Philosophical Bias in Science

Term coined by Andersen, Anjum & Rocca 2019 ("Philosophical bias is the one bias that science cannot avoid", eLife): basic implicit assumptions in science of an ontological, epistemological, or normative nature, which skew hypothesis formation, experiment design, evidence evaluation and interpretation of results.

Unlike confirmation bias or cognitive bias, philosophical bias cannot be eliminated — only made explicit and critically examined. "Philosophy-free science is simply not an option" (sources/2023-anjum-rocca-phi403-lecture-01-metascience).

Types

Ontological — implicit commitments about what exists (e.g. Descartes' mind–body dualism still structuring the division between psychology and medicine).
Epistemological — implicit commitments about what counts as knowledge (e.g. data must be observable; methods define evidence).
Normative — implicit commitments about how science should be done (e.g. "more data is better", "RCTs trump mechanisms").

Philosophical biases embedded in scientific methods (the PHI403 tour)

Scientific methods carry implicit commitments about causation, probability, and complexity:

Method	Implicit theory of causation
Statistical correlation	[[concepts/regularity-theory-of-causation
Comparative studies	Difference-making (Lewis)
Experimental methods	[[concepts/interventionist-theory-of-causation
RCT-based EBM	[[concepts/evidence-hierarchy
Bayesian reasoning	Credence, ratio analysis (sources/2023-anjum-rocca-phi403-lecture-17-conditional-probability)
Mechanistic reasoning	Productive-mechanism theory (concepts/mechanisms-in-science)
Case studies / N=1	Singularist concepts/dispositionalism

Detecting philosophical bias

Notice causal vocabulary ("increase", "prevent", "affect") smuggled into correlational conclusions (sources/2023-anjum-rocca-phi403-lecture-12-explanatory-power-of-mechanisms).
Ask what the method assumes about causation, and whether that matches the target phenomenon.
Thomas Kuhn: paradigmatic crisis = scientists starting to do philosophy about their own norms.

Relevance to this wiki

Any choice of BPM / PPM / APM method carries philosophical biases — for instance, benchmarking on aggregate metrics smuggles in a frequentist / Humean ontology. See concepts/rct-limitations, concepts/probabilistic-causation, concepts/causal-process-discovery.

concepts/causation · concepts/evidence-hierarchy · concepts/methodological-pluralism · concepts/rct-limitations