problem-solving 1 &

well-defined problems & examples, who can solve?

requirements are clear, all info needed is present

• puzzles, applying algorithms

• AI can solve

problem space

initial & goal states, intermediate paths, operators, task constraints

generalization in problem-solving

storing & applying old scenarios w/o details to be able to generalize solutions to new problem s

ill-defined problems: what type of solutions & examples

how to overcome abiguous situations, requires added info, situational

• can be multiple solutions & expected outcomes

  • social or self problem solving

evidence of episodic memory assisting in ill-defined problem solving

help imagine future, hypothetical outcomes

what portion of brain is more active during ill-defined tasks & why

right lateral PFC, ill-defined tasks require cognitive load bc you have to make ur own constraints in loose-ended situations & working memory capacity is alr limited

moravec’s paradox & its significance to AI

everything thats hard is easy & everything thats easy is hard

• AI can solve well-defined problems well but not ill-defined problems & simple skills

tower of hanoi

used to study well-defined problem (rings need to be moved correctly to solve, algorithm can be used)

brute force & pros/cons

represents all possible steps from problem to goal state

• pros: guaranteed to find solution

• cons: decision fatigue

combinatorial explosion

computing too many alternatives, linked to decision fatigue

trial & error

considered lower-level thinking, ruling out solutions that don’t work, only suitable for limited # of possible outcomes

hill climbing strategy, cons & pros

reduces distance btwn you & endgoal, takes you as close to goal as you want— can lead to false outcome

• subgoal is mistaken as final goal

doesn’t always work bc some problems require u to move away from goal to fully solve it

• ex: me making to-do lists (get me closer to goal) instead of just doing actual work

3 heuristics useful for evading decision fatigue

trial & error, hill-climbing strategy, means-end analysis

means-end analysis & example, pros & cons

most flexible approach, identifying sub-problems to complete goal

• include forward & backward movements— importance of recursion

• constantly evaluating difference btwn current & goal states

  • ex: hobbits & orcs across river

thinking-aloud procedures

used to measure complex thinking to understand strategies

• concurrent verbalizations, retrospective verbalizations

concurrent verbalizations

actively describe what you’re doing

retrospective verbalization

describe how you solved problem afterwards

difference btwn how experts vs non-experts evaluate/solve problems w/ applied example

experts take longer with defining the problem appropriately, whereas non-experts take longer trying to develop solution

• more familiar w/ certain info & therefore can represent problem differently than non-experts

• radiologists use more holistic visual processes to analyze chest x-rays than non-experts

how does chunking relate to expertise when approaching problems

frees up space in short-term by having vast long-term— episodic memory helps

• remembering more chess board pieces

negative transfer effect

there’s bias for person to solve a problem a particular way when actually a diff approach would be more effective