Problem Solving (2)

Question 1

Define the general problem solver

Answer 1

Information processing approach to solving problems, it is a computer programme

Question 2

Define operator

Answer 2

An action towards solving a problem

Question 3

Define sub-goal

Answer 3

A step on the way from the current state to the end (goal) state

Question 4

What is the problem space theory?

Answer 4

Newell and Simon (1972):
- More steps = harder problem
- navigate through steps
Heuristics to help:
- Means-end analysis
- depth-first approaches
- backup avoidance (don’t go back to square 1)
- differences reduction/hill-climbing (get closer to goal)
- anti-looping (don’t keep making the same moves)

Question 5

What is meant by sub-goal creation?

Answer 5

• Note difference between ‘current state’ and ‘goal state’ (means-end analysis)
• create a sub goal to reduce this difference
• select operator that will solve this sub goal

Question 6

What experimental evidence is there for ‘sub-goal creation’?

Answer 6

Egan and Greeno, (1974):

• 5 and 6 disc Tower of Hanoi
• experimental group = no prior experience (novices)
• findings = experimental group gave better performances and better memory for the top-level goals - (evidence for sub-goals?)

Question 7

What is limitation 1?

Answer 7

• The general problem solver didn’t have a problem with moving away from the goal state
• BUT humans do struggle, so it’s love completely transferrable to humans and their processes/behaviour of/for problem solving
• (is this a good model of human performance?)

Question 8

Explain Atwood & Polson’s (1976) study

Answer 8

The water-jugs problem:

1. Means-end analysis
2. anti-looping strategy
3. 1 move look-ahead strategy

• Limited by working memory constraints
• Supports general problem solver - but working memory limits how far we can look ahead

Question 9

Do people make use of means-end analysis?

Answer 9

Thomas (1947):

Findings suggest:

• people do originally make use of a means-end analysis
• people must construct new sub-goals that seemingly increase distance between the current state and the goal state (the correct move in a few places seemingly moves one further away from the problem, requires a look-ahead ability and draws on working memory)

Question 10

What did Greeno et al., (1974) do/study?

Answer 10

• GPS - means-end analysis
• forward search, hill climbing, reasoning by analogy, other strategies
• people use small forward steps, not means-end analysis

Question 11

Does means-end analysis always work?

Answer 11

• People use various strategies, not just means-end analysis
• people don’t like moving away from the goal
• what about memory effects?

Question 12

What are some differences between human performance and GPS performance?

Answer 12

Some important differences between human performance and GPS performance:

• GPS ‘remembers better’ which moves have previously been made
• GPS is more short sighted than humans (it only looks 1 move ahead)
• GPS has no problem moving away from a goal

GPS:

• Only applies to ‘well-defined’ problems
• it doesn’t explain insight
• it doesn’t look at individual differencesWh

Question 13

What is the ‘nine-dot’ problem?

Answer 13

• Gestalt theory explanation
• the 9 dots form a square shape and people assume the edges of the square are the boundaries of the problem (but they aren’t)
• participants ‘misrepresent’ the operators of the problem (Ilsaak & Just, 1995)

Question 14

What did Macgregor et al., (2001) do/study?

Answer 14

• Post-Gestalt theory don’t predict the difficulty of the nine-dot problem
• they also don’t predict how helpful clues will be/are to solve the nine-dot problem
• this is unlike similar ‘move’ problems e.g. Tower of Hanoi and Water Jugs)
• 1 source of difficulty may be: the goal-state of the nine-dot problem isn’t specified - unlike other ‘move’ problems (Weisberg & Alba, 1981a)

Question 15

What are the problems of ill-defined problems?

Answer 15

• The nine-dot problem - what is the goal state?
• if Newell & Simon (researchers) were right then without a goal state it becomes almost impossible to apply means-end analysis (you’re comparing your current state with…what?)
• people (not many) solve the nine-dot problem (first time as well): how?

Question 16

What is the progress monitoring theory?

Answer 16

It is a computer model based on 2 theories:

1. Maximisation heuristic aim at a sub-goal (bit like hill-climbing)
2. progress monitoring how successful is each operator at getting to the goal (like means-end analysis) - repeat successful operator. If the operator fails = become stuck (criterion failure)
   - Criterion Failure occurs when PPS seems to be failing to make headway towards the goal state
   - criterion failure is important because it signals when and why participants decide to change (their) strategy
   - this may lead to insight
   - (Kaplan and Simon, 1990)

Question 17

Why may we need to apply a new solution?

Answer 17

Like Ohlsson et al suggested - constraint relaxation (realising that you don’t have to end the line at a dot) is critical

Question 18

What evidence is there for the progress monitoring theory?

Answer 18

• A ‘locally rational operator’ that moves the problem solver towards the goal (connecting all of the dots), may be to draw lines that connect as many dots as possible

Dominowski (1985) found:

• 95% of participants chose 1st lines connecting 3 dots
• 83% of 2nd lines connecting 2 dots (the maximum possible in each case)

Question 19

What are the strengths of GPS?

Answer 19

• Has created a normative model (base-line)
• provides a solid foundation for research
• explains puzzle-solving reasonably well
• explains how we go from reasoning about new problems to applying earlier situations to them (learning/analogical transfer)
• ties in with existing models of cognition e.g. memory
• has resulted in several successful Al expert system e.g. MYCIN

Question 20

What are the weaknesses of GPS?

Answer 20

• Doesn’t have much ecological validity (often doesn’t work that well in the real-world/real-world situations)
• most ‘real’ problems aren’t well defined
• choice of which ‘operators’ to choose from isn’t well specified
• heuristics are vague - i.e. people use many different and (often) ineffective heuristics that interfere with one another (e.g. balancing interferes with means-end)
• (what causes us to abandon a heuristic and use another one?)
• (homunculus effect)