1. PH201/400 – Week 14
Causation

2. A Word of Caution

3. Disambiguation Unless stated otherwise we focus on discrete events:
Cause C and effect E
C and E are taken to be types
Variables and causal process appear later.

4. Conditions of Adequacy
Condition 1: Rule out accidental relations
A theory of causation must be able to distinguish between real causal relations and accidental relations.
Example:
Every time I buy a ticket to hear Martha Argerich she cancels.
But my buying ticket does no cause her cancelling.

5. Condition 2: Rule out common causes
It may be that two events are perfectly correlated not because C causes E but rather because they are concomitant effects of a common cause: A C E
versus C E
Examples:
Lightning, thunder and electrically the discharge of charged regions between a cloud and the ground.
Stained teeth, lung cancer and smoking

6. Condition 3: Causal Asymmetry
If C causes E, then (usually) E does not also cause C.
What explains this asymmetry and wherein exactly does this asymmetry lie?
Example:
Discharge causes thunder, but thunder does not cause discharge.

7. Condition 4: Deal with pre-emption
C causes E. A pre-empted (potential) cause P is an even that brings about E in the absence of C but has been ‘pre-empted’ to act by C.
Examples:
National grid and the power generator
Two assassins
Challenge: Distinguish between real causes and pre-empted backups

8. Types of Analyses Reductive:
Provide truth conditions for causal claims in non-causal terms.
Example: regularity accounts.
Motivation:
Epistemic: we don’t observe causality directly and so to assess the truth of causal claim we to have to translate it into something non-causal (remember Hume!).
Metaphysical: ontological parsimony favours keeping the number of basic object low.

9. Example: Manipulability accounts.
Non-reductive:
Causality is basic. Causal claims can therefore not be rephrased in non-causal terms.
Example: Manipulability accounts.
Resist motivation for reductive accounts:
Epistemic: either deny that we don’t observe causality directly or understand causal claims as being on par with claims about theoretical entities (remember weeks 1-4!).
Metaphysical: causal relation are indispensible.
Aim of an analysis: understand the relation between causality and other concepts of interest: laws of nature, intervention, etc.

10. Regularity Theories Hume’s Account C causes E iff
(i) C is (spatio-temporally) contiguous to E
(ii) C occurs before E and
(iii) All C’s are invariably followed by E’s (constant conjunction).
(cf. Hume readings in week 5)

11. Problems I: Non-sufficiency
Relations can be non-causal and yet satisfy Hume’s conditions:
Accidental causes: Martha Argerich’s cancelling her concerts.
b) Concomitant effects which have a common cause satisfy (i)-(iii), but they aren’t real causes. (cf. Condition 2): Lightning and thunder.

12. c) Problems with pre-emption: Whenever the generator is in good working order the lights in the house are on. Either this is because the national grid provides electricity or because the generator does. But only in the latter case would the generator be the cause of the light being on.

13. Problems II: Non-necessity
There are relations that are causal, but do not satisfy Hume’s conditions.
Imperfect regularities: C can cause E but not all occurrences of a C are followed by an E: smoking and lung cancer
Non-temporally ordered causes: Not all causes are prior to their effects. (cf. Condition 3): functional relations, e.g. turning the amplifier loudness button and the music getting louder.

14. Often, a plurality of causes can be responsible for an effect.
2. Mill’s Account
Like Hume, Mill held a regularity view but qualified Hume’s account in four ways:
It is never a single circumstance responsible for an effect but a set of conditions.
Often, a plurality of causes can be responsible for an effect.
Example: the house does not burn just because fell; there has to be oxygen, inflammable material, etc.

15. Hence: A cause is a set of antecedents or causal conditions which is invariably followed by its effect: ABCD … E
Assessment: This somewhat improves on the problem of imperfect regularities, but doesn’t solve the other problems of Hume’s account.
Exercise: which, if any, of the problems of Hume’s account are solved in Mill’s account?

16. Sometimes this is called the ‘plurality principle’.
3. Mackie’s Account
Mackie added to Mill’s account that there can be several sets of conditions that lead to E:
ABC… or A’B’C’ or … or … E
Sometimes this is called the ‘plurality principle’.
Example: The house can burn because of candle (and circumstances) or a short circuit (and circumstances) or …

17. A cause then is one of the factors in one of disjuncts, for instance B, in
ABC… or A’B’C’ or … or … E
such that if ABC… occur together E follows, but if only AC… occur then E does not follow . This allows for B to occur without E.
B then is an:
Insufficient but
Non-redundant part of an
Unnecessary but
Sufficient condition

18. So a cause is an ‘INUS’ condition.
A cause then is one of the factors in one of disjuncts, for instance B, in
ABC… or A’B’C’ or … or … E
such that if ABC… occur together E follows, but if only AC… occur then E does not follow . This allows for B to occur without E.
B then is an:
Insufficient but
Non-redundant part of an
Unnecessary but
Sufficient condition
So a cause is an ‘INUS’ condition.

