Problem Solving-Lecture17

Ch 11: Problem Solving
04/09/07
Problem solving
• Richard Feynman:You write down the problem.You

think very hard. Then you write down the answer.
(Kohler, 1925)
(Kohler, 1925)
(Kohler, 1925)
Insight & problem solving
• Metcalfe & Wiebe (1987)
• Non-insight problems:
perceived ‘warmth’
increases gradually
• Insight problems: Feel

‘cold’ right until the
point where you solve
the problem
A checkerboard consists of 64 squares. These 64 squares
can be completely covered by placing 32 dominos on the
board so each domino covers two squares. If we eliminate
two corners of the checkerboard, can we cover the
remaining squares with 31 dominos?
Gestalt approach
• Problem solving =
• Problem perception
• Insight through reorganization & restructuring of a

problem representation
Gestalt approach
• Weaknesses:
• Explanations mostly focus on visual

representations
• Can describe restructuring after the fact, but

can’t predict insight
• i.e., no mechanism behind the theory

7x + 3 = 38
• Why is that interesting?
• How do we do that?
Procedural knowledge
IF the goal is to solve an equation

THEN retrieve instructions how to solve it
IF I’ve retrieved a set of instructions
THEN retrieve the first step of the instructions
IF I’m evaluating the expression ‘a + b’
THEN try to retrieve a fact of the form ‘a + b = c’
...
Procedural knowledge
Means-Ends Analysis
• Much of problem solving can be understood

through the application of something called means-
ends analysis
• "A problem-solving strategy in which the goal is to

reduce the difference between the initial and goal
state. This is achieved by working to achieve
subgoals that move the process of solution closer
to the goal."
The secret of humanity
• (Qin, Anderson, Silk, Stenger, & Carter, 2004)
• Examined 11–14yr olds about to start Algebra 1
• Students spent 6 days practicing simple algebra

equations
• 0-, 1-, or 2-step equations

Average solution time
8000
7000
6000
Time to Solve (msec.)
5000
4000
3000
2000
1000 0 Step: Data 1 Step: Data 2 Step: Data

O Step: Theory 1 Step: Theory 2 Step: Theory
0
0 1 2 3 4 5
Days
Modeling human problem
solving in ACT-R
Press “5” 7x+3=38
35/7=5 7x=35
Unwind-left
Procedural module
• Coordinates the activity of independent modules
• Executes procedural knowledge by sending

commands to other modules and harvesting results
• Only one “production rule” can execute at a time
• Each rule takes 50ms to initiate

Trip down declarative memory lane
Retrieve: 35 / 7 =__
( )
ai = ∑ ln t j − d
j
RT = F ⋅ e− ai
35 / 7 = 5
Human problem solving
• Trace of the steps involved in

solving 7x+3=38
• Left: Steps required on day 1
• Right: Steps on day 4

From theory to data
8000
7000
6000
Time to Solve (msec.)

5000
4000
3000
2000
1000 0 Step: Data 1 Step: Data 2 Step: Data

O Step: Theory 1 Step: Theory 2 Step: Theory
0
0 1 2 3 4 5
Days
From theory to brain
• What would you expect if you studied children

solving these problems in an fMRI machine?
• What does fMRI even measure?

The brain is sloshy
• BOLD response for various regions following
activation at time 0
Predicting BOLD response
So what makes us human?
• Solving algebra problems requires:
• Concept of numbers
• Ability to manipulate abstract symbols
• Generativity of rules (?)
• Ability to mentally construct & react to problem

structures that conflict with information in the
world
Pyramid arithmetic
• Pyramid problems: A notation for writing repeated

addition where each term added is one less than
the previous
• For example, 5+4+3 can be written 5∆2
• Since 5+4+3=12, can write 5∆2=12

Pyramid arithmetic
• 5∆3 =
• 8∆1 =
• 3∆4 =
• 5∆7 =
• 13∆0 =
• 1000∆2000 =
• 100∆X = -101
• X∆6 = 0
1000∆2000
• No students attempted to use their current

method, despite the fact that
• It was the only method instructed,
• It worked well on past problems,
• & was highly practiced.

1000∆2000
• Students seemingly inferred that performing 2000

additions isn’t the correct solution to the problem
• Also requires knowing that:
• There are positive and negative values that

cancel
• Last number in the sum is the difference of the

base and height
• None of which was instructed to the students!

Metacognitive parallel reflection
• While proceeding with routine problem solving

procedures we are nondeliberately reflecting on
what we are doing
• Simultaneously inventing and teaching ourselves

new knowledge that might be useful later on...all
without realizing it
Insight & strategy discovery
• Siegler & Stern (1998) studied 2nd graders solving
simple arithmetic problems:
• (a+b-c)
• (a+b-b)
• At the start of the study, none of the children used

the shortcut strategy
• Recorded RT for each problem, and got verbal

report of strategy children used on the problem
• Found that children started using the shortcut

strategy before they realized that they were doing
so
Wrap-up
• Need to see one person from each project to sign

up for a presentation slot
• First presentations are next Thursday (4/19)
• Quiz this Thurs on Ch. 11

Problem Solving-Lecture17

Uploaded by

Copyright:

Available Formats

Problem Solving-Lecture17

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Problem Solving-Lecture17

Uploaded by

Copyright:

Available Formats

Ch 11: Problem Solving

• Richard Feynman:You write down the problem.You

• Metcalfe & Wiebe (1987)

• Insight problems: Feel

• Insight through reorganization & restructuring of a

• Explanations mostly focus on visual

• Can describe restructuring after the fact, but

• i.e., no mechanism behind the theory

IF the goal is to solve an equation

• Much of problem solving can be understood

• "A problem-solving strategy in which the goal is to

• (Qin, Anderson, Silk, Stenger, & Carter, 2004)

• Examined 11–14yr olds about to start Algebra 1

• Students spent 6 days practicing simple algebra

• 0-, 1-, or 2-step equations

1000 0 Step: Data 1 Step: Data 2 Step: Data

Press “5” 7x+3=38

• Coordinates the activity of independent modules

• Executes procedural knowledge by sending

• Only one “production rule” can execute at a time

• Each rule takes 50ms to initiate

• Trace of the steps involved in

• Left: Steps required on day 1

• Right: Steps on day 4

Time to Solve (msec.)

1000 0 Step: Data 1 Step: Data 2 Step: Data

• What would you expect if you studied children

• What does fMRI even measure?

• Solving algebra problems requires:

• Ability to manipulate abstract symbols

• Generativity of rules (?)

• Ability to mentally construct & react to problem

• Pyramid problems: A notation for writing repeated

• For example, 5+4+3 can be written 5∆2

• Since 5+4+3=12, can write 5∆2=12

• No students attempted to use their current

• It was the only method instructed,

• It worked well on past problems,

• & was highly practiced.

• Students seemingly inferred that performing 2000

• Also requires knowing that:

• There are positive and negative values that

• Last number in the sum is the difference of the

• None of which was instructed to the students!

• While proceeding with routine problem solving

• Simultaneously inventing and teaching ourselves

• At the start of the study, none of the children used

• Recorded RT for each problem, and got verbal

• Found that children started using the shortcut

• Need to see one person from each project to sign

• First presentations are next Thursday (4/19)

• Quiz this Thurs on Ch. 11

You might also like