The psychology of planning

Question 1

CCC review- problem solving

Answer 1

Three aspects:
1. goal directed
2. an immediate solution is not available
3. involves conscious cognitive processes (however it doesn’t just involve conscious cognitive processes)

Question 2

CCC review- the three parts of a problem

Answer 2

1. The problem itself (the start state)
2. The things you might do (the operators)
3. The solution (goal state)

If you’re clear about all of these, the problem is well-specified

However…
- you don’t necessarily know what you’re start state is
- research doesn’t just focus on well-specified problems
- ill defined problems are much more realistic of the problems we tackle everyday

Question 3

CCC review- the tower of Hanoi

Answer 3

‘The hardest problems are those requiring a move that takes you (temporarily) further away from the goal state’

This is not really the reason why some problems are hard and other problems are easy- this point is only kind of true

Question 4

What is planning?

Answer 4

deciding on the order and intensity of decomposition of problem, and determining consequences of alternative plans

Question 5

Planning:
1- what does planning involve?
2- what is search guided by?
3- what is planning?
4- what is planning mediated by?

Answer 5

1- Planning involves search through the problem space (mental representation of the problem and prior knowledge and motivations and environment in which you’re trying to solve the problem).

Moving from a start state in a series of steps you search for possibilities

For ill defined problems- the possibilities could be infinite. SO what people tend to do is form heuristics

2- Search is guided by heuristics (yields the results you want to yield but doesn’t necessarily do so eg. airport security- to check people the best thing to do would be strip them. This is an algorithmic approach- guarantees the same outcome every time you run the analysis). Heuristic doesn’t guarantee the same behaviour so it could involve interviewing people in relation to airport security. However there are biases such as gender, racial profiling, age.

3- Planning is constrained
example = chess- if you plan exhaustively, within 2 moves, the number of possibilities will exhaust good people.

4- Planning is mediated by external environments

Question 6

Problem decomposition + example

Answer 6

Sub-goal specification

Essay example- 2 to do:

1. what are the tasks (understanding the question, finding relevant literature, writing a plan, writing a draft)
2. what is the essay content going to be, what is the plan how the contents is going to go

Question 7

Negatives for compare and contrast

Answer 7

Two fold- writing about A and B is obvious from writer pov that they know the contrast when they don’t

Also have to repeat what you said

To avoid this, need a structure and have a narrative about A vs B all the way through your essay

Question 8

What is planning: problem components

Answer 8

Initial state
* The problem as presented
Goal state
* The aim/intention/desired outcome
Operators
* Things you can do/try/execute
Constraints
* Limitations on what you can do/try/execute
* Additional requirements/rules (e.g., accuracy, latency, etc.)

eg. can’t press a bigger disc on top of a smaller disc

Question 9

What is planning: problem decomposition
What did H.A. Simon, Sciences of the Artificial, 1981 come up with?

Answer 9

“..Design may be the ultimate expression of human thought..”
(H.A. Simon, Sciences of the Artificial, 1981)
* Complex, multi-faceted.
* Constrained by domain, brief, market & tradition.
* Requiring an element of creativity, sometimes novelty.

Simon also introduced the idea of satisficing- the idea that we don’t try to optimise the outcome of our decisions, we just try to come up with something that satisfies the problem we have at that time. Its an adequate rather than perfect solution.

Question 10

PROBLEM DECOMPOSITION
Thinking about the problem of designing a car

Answer 10

Each item has problems and sub-sub problems
eg.
chassis - frame and axles
axles- wheels and transmission
wheels- tyres and hubs
tyres- tread and profile

You decompose these things until you get to the bottom of this hierarchy where you can actually do things

Question 11

Decomposition orders

Answer 11

Breadth-first
Going from unpacking the problem across the .. all the way down
* Advantage – minimal commitment- don’t commit yourself to something you will later have to reverse
* disadvantage- don’t get feedback from anything until you start doing it

Depth-first
* Advantage – immediate feedback; lower cognitive load
* Disadvantage- commit yourself so can’t undo things

Opportunistic (take things as they arrive)
* Capitalising on current state

Question 12

The problem space
What is it?

Answer 12

The mental representation of a problem

Question 13

The problem space:
What makes it up?

