PHI3400: Philosophy of Science

Course Description

PHI3400 Philosophy of Science asks what makes science different from other ways of claiming to know things — how scientific explanation works, what evidence establishes, and why a method that is demonstrably successful is so hard to characterise precisely.

The course is offered at approximately four Florida institutions: Florida International University, Florida State University, the University of Florida and the University of West Florida.

The University of West Florida places it in the College of Arts, Social Sciences and Humanities, Department of History and Philosophy at 3 semester hours, covering "concepts and types of explanation used in sciences," and noting it "may include differences between natural and social sciences, inductive reasoning and scientific explanation, and relation of science to society." It meets the College-Level Communication Skills Requirement.

⚠ Florida State University titles it History and Philosophy of Science and describes "a close look at some of the crucial philosophical problems of the sciences as they have developed throughout history, from Aristotle through Galileo, Pasteur, and Einstein, including what methods count as scientific, along with a consideration of how science has changed the world and the role of values."

Same subject, different centre of gravity. UWF's is organised around problems — explanation, induction, the natural/social science distinction. FSU's is organised around history, working through episodes to reach the same problems. Both cover the core; the historical version spends more time on cases and less on the analytic apparatus. Neither affects transfer, but the reading list will differ noticeably.

The problem of induction is where the course starts and where it never entirely finishes. Science generalises from observed cases to unobserved ones, and Hume's argument that this cannot be justified without circularity has never been satisfactorily answered. Every attempt to justify induction either assumes the uniformity of nature — which is what needs justifying — or appeals to induction's past success, which is itself an inductive argument. ⚠ Students find it genuinely unsettling that the most successful knowledge-producing enterprise in history rests on an inference nobody can fully vindicate, and the correct response is not to resolve the discomfort but to understand what does and does not follow from it.

The demarcation problem is the course's most publicly useful content. What separates science from pseudoscience? Popper's answer — falsifiability — is the one most students arrive already knowing, usually in a simplified form, and the course's job is to show both why it was a genuine advance and why it does not work as a criterion. Real theories are not abandoned when a prediction fails; auxiliary assumptions are adjusted, instruments are questioned, anomalies are set aside. ⚠ The Duhem-Quine thesis — that hypotheses face evidence only in bundles, never singly — is the reason no experiment ever refutes exactly one thing. Nothing else in an undergraduate curriculum equips a student as directly to evaluate a public claim that something is or is not "scientific."

Kuhn is the pivot of the course. The Structure of Scientific Revolutions (1962) replaced the picture of science as steady accumulation with normal science punctuated by revolutions in which a paradigm — a whole framework of assumptions, methods and exemplars — is replaced rather than corrected. The claim that competing paradigms are incommensurable, so that adherents partly talk past one another, raises the question the rest of the course circles: does science progress toward truth, or merely change? ⚠ Kuhn's argument is routinely misused in popular writing as a licence for relativism about scientific claims, and one of the more valuable things this course does is show carefully what he did and did not argue.

Realism versus anti-realism is the metaphysical question. Do the unobservable entities of successful theories — electrons, fields, genes — exist, or are theories instruments that predict well? The no-miracles argument holds that realism is the only explanation of science's success that does not make it a miracle. The pessimistic meta-induction replies that past successful theories posited entities we no longer believe in — phlogiston, caloric, the luminiferous ether — so present success is weak evidence of truth.

⚠ The values material has become the most contemporary part of the course. Whether science can or should be value-free, how values enter through problem choice and standards of evidence, and — increasingly — the replication crisis, publication bias, and the sociology of how scientific consensus is actually produced. This is where the course connects directly to public disputes about scientific authority, and where a philosophically careful treatment is most obviously worth having.

