10:06 Theory of causal explanation, not causation. Something in our - - PowerPoint PPT Presentation

James Woodward’s Manipulability Theory of Causal Explanation

The Scientific Revolution, Experimentation, and Causation HILR—Fall, 2017

1

10:06 Theory of causal explanation, not causation. Something in our heads, not in the world.

Acceleration of block on inclined plane

• Angle of plane
• Color of block
• Strength of gravity
• Mass of block
• Roughness of surfaces
• Material of block

10:08 = Before stripping:

• Not go into math
• Use your intuitions

= Next: Woodward’s theory concerns causal explanation

Causal Explanation A causal explanation proceeds by showing how an outcome depends (not logically or conceptually)

• n other variables or factors,

thus furnishing information relevant to manipulation and control.

10:13 = At end of stripping: > Why “not logically or conceptually”? > Why that instead of “physically”? > Also: substitute event for:

• outcome ?
• other variables or factors ?

= Next 3 slides are basic concepts we need [correlation]

Correlation vs Causation Atmospheric pressure Weather Barometer reading

10:18 = First concept concerns relation of causation to correlation > Reading and weather are correlated > Reading is not a cause of the weather (or vice versa) > The reason for correlation is the common cause: atmospheric pressure > Hypothesis: all “true” correlations are result of deep causal connections > Prediction and control

• Correlation is good enough for prediction
• But causation is needed for control

> Classical and operant conditioning > Understanding and action = Second basic concept is distinction between type and token

Type-level vs. token (actual) causal explanations

• Examples
• Precedence in reality
• Precedence in thought

10:24 = What is the distinction? TAP > smoking; dinosaurs TAP talk TAP talk,

• so must understand type first
• everything will be about type-level until near the end

= Third concept is directed graph

A directed graph X A C Y D B

10:27 = We are interested in X as a possible cause of Y TAP > Arrow represents a direct causes (to be defined) > Letter represents an event > So there can be prior causes of our candidate cause TAP > There can be events between our cause and the effect TAP > There can be more than one path from cause to effect TAP > And there can be other causes of the effect = Candidate definition of cause

Candidate Definition of Cause X is a cause of Y if and only if there is a possible intervention on X that will change Y

• r the probability distribution of Y

in some background circumstances.

10:30 = At end of stripping slide > Wiggle wobble > Two problems:

• Meaning of intervention; possible
• Definition is inadequate

> Approach

• Intervention
• Then why inadequate, and what to do about it
• In process, explain “some background circumstances”
• Will get to “possible” at very end, but for now, obviously:

Has to be very broad Events in past Astronomical events = So, what’s an intervention?

Intervention I is an intervention on X with respect to Y if and only if:

• 1. I causes X
• 2. I acts as a switch for all the other variables that cause X
• 3. Any directed path from I to Y goes through X
• 4. I is (statistically) independent of any variable that causes Y

and that is on a directed path that does not go through X.

10:33 = Will read quickly then walk through example then walk through this definition again

A prototypical intervention random assignment = Intervention B placebo = X A C Y D = access ; attitude = germs = immune system = medicine = health administration by shotgun The prototypical intervention is the randomized, blind, controlled experiment, as for a possible drug.

10:35 = Read prototype TAP > Explain all causal factors TAP

• 1. I causes X

> Intervention is randomized determination of who gets the medicine TAP

• 2. Intervention switches off other causes of X:

> A does not cause X

• 3. No path from I to Y except through X

> Consider the potential placebo effect:

• There isn’t really a causal arrow from the medicine itself to the effect
• It’s from the belief that you are taking the medicine

TAP > So have to blind subjects to which group they are in > To avoid creating a path from the I to Y that doesn’t go through X > To pick a more extreme example, TAP > Administering drug by shotgun would create path not through X

• 4. I cannot be correlated with causes of Y not on path through X

> OK for effect on germs to be correlated with I, since it’s on the path. > For example

• Don’t want I correlated with D, the immune system
• which is not on the path from X to Y
• so need large enough sample to assure decorrelation

= So let’s walk back through the definition with example in mind

Intervention I is an intervention on X with respect to Y if and only if:

• 1. I causes X

random assignment

• 2. I acts as a switch for all the other variables that cause X

access and attitude

• 3. Any directed path from I to Y goes through X

placebo shotgun

• 4. I is (statistically) independent of any variable that causes Y

and that is on a directed path that does not go through X. germs immune system

10:45 = Strip and talk = Now that we completely understand what an intervention is, let’s return to the candidate definition of a cause

Candidate Definition of Cause—Again X is a cause of Y if and only if there is a possible intervention on X that will change Y

• r the probability distribution of Y

in some background circumstances.

