Mechanisms, propensities, causation Isabelle Drouet Université Catholique de Louvain.

Mechanisms, propensities, causation

Isabelle Drouet

Université Catholique de Louvain

Two approaches to mechanisms (Glennan 2002)

higher levelslevel of physics

thing, with stable dispositions

sequence of singular events

MDC, GlennanSalmon, Dowe

complex-system approachprocess approach

c-s mechanisms and difference-making

• several papers dealing with the relationship between c-s mechanisms and causality as counterfactual dependence (Woodward 2002, Psillos 2004, Glennan 2009)

• question: is it mechanisms or counterfactuals that are fundamental?

p mechanisms and counterfactuals

• Counterfactuals involved in the definitions of mark transmission and of causal interaction in Salmon 1984

• Reference dropped in Dowe and Salmon 1994: conserved quantity theory

Questions

• Can't we say something more concerning the relationship between p mechanisms and– counterfactuals – more generally, difference-making?

• Remark: p mechanisms / c-s mechanisms vs. possible / impossible to explain “genuinely singular events” (Glennan 2002)

Approach

• Focus on propensities

• Justification: – provide the only physical interpretation for

probabilities of singular events– have to do with p mechanisms

• NB. One goal of the talk is to explicate this justification.

Outline

1. Propensities and p mechanisms

2. Propensities and difference-making

3. p mechanisms and difference-making

Outline

1. Propensities and p mechanisms

2. Propensities and difference-making

3. p mechanisms and difference-making

Popper’s propensity theory

• An objective, physical interpretation of singular probabilities

• Propensities as tendencies of physical situations to produce singular events

• Physical reality of propensities• Probabilities as measures of propensities

Propensities and processes

• Ontological similarities: physically real, dynamic / active, causally productive, somehow continuous

• Yet no identity of (some) processes with propensities: processes ≠ tendencies

Salmon (1984) on propensities and processes

• Propensities = probas that are carried by indeterministic causal processes (203)

• Problems:- difficult to express in terms of conserved

quantities- misses the Popperian probability / propensity

distinction

Dowe (1992) on propensities and processes

• “propensities should be regarded as referring to the operation of indeterministic causal processes and interactions” (213)

• Relies only on (Popperian) propensities being attached to the level of physics

• Implies there are no propensities (although there may be objective, physical probas) but at the level of physics

Outline

1. Propensities and p mechanisms

2. Propensities and difference-making

3. p mechanisms and difference-making

Probabilistic causality

• Analyses of causality in terms of inequalities between absolute and conditional probabilities

• At the level of physics, problem with the interpretation of conditional probabilities

Humphreys' paradox (1985)

• (CI):if p(A / B) is temporally inverse and p is given

a propensity interpretation,then p(A / B) = p(A / non-B) = p(A)

• logically incompatible with the calculus of probability

• conclusion: no propensity interpretation of conditional probabilities.

Against HP

• Justification of (CI): if p(A / B) is inverse, impossible for the propensity tending to realize A to be (physically) modified by the occurrence of B

• But: according to the propensity theory, the production relation is between a physical situation and sg events, not between two sg events (conditioning and conditioned).

Towards a propensity interpretation of p(A / B)

• Conditionalization as a modification of p• A propensity interpretation tells how a new

probability function, meaning a new physical situation, is defined out the initial one and B

• The definition can be such that the properties of conditionalization are accounted for

• Csq: one can define probabilistic notions of singular causality at the level of physics

Link with counterfactuals

• The new situation has to be different from the initial, actual one, i.e. p(A / B)- measures the propensity of a counterfactual

situation to produce A- measures a counterfactual propensity to produce

A- is a counterfactual probability

• At the level of physics, notions of probabilistic causality collapse into counterfactual analyses.

Outline

1. Propensities and p mechanisms

2. Propensities and difference-making

3. p mechanisms and difference-making

What I have claimed

• propensities pertain to the level of physics and they refer to indeterministic causal processes and interactions

• probabilistic causality can be defined at this level and in terms of propensities, and there it collapses with the counterfactual approach

p mechanisms and difference-making

• p mechanisms are more fundamental than difference-making:• they give reference to the propensity talk• propensities are needed to ground a

difference-making notion of singular physical causality

• Conceptual and ontological priority

Scope

• Not a conclusion specific to p mech.• A conclusion

– about mechanisms and difference-making at the same level

– concerning mechanisms• What is specific to p mechanisms:

– irrelevance of the debate concerning counterfactuals and characterizations of mechanisms

– problem with the interpretation of conditional probabilities.