Scientific Realism and the No Miracles Argument | Philosophy of Science

The realism-antirealism debate in contemporary analytic philosophy concerns whether reality, truth, and meaning exist independently of human cognition or are constructed within our conceptual frameworks; realism holds that truth is independent and eternal (requiring discovery), while anti-realism argues that truth must be cognition-immanent and decidable within finite time, with reality being shaped by human language and conceptual schemes, as exemplified by Wittgenstein's later philosophy, Dummett's verificationist semantics, and Putnam's internal realism.

The realism versus anti-realism debate concerns whether scientific theories aim to describe truth about unobservable reality. Realists believe there are facts about what exists beyond observation, and that successful scientific theories reveal these truths. Anti-realists (like constructive empiricists) argue only observable predictions matter; theories need not be true about unobservables. Historical examples illustrate this: Galileo's heliocentric model was initially rejected by religious authorities who preferred geocentrism, yet it proved correct. The debate raises profound questions about whether science reveals reality or merely constructs useful models. Realists see science as progressively approximating truth, while anti-realists view it as a pragmatic tool for prediction.

Academic disciplines are often polemically organized between realists and anti-realists, but they frequently talk past each other. Berkeley's idealism denied a 'real world behind appearances' but did not deny that rocks exist or that people stub their toes. Logical positivists like Carnap argued that realism claims are pseudo-statements without factual content, yet this position is equivalent to Nietzsche's view that the real world became a fable. The word 'real' is primarily a contrastive term, and most debates fail to spell out their contrast classes. When someone says 'race isn't real,' they might mean it's not a biological category or that people don't experience racism. Most anti-realists do not deny experiencing cups, and the distinction between mind-dependent and mind-independent is more nuanced than a simple opposition. The realism-anti-realism debate is often pathological and prevents progress in other areas.

Philosophers have called reality into question with two main sides to this argument. Realists believe that reality is independent of any beliefs, perceptions, or cultural artifacts. Anti-realists believe the exact opposite, that reality is dependent upon the mind or cultural artifacts. This argument is essentially about the objectivity of reality.

There is a debate between realism and anti-realism about kinds. Realists believe kinds like tables really exist, while anti-realists argue that tables are merely ways of world-making—our projections that constitute reality. The argument for realism is that if there is something in reality, it has some nature independent of us. We cannot understand a position in which there is something but the entity has no nature at all. This supports the view that kinds are real features of reality, not merely linguistic constructs.

Scientific realists use abductive reasoning (inference to the best explanation) to argue for their position. The process involves: identifying what needs to be explained, generating multiple possible explanations, evaluating which is most probable, and concluding the best explanation is likely true. The detective footprint example illustrates this: given a footprint, the best explanation is that a human walked there, not that a cow with shoes or an alien did. Criteria include parsimony (fewest assumptions), coherence with existing knowledge, and explanatory power.

Abductive reasoning (also called adaptive reasoning or inference to the best explanation) is a form of logical inference that starts with observations and seeks the most likely or best explanation for them, unlike deductive reasoning (which moves from general rules to certain specific conclusions) or inductive reasoning (which moves from specific instances to probable generalizations). This reasoning process is hypothetical in nature, not certain, and is used in fields like medicine, detective work, scientific research, and everyday decision-making to generate new ideas and theories.

This comprehensive segment establishes abductive reasoning as a distinct form of inference beyond inductive arguments. The instructor explains that while causal arguments identify patterns of association, abductive reasoning focuses on explaining why such links exist. The core concept is that abductive inference derives strength from the idea that if a conclusion were true, it would explain all premises—this is also called 'inference to the best explanation' (ITBE). The lecture emphasizes that unlike deductive or inductive reasoning, abductive reasoning begins with phenomena demanding explanatory satisfaction and works backward to propose hypotheses. Students learn that effective abductive reasoning requires entertaining multiple competing explanations before selecting the most adequate one. The segment introduces Vaughn's six criteria of adequacy: consistency, testability, fruitfulness, scope, simplicity, and conservatism, which operate under the principle 'ceteris paribus' (all other things being equal).