Learning Outcomes

Required Outcomes

• Explain what philosophy of science studies, and distinguish it from the history and the sociology of science.
• Explain the problem of induction and evaluate the principal responses to it.
• Explain the demarcation problem and evaluate falsificationism as a solution, including its known failures.
• Explain the hypothetico-deductive model of scientific method and its limitations.
• Explain the Duhem-Quine thesis and the underdetermination of theory by evidence.
• Distinguish and evaluate models of scientific explanation — deductive-nomological, statistical, causal, unificationist, and explanation by mechanism.
• Explain the relationship between explanation, prediction and causation, and why the three come apart.
• Explain the role of laws of nature and evaluate competing accounts of what a law is.
• Explain the nature and function of scientific models and idealisation.
• Explain confirmation theory, including Bayesian approaches, and the paradoxes of confirmation.
• Explain Kuhn's account of normal science, paradigms, anomalies, revolutions and incommensurability, and evaluate it.
• Compare Kuhn with Popper, Lakatos and Feyerabend on how theory change occurs and whether it is rational.
• Explain and evaluate scientific realism and anti-realism, including the no-miracles argument and the pessimistic meta-induction.
• Explain theory-ladenness of observation and its consequences for the idea of neutral evidence.
• Analyse the relationship between science and values — the value-free ideal, inductive risk, and how values legitimately and illegitimately enter inquiry.
• Explain the distinction between the natural and social sciences and evaluate arguments about whether social inquiry can or should follow natural-scientific method.
• Apply the course's apparatus to a real scientific episode or a live public dispute.
• Write a sustained philosophical argument engaging a serious objection.

Optional Outcomes

• Analyse historical case studies — the Copernican revolution, Galileo, Newton, Darwin, Pasteur and germ theory, relativity, plate tectonics, quantum mechanics.
• Analyse philosophical problems specific to biology — the species concept, function and teleology, levels of selection, reductionism.
• Analyse philosophical problems specific to physics — space and time, the interpretation of quantum mechanics, determinism.
• Analyse philosophical problems in the social sciences — intentional explanation, methodological individualism, prediction and reflexivity.
• Analyse the replication crisis, publication bias, p-hacking and research integrity philosophically.
• Analyse science and public policy — expertise, the precautionary principle, communicating uncertainty, manufactured doubt.
• Analyse feminist and social epistemology of science, including standpoint theory and the role of diversity in inquiry.
• Analyse reductionism and emergence across the sciences.
• Analyse scientific ethics — human and animal subjects, dual use, conflicts of interest.

Major Topics

Required Topics

• What philosophy of science is; the field's questions and methods.
• Induction — Hume's problem, Goodman's new riddle, responses.
• Demarcation — Popper, falsifiability, and its failures; pseudoscience as a case.
• Scientific method — hypothetico-deductive model, inference to the best explanation.
• Underdetermination — the Duhem-Quine thesis, auxiliary hypotheses.
• Explanation — covering-law, statistical, causal, unificationist, mechanistic models.
• Laws of nature.
• Models, idealisation and representation.
• Confirmation and evidence — Bayesian confirmation, the raven paradox, the base rate problem.
• Theory change — Kuhn, Lakatos, Feyerabend; normal science and revolutions.
• Incommensurability and progress.
• Realism and anti-realism.
• Observation and theory-ladenness.
• Science, values and society.
• Natural versus social sciences.
• Philosophical writing and argument.

Optional Topics

• Historical case studies — Copernicus to Einstein.
• Philosophy of biology.
• Philosophy of physics.
• Philosophy of social science.
• The replication crisis and research integrity.
• Science, expertise and public policy.
• Feminist and social epistemology of science.
• Reductionism and emergence.
• Research ethics.
• Philosophy of medicine — evidence-based medicine, disease concepts.

Resources & Tools

• Standard textbooks: Godfrey-Smith, Theory and Reality — the most widely adopted introduction, clear and genuinely good; Okasha, Philosophy of Science: A Very Short Introduction — short, excellent, and a good pre-term read; Chalmers, What Is This Thing Called Science?; Ladyman, Understanding Philosophy of Science.
• Primary texts: Kuhn, The Structure of Scientific Revolutions — frequently assigned entire, and the single most influential book in the field; Popper, The Logic of Scientific Discovery and Conjectures and Refutations (the latter more readable); Hempel, Aspects of Scientific Explanation; Lakatos, "Falsification and the Methodology of Scientific Research Programmes"; Feyerabend, Against Method; van Fraassen, The Scientific Image; Hacking, Representing and Intervening.
• Anthologies: Curd and Cover, Philosophy of Science: The Central Issues — the standard reader, with excellent commentaries.
• Where the course is historical (FSU's version): Kuhn, The Copernican Revolution; Galileo's Dialogue in selection; Darwin's Origin in selection; Shapin, The Scientific Revolution.
• Stanford Encyclopedia of Philosophy (plato.stanford.edu) — free and peer-reviewed; its entries on scientific explanation, confirmation, realism, Kuhn and the demarcation problem are the standard reference and are frequently assigned directly. PhilPapers for locating literature.
• Journals: Philosophy of Science, The British Journal for the Philosophy of Science, Studies in History and Philosophy of Science, Synthese. The Philosophy of Science Association is the relevant body.
• ⚠ The most useful supplementary resource is a real scientific controversy you know something about. Students with a science background should bring it — the course is markedly better when someone in the room can say what actually happens in a laboratory, and instructors welcome it.

