van Fraassen's The Scientific Image: Constructive Empiricism

Michael BrenndoerferUpdated September 1, 202613 min read

Bas van Fraassen's 1980 defense of constructive empiricism against scientific realism, arguing that science aims at empirical adequacy rather than truth about unobservables.

PhilosophyAll chapters
Reading tools Notes & comments
Notes & comments
Part 9 of 12Science and Explanation
Chapter 115Evidence and scientific realism › What Are You Accepting When You Accept a Theory?
Reading the full chapter · 7 sections · 13 min7 sections · 13 min

All 7 sections · 13 min read

Evidence and scientific realism › What Are You Accepting When You Accept a Theory?Full chapter · Sections in this chapter1 / 7 ↓

What Are You Accepting When You Accept a Theory?

Suppose a physicist tells you that a thin white line in a chamber is the track of a particle. You can see the white line. The theory tells you how a charged particle in the chamber can leave a visible trail of droplets, and it may tell you why different tracks have different shapes. Accepting the theory could mean treating the line as a phenomenon the theory accurately describes and accounts for. Does it also commit you to believing everything the theory says about what produced the line?

Bas van Fraassen puts that question at the center of The Scientific Image, published in 1980. He takes the language of scientific theories seriously enough to ask what accepting those theories commits a person to. His proposal, which he calls constructive empiricism, is that science aims at empirical adequacy, and belief in a theory's empirical adequacy is the belief involved in accepting it. Acceptance also involves decisions about how to continue working with the theory. The difference between those two parts of acceptance helps explain why people who use the same theory can have different views about how much of it they believe. The Scientific Image, ch. II, §§1.1–1.3 · Author’s publication record

Literal Claims and the Aim of Science

Imagine that one theory says ordinary matter consists of atoms and another says it is fundamentally continuous. Suppose for the sake of argument that the theories yield precisely the same predictions about the observable cases under study. Their success in those cases gives us a way to compare what each can account for. It leaves their claims about the underlying constitution of matter opposed. A literal understanding of a theory preserves that disagreement. If the theory says that something exists, an interpretation can clarify what it says exists. Removing the implication of existence would change what the theory says. The Scientific Image, ch. II, §1.2

Van Fraassen distinguishes this point about meaning from one about belief. Understanding what a theory asserts is part of deciding which of its claims to accept. Someone can make sense of incompatible assertions about matter without having enough reason to pick one as true. The same distinction applies to the more ordinary decision to work with one of the accounts. A group might choose it for its capacity to guide a new investigation. That decision does not settle every claim the account makes. The Scientific Image, ch. II, §§1.2–1.3

Both a scientific realist and a constructive empiricist can read scientific claims literally. In the position van Fraassen sets out as scientific realism, science aims to arrive at theories that give a literally true account of the world, and accepting a theory involves believing it is true. The constructive empiricist allows the reading while changing the account of the aim. Empirical adequacy requires getting the observable phenomena right. A person who accepts a theory can therefore take it to be adequate while leaving open whether its account of what cannot be observed is true. The Scientific Image, ch. II, §§1.1–1.3 · Oxford chapter summary

There is a substantial commitment in doing that. Imagine that an account of the lines in the chamber has handled every run made so far. Believing that the account is empirically adequate goes beyond endorsing a summary of those runs. It involves believing that the observable phenomena, including ones that no experimenter has yet encountered, have the features the account gives them. A future sighting can therefore show that an accepted theory falls short. Someone can be cautious about one part of the theory's description and still take an epistemic risk in accepting its adequacy. The Scientific Image, ch. II, §1.3 and ch. III, discussion of empirical adequacy

How a Model Can Fit the Phenomena

Van Fraassen develops a way to state empirical adequacy using models. A theory specifies a family of structures. It also identifies parts of its models that are candidates for representing observable phenomena. Experimental and measurement reports can themselves describe structures, which he calls appearances. The theory is adequate if some model contains empirical substructures with the same relevant structure as all the appearances it needs to accommodate. The fit is a relationship between structures. The model can contain more than is used in representing those appearances. The Scientific Image, ch. III, “To Save the Phenomena”

For example, imagine a device that produces a record of three flashes at known intervals. An account of the device may include an underlying mechanism that goes through several stages between flashes. Part of a model can represent the order, spacing, and duration of all observable flashes. Whether that part fits the record can be addressed without establishing that the model's internal stages happen exactly as described. The observable behavior may give people a good reason to work further with the account. If an independent experiment reveals a persistent difference in the spacing of flashes under conditions the model covers, that difference bears on how well the model represents the phenomenon. The Scientific Image, ch. III, on models and empirical substructures

This example only uses a small set of records. The general requirement concerns the phenomena a theory is meant to accommodate. It is possible to get one measured result right by changing a parameter for that case and to adjust a second result by changing the parameter again. If the two observations have to be represented together, producing that pair of fits does not tell you whether there is one model in which both can be represented. Van Fraassen's formulation asks for a model in which the appearances can fit together. That is a stronger requirement than being able to match each isolated outcome with a different description chosen after the event. The Scientific Image, ch. III, on empirical adequacy and appearances