Abductive reasoning, or inference to the best explanation, is a method of reasoning where one selects the most likely explanation for an observation based on available evidence. The theist claims to use this by stating that since there is no other explanation for their experiences, supernatural causation must be the best explanation. However, this reasoning is flawed because it assumes that the absence of a known natural explanation automatically makes supernatural explanation the best one. In reality, there may be natural explanations that have not yet been discovered, or the experience may be a hallucination or illusion.

Abductive reasoning, also known as inference to the best explanation, is an inductive reasoning process where one generates multiple plausible explanations for a phenomenon and selects the best one based on criteria such as scope, explanatory power, simplicity, plausibility, concordance, and superiority; this method, used by scientists like Darwin and Lavoisier, does not provide certainty but rather points toward the most probable explanation given the available evidence.

Empirical success in science should not be defined solely by result-oriented measures such as data accommodation, instrumental reliability, or predictive accuracy. Instead, empirical success should be evaluated by whether a model or theory allows us to derive new questions and integrate findings into broader scientific frameworks. The replication crisis in science reveals problems with overemphasizing results as endpoints rather than recognizing them as starting points for further investigation. Good scientific models are productive question-generators that open up new avenues of inquiry rather than providing final answers. This shifts the focus from achieving definitive results to using scientific findings as foundations for continued exploration and refinement of our understanding.

This segment explores the boundaries of scientific empiricism. The speaker argues that science cannot prove foundational claims such as the existence of the external world, reason, or logic - these are self-evident premises that science itself depends upon. The claim that 'science is the best representation of truth' is self-refuting because it cannot be empirically tested. The speaker also discusses how empirical success does not guarantee truth, citing Larry Laudan's presentation of 30 empirically successful theories that turned out to be false. This demonstrates that scientific validity requires more than just empirical success.

The argument from empirical success suggests that if a scientific theory works extraordinarily well empirically, we should be realist about it—that is, we should believe the theory approximately describes how the world actually is. This argument is commonly used to justify taking successful scientific theories seriously as representations of reality.

Science has achieved remarkable empirical success, transforming medicine, technology, and daily life. Cancer survival rates for children have improved dramatically—from one in ten surviving leukemia in the 1960s to eight in ten today. Vaccines, antibiotics, and fertility treatments have saved countless lives. Technologies like internal combustion engines, aircraft, computers, and the internet emerged from scientific understanding. Even everyday items like autofocus cameras and hair straighteners rely on scientific discoveries. Science is self-correcting—when theories prove inadequate, better explanations emerge. However, excessive trust in science leads to significant problems. Science is often taught as fixed facts, creating misunderstanding of its probabilistic nature. Medical treatments' effectiveness data is often decades old—15-year life expectancy data for cancer treatments is typically unavailable because analyses take that long. Science can discard ideas like homeopathy but cannot determine policy values. For drugs policy, science can provide death rate comparisons but cannot decide which substances should be legal or illegal, as that requires value judgments beyond empirical data. The Italian earthquake prediction failure demonstrates consequences of over-trusting scientific predictions—scientists were prosecuted not because science failed but because they communicated risks inadequately. Distinguishing between science (the body of knowledge) and the scientific method (the process of inquiry) is crucial. The scientific method emphasizes skepticism and questioning, which actually supports the proposition that we shouldn't blindly trust science. Using the scientific method means constantly evaluating and testing claims rather than accepting them uncritically.

Petzschner argues that scientific terms can have heterogeneous reference potentials, meaning they can refer to different things in different contexts. The term 'phlogiston' had multiple reference tokens: one referring to oxygen (actual entity) and another referring to nothing (the theoretical substance). He argues that empirical success is connected to reference in scientific theories, contrary to Laudan's pessimistic meta-induction. The phlogiston theory was successful because it did refer to actual empirical experiences, such as the discovery of oxygen. This challenges the view that there is no necessary connection between empirical success and reference to actual entities, suggesting that successful theories typically do have some connection to reality.