ones will be too.” The paper adds an optimistic twist: “New physics keeps and improves the successful, tested distinctions of older physics.” Over time, physics tends to refine empirical structure—keeping all the old successes and adding finer distinctions. This optimism can coexist with pessimism about strict truth.
- Empirical structure physicalism is genuinely a kind of physicalism. It respects common goals of physicalism: it roots explanations in physics and says everything depends on physics for empirical distinctions, while being flexible about how physics changes.
- It is compatible with multiple realization. Multiple realization means the same higher-level pattern (like “having pain” or “being a memory”) can show up in different physical ways. Empirical structure physicalism allows that: physics can still tell the relevant situations apart, even if the lower-level details vary.
- It can address the knowledge argument. The “Mary” thought experiment says someone who knows all physical facts about color but has never seen color learns something new when she sees red. The paper suggests its view can respond: the empirical structures and the way physics sorts experiences can account for what’s learnable and observable, without abandoning physicalism.
Implications and impact
- A safer path for physicalism: By focusing on tested, observable distinctions rather than committing to a fixed list of physical entities, this view stays steady even if physics evolves.
- Unity without rigidity: The special sciences keep their usefulness for modeling and understanding, but physics underwrites and refines the empirical distinctions they rely on.
- Better picture of theory change: We can be pessimistic about exact truth while being optimistic about cumulative empirical success—new theories preserve and sharpen what worked.
- Mind and science: Even for mental phenomena, future physics is expected to match the distinguishing power of mental theories, keeping a physicalist outlook intact.
- Flexibility for real-world complexity: Multiple realization is welcome; what matters is that physics can sort the situations we can observe.
In short, the paper offers a practical, observation-first way to connect physics to all other sciences that avoids old philosophical traps, matches how science actually progresses, and keeps physicalism robust as our best theories improve.
Collections
Sign up for free to add this paper to one or more collections.