Career Pathways

This is a philosophy elective, and its value is transferable reasoning rather than a credential. But it has an unusually specific audience: science students who will spend a career making evidential claims.

• Research careers in any science (SOC 19-1xxx, 19-2xxx) — ⚠ this course is the one place in most curricula where the concepts scientists use daily — evidence, explanation, confirmation, significance — are examined rather than assumed. Researchers who have thought about underdetermination and inductive risk design and interpret studies better.
• Medicine and the health professions — evidence-based medicine is a philosophical position with known difficulties, and diagnostic reasoning is inference to the best explanation under uncertainty. Philosophy is a well-regarded pre-medical major; Florida medical schools include UF, FSU, USF Morsani, UCF, FIU Wertheim and FAU Schmidt.
• Science policy and regulatory analysis (SOC 19-3094, 13-1041) — ⚠ the values-in-science and expertise material is directly the substance of this work, which involves deciding what evidence suffices for action under uncertainty.
• Science journalism and communication (SOC 27-3023, 27-3042) — the ability to characterise what a study establishes, and what it does not, is the core skill and is in short supply.
• Law (SOC 23-1011) — ⚠ expert evidence admissibility under the Daubert standard turns explicitly on demarcation criteria, including falsifiability and peer review. This course is unusually direct preparation for scientific-evidence practice, and philosophy majors post among the highest mean LSAT scores of any major.
• Data science and analytics (SOC 15-2051) — causal inference, model selection and the interpretation of statistical evidence are the applied edge of this course's content.
• Research ethics and integrity (SOC 11-9121, 13-1041) — institutional review boards, research compliance offices, and integrity roles at universities and health systems.
• Science education (SOC 25-2031, 25-1052) — the nature of science is an explicit component of science education standards, and it is the part teachers are least prepared for.
• Museums and public science (SOC 25-4012, 25-9031).
• Postsecondary teaching and academic philosophy (SOC 25-1126) — requires a PhD; philosophy of science is one of the field's larger areas and is often housed jointly with history of science.

Special Information

Title and emphasis — problems or history?

	UWF	FSU
Title	Philosophy of Science (= statewide)	History and Philosophy of Science
Organisation	by problem — explanation, induction, natural vs social science	by history — Aristotle, Galileo, Pasteur, Einstein
Also named	relation of science to society	what methods count as scientific; the role of values

The number is PHI3400 at all four institutions, so articulation is clean. Both versions cover the core problems; the historical version reaches them through cases and spends less time on the analytic apparatus, while the problem-based version does the reverse.

⚠ If you have a reason to prefer one, check the syllabus rather than the title. For graduate study in philosophy the problem-based version is closer to the professional literature; for a science student wanting context for their own field, the historical version is often more immediately illuminating.

Prerequisites and position in the curriculum

UWF lists no prerequisite, which is typical.

⚠ "No prerequisite" is not "no useful preparation", and this course has two distinct useful backgrounds that different students bring.

• Philosophy students benefit from logic (PHI3130) — much of the course is argument reconstruction, and the confirmation material uses formal notation — and from epistemology (PHI3300), since the questions are continuous with it.
• Science students benefit from having actually taken laboratory science, and this is worth stating because they frequently underrate it. Having designed an experiment, seen data fail to cooperate, and made a judgement about an anomaly gives you something no amount of reading supplies. ⚠ A statistics course is particularly valuable for the confirmation and replication material.

⚠ The two groups find different halves hard, and knowing which you are helps. Philosophy students handle the arguments comfortably and sometimes lack a feel for how science is actually done. Science students recognise the practice immediately and are less used to the demand that every claim be argued rather than cited. The course is at its best when both are in the room.

The course is a 3000-level upper-division elective, normally junior or senior year. It is a topic elective in philosophy and a common elective for biology, physics, chemistry, psychology, computer science and pre-medical students.

Course format and workload

3 credits, 45 contact hours — lecture and discussion, three hours per week, with substantial discussion.

Expect 6–9 hours per week outside class. ⚠ The reading is not long but it is slow — a twenty-page paper in this literature commonly needs two passes, and Kuhn in particular rewards re-reading. Assessment is normally two or three papers plus examinations, sometimes with short response papers or a presentation.

