Untitled_2

Question	Answer
What is a problem?	Used everyday Draws on many cognitive functions, e.g. visual perception, language, memory, attention, reasoning, judgment, decision making Problems may share characteristics but may be affected by different factors, internal (motivation, personality) and external (social/cultural factors) Duncker (45) Problem = having a goal but not knowing how to reach it, a gap between start state and goal state
Protocol analysis	'Think-aloud' method. Solver gives running commentary. Assumes: 1. Cognition is information processing 2. Inf stored in diff memory stores 3. Recently acquired info proc in WM + PA gives data re thought proc and diffs that would not be avail vform behavioural / perf errors, IDs diff strategies - not a direct externalisation of cognition. Act of introspection may affect processing?
Protocol analysis in action	Gilhooly '97 - previous research debated what kind of info Drs used in diag patients. Experts, intermediates, novices thought aloud to diag ECG. Experts used biomedical knowledge, less skilled did not.
Representation	Core to simple problems A problem is solved when an appropriate representation is found Restructuring or re-representing a problem can bring a solution - INSIGHT (an 'aha' moment if sudden) Zhang & Norman (94) distinguished btwn internal (e.g. only one disc at a time) and external (implied but not explicit, e.g. cups have coffee in) representations
GESTALTIST THINKING	Studied PS from 1910 Importance of representation, and problem solution in terms of representational restructuring (insight if sudden). Problem solving = exploring and testing diff reps
GESTALTIST THINKING	Gauss (Hall, 70) 5050. e.g. of good structuring, or representation of a problem. RESTRUCTURING = can lead to solution, e.g. X-ray, creating sub-goals can lead to solution. WHY DO PEOPLE NOT SOLVE?Duncker's (45) X-ray problem and Luchins (59) water-jugs and 9-dot problem e.g's of ppts 'SET' effects (learned /habitual ways) preventing solver of using better methods or cause them to make unwarranted assumptions. FUNCTIONAL FIXITY = difficulty in imagining object being used in new way
REPRESENTATION - ISOMORPHIC PROBLEMS	Problems with ID underlying structure, e.g.Simon & Hayes (76) 'monster problem, same as TOH. 'Move' version where ppts had to transfer according to size easier than 'change' version where ppts had to change size of globe. Because representation for move problem simpler ops than for change representation
INFORMATION-PROCESSING APPROACH: SEARCH	Problem solving as search process. Search can be forwards by generating possible actions or backwards - means-ends approach which breaks into subgoals Gilhooly (99) ToL problem - strategy tended to be means-ends. M-E analysis involves reducing diff between start and goal state, so moves that bring closer to goal preferred. Thomas (74) similar finding with Hobbits & Orcs. Data form H&O and Water Jars suggest solvers look ahead, eval few poss steps to see whether they appear to move closer to goal
INFORMATION PROCESSING APPROACH: INSIGHT	Ohlsoon (92) clarified Gestaltist 'restructuring' People generate OPERATORS (poss actions) from LTM, cued by problem representation. If they reach an impasse: 1) ELABORATION adding info, noticing additional features 2) RE-ENCODING changing the encoding e.g. of 'married' to 'the person who conducted the marriage' 3) CONSTRAINT RELAXATION making goal req less restrictive, e.g. relaxing idea that lines stay inside for 9-dot Chronicle (01) because ppts apply heuristic search proc w/ 9-dot to cancel as many dots as possible RE-ENCODING is necessary Knoblich (99) e.g. of CONSTRAINT RELAXATION - harder to break constraint on changing operators than numerals
ANALOGIES	Share same deep structure. Problems solved when similarity of structure is realised Analogies help understanding Duncker's x-ray problem - analogy with general attacking castle. Analogue + hint = highest solve rate The closer the analogy the more transfer likely. Dunbar 'analogical paradox' = in RL use of abstract analogies common, in exps ppts don't. Blanchette & Dunbar (00) generating analogies requires use of structural not superficial features. In exp studies, materials are not fam so structural difficult. Explains why hint + analogy effectve
HOW DO ANALOGIES WORK?	Gentner (01) theory of structure mapping. Anological mapping is structural alignment between base and target. Explicit correspondences established. Has been successfully modelled (SME)
COMPLEX PROBLEM SOLVING	Types: Adversary problem solving e.g. chess Non-adversary PS e.g. code-breaking
