Problem-solving 1

Allen Newell & Herbert Simon introduce an

abstract way of thinking about problems

an abstract way of thinking about problems called

problem space

Newell & Simon worked at a time after

The decline of behaviorism due to computers became more widespread

Cog psychologist start studying the mind as

an informationa-processing device

EX of studying the mind as information-processing device

Broadbent conceptualized attention as a “filter” between input and memory

Newell & Simon were

computer scientist and psychologist

Newell & Simon wanted to make computer programs that

solved problems and understand human problem solving

problem space

problem can be thought of as a number of states and to solve the problem, you need to move between states

2 special states

the initial state and the goal state

initial state is

the current situation (with problem)

ex of initial state

a mouse is at the entrance of a maze and wants to solve the problem of reaching the cheese at the exist

goal state is

the situation in which the problem is solved and you have reached the goal

ex of goal state

mouse has solved the problem and is eating the cheese at the exist

3rd state is

all the other ones that help moves from the initial state to the goal state

ex of 3rd state

mouse is at a dead end, mouse is on its way, mouse is nearly there

the collection of all states the problems can be in

initial state, goal state, and all the other ones

solving a problem comes down to finding a

path through the problem space from initial state to goal state

2 methods that humans and other animals use for finding one’s way from initial state to goal state

Hillclimbing and Subgoals

Hillclimbing is

a primitive way that doesn’t work for more complicated problems

Hillclimbing is only do things that

bring the current state closer to the goal state (move straight toward the ultimate goal without changing any path)

ex of hillclimbing: mouse

the mouse can move rightward, bringing it closer to the cheese

the chicken shows that the usage of hillclimbing often

results in being stuck in a local maximum (couldn’t get the food but still trying really hard)

hillclimbing method doesn’t allow the agent to

step away from its current path to explore better options. so they stuck at local maxima

Subgoals is more

sophisticated and flexible

Subgoals is often the solution of

local maximum problem

3 basic component of Subgoal: 1

Hillclimbing: move closer to the goal state

3 basic component of Subgoal: 2

if you can’t get closer to the goal, identify what obstacle is stopping you

3 basic component of Subgoal: 3

formulate a new goal: Subgoal - removing the obstacle to get to goal state

Ape using subgoal: initial state

ape is sitting next to tube, peanut is in the tube

Ape using subgoal: first action

hillclimbing: move hand closer to peanut

Ape using subgoal: obstacle

peanut is too far down in the tube to reach it

Ape using subgoal: obstacle leads to subgoal

need to reach a state where the peanut isn’t as far down the tube

Ape using subgoal: action to reach subgoal

hillclimbing: add water to move the peanut closer

Ape using subgoal: goal state

ape has its hand on the peanut

once a subgoal is formulated, the new problem is

to reach subgoal

while pursuing subgoal, the main goal is

forgotten for a sec

after the subgoal has been reached, the main goal becomes

the objective again