⚠ The papers are graded on argument, not on position, and this catches science students out more than philosophy students. Defending realism and attacking it are equally available; what is not available is asserting a view without engaging its strongest objection. A paper that reports what several philosophers said, without arguing, scores poorly everywhere.

⚠ What students find hardest

• Accepting that induction has no clean justification. Science students especially resist this, often by re-deriving one of the responses the course has already examined. The point is not that science is unjustified — it is understanding precisely what kind of justification it has.
• Kuhn without relativism. The step from "paradigms shape observation" to "science is just one story among many" is fast, popular, and not what Kuhn argued. Getting this right is one of the course's genuine achievements.
• Falsifiability is not a checklist. Students arrive with a simplified Popper and are surprised to find that the criterion does not do what it is popularly taken to do.
• Formal confirmation theory. Bayesian material involves conditional probability, and students without statistics find it a step up. ⚠ The base rate material is worth the effort regardless — it is the single most transferable thing in the course, and it explains why a positive result on a highly accurate test for a rare condition is usually a false positive.

Articulation and transfer

PHI3400 is a 3000-level upper-division course, not offered at Florida College System institutions, and taken after transfer. The number is consistent across the four Florida institutions that carry it, so SCNS articulation is clean.

⚠ Prefix note. PHI is philosophy by topic; PHH is history of philosophy; PHM social and political philosophy; PHP individual philosophers. These are not interchangeable for requirement purposes. Related PHI numbers include PHI3130 (logic — ⚠ and see that guide: it carries a symbolic version and a critical-reasoning version at different institutions), PHI3300 (epistemology), PHI3500/PHI4500 (metaphysics — ⚠ and note FSU numbers it 4500 where others use 3500), and separate numbers for philosophy of biology (PHI3452 at UWF) and philosophy of mind.

General-education note

UWF records that this course meets the College-Level Communication Skills Requirement. ⚠ That designation is institution-specific and does not automatically travel with the credit; confirm how a receiving institution classifies it rather than assuming.

AI Integration

⚠ This course has an unusually direct claim on AI as subject matter, because machine learning raises its central questions in a new and concrete form.

• Prediction without explanation. A model can predict accurately while offering nothing that counts as an explanation under any of the accounts this course studies — no covering law, no mechanism, no unification. ⚠ That forces the question of whether explanation was ever the point of science, or only prediction, which is the realism/instrumentalism debate arriving in a laboratory rather than a seminar.
• Induction, mechanised. Machine learning is inductive inference at scale, and it inherits every one of induction's problems — including the new riddle: a model that fits all past data can still generalise wrongly, which is what out-of-distribution failure is. Goodman's problem is now an engineering concern.
• Theory-ladenness of data. Training sets embody choices about what to measure and how to label it. There is no neutral dataset, which is the same point Hanson and Kuhn made about observation.
• Underdetermination made vivid. Many models fit the same data equally well and disagree about new cases. Duhem and Quine described this; model selection is the practical version.
• AI in scientific discovery. Systems that generate hypotheses or design experiments raise the question of what the scientist's role then is, and whether understanding is required for knowledge.
• Values in classification. The inductive-risk literature — that choosing an evidential threshold means choosing whose errors matter — applies directly to setting a decision threshold in a deployed model.

A course taught now can hardly avoid these, and instructors increasingly use them as the motivating cases.

Using AI tools for coursework. Models are competent at summarising a position — what falsificationism claims, how Lakatos differs from Kuhn — and reasonable at generating objections to a thesis you have written, which is the hardest part of a philosophy paper and a legitimate use.

⚠ Where they fail, and one failure is specific enough to be worth stating precisely: models reproduce the popular simplifications this course exists to correct. Ask about falsifiability and you will get the textbook criterion without the objections; ask about Kuhn and you may well get the relativist reading. The consensus in the training data on these questions is precisely the consensus the course is trying to displace.

They also fabricate citations and misattribute positions — assigning an argument to Popper that belongs to Lakatos, or inventing a paper title that sounds right — and they flatten interpretive disagreement where the scholarly dispute is the assignment.

Academic integrity. Read the syllabus; policies vary. Submitting generated prose as your own violates every Florida institution's policy. ⚠ One argument specific to this course: its value is the capacity to look at a claim — in a paper, a news report, a courtroom, a policy brief — and say precisely what the evidence supports and how strongly. That is built only by working through the arguments yourself, and it is a capacity, not a body of information.