Answer 13

• State space (the number of things you can do to the problems that are in your head)
• Task environment
• Information processing system (idea that we have a limited capacity of memory and a limited processing capacity for thinking)

Question 14

The state space:
1- What is the state space of a problem?
2- What happens the larger it is?
3- What 2 concepts did Newell & Simon (1972) introduce
4- Task

Answer 14

1- All possible paths between initial and goal states
All the possible places you can visit to get to the solution

2- The larger it is, the harder a problem will be to solve

3- Newell & Simon (1972)
a) ‘bounded rationality’ - because problems tend to be too big, we place rational, sensible boundaries on what we tend to explore
b) satisficing

4- Towers of Hanoi

Question 15

The task environment
1- what does the ability to solve problems depend on?
2- what are the ways a problem is presented to the solver?

Answer 15

1- the world around us

2-
* Format (display type)
* Thematic content (e.g., familiarity)
* Conditions (e.g., criticality; risk)

Question 16

The task environment:
What did Zhang and Norman (1994) do?

Answer 16

Took the Towers of Hanoi problem and changed the way it was presented.

3 beakers of water- large balancing on medium, balancing on small
Have to move beakers from one bin to another under the constraint that you cant place a small beaker over a large one because it will fall over.

Simply by shifting the context of the task and the environment being solid disks on top of each other to cups that could fall into each other, people found this cup version easier to solve.

Their explanation was that he task environment encodes the constraint about whats cup can be placed don’t think about placing small cup on top of large because you know it would fall in. This makes it easier.

Question 17

Information Processing System

Answer 17

Working memory
* Constraint on planning steps
* Chess: From any given position there are on average 35 possible moves. If a chess game lasts 100 moves = 35^100 possible moves. Impossible to calculate

Long-term memory
* Knowledge of solutions, operators and constraints
* Expertise

Question 18

Search using heuristics
2 methods

Answer 18

Means-ends analysis
* E.g., fix a car tyre
* Make situation safe; Remove wheel; Loosen nuts; Raise car; Undo nuts; Slide wheel off; Replace wheel, lower car; tighten nuts, etc.

Analyse what means are going to get you the furthest and the order
Allows you to understand that removing the wheel enables you to remove the nuts ect.

Operator selection
* Select operator that maximises reduction of distance between initial and goal states
* Set as ‘sub-goal’ to apply the operator

Question 19

Means-ends analysis and the Tower of Hanoi

Answer 19

• Goal – move all three disks from A to C
• Operator – move disks 3 then 2 then 1 to C
  
  • Subgoal – move 2 then 1 to C
  • Subgoal – move 1 to C
    BUT this can lead to a wrong move (1 to B so you can move 2 to C faster)

-> planning (i.e. thinking ahead)

Question 20

What are other heuristics?

Answer 20

• Means-ends analysis
• Hill-climbing
• Trial and error
• Heuristics for sampling
  — Anchoring
  — Representativeness
  — Etc…

Question 21

The nine-ball problem

Answer 21

You have 7 / 9 balls ….
It costs £1 to weigh each ball.
You have £8 / £12….

% first weighs for 9-ball:
4 vs. 4 – 42%
4vs. 5 – 37%

People tend to weigh 4 v 4 when they should be weighing 3 v 3

Many people weight 4 v 5 but this would mean one side is bound to go down because it has more balls on it. When people make this mistake, they are more likely to solve compared to people who don’t.
Therefore this failure triggers you to think more carefully.

If you have £12 you can weigh 4 v 4 and 4 v 5, there is no indication you’re getting into problems. SO theres no incentive to think ahead. Where as for £8 you run out of money. This triggers people to plan more carefully.

Question 22

The nine-ball problem results

Answer 22

For the 7 ball version, the solution is the same- you weigh 3 v 3 if it balances you know its in the one you haven’t weighed. You would predict this is easier to solve because the state space is smaller

For the 7 ball- The moves that make the most progress towards the solution are the correct moves to make. This is not the case for the 9 ball.

This is why the results proved the 7 ball to have a higher % of correct solutions.

But it also matters how much money you have been given:

If you have been given £12 for the 9 ball problem it makes no difference compared to having no cost. But for the 7 ball, it makes you worse.

If given £8 for the 9 ball it makes you much better but for the 7 ball it doesn’t make much difference and makes you slightly worse.

7 ball- adding money seems to be an extra piece of info that confuses people.- problem with task environment

More solutions to people who have constraint that makes them fail early.

Design philosophy- fail fast, fail often- you want to try things out and get feedback quickly so you take risks to fail.