Empirical adequacy

An empirically adequate theory has a model whose designated empirical parts represent the relevant observable phenomena. Matching observations already made gives evidence for adequacy. The claim of adequacy extends to observable phenomena beyond that record. The Scientific Image, ch. II, §1.3 and ch. III

Van Fraassen also distinguishes things that actually happen but nobody observes from things that would happen if the world were different. A completed record could leave out real events because no instrument captured them. Those events still matter to empirical adequacy if they are observable by us in the relevant sense. That does not turn the claim of adequacy into an assertion that the same model describes the outcome in every world we can imagine. In Chapter III, van Fraassen ties adequacy to the actual phenomena. The difference is part of why testing a theory takes a particular event seriously without treating every counterfactual implication of the theory as another observation. The Scientific Image, ch. III, §§8–9

Sometimes theories that agree on the observed cases differ in where they direct the next one. Suppose both models fit the sequence of flashes under current conditions, while one gives a clear prediction for a device with a redesigned switch. The experiments so far leave both models in play. A test on the redesigned device would bear on whether their empirical commitments really coincide. If one merely gives a more convenient way to modify the mathematics while making the same empirical commitments, its advantage is different. Van Fraassen discusses this difference when considering how otherwise equivalent theories could be extended. It is a reason to identify the scope of the comparison before calling the theories empirically equivalent. The Scientific Image, ch. III, §4, “Theories and Their Extensions”

Seeing the Object and Seeing Its Trace

A distinction between observable and unobservable things has to earn a place in this account. Van Fraassen considers Grover Maxwell's challenge that seeing passes through many kinds of assistance. We can look through a window, through glasses, and through more complex optical instruments. Drawing a sharp line between every permissible and impermissible way of seeing would be difficult. Van Fraassen shifts attention from ordering acts of seeing to asking whether their objects are observable. An observer who cannot see an object where they are standing might see that same object under other conditions. The Scientific Image, ch. II, §2

The moons of Jupiter give him a clear case. Looking through a telescope to see them from Earth does not establish that they could only ever show up as something made by a telescope. A suitably located observer could see them from closer up. Their present distance is a practical obstacle to unaided observation from the ground. Van Fraassen accepts this example as one of observation. His contrast is with the track in a cloud chamber. There, according to the theory, a particle makes a series of events possible that produce a visible trail. Seeing the trail and identifying what produced it are distinguishable achievements. The Scientific Image, ch. II, §2, Jupiter and cloud chamber examples

It is possible to learn a great deal by identifying what produced a trace. The theory may lead an experimenter to predict how the track changes as conditions in the chamber vary. The experimenter can use further observations to test the prediction. Van Fraassen's distinction asks what that evidence warrants believing about the visible phenomena and what additional belief would be involved in claiming that the theory's entire account of the cause is true. It does not require refusing to rely on a well-understood instrument or declining to say what an accepted theory represents as happening inside one. The Scientific Image, ch. II, §§1.3, 2

The argument leaves a real question about what humans can observe in the relevant sense. Van Fraassen acknowledges that the predicate is vague. A useful distinction can have good examples on each side even if a succession of intermediate cases gives no obvious place for a boundary. How much this concession helps depends on what one needs the boundary to do. If a particular inference depends on treating an object as observable, the critic can still ask whether the account says enough about that object and the conditions under which we might encounter it. The Scientific Image, ch. II, §2

What Acceptance Lets People Do

Researchers may accept a theory by setting up experiments with it in mind, framing questions in its terms, and treating its explanations as useful resources. They can spend months working out what the theory says about a phenomenon it has never faced before. Van Fraassen calls this the pragmatic dimension of acceptance. The decision to pursue the work is more than a prediction that the theory will pass the next test. It is a commitment to confronting further phenomena through the theory's resources, including the possibility of discovering where it fails. The Scientific Image, ch. II, §1.3 and ch. IV

Suppose the theory of the chamber gives researchers a convenient way to classify tracks so they can compare records from several laboratories. They can use that vocabulary, ask how the tracks depend on conditions, and teach someone else to run the instruments. Those activities can matter even when there is reason to stay uncertain between different accounts of the underlying processes. Choosing the vocabulary can shape which questions people spend time on and which experiments they can formulate. Believing all of the theory's claims would be another thing a person could do, over and above their use of it in those activities. The Scientific Image, ch. II, §1.3 and ch. IV

One can see the difference by asking why two devices produced different readings. Somebody examining whether the devices were put in different places will be interested in a different explanation from a person checking whether one had an exhausted battery. Van Fraassen develops this point in his account of explanation. What adequately answers a why-question partly depends on what the person asks and what they are taking for granted. That context determines which information is relevant. Offering one good answer does not settle every question about the processes described by the theory used to give it. The Scientific Image, ch. V, “The Pragmatics of Explanation”

Does Success Give Us Reason to Believe More?

