Reasoning and skill

Question 1

dual process theory

Answer 1

Evans (2003) proposed reasoning uses two processes:

1. Ancient system that relies on associative learning
(implicit)
– Biased by prior beliefs and knowledge
– Heuristic
– Lead to cognitive illusions
– Built by experience

2. System using working memory resources and is
   therefore slow and sequential
   – Evolutionarily recent
   – Related to IQ (Stanovich & West, 1998)
   – Requires effortful intervention (so instructions important)
   – Built by training

 (1) is default system, but reasoning tasks require
intervention of (2) to succeed consistently

Question 2

dual process approach vs other approaches

Answer 2

Evans (2003) sees
– System 1 as a set of autonomous subsystems using domain specific knowledge
– System 2 permits abstract reasoning but constrained by working memory capacity

Comparison to other approaches
– Mental models and mental logic approaches both seem to emphasis working things through, so more System 2
– Errors due to too little System 2, not too much.
– But initial models or application of mental logic rules often described as effortless, so imply System 1.
– Bayesian approaches seems to emphasis System 1.

Question 3

evidence about believability bias

Answer 3

believable and invalid: only 31% correct

Evans (2003) see it as support for dual-processes
– Affected by instructions to focus on deduction (invoke System 2)
– Shift from logic to belief-based reasoning under time pressure (Evans & Curtis-Holmes, 2005)
– Logic performance declines with age (Gilinsky & Judd, 1994)

But mental models may also explain it
– Oakhill et al (1989) tested mental model account of belief bias by showing that initial model produces bias.
»Need more models to get right result.
– Klauer et al (2000) found evidence of people producing just one model, and then applying other processes.

Question 4

are there 2 systems of skill and reasoning

Answer 4

Stanovich (1999) identifies 12 sets of researchers who see reasoning as involving two systems:
– System 1: associative, holistic, and automatic
– System 2: rule-based, analytic, and controlled.

System 1 keeps providing “answers” that System 2 may accept or reject.
– Systems usually cooperate.

Different pattern of errors depending on which system is being used.

Most theories of thinking to some extent are incorporating idea of multiple systems.
– Decision making (heuristics vs calculation)
– Problem solving (hypothesis testing vs push towards goal)
– Skill acquisition as development of System 2 then 1
– Reasoning (heuristics vs calculation)

Question 5

IQ and reasoning

Answer 5

Stanovich & West (1998) gave 3 versions of the Wason selection task and asked for SAT scores

Many studies of which reasoning tasks do
and don’t correlate with IQ

Applications of heuristics (e.g., anchoring
effect) tend not to, logical reasoning does.

System 2 needs both capacity and metacognition to override System 1.

System 1 is often right

Question 6

summary

Answer 6

We are imperfect, we make errors.
– We operate under constraints (bounded rationality)
– Make trade-offs

Representation is critical.
– Changes as we gain skill/expertise
– Critical to reasoning

We use short-cuts (heuristics).
– Expertise partly gaining right heuristics
– Conform to constraints

We use old information in new ways so experience helps.
– Skill as automatization of results of problem solving
– Reason better about familiar situations
– Deduction as induction

We have to deal with uncertainty.
– Reasoning not always clear cut in our environment
– So constrained by environment