KNOWLEDGE AND EXPERT PROBLEM SOLVING	Experts remember more than novices , have greater domain expertise that is organised in a meaningful and accessible way (Chase and Simon 73 chess) But experts do not appear to be thinking ahead more than novices (De Groot 65) So expertise is linked to how much you remember and domain-specific knowledge
EXPERTS WORK FORWARD	Experts worked forwards and novices backwards (means-ends analysis) (Larkin physics problems 80). Experts use domain knowledge to generate good problem representation which supports the use of working forwards. The absence of detailed knowledge means novices fall back on means-ends analysis
EXPERTS HAVE BETTER PROBLEM REPRESENTATIONS	Chi et al (82) asked expert and novice physicists to categorise problems in terms of similarity in solving. No diff in quant measures e.g. no. categories or time taken Clear qual diffs in nature of categories. Novices grouped in terms of surface structures, experts in terms of deep structure, e.g. physics laws that were needed to solve problems. Experts aware of commonalities between problems. (Similar findings with maths) Experts able to perceive appropriate solution within 45 secs suggesting that when knowledge useful for particular problem is accessed when a problem is characterised as a particular type
EXPERTS PRACTICE	Practice is necessary but not sufficient to ensure mastery. Expertise takes time (S&C 73) - 3k hours to chess expert, 30k hours to master Future experts seem to be better learners and to encode more optimally than other novices. But Ericcson and Harris 90 trained subjects to improve on previous best performance - non-chess player took 50hrs to recognise chess positions almost as acc as masters Waiters most skilled in remembering orders used more effective encoding strategies. Something must mediate between practice and performance.
MODAL MODEL?	Results consistently show link between expertise and domain-relevant knowledge (Chi etc). Therefore modal model emerged expertise depends on acquisition and organisation in LTM of domain-relevant knowledge and skill. Although a relationship exists and is supported by data, model lacked explanatory power (Sternberg 95).
DIRECTIONS: CHESS SKILL & MEMORY	Holding and Reynolds (82) are skill differences there in absence of memory diffs? Skill level unrelated to recall of random positions (like de Groot 65) But number of best moves correlated with playing strength. Something other than memory must be in play (ability to evaluate given position) Holding (79) chess players A-E confirmed this - As made more good move and fewer evaluation errors
DIRECTIONS: GENERAL VS SPECIFIC METHODS	Shraagen (83) sensory psy. exp. Domain-experts = better solutions but reasoning comparable - skills seem to transfer. This is at odds with idea that expertise is domain-specific. Shu & Anderson (99) - do expert scientists share skills? Domain experts designed best experiment. But a larger set of domain-general skills important to scientific reasoning Skill in problem solving is more than domain-specific knowledge
INDIVIDUAL DIFFERENCES	Novices Only some novices become experts Good learners use better encoding strategies, e.g. for map info (Thorndyke and Stasz). Good learners use exploratory approach (Green and Gilhooly 90) Good learners seem to self-explain better (Chi 94) and prompting learners to self-explain leads to better problem solving. This leads to elaboration and aids understanding and schema development EXPERTS Expert computer programmers differ in their extent and use of knowledge (Draper 84). Chess masters show different sub-domains of specialisation (Charness 91). Neither experts or novices are homogonous.
ENHANCING SKILL ACQUISITION	Sweller (88) paradox of M-E strategy Emphasis on goal = limitation ppts with no goal workd forward, therefore presence of goal creates bias to M-E analysis de-emphasising goal frees WM resources. Vollmeyer (96) non specific goals seem to aid learning because it encourages more hypothesis testing. Hypothesis testing, not reduction in goal specificity encourages learning. Green (02) instructions that led to swift learning seemed to result in poor problem solving, instructions that led to slower learning seemed to result in better PS. Different instructions influence nature of task or problem representation that effects learning and problem solving. See internal and external representations. Haider & Frensch (96) as we become experienced we learn to ignore task-irrelevant info. But not all individuals can do this

Next up

Description

Resource summary

Similar

	Created by katharinemlee over 12 years ago