9. Problem-solving: Animals, Humans, and Machines Flashcards
problem-solving
- type of thinking requires realization that a problem exists
- appears to involve conscious manipulation of knowledge and involves:
1. sensemaking
2. reasoning
3. decision-making - problem cannot be solved by an automatic process of perception or recognition (we haven’t encountered this problem before, and therefore we cannot retrieve a solution)
constrained problem-solving: Scrabble
goal
- win the scrabble game
relevant objects (resources, problem-solver)
- knowledge of players, dictionary
operations available
- combine letters
constraints
- spell legal words
- words must intersect with other words
- maximise points
puzzling problems: Thorndike’s cat in a box
- refer to page 57 of notebook
perceptual reorganization analogy: Getsalt Principles
- if perception is more than just compiling lists of features, problem-solving is more than just listing of info relevant to problem
- gestalt psychologists specified processes relating info used during problem-solving followed by insight (a-ha moment)
animal problem-solving: Kohler’s (1925) chimps
- chimps try to solve a problem, pause for a period and then go directly to the solution
- chimps’ B suggested that problem solving sometimes require a reorganization of objects + their relationship between one another in the problem solver’s environment
Kohler’s Gestalt problem-solving
in Gestalt terms, problem-solving involves
- attention to the structure of the problem - restructuring may be necessary
- insight involving the sudden realization of solution to a problem
- use of productive and reproductive strategies - interference can result from use of reproductive strategies
- problem solving set
Kohler’s Gestalt problem-solving: problem-solving set
- tendency to repeat a solution process that has been previously successful
- e.g. taking a circuitous route to an unfamiliar city because it is easier than learning a new more direct route
Kohler’s Gestalt problem-solving: functional fixedness
- if an object has one established use in a situation, people have difficulty using the object in another way
- e.g. glue is for sticking one object to another, more or less permanently - post it notes
Kohler’s Gestalt problem-solving: Duncker’s candle-and-tack problem example
- refer to slide 14 of powerpoint
problem-restructuring
- refer to slide 15 of powerpoint for Maier’s two-string problem
- refer to slides 16-17 of powerpoint for the Tumour Problem
- refer to slides 18-19 of powerpoint for Die Hard jug problem
mental sets
- formed, which aids them in completion of subsequent tasks
- when given different examples, participants had an incompatible mental set that created performance decrements
- i.e. less ACC in finding solution
general features of problem-solving
- previously learned responses can interfere with problem-solving (e.g. compatibility problems)
- Lunchin’s work
general features of probem-solving: Lunchin’s work
- provided important insights into problem solving, producing a body of experimental data for evaluation by later researchers
- neurological evidence suggests that insight (vs non-insight) problems activate specific regions (right anterior superior temporal gyrus)
issue with general features of problem solving
- ‘insight’ and ‘restructuring’ are vague concepts
- more descriptive that explanatory
alternatives to Gestalt approach
- procedural (step by step) approach
- Wallace’s stages of problem solving
1. preparation: formulation of the problem
2. incubation: leave problem temporarily
3. illumination: insight into correct solution
4. verification: make sure solution actually works
problem space theory:
info processing approach to problem-solving
- problem solver searches a problem space filled with possible solutions
- some work, some don’t
computer simulation of problem solving by humans based on protocols collected form subjects as they solved problems
problem space
- combination of initial state, goal state and intermediate (states necessary for solution and other potential states)
- intermediate states contain both the optimal solution and other potential states
- multiple solutions can achieve the goal state (depending on constraints)
- refer to the maze example, page 58 in notebook
problem space theory: components
operators
- actions that cna be performed to move from one problem state to another state
heuristics
- strategies for the efficient application of operators that generally lead to the goal state
- contrast with algorithm
constraints
- limitations that are imposed by info processing system
- e.g. working memory and attention limitations
domain-independent heuristics: means-end analysis
- note difference between current sate and goal state
- e.g. I’m at the top left corner of the maze and I want to be in the bottom right corner - create sub goals to reduce differences
- e.g. I need to get to the middle because there appears to be a path leading to the bottom right - select operator that will lead to the subgoal
- e.g. this path appears to be heading in the right direction; I’ll take it!
domain-independent heuristics: loop avoidance strategy
- note how often steps are repeated
- if they are repeated and no further advancement is made toward the goal state, do something else
- e.g. if I keep coming back to the same part of the maze, I’m lost
domain-independent heuristics: working backward
- start at the goal state and move towards initial state
- RECALL water lily problem
domain-dependent heuristics
heuristic that can be used in a fixed context for a limited number of problems
domain-independent heuristic
heuristic for use in any problem
domain-dependent heuristic examples
- counting tiles to measure a surface area of a lorr
- tiles won’t allow you to measure the size of an atom - rounding numbers when performing calculations
- you can’t actually have the average number of children - bringing a doctor’s note for a missed test
- if you’ve missed someone’s birthday (or job interview) you’re out of luck
domain-dependent heuristics: Tower of Hanoi
- refer to slides 31-32-33 of PowerPoint
- benefit: limited nb of intermediate stages - easy to model
- we want to avoid the suboptimal path
Tower Power: Kotovsky (1985)
- suggested that working memory = required (hints only in ‘no load’ condition), unless participants have automated their response selection
Tower Power: Goel and Grafman (1995)
examined performance of patients with frontal lobe lesions compared to controls (no difference in IQ/memory)
- both used same general strategies with no evidence that planning was impaired
- patients had issues in identifying and resolving conflicts between goals and subgoals
cognitive mechanisms of problem solving: domain general mechanisms
working memory
- central executive
- phonological loop
- visuo-spatial sketchpad
cognitive mechanisms of problem solving: domain specific mechanisms
long-term memory
- procedural knowledge
- schemas
- templates
- knowledge/organization
cognitive mechanisms of problem solving: problem space theory
- useful for understanding well-defined problems
- HOWEVER, majority of problems are ill defined
- e.g. raising children, learning the piano, writing a novel
What do we need to do with other situations and problems from our experience?
We need to find similarities