A realist can press the significance of a theory's success. When a theory keeps producing correct predictions in unfamiliar cases, it is reasonable to ask what accounts for the success. Van Fraassen discusses this through the argument that we infer the best explanation. The ordinary example is a mouse in the wall. Noises and the disappearance of food can give you reason to believe that a mouse is present. The same pattern of inference might seem to support belief in the unobservable things scientists use in giving explanations. The Scientific Image, ch. II, §§3–4

Van Fraassen asks whether the successful cases show that people infer a theory's truth or that they infer what it gets right about the observable phenomena. Both descriptions can fit the mouse example, because the mouse is an animal that could be seen. The two ways of describing the conclusion diverge when one accepts an account of what cannot be observed in order to account for an observation. Relying on its explanatory success leaves a question about whether that success warrants belief in everything the explanation uses. His response is that a claim about what researchers actually infer needs to be compared with rival accounts of their practice. The Scientific Image, ch. II, §3, “Inference to the Best Explanation”

A critic can reply that a sustained pattern of precise success calls for more explanation than saying the models fit. For instance, a theory can generate successful predictions about both the shape of tracks in the chamber and what happens in a very different apparatus. That makes the relationship between the lines of evidence worth understanding. The constructive empiricist can recognize the success, compare it with alternatives, and use the theory to investigate how it occurred. The dispute concerns the further step from doing that work to treating the entire description as true. In van Fraassen's argument with J. J. C. Smart, even an account of why one theory reproduces the usable predictions of another need not start by treating the favored theory as wholly true. The Scientific Image, ch. II, §4, discussion of Smart

Return to the line in the chamber. Someone can describe what was observed, state what an accepted model says produced it, and specify the observable results the model predicts elsewhere. Those are different things to establish. Asking which of them the evidence warrants, and where the additional commitment of taking the whole account as true begins, makes the realist and constructive empiricist disagreement concrete. It also gives a scientist who uses the model something to examine before deciding how far to stand behind it. The Scientific Image, ch. II, §§1.3, 2–4, and ch. III

Further Reading

  • Bas C. van Fraassen, The Scientific Image. Chapter II sets out the debate about realism. Read Chapter III for models and empirical adequacy, and Chapter V for explanation.
  • Bas C. van Fraassen, The Scientific Image, 1980 text. Start with Chapter II, §§1–2, for theory acceptance, observability, and the cloud chamber example.
Join the discussionAsk a question or share what you learned.

Comments

Add to the conversation

A question, a correction, or an example from your own work.

No comments yet. Be the first to share your thoughts!

A different angle
All writing
Philosophy

Ayer's Language, Truth and Logic: Logical Positivism

A. J. Ayer's account of empirical verification, its limits for metaphysical claims, and the different work done by logic and ethical language.

13 min read
Philosophy

Feyerabend's Against Method: Epistemological Anarchism

Paul Feyerabend's 1975 challenge to a universal scientific method, using Galileo and other historical cases to show why rival approaches matter.

13 min read
Philosophy

Kuhn's Structure of Scientific Revolutions: Paradigms and Normal Science

Thomas Kuhn's 1962 account of how scientific communities solve problems, respond to anomalies, and change the frameworks that guide their research.

13 min read

About the author

Reference

Citation details

Cite or share this article.

BIBTEXAcademic
@misc{brenndoerfer2026vanfraassen, author = {Michael Brenndoerfer}, title = {van Fraassen's The Scientific Image: Constructive Empiricism}, year = {2026}, url = {https://mbrenndoerfer.com/writing/van-fraassen-scientific-image-constructive-empiricism-guide}, organization = {mbrenndoerfer.com}, note = {Accessed: 2026-11-19} }
APAAcademic
Michael Brenndoerfer (2026). van Fraassen's The Scientific Image: Constructive Empiricism. Retrieved from https://mbrenndoerfer.com/writing/van-fraassen-scientific-image-constructive-empiricism-guide
MLAAcademic
Michael Brenndoerfer. "van Fraassen's The Scientific Image: Constructive Empiricism." 2026. Web. November 19, 2026. <https://mbrenndoerfer.com/writing/van-fraassen-scientific-image-constructive-empiricism-guide>.
CHICAGOAcademic
Michael Brenndoerfer. "van Fraassen's The Scientific Image: Constructive Empiricism." Accessed November 19, 2026. https://mbrenndoerfer.com/writing/van-fraassen-scientific-image-constructive-empiricism-guide.
HARVARDAcademic
Michael Brenndoerfer (2026) 'van Fraassen's The Scientific Image: Constructive Empiricism'. Available at: https://mbrenndoerfer.com/writing/van-fraassen-scientific-image-constructive-empiricism-guide (Accessed: November 19, 2026).
SimpleBasic
Michael Brenndoerfer (2026). van Fraassen's The Scientific Image: Constructive Empiricism. https://mbrenndoerfer.com/writing/van-fraassen-scientific-image-constructive-empiricism-guide
Newsletter

Stay up to date

Get new articles, book updates, and notes on AI and technology.

Unsubscribe at any time. Privacy policy