10:49 =Makes sense =But there’s a problem

Pill Pregnancy Thrombosis

10:51 = Explain relationships > At end:

• Pill doesn’t meet candidate definition of cause
• But in fact the pill is a cause of thrombosis
• Have to modify definition

> What we want is the idea of a contributing cause > To get there, we need the idea of a direct cause = Direct cause

Direct Cause A necessary and sufficient condition for X to be a (type-level) direct cause of Y with respect to some variable set V is that there be a possible intervention I on X that will change Y (or the probability distribution of Y) when all other variables in V besides X and Y are held fixed at some value by additional interventions that are independent of I.

10:53 = Do slide > So, it’s really very simple:

• hold everything else constant
• wiggle X
• if Y wobbles, X is a direct cause of Y

= Back to example of thrombosis

Pill Pregnancy Thrombosis Intervention1 Intervention2

10:56 = Here’s how: > The only variable except X (pill) and Y (thrombosis) is pregnancy > So we have women either take or not take the pill (intervention 1) > While holding pregnancy fixed at some value (intervention 2)

• Yes: they are already pregnant
• No: they can’t get pregnant

> And we find the pill is a direct cause of thrombosis > Notice that whether X is direct cause of Y depends on choice of other variables for analysis = Another example

Gas pedal Injector Cylinders . . . Acceleration Gears Brakes

10:59 = Do slide > First,

• There is some value of Gears (engaged)
• and some value of Brakes (not engaged)
• for which intervening on Gas Pedal will change Acceleration

> But,

• if we fix the value of any variable intervening btw Gas and Accel
• intervening on Gas doesn’t change Accel.

> So Gas is not a direct cause of Acceleration > Rather, it is a contributing cause = So what is a contribution cause?

Break?

11:03

Contributing Cause A necessary and sufficient condition for X to be a (type‐level) contributing cause of Y with respect to variable set V is that (i) there be a directed path from X to Y and that (ii) there be some intervention on X that will change Y when all variables in V that are not on this path are fixed at some value.

11:13 = Do slide > Define directed path > Some values for variables not on path: not all values

• Gear
• Brake

> Some intervention on X: not all interventiions

• Lightswitch
• Sound system

= Now ready to define cause

Definition of cause A necessary and sufficient condition for X to be a (type-level) cause of Y is for X to be either a direct cause of Y

• r a contributing cause of Y.

11:16 = Do slide > In fact, direct is special case of contributing, so just contributing = At end: > Thus endeth the explication of type-level causes > Next, consider token causes, in actual situations > To do this, think about the French Foreign Legion example > [next slide]

Poison in water Death Poison in body Hole in canteen Dehydration

11:18 = But eliminate the sand man: just poison and hole > Who did it? = Actual causation defined

Actual Causation

• 1. The actual value of X = x and the

actual value of Y = y.

• 2. There is at least one route from X to Y

for which an intervention on X will change the value of Y, given that other direct causes of Y that are not on this route have been fixed at their actual values. X = x is an actual cause of Y = y if and only if: Based on a type-level graph of dependency relationships,

11:26 = Do slide > Key is actual values of other direct causes of Y > Rather than just any value that makes Y depend on X = Back to example

Poison in water Death Poison in body Hole in canteen Dehydration = No

11:30 = What happens if we hold the other direct cause of Y at its actual value and wiggle the hole in the canteen? > Discussion > Woodward:

• If, by some intervention, we assured he didn’t have poison in his body,
• Then if, by some intervention, there was no hole in his canteen,
• Then he would not have died.

= The 3-murderer example is tougher

Poison in water Death Poison in body Hole in canteen Dehydration = No Sand in canteen

11:35 = The 3-murderer case look like this: > Explain > TAP > Hold Poison in Body = No > TAP > Did he die of dehydration because of the sand or the hole?

