Explanations and Causes Flashcards
An argument can be made that explanations should be asymmetric. If so, the Deductive-Nomological account of explanation is in trouble, since it is structurally symmetric. What does this imply? Mark all correct options.
a) It implies that, on the account of the DN-model, it is sometimes possible to explain a phenomenon by pointing to its effects (in conjunction with relevant law-like generalities), rather than to its causes. This is quite counter-intuitive.
b) It implies that there are cases that fail to satisfy the conditions of the Deductive-Nomological account, even though they intuitively seem to constitute genuine explanations.
c) It implies that there are cases that satisfy the conditions of the Deductive-Nomological account, but intuitively do not constitute genuine explanations.
Answer: a), c)
General Feedback
A symmetric relation is a relation whose relata (the things that stands in the relation) relate to each other in the same way e.g. “being of the same age” is a symmetrical relation. If A is the same age as B, then B is the same age as A. An example of an asymmetric is “being older than”: If A is older than B, then B is not older than A.
What is true about the problem of causal irrelevance? Mark all correct options.
a) If an explanation is causally irrelevant, then its explanans doesn’t point out the difference-making causes.
b) The problem of causal irrelevance shows that certain accounts of explanation are not necessary for all situations.
c) The problem of causal irrelevance shows that certain accounts of explanation are not sufficient for all situations.
d) If a DN explanation is causally irrelevant, then the deduction in the explanation is not valid.
Answer: a), c)
Besides avoiding the problems outlined in the previous questions, what advantages do causal explanations have over Deductive-Nomological explanations? Mark the correct option.
a) Causal explanations, but not DN-explanations, can provide us with understanding of why a phenomenon occurred without making reference to any general, law-like regularities.
b) Causal explanations, but not DN-explanations, can provide us with an understanding of why a singular phenomenon occurred.
c) Causal explanations are always easier to understand.
Answer: a)
Is it problematic if an explanandum isn’t contrastive? Mark all correct options.
a) Yes, because then the explanans might not provide the relevant kind of contrast.
b) No, it is sufficient that the explanans is contrastive. Contrast in the explanandum is superfluous (and therefore often omitted).
c) Yes, because explanations that aren’t contrastive are false.
Answer: a)
What is correct about the following graph of a causal scenario? Mark all correct options.
X1
|
v
Y1 <—– X2
|
v
Z <—– Y2
a) If an intervention is done on Y2, keeping all other Xs and Ys fixed, then Z will change.
b) If an intervention is done on X1, keeping all other Xs and Ys fixed, then Z will change.
c) X2 is a contributing cause of X1.
d) X1 is a contributing cause of Z.
e) The graph could be describing the following scenario: “If the knob is turned (X1) and the pan of soup is on the stove (Y1), then the stove is connected to electricity (X2); and if also the stove heats up (Z), then the soup is cooked (Y2).
f) The graph could be describing the following scenario: “If the knob is turned (X1) and the stove is connected to electricity (X2), then the stove heats up (Y1); and if also the pan of soup is on the stove (Y2), then the soup is cooked (Z).
Answer: a), d), f)
What is true regarding causation? Mark all correct options.
a) Correlation in a data set is necessary for causation.
b) Causation is an asymmetric relation.
c) Correlation is sufficient for proving causation.
d) Causation measures the productive influence of one variable on another.
Answer: b), d)
What is true regarding correlation? Mark all correct options.
a) Correlation measures the productive influence of one variable on another.
b) Correlation can be a sign of causation.
c) Correlation is not sufficient for proving causation.
d) Correlation measures the association between two variables.
e) Correlation is a symmetric relation.
Answer: b), c), d), e)
Which of these are examples of causal models (causal scenarios) that are compatible with the observation that ice cream sales and the rate of murder in a population are correlated in a data set? Mark all correct options.
a) When someone has been murdered, people tend to console themselves by eating ice cream.
b) Both ice cream and murder rates increase when the summer comes because people like to eat cold food when it is hot, and people are out and about more in the summer, making them easier to kill.
c) Eating ice cream makes a person more likely to kill another person.
d) The observation is an artefact of the study and there is no causal relationship at all between the events, nor is there any other common cause.
e) When someone has been murdered, this increases the likelihood of revenge murders, increasing the total amount of murders. However, when someone has eaten a lot of ice cream, they become less interested in eating an additional ice cream.
Answer: a), b), c)
General Feedback
There are three basic types of causal scenarios or causal models.
A–>B
B–>A
C–>(A&B)
Although there might instead be a spurious correlation between A and B, we don’t consider that a causal model (scenario).
Suppose that we have found that X and Y are correlated. Now we want to find out whether X causes Y. Which of the following is true? Mark all correct options.
a) The only way to prove that X causes Y is by performing some kind of experiment where we intervene on X.
b) Instrumental variable analysis can only be used to show a correlation between X and Y, not a causal relation.
c) Mill’s method of difference can be used to exclude that X and Y have a common cause.
d) Mill’s method of difference can be used to investigate whether X causes Y.
Answer: c), d)
