Singular and General Causal Relations - PowerPoint Presentation

Slide1

Singular and General Causal Relations: A Mechanist Perspective

Stuart Glennan

Butler

UniversitySlide2

The singularist and generalist view of causation

The

generalist view: Particular events are causally related because they fall under general laws

The singularist view: Causal relations obtain between particular entities, and causal generalizations are true, to the extent they are true, in virtue of generalizing over singular causal facts.Slide3

Talk Outline

Productivity

, Relevance and Singularism

The grounds of the Singularist Intuition

Three “mechanistic “theories

Process theories

Mechanical theories

Manipulability theories

The character of fundamental interactions and the case for singularismSlide4

Productivity & Relevance

Sober, Elliott. 1984. Two concepts of cause.

PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association

Hall, Ned. 2004. Two concepts of causation. In

Causation and counterfactuals.

Godfrey-Smith, Peter. 2010. Causal pluralism. In

Oxford handbook of causation

Glennan, Stuart. forthcoming. Mechanisms, causes and the layered model of the world.

Philosophy and Phenomenological Research

. Slide5

Productivity and Relevance

Sometimes we have production without relevance (difference making): e.g. – overdetermination cases like firing squads

Sometimes we have relevance without production: e.g., omissions like the failure to break at a stop light.

Generalist or type level causal theories focus on relevance, while singularist theories focus on production – but there

are exceptionsSlide6

The Singularist Intuition

The Nature and Observability of the Causal Relationship

C.J. Ducasse, 1926

Causality and Determination

Elizabeth Anscombe, 1971Slide7

The Singularist Intuition

… [T]he cause of a particular event [is defined] in terms of but a single occurrence of it, and thus in no way involves the supposition that it, or one like it, ever has occurred before or ever will again. … And recurrence becomes related at all to causation only when a law is considered which happens to be a generalization of facts themselves individually causal to begin with.

Ducasse 1926Slide8

The Singularist Intuition

Causality consists in the derivedness of an effect from its causes . This is the core, the common feature, of causality in its various kinds. Effects derive from, arise out of, come of, their causes. … Now analysis in terms of necessity or universality does not tell us of this derivedness of the effect; rather it forgets about that.

Anscombe 1971Slide9

Three Theories of Causality

Process

Theories

Wesley Salmon, Phil Dowe

Manipulability

Theories

Jim Woodward, Judea Pearl, also

Spirtes, Glymour and Scheines

Mechanical Theories

Glennan,

MDC, BechtelSlide10

Process Theories

Causal

p

rocesses are understood as world-lines of objects that propagate

causal influence

through

space-time.

When causal processes intersect they may interact, changing the properties of each process.

Causally related events must be connected by a continuous network of intersecting causal processes.

Process theories are clearly singularistSlide11

Problems for Process Theories

Process theories provide

an analysis of productivity, but have problems with relevance, including:

Irrelevant interactions

Relevant negative causes (omissions, preventions)

Reductive character of analysis of interactionsSlide12

Mechanical Theories

Mechanisms are systems consisting of a set of parts, entities, components.

Activities of and interactions between parts of mechanisms are regular -- characterized

by

counterfactually

invariant generalizations

Mechanisms are hierarchical in the sense that the parts of mechanisms may

themselves be partsSlide13

Advantages of Mechanical Theories over Process Theories

Mechanical theories seem to address the causal relevance problem, in part because of their reliance on counterfactual-supporting generalizations in describing relations between parts.

Because mechanical theories are hierarchical, they seem better suited than process theories to handle the fact that higher level properties are often the causally relevant ones.Slide14

Manipulability Theories

Causal relations are represented by directed acyclic graphs. Direct Causal Relations between nodes are characterized by functional relations

If X causes Y then an intervention on X will cause a change in Y in accordance with these functional relations.Slide15

A MechanismSlide16

DAG for a toiletSlide17

The character of fundamental interactions and the case for singularismSlide18

Some important points about generalizations, counterfactuals and mechanisms:

Interactions

between parts of mechanisms “can be characterized by direct, invariant, change-relating generalizations” (Glennan

2002)

These generalizations support counterfactuals.

Outside of fundamental physics, these generalizations are mechanically explicable, meaning that the truth makers for these generalizations are further mechanisms

But eventually we bottom out and have to wonder what the truth maker of the generalization

is.Slide19

A Quick apology to modern physics

This picture of mechanisms presupposes a “classical” view of the fundamental physical entities and interactions

Entities must have definite properties and must be distinct from other entities. Interactions must in some sense be local.

Quantum mechanics tells us this is wrong

Somehow classical entities and properties emerge at some level of organization, and this is our “fundamental” level.Slide20

Psillos’ asymmetry thesis:

Because of this, Psillos (2004)

argues that there is a certain asymmetry between

counterfactual and mechanistic approaches

Counterfactuals are needed to underwrite fundamental interactions between mechanisms, but mechanisms can’t always be the truth-makers for counterfactuals.

A genuine case for the priority of counterfactuals over mechanisms would require us to have a reductive account of the truth makers for counterfactuals, but we (or Woodward at least) doesn’t have that.Slide21

Our key question

What in the hell does it mean

that, an interaction between fundamental parts of a mechanism “can be characterized by change relating generalization”?Slide22

Fundamental Interactions

Humean

Lawlessness

The interaction is nothing more than an instance of a pattern that is described by a generalization

Nomological

Determination

The interaction is governed by the generalization (law)

Singular Determination

The interaction is a singular case of causal determination and any generalizations describing interactions are true in virtue of there being a general pattern of such singular instances.Slide23

Taking the Positivist Route

One option here is to simply reject the question of which interpretation is right, since it is empirically

undecidable

.

This is mechanisms sans metaphysics, which is good if metaphysics is nonsense.Slide24

Against Humean Lawlessness

Humean

lawlessness is antirealist about laws and causes.

There are no genuine modal relationships.

Singular counterfactual claims are not really claims about what would have happened in a single case.

Manipulation, like other forms of causing, is a fiction.

The view is, as Mumford claims, “irrefutable, but neither compelling, appealing, nor intuitive.”Slide25

Against Nomological Determination

The main argument for the nomological determination view is that it explains why there are fundamental level regularities.

The singularist responds that the fact that an interaction at one place is productive should not depend upon what happens elsewhere.

It does not follow from the fact that we live in a world in which fundamental interactions fall under patterns that it is in virtue of these patterns that the productive relationship holds.

We could live in a higgledy-piggledy worldSlide26

For singular determination

The singularist view of determination that is consistent between the fundamental and higher-levels.

Laws are typically understood to be relations that hold in virtue of the properties of the things related.

But the properties of complex things are not basic facts about those things, but are mechanically explicable. Higher level properties and laws depend upon particulars

Consequently, it would be good if the same held at the fundamental level.