19. Assessment:
INUS conditions improve the situation with imperfect regularities.
But they still suffer from problems with:
Accidental Causes
Common causes
Non-temporal causes
Pre-emption
Exercise: Give examples.

20. Manipulability Accounts
Main idea:
Causes are a means to produce an effect.
Agents can therefore use causes to manipulate and control effects: intervene on C to bring about E.
Long pedigree: Collingwood (1940), Gasking (1955), von Wright (1971)
This lecture: Woodward (2003).

21. A first shot:
C causes E iff someone manipulates C and thereby brings about E.
Example: We flip the switch and the light goes on.
Immediate objection: Causality can present in things that are in the past, are far away and are too large to manipulate.
So this idea needs elaboration.

22. Woodward states the main idea as follows:
Woodward’s Account
Woodward states the main idea as follows:
‘my idea is that one ought to be able to associate with any successful [causal] explanation a hypothetical or counterfactual experiment that shows us that and how manipulation of the factors mentioned in the explanation […] would be a way of manipulating or altering the phenemena explained. […] an explanation ought to be such that it can be used to answer what I call a what-if-things-had-been-different question: the explanation must enable us to see what sort of difference it would have made for the explanadum if the factors cited in the explanans had been different in various possible ways.’ (2003, p. 11)

23. This is in effect a definition of causation.
Let X and Y be variables.
Example: X is the position of the oven dial and Y is the temperature in the oven.
Causation: X causes Y iff the value of Y would change if we were to intervene on the value of X (in some background circumstances).

24. Example: Inclined plane
Friction: f = c m g cos(θ), where c is the friction coefficient.
Net accelerating force Fa= m g [sin(θ) – c cos(θ)]
Acceleration: a = g [sin(θ) – c cos(θ)]

25. Relations:
One can change the acceleration by changing the inclination θ: If θ had been different, then a would have been different.
 causal relation.
One cannot change the acceleration by changing the mass m: If m had been different, then a would not have been different
 non-causal relation.
One can change the acceleration by changing the gravitational constant g: If g had been different, then a would have been different.

26. Counterfactuals
An intervention need not be feasible. All we need is counterfactual dependence.
This allow us to account for causal relations in things cannot manipulate.
Example: The mass extinction of the dinosaurs is caused by asteroid: the dinosaurs would not have died out when the did if we had manipulated the trajectory of the asteroid so that it had missed the earth.

27. But: not every counterfactual dependence is indicative of causality.
Invariance
But: not every counterfactual dependence is indicative of causality.
Example of inclined plane: If θ had been different, then a would have been different.
More: The formula a = g [sin(θ) – c cos(θ)] tells us by how much a change in g changes a, and this functional relation does not change when we intervene.
This is invariance: a functional relationship is invariant (under an intervention) if it continues to hold under the intervention.

28. Example: Barometer
Functional relation between barometer readings B and the weather:
960mb corresponds to stormy weather
1030mb to sunshine.

29. Now intervene on B: and turn the needle from 960 to 1030:
Counterfactual dependence: It’s raining, but had the dial been at 1030, the weather would have been sunny.
Now intervene on B: and turn the needle from 960 to 1030:
It fails to bring about sunshine
It also destroys the functional relation (it’s still raining but the pointer is now on 1030)  Failure of invariance.
Hence: Counterfactual dependences have to be invariant to for the relation to be causal.

30. Type of analysis:
This is non-reductive account since intervention is itself a causal concept.
But this does not render the account circular or empty: it analyses the relation between cause, intervention, invariance, dependence, and explanation in particular way, and this way is neither obvious nor uncontroversial.,

31. Counterfactual Accounts
David Lewis an influential analysis of causation in terms of counterfactual dependence.
For a discussion see the entry Counterfactual Theories of Causation in SEP.

32. Probabilistic Accounts
We have encountered the main idea in connection with probabilistic explanation. For a discussion see see the entry Counterfactual Probabilistic Causation in SEP.

33. Causal Processes
The classical source is Salmon For a discussion see the entry Causal Processes in SEP.