Interpreting Causality
0 likes893 views
The document discusses causality in health sciences. It argues that determining causality requires both probabilistic evidence from statistical analyses of studies as well as mechanistic evidence about biological mechanisms. A monistic view that only considers one type of evidence is insufficient, as is a pluralistic view that treats different types of evidence as completely separate. Instead, the document advocates an epistemic view of causality, where causal beliefs are determined by all the available evidence and must account for all known dependencies.
Interpreting Causality
- 1. Interpreting causality in the health sciences Federica Russo & Jon Williamson Philosophy, University of Kent
- 2. Causality and probability in the sciences An ongoing project at Kent www.kent.ac.uk/secl/philosophy/jw/2006/CausalityProbability.htm In the sciences About probability We defend an ObjectiveBayesian-cum-Frequency interpretation About causality We defend an epistemic interpretation
- 3. Overview Two types of evidence Probabilistic Mechanistic Against causal monism Against causal pluralism The case for epistemic causality
- 4. Probabilistic evidence Observed dependencies in a range of similar studies Coherent results Tests for stability in structural models …
- 5. Mechanistic evidence Biomedical mechanisms Chemical reactions, electric signals, alterations at the cellular level, … A plausible (physiological) link from the cause to the effect
- 6. We need both types of evidence Semmelweis and puerperal fever He had statistics but the link wasn’t accepted till backed with the mechanism Helicobacter pylory The causal relation was hypothesised based on probabilistic evidence and accepted when backed with the mechanism
- 7. A classic: Bradford Hill’s criteria Strength of association Temporality Consistency Theoretical plausibility Coherence Specificity in the causes Dose response relationship Experimental evidence Analogy
- 8. Contemporary medicine: IARC IARC reviews published studies Assessment of causality depends on: Presence of a plausible mechanism Probabilistic evidence (e.g. frequencies, risks)
- 9. Monistic accounts won’t do Mechanistic accounts Causal processes intersect with each other in interactive forks Probabilistic accounts Causes make a difference in the probability of the effect (ceteris paribus) Problem: Neither can handle the dual aspect of causal epistemology
- 10. Pluralistic accounts won’t do … either Uniformity of causal language: A single notion of cause is used The pluralist rebuts: A mechanistic cause 1 A probabilistic cause 2 Different meanings of cause But each refers to a single concept! Therefore, the pluralist has twice as the problems of the monist!
- 11. The way out: epistemic causality Rational causal beliefs: The agent’s evidence determines which beliefs to adopt A causal relation is the set of causal beliefs that an agent with total evidence should adopt
- 12. Constraints on causal beliefs The agent’s causal beliefs should account for all known dependencies that are not already accounted for by non-causal dependencies The agent’s causal beliefs should be compatible with other knowledge The agent should not have causal beliefs that are not warranted by her evidence
- 13. An application: epistemic causality in cancer science Dataset of clinical observations of past patients Dataset of observations at molecular level Probabilistic evidence Knowledge of biological mechanisms Mechanistic evidence Knowledge of semantic relations between variables Pro/contra mechanistic evidence
- 14. To sum up The health sciences need and employ two types of evidence Monistic accounts won’t do Pluralistic accounts won’t do either The epistemic account succeeds
- 15. To conclude … and to research ahead There is a key distinction between Evidence from which we draw causal conclusions and the concept of cause A single epistemic concept suits the case of the health sciences … and possibly the social and natural sciences too