Theories of Reasoning Flashcards

You may prefer our related Brainscape-certified flashcards:
1
Q

What does the data suggest reasoning involves?

A

Reasoning is not purely logic, but logic is not irrelevant
Reasoning is influenced by context
Reasoning evokes heuristics

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What are the three main theoretical approaches that account for reasoning?

A
  1. Mental Models (Johnson-Laird)
  2. Mental Logics (Braine, Rips)
  3. Bayesian reasoning (Chater, Oaksford)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What does the Mental Logics model posit about reasoning? What are its assumptions? What is an example of this rule?

A

People often respond inconsistently with formal logic, but does not mean there isn’t a mental logic
Braine and O’Brien (1991) and Rips (1994) - both assume propositional representations: represent knowledge as statements, describing the knowledge
• Errors occur because the rules in mental logics are incomplete, and requires taking different steps to reach conclusions

E.g. Modus Ponens: Proof by confirmation. Turning over the “A” card
Modus Tollens: Proof by contradiction. Turning over the “7” card
- We don’t have a direct rule for Modus Tollens/it’s not part of our set –> hard

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

How are mental logics used?

A

Use a limited set of logical rules, rather than all the ones in formal logic
(e.g. We have Modus Ponens, but not Modus Tollens)
Chains of inferences can be forward (premises –> concl) or backward (concl –> premises), but eventually all suppositions are rejected/accepted - reasoning.
Rules in mental logic depends on individuals
- e.g. Musicians may reason differently to others, due to their skill set?

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

How does the Mental Logics model explain errors?

A
  • Chain of inferences –> greater chance of error
  • Exceed WM
  • Exec control limitations: can’t keep track of chain in argument
  • Wrong rule: Lack of appropriate inference rule, failure to apply appropriate rule, or application of inappropriate rule
  • Garbled output of rule.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What evidence is there to support the Mental Logics model?

A

Tasks where people judge the logical correctness of a series of arguments.

Braine et al (1984): When asked to judge 85 syllogisms, participants applied more complex rules as the difficulty of syllogisms increased. (r=0.79)

Rips (1994): People found problems more difficult if backwards reasoning was required

Forward reasoning seems more automatic
- Lea et al (1990) found more false recog for forward reasoning conclusions

Rips (1989) found evidence of mental logic type reasoning in verbal protocols

Good evidence for this approach.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What does the Mental Models approach suggest about how we reason?

A

Reasoning involves building >1 representation (model) of the situation, then making conclusions from introspection

Mental model = a representation of a possible state-of-affairs in the world

E.g. Black behind cue
Green on right of cue
Red ball between them

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

How can Euler circles be used? (Erickson, 1974)

A

Euler circles: Creating circles to represent info (sort of like Venn diagram)
e.g. Some artists (A) are beekeepers (B). All the beekeepers (B) are chemists (C). Therefore some of the artists (A) are chemists (C). (refer to notes for actual circles)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Do people use Euler circles to reason?

A
  • Can be ambiguous when used in situations.

- Need to be taught this in maths - no evidence that people do this spontaneously

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What form of Mental Models does Johnson-Laird (1983) propose?

A

Comprehending text –> Construct different critical, concrete possibilities

e.g. lots of ways of tossing a coin, but only two models of outcomes: H/T
e.g. Some of the artists are beekeepers
All the beekeepers are chemists
Therefore some of the artists are chemists
Set up the following model:
A B C
A
B C, etc.

However, one model is not always enough.
E.g. None of the artists are chemists. All the beekeepers are chemists. Therefore some of the artists are beekeepers
A
B C … no valid conclusion?
More possible models:
A
B C
C
Valid conclusion: Some chemists are not artists.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What evidence is there to support mental models?

A

Johnson-Laird and Byrne (1991) predicted:
Inferences based on one model will be easier than those based on multiple models
Systematic errors are likely to correspond to initial models of premises - knowledge can influence the process of inference

Copeland and Radvansky (2004): When participants asked to rate validity of syllogisms, depending on the number of mental models that needed to be constructed for the correct response,
- Fewer models necessary –> More accurate, faster, more confident conclusions
- WM capacity corr. .42 with accuracy (higher WM –> more space to store more models)
- Higher corr with accuracy for multiple than single model syllogisms (more models = more accurate)
Supports Mental Models theory.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

How does the Mental Models approach explain for contextual effects on reasoning?

A

The context of the syllogism influences the first model that is created.

  • If the first model is incorrect, it can lead the person to make the wrong conclusion, and also hinder the person from creating other models that lead to different conclusions.
  • Whether or not the conclusions reflect reality also influences the endorsement of conclusions

E.g. All athletes are bakers. Some of the bakers are chemists. Therefore some of the athletes are chemists.
People tend to endorse this invalid conclusion.
Woodsworth and Sells (1935) proposed the “Atmosphere of the premises” hypothesis: “Some” in the premise biases us towards “some” in the conclusion
- Suggests people do not reason, but take a guess at what looks right - not well supported.

Mental model to this approach: 
Model 1:
A      B      C
A      B
                 C --> statement is correct (false)
Model 2: 
A       B
          B       C
                    C
--> Contradicts conclusion, but requires going beyond first model
--> First model created is critical
Knowledge influences first model created --> Knowledge can influence whether or not the correct conclusion is reached
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

How does the Mental Model approach explain the error in Wason selection task?

