Reasoning and Decision Making - Lecture 1

Gestalt Approach - Representational Restructuring

solve problem by restructuring it - because we encounter a block when represented in the wrong way (Ohlsson, 1992).

Gestalt Approach - Insight

the point at which solution is suddenly seen/problem becomes clearer.

argue insight caused by sudden restructuring. may not be separate cognitive process - may arrive gradually at solution, but only aware once certain threshold crossed.

Gestalt Approach - Functional Fixedness

ideas about object function interferes with using object more usefully e.g. Candle Problem (Duncker, 1949).

Adamson (1952) - 80% solved candle problem correctly if matchbox was empty, versus 40% when not.

Luchins (1942)

Water jug problem.

Ps who did easy problem first - 95% correct on mid-level problems.

Ps who did difficult problem first - 35% correct on mid-level problems.

because of mental set - continuing to use a strategy that is not the most helpful.

Incubation and Sleep

Incubation = stop thinking about problem for a while.

• developing a new strategy (Simon, 1966)

• forgetting irrelevant/misleading info (Penaloza & Calvillo, 2012)

Sleep

• increased performance for identifying a ‘hidden rule’ amongst letter strings (Wagner et al., 2004)

Gestalt Approach - Criticism

doesn’t explain how brain is involved in solving problems - what happens in brain during representational restructuring.

Information Processing Approach (Newell & Simon, 1970s)

problems solved using search process:

• initial state

• goal state

• many subgoals (intermediate states)

intermediate states - we perform search of all possible choices.

methods for problem solving:

• planning

• heuristic methods

• progress monitoring

Planning

Tower of Hanoi Problem - Ps spent longer planning moves performed better with fewer errors (Koppenol-Gonzalez et al., 2010).

limit on amount of planning we can engage in due to limitations in STM.

Heuristic Methods - Hill-Climbing

change present state to be one step closer to goal.

mostly used when we have no clear understanding of how to achieve goal.

relatively unsophisticated, little planning involved.

Heuristic Methods - Means-End Analysis

look for difference between current state and goal.

find action to reduce difference and perform it.

repeat until final goal achieved.

Progress Monitoring

track progress towards goal, switch strategy if progress slow.

performance worse if Ps think progress is being made. realise progress slow = more likely to switch strategies (MacGregor et al., 2001).

Evidence for information processing

damage to PFC perform worse:

• on Tower of Hanoi than control, particularly for moves which take away from end goal (Goel & Grafman, 1995)

• on water-jug problem - patients using relatively unsophisticated hill-climbing strategy (Colvin et al., 2001)

Kleibeuker et al., 2013

fMRI looking at regions involved in planning in adolescents and adults.

response in lateral PFC during problem solving - activity correlated with performance. better problem solving → more activity.

Kounios et al., 2006

EEG and fMRI - neural activity before seeing problem predicts whether problem will be solved with or without insight.

activity in posterior middle temporal gyrus and cingulate.

suggests areas related to cognitive control (strategy selection) and conceptual processing (meaning).

Qiu et al., 2010

fMRI - neural activity correlated with insight during solving CHinese logographs (puzzle, like an anagram).

activity in precuneus, left inferior/middle frontal gyrus, inferior occipital gyrus.

suggests areas related to memory retrieval, changing strategies, visual imagery, and attention.

Crescentini et al., 2012

fMRI - look at regions involved in planning during problem solving of Tower of Hanoi task.

dorsolateral prefrontal cortex (DLPFC) - key area involved in initial planning more than memory.

Information Processing Approach - Pros

• precise account of how we solve problems by applying rules to reduce complexity, learn diff. strategies

• general enough to apply to variety of problems

• fits with standard models of memory i.e. limited capacity of STM

Information Processing Approach - Cons

• assumes problem solving is serial (happens in a series)

• seems to apply to well-defined problems, unclear if can be applied to ill-defined problems

Analogies

using the solution of one problem to guide the solution to a new, similar, problem.

Duncker’s Radiation Problem (Gick & Holyoak, 1980)

high laser = kill tumour + healthy tissue

low laser = harmless to healthy, too weak to kill tumour

solved using fortress analogy - multiple low intensity lasers.

• 10% could solve no help

• 30% could solve hearing fortress story

• 75% solve being told two stories related

Explicit and Implicit Reasoning

explicit = slow, deliberate, consciously aware

implicit = fast, automatic, not always aware

medical experts largely utilise implicit reasoning - can lead to errors e.g. radiologists can fail to detect diseases in ~30% of cases (Krupinski, 2011).

Eye Tracking Studies

Krupinski et al. (2013) - breast biopsies - look less at non-diagnostic regions as expertise increased.

Kundel et al. (2007) - mammograms - more likely to identify cancer if make eye movement to the cancer in under 1s.

• most experience = fixate on cancer quickest → use holistic, implicit process rather than serial search.

Chase & Simon (1973)

task - memorise location of chess pieces - either randomly or in game positions.

experts → remember game positions better, not random. do not have overall better memory.

do this by chunking relevant information.