L17 - Category Learning 1

Question 1

Describe Ashby and Madox’s 2 Systems of Category Learning

Answer 1

Explicit Hypothesis Testing System

• Working memory and executive function dependent.
• Considers explicit rules that treat stimulus dimensions one at a time.
• Doesn’t require feedback.
• Explicit rules mediate specific S-R associations.
  e. g. medical diagnosis

Implicit Procedural System

• Does not require working memory.
• Integrated across multiple stimulus dimensions
• Learns via prediction and reinforcement
• Specific stimulus and response associations.

Two systems are in competition and type of task can determine which wins.

Question 2

What do rule tasks elicit better performance from?

Answer 2

Explicit hypothesis testing system

Question 3

What tasks require Implicit Procedural Learning?

Answer 3

Information-Integration tasks

Question 4

Describe the study Fileto et al. (2010) did about working memory on category learning tasks?

Answer 4

• A working memory dual task using either conjunctive rule or information-integration.
• 2D task: only length and orientation
• 3D task: also position (doesn’t matter for classification)
• Having extra information decreased performance, but in I-I working memory load increased performance as it shuts of the explicit rule system earlier.

Question 5

What did Markman et al (2006) find about social pressure on category learning tasks?

Answer 5

Adding social pressure of ($20 to both of you if you do well), helps Information-integration and harms rule learning.

Question 6

What affects performance for the information-integration system?

Answer 6

Boosts:

• working memory load
• social pressure

Decreases:

• feedback delay
• button-switching

(opposite for rule based learning)

Question 7

What did Chin-Parker and Ross (2002) conclude about category learning using description of employee files?

Answer 7

• inference learning supports acquisition of internal structure of the category than classification.
• we only learn what we need to learn.

Question 8

What are the differences between inference and classification learners?

Answer 8

• inference learners are sensitive to all features equally.

- classification learners only care about diagnostic features.

Question 9

What is inference learning bad for?

Answer 9

• Good at prototypes and feature-correlations.

- Bad at learning examples that share little with other examples (lions, bears, platypuses).