A

Model 1 created:
[Vowel] Even ….
- Easy to see that need to turn over vowel card (Modus Ponens)
- Also makes sense to look at the even card, due to model –> error (due to no other models having been created)
To turn over odd card, another model needs to be developed:
[Vowel] Even
Not vowel Even
Not vowel not even

Error in Wason’s because we don’t expand on the first model enough.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Mental logics vs Mental models?

A
  • Testing theories requires critical tests that can distinguish them.
  • Hard to do for mental logics and mental models - Tend to be able to explain each others’ results post-hoc
    May be due to fact that each is under-constrained
  • What is a model and which ones do we build?
  • What mental logics do we have and not have?

Thus argument continues, but both provide possible explanation of reasoning phenomena.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What does the Bayesian approach propose about reasoning?

A

People reason about syllogisms based on the probability that a conclusion is correct.
- Probabilistic strategy in an uncertain world
- Probability theory better than logic, because it’s a computational level of reasoning
But: Judgements of probability are hard, so this makes reasoning fundamentally subjective.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is Bayes’ Theorem?

A

Can be used to see how we update data and hypotheses
When we learn, we acquire data about the world
Data can update our theories about the world
Belief –> Data –> Updated belief
Reasoning is updating info about world

P (H | D) = (P(D | H) P(H)’ ) / P(D)
H = Hypothesis
D = Data
P(H) = Prob that H is correct before D is seen
P(D | H) = Likelihood. Conditional probability of seeing data D given that hypothesis H is true
P(D) = Marginal probability of D
P(H | D) = Probability that H is true, given D and previous belief about hypothesis

17
Q

How is the Bayes’ Theorem also a philosophy of science?

A

Popper (1935): Science should focus on falsifiability
Critics: Science is often about increasing confidence in what we think is the right hypothesis - but we should be focused on decreasing confidence in the null hypothesis (stats)

E.g. Leverrier: Didn’t reject the theory of Universal Gravitation, but reject hypothesis that there were only 7 planets

18
Q

How does Bayesian theory explain the error in the Wason selection task?

A

Is turning over the “4” card an error?
Hypothesis: If you eat tripe, you will feel sick.
How to test out hypothesis?
A - Ate tripe
B - Not ate tripe
4 - Feel sick
7 - Not feel sick
You would check the sick people, not the healthy - don’t usually learn about diseases from examining the healthy.
Asking sick person if they ate tripe is useful - if they did, this would increase confidence in rule (due to data)

Optimality, as measured by info gain, depends on the distribution of information in the environment. - Investigating rare cases yields optimal info.
Most of the time it is the “4” card that is rare - Positive case (A, 4) often more informative than negative case, when (A, 4) occurrence is low.
In fact, we state rules by rarity, because rare case are more informative
- although nothing about the problem suggests that rarity is involved
Reasoning could involve people updating their current beliefs based on prior beliefs and new data.

19
Q

Based on the Bayesian theory, what does Oaksford and Chater (1994) argue about reasoning and optimality?

A

Optimality - Reasoning involves heuristics that maximises information gain.
- Not necessarily equivalent to aiming for certainty
- Can lead to diff conclusions about how people should reason
Goal of reasoning: Operate effectively in environment
Explanation more important than validity

20
Q

How did Oaksford et al (1997) manipulate probability using the Reduced Array Selection Task (RAST)?

A

If people reason probabilistically, then manipulating probabilities should lead people to change behav.
Reduced Array Selection Task (RAST): Reduced Wason task to just “numbers”
Test: All triangles (P) are blue (Q)
Examine from red vs blue shapes card stack? Turn over cards until a certain rule is right/wrong.
- Varied sizes of stack of cards
FOUND: Smaller the stack, the more likely the cards were selected.
Fits rarity assumption - supports Bayesian approach.

21
Q

What is the rarity assumption, and how does evidence from the two experiments conducted by McKenzie et al (2001) support Bayesian theory?

A

Rarity assumption: People prefer rare evidence
McKenzie et al (2001): “If it is a raven, then it is black”
Had Black/Raven, not black/not raven
FOUND: Both logically correct, but people see black raven as more supportive
Tendency to phrase hypotheses in terms of rare events
Supports Bayesian theory, in that people prefer to use probabilistically rare information to support their hypotheses

22
Q

How does McKenzie et al’s (2001) experiment reveal how we use language in reasoning?

A

Showed data about SAT scores and college admission
- One student admitted had high SAT
- Participants had to fill in “If___ then __” sentence
Preferred “If applicants have high SAT scores, then they will be admitted” over “If applicants have low SAT scores, then they will be rejected”
Reversed responses if most students were accepted, but not those with low SAT scores.

We use language to be maximally informative.
Wason: stated as max informative.
4 card talked about –> rarer
Wason error is biased in terms of rarity principle.

23
Q

What does Pennington and Hastie’s (1988) experiment reveal about how language can affect reasoning?

A

Reasoning not just based on evidence, but also on what seems to be explanatorily coherent.

Studied juries
Presented same evidence, but manipulated coherence of story presented

FOUND: The more coherent the evidence, the more favourable the jury were to the groups.
- Same info, but changed order –> changed conviction rate!

24
Q

How does rarity principle (in Bayesian theory) explain errors?

A

They aren’t all errors.
Classic: memory constraints, misperceptions
Emphasis on Envr constraints:
- Inaccurate info about envr
- Envr may be noisy
- Priming of wrong elements of Envr
Experts make fewer errors due to learning the right probabilities - but how do we know what the right ones are?