Depression and causal reasoning

Question 1

Uncertainty coefficient

Answer 1

The per cent reduction in uncertainty about whether and/or when a predicted event will occur produced by the occurrence of a predictor event is

a broadly useable measure of contingency or association. (how much do you feel that one stimulus predict the other)

Question 2

A normative metric for contingency

Answer 2

(normative= what ppl should be doing based on theory, not what they are doing):

Allan (1980) developed a normative metric for contingency:

Δp = P(O|R) - P(O|no R)

Question 3

Deductive reasoning

Answer 3

Wason (1966) Selection task requires it.
Modus Ponens for picking A [if P then Q]
Modus Tollens for picking 7 [if not Q then not P]

Using logical rules to deduce true, linking premises with conclusions.

Question 4

Modus Ponens

Answer 4

We can confirm Q by confirming P, as P implies Q;

If we know that P is true, we can conclude Q must be true. This is a valid argument and a rule of inference

Question 5

Modus Tollens

Answer 5

We can deny P by denying Q, as not-Q implies not-P (denying the consequent).
This is a valid argument and a rule of inference.

If I am the axe murderer, then I can use an axe. (if P then Q)
I cannot use an axe. (Not Q)
Therefore, I am not an axe murderer. (therefore, Not P)

Question 6

Dependence hypothesis

Answer 6

Checking if q depends on p

Question 7

Independence hypothesis

Answer 7

Checking if q and p are statistically independent

Question 8

Information gain model

Answer 8

Assumes that participants are looking for evidence that provides the most discrimination between the dependency hypothesis (that Q depends on P) and the independence hypothesis (that P and Q are statistically independent)

Question 9

Rarity assumption

Answer 9

That the probability of the antecedent and consequent are small.
- In the world around us, the rarity assumption generally holds true because natural language divides the world up quite finely - very few things belong to any given category.
(Mckenzie & Mikkelsen (2000)) - rarity appears to be the default assumption.

Question 10

Inter-trial interval (ITI)

Answer 10

Is a time gap programmed in by the experiment, between the trials.

Question 11

Probability of outcome p(O)

Answer 11

Same as outcome density

Question 12

Biases in probability estimations

Answer 12

We over-estimate the likelihood of events that have low probability and under-estimate the likelihood of events that have high probability (Lichtenstien et al. 1978)

Question 13

Outcome density

Answer 13

Means the number of trials on which an outcome occurs = probability of outcome

Question 14

Depressive realism

Answer 14

The hypothesis developed by Alloy and Abramson 1979 with their finding with difference between D and ND ppts.

They concluded that D ppts are “sader but wiser”

Question 15

Are ppl irrational? Results of Wason selection taks

Answer 15

Oaksford & Chater (1994) Summarize the results of more than 30 studies from 13 authors:
- 89% choose A
- 25% choose 7
- 62% choose 4
- 16% choose D
=> We could conclude, therefore, that people are irrational!
=> Ppl select 4 b.c. Affirming the consequent: if Q then P (reverse to normal if P then Q) but not all Qs imply Ps

Question 16

Are ppl irrational? Results of Wason selection task

Answer 16

Oaksford & Chater (1994) Summarize the results of more than 30 studies from 13 authors:
- 89% choose A
- 25% choose 7
- 62% choose 4
- 16% choose D
=> We could conclude, therefore, that people are irrational!
=> Ppl select 4 b.c. Affirming the consequent: if Q then P (reverse to normal if P then Q) but not all Qs imply Ps

Question 17

Hempel’s (1965) Black Raven Paradox

Answer 17

If it is a raven then it is black
To confirm:
More efficient to test all Ps (all ravens) than not-Qs (all black things.

Question 18

Why, hypothetically, would non-depressed people be overestimating the control over actions? (according to Alloy and Abramson, 1979)

Answer 18

Non-depressed ppl are motivated to maintain self-esteem; not having control lowers self-esteem; they over-estimate control to feel better.

Ppl with depression lack the motivation to maintain self-esteem and hence show more accurate judgements of control.

Question 19

What factors are affecting the contingency judgement? short verso deleted

Answer 19

1. Outcome density (Alloy and Abramson, 1979 critique ish)

2. Probability of response

Question 20

What factors are affecting the contingency judgement?

Answer 20

1. Outcome density (Alloy and Abramson, 1979 critique ish)
2. Probability of response
3. Duration of ITI
4. Activity rates

Question 21

Activity rates

Answer 21

Activity rates both influence causal learning and may be influenced by causal learning, in a mutually reinforcing relationship (Byrom, Msetfi & Murphy, 2015).

While some studies suggest that accurate judgements of causal control may arise due to low activity rates associated with low mood (Blanco, Matute, Vadillo, 2012), there is also evidence of high activity rates in individuals with low mood (Byrom et al., 2015).

So mixed findings when trying to explain depressed ppl answering differently

Question 22

Associative learning theory

Answer 22

Associative learning theory explains the process of learning about contingencies through the development of associations between events an outcome

(e.g. RW 1972):
ΔV = αβ(λ – ΣV)

Mental associations gradually develop and become stronger when events and outcomes occur repeatedly at around the same time.

• The event-outcome association is directly strengthened by co-occurrences (Cell A)
• And weakened by the occurrence of the event on its own (Cell B)
• Indirect influences take place when the outcome occurs unaccompanied (Cell C) - e.g., the association between context and outcome is strengthened.
• Or when neither happens (cell D)- e.g., the association between context and outcome is weakened.

BUT because the changes to associative strength for any stimulus on any trial are governed by the total associative strength –> if you add more C cell trials and strengthen the context-outcome association, and weaken the CS- response association
AND adding more D cell trials should weak both associations