• Discussion
• IMHO, this is symmetrical overdetermination

= It’s easier to analyze in a simpler case [next slide]

Symmetrical Overdetermination Smoker 1 Smoker 2 Forest fire Spark Oxygen House fire

11:39 = Symmetrical overdetermination is about the most basic problem case for theories. > Usual example is firing squad. TAP > Explain forest fire graph > What do you think? > What analysis does our definition give? [neither is cause] > Woodward’s intuition is that they both caused the fire > His solution

• OK to set off-path variables to non-actual values that do not change result
• If 2 had not tossed cigarette, 1 would have caused fire, and vice versa
• But you have to be careful

TAP > Take the spark that caused the fire when oxygen was in the air

• If we say there had been no oxygen, the spark would not have been a

cause

• But we want spark to be a cause (along with the oxygen)
• Explain range of redundancy

> The more important point: when common sense is conflicted, what matters are the counterfactual relationships, they are the reality. = So much for actual causation. Now, finally, possible intervention.

Possible Intervention

• Does not require human action
• Must be logically possible
• Must be conceptually clear
• Physical possibility ?
• Regulative ideal

11:44 = Reintroduce issue: definitions mention “possible interventions” > TAP If there had never been any humans, some events in nature could be considered interventions > TAP A candidate intervention must be logically possible > TAP Also conceptually clear: Example of Julius Caesar, UN forces Korea, bomb or catapult > TAP Physical possibility?

• Depends on what that means
• If laws of nature, initial condition, and strict determinism, no interventions

are possible

• Different initial conditions: intervene on orbit of moon to gauge effect on

tides?

• Not without other effects
• But Newton tells us what would happen
• Like the block sliding on the inclined plane: if you know the physics, you

can figure out the causation

• But theory can require possibility of interventions that are physically

impossible > One key idea:

• if you don’t know the physics, you have to be able to do the interventions

to learn

• if you do know the physics, you don’t have to be able to do the

interventions > TAP The idea of intervention is not hard and fast, but a regulative ideal = Another factor: what possibilities are serious?

Serious Possibilities

• Spark - oxygen - fire
• Doctor and plumber
• Ball - wall - window

11:50 = Causal explanation depends on the possibilities we are prepared to take seriously > TAP Why do we think the spark is the cause of the fire, rather than the

• xygen?

> TAP Doctor plumber > TAP Ball, wall, window

• What if wrecking ball hit wall just before ball?
• What if wall replaced with second person?
• What if second person 10 years old?
• What if using bat rather than catching?
• As the possibility of failure increases, the early preventer is more causal

>Judgement of seriousness of possibilities, which governs which interventions will be considered, is subjective. > Woodward considers that, once this issue is decided, the rest is objective = I would like to add another subjective factor, however

Causation and Human Purposes The Titanic sank because:

• A. The hull filled with water
• B. An iceberg penetrated the hull
• C. The ship hit an iceberg
• D. The captain took the great circle route
• E. The captain believed the ship was unsinkable

F . The designers didn’t understand the physics

• G. National pride overrode design caution

Note: this slide is not based on Woodward’s work

11:54 = The purposes for which we want causal explanations matter for the kind of cause that will be accepted as explanatory. Do entire slide, then: > The designer would be concerned with the penetration > Company management might be concerned about the psychology of route planning > The shipowner might want to know about designer competence > The historian might probe national psyche So level of analysis relates to purposes, which are also subjective. = Is Woodward a realist about causes?

Causation, Realism, and Metaphysics

• Causal explanations are shaped by human

judgements of possibility (and human purposes).

• Given a set of serious possibilities,

counterfactual relationships among variables are

• bjective facts.
• When intuition of causes is unclear, look to the

counterfactual relations, which are what matter scientifically.

• The manipulationist theory of causation

requires no other metaphysical commitments. “ I [Woodward] leave it to the reader to decide whether it [his project] counts as discovering ‘what causation is’. ” (page 7)

11:58 =Read quote > Has Woodward told us “what causation really is”? Discuss Strip > At end, example of double prevention:

• You tackle someone who would have caught ball
• The window breaks
• You caused it
• But no spatiotemporal path and no energy transfer

= How does causation relate to this course?

What is the relationship between

• This theory of causation
• Experimentation
• Scientific revolution

?

12:06 to 12:11.