Problem Solving

The areas of research in problem solving include:

• problem solving tasks
• transfer of knowledge
• expertise (development of skills, from controlled to automatic processing

Problem solving is:

1. Goal Directed
2. Cognitive - not an automatic process
3. There is only a problem to solve with the individual lacks the relevant knowledge

Gestalt approach

• this distinguishes between reproductive problem solving which involves the use of experience, and productive problem solving which involves novel restructuring.
• Restructuring and insight was tested by Maier (1931) using the two string problem. Participants were given the problem of two strings hanging from the ceiling too far apart to hold at the same time. Maier observed how they overcame this problem
• Evaluation: concepts of insight and restructuring are quite vague and hard to measure, particularly as we dont know the processes underlying them,
• Novick and Sherman 2003, suggested that insight may be based on the accumulation of knowledge, Gestaltists did not use problems such as skilled games which involve this however.

Functional Fixedness

• Past experience does not always help to find a solution to a problem
• eg Duncker (1945) used the candle problem. The task is to attach a candle to the wall next to a table so that the wax doesnt drip onto the table below.

Computational Approach: Newell and Simon's (1972) General Problem Solver (GPS)

• This involved a systematic computer simulation of human problem solving.
• It had several assumptions about the human mind however, for example, serial processing, limited short term memory capacity and the ability to retrieve information from long term memory
• They investigated the strategies that are used by asking people to think aloud while solving problems
• In the GPS a problem is represented as a problem space
• Initial state --> intermediate states --> Goal State
• The towers of Hanio/London are an example of a problem state. They give the participant a starting position and a goal position which they must rearrange the blocks into
• Newell and Simon concluded from their observations that we select manual operators (moves) by relying on the 'rules of thumb'. This is also known as heuristics
• Means-end analysis: this requires the difference between the goal state and the initial state to be observed. A sub-goal is then formed to reduce this difference. Then, a mental operator or move is selected to reach the sub-goal. This is a contrast with algorithms which are complex methods that are guaranteed to solve problems
• Evaluation: Newell and Simon applied the general problem solver (GPS) to 11 different problems. It managed to solve all of them, but not necessarily in the same way as humans. The computational model allows us to see when and how performance deviates from the ideal. FInally, GPS is consistent with knowledge of human information processing, eg limited working memory capacity.
• Limitations of the computational approach: the general problem solver is better than humans at remembering previous moves, but worse at planning future moves.
• Everyday problems are ill-defined and so some solutions might cause other problems. Specific knowledge might be needed. The problems used as examples in the computational model are well defined and have a clearly specified initial state, goal state and range of moves
• Computational models are best suited to serial processing tasks, and not so well suited to insight.