Chapter 10: Problem Solving

- Gestalt psychologists argue that consciousness tends to be organized into a coherent whole or
Gestalt

- The figure above is a bi-stable figure; you either see an old lady or a young woman looking
over her shoulder → you only see one figure at a time not both
- Gestalt switch: a sudden change in the way information is organized
- Gestalt switches can occur verbally as well;
- E.g Swimming under a bridge came two ducks in front of two ducks, two ducks behind
two ducks, and two ducks in the middle. How many ducks were there in all?
- If you’re like most people your answer is six, your organized the duck this way:
O O
O O
O O
- The phrase two duck makes you think of pair but suppose you were told they are
swimming in a single file, your answer might be four;
O
O
O
O

- Insight problem: a problem that we must look at from a diff angle before we can see how to
solve it
- An insight problem typically gives us all the information we need to solve it: no
additional information is required but we may have to look at it from a diff way

Kohler and the mentality of apes

- Kohler studied “whether chimpanzees do not behave with intelligence and insight under
conditions which require such behavior”
 - Chimpanzees were useful participants because they could be placed in an experimental situation
and presented with problems that they might never have faced before.

- Sultan the chimpanzee:

- He was in a cage with fruit outside it, beyond his reach. There was a short stick in the
cage, and a longer stick just outside the bars.
- The longer stick was out of reach but the short one could be used to drag it nearer
- After vainly trying to use first the short stick and then a bit of wire, Sultan studied the
whole scene.
- Finally he used the short stick to obtain the large stick and then used the latter to get the
banana.
- Sultan’s behavior displayed insight, by which he meant the ability to understand how the parts of
a situation are related to one another.

Wertheimer and Productive thinking

Altar window problem:
- A circular altar window is to be surrounded with gold paint, like the image below:

- The area to be painted gold is bounded by two parallel vertical lines tangent to the circle and
equal in length to the diameter of the circle. These lines are joined by semicircles. To figure out
how much paint is required, you need to know the size of the area inside the lines but outside the
window.
- Wertheimer described several attempts to solve this problem:
- Adult participants interpreted the problem in terms of what they had learned from similar
problems in the past.
- Some of them felt certain that they could solve such an apparently simple problem.
- they attempted to apply solution procedures blindly, without any real conception of what
the problem required.
- it was easy for them to find the area of the window itself because they already knew the
formula for finding the area of a circle.
- Children participants w/ not math skills → reaction to the problem to say they dont
know how to solve it
- Then he looks at the figure for a moment and realizes that the two top and bottom
semicircles fit inside the window.
- Thus the area required is simply the area of a square with sides the same size as the
diameter of the circle.
 - Productive thinking (Wertheimer): thinking based on a grasp of the general principles that
apply in the situation at hand
- Structurally blind / reproductive thinking: the tendency to use familiar or routine procedures,
reproducing thinking that was appropriate for other situations, but is not appropriate for the
current situation
- In order to think productively, you need to go beyond having a little knowledge that you can
misapply: you need to look at the situation with fresh eyes in order to recognize and apply the
general principles that are relevant to it

Duncker and Functional Fixedness

- Analysis of the situation: determining what functions the object in the situation have and how
they can be used to solve the problem
- E.g When we have a problem, often our first impulse is to ask ourselves, “What did I do
in similar situations in the past?”
- Functional fixedness: the inability to see beyond the most common use of a particular object and
recognize that it could also perform the function needed to solve a problem; also, the tendency to
think about objects based on the function for which they were designed
- Here is an example of functional fixedness called the coin problem (Simmel, 1953).
- Suppose you have eight coins and a balance. One of the coins is counterfeit and therefore
lighter than the others. How can you find the counterfeit coin by using the balance only
twice?
- Most people initially think of dividing the coins into two groups of four coins
each. One of the two groups will be lighter and so must contain the counterfeit
coin. Then you can take the four coins from that group and weigh them two
against two. Of course, one of the groups of two will be lighter. But you can’t
determine which of the two remaining coins is the counterfeit one because you
have already used the balance twice.

Maier and the concept of direction

- Maier introduced the nine dot problem:

- It requires you to connect all the dots with four straight lines without lifting your pencil
from the paper (the solution is given in the second part of the figure).
- Even with the hint, however, many participants still did not solve the problem.
- “The Gestalt view holds that once fixation is broken, the solution either appears whole in
a flash of insight or is produced smoothly as one step leads to another”

- One of Maier’s best-known problems is the two-string problem:

 - There are two strings hanging from the ceiling. The task is to tie the two strings together,
but they are too short for the participant to reach one while holding onto the other.
- The productive solution is to tie a weight to one of the strings, set it swinging, go and get
the second string, walk over to the middle of the room, and wait for the first string to
swing over to you. Then you can tie them together.
- If participants did not figure out the solution Maier gave them a hint → He brushed
past one of the strings, setting it swinging.
- Maier argued that, in order to be effective, a hint must be consistent with the direction
that the person’s thinking is taking.

- Hint (maier’s view): a hint must be consistent w/ the direction that the person’s thinking is
taking, and cannot be useful unless it responds to a difficulty that the person has already
experienced.

Insight is involuntary
- Feeling of warmth: the feeling that many people have as they approach solution to a problem (i.e
getting warm)
- solution. This is because non-insight problems are solved step-by-step, and with each
step they feel they are “getting warmer.”
- with insight problems participants should not feel that they are getting warmer until the
solution actually appears; thus the feeling-of-warmth with insight problems should stay
more or less level until the solution is reached, at which time it should rise dramatically.
- Feeling of knowing: the feeling that you will be able to solve a particular problem
- participants were asked to rank in order the set of problems they would be working on,
from those they thought would be the easiest to handle to those they thought they might
not be able to solve.
- For non-insight problems, participants were able to predict fairly accurately which ones
they would be able to solve and which ones they would not.

Progression Monitoring theory

- Progress monitoring theory: the theory that we monitor our progress on a problem, and when
we reach an impasse we are open to an insightful
- Participants in their research took what seemed to be the most straight forward route to a solution,
but with insight problems this approach often led to failure.
- It was only when the participants realized that they had gone down a blind alley that they
considered alternative possibilities. The participants monitored their progress on a problem, and
when they reached an impasse they were open to an insightful solution.
 - In one experiment participants were given the nine-dot problem with one correct line already
drawn.
- One group was given the problem with a line connecting three dots horizontally and
extending outside the area of the square. Another group was given the problem with a
diagonal line that did not extend outside the area of the square.
- Another group was given the problem with a diagonal line that did not extend outside the
area of the square. Since drawing lines outside the square is crucial to the solution, you
might think that the horizontal line would be more effective.
- The reason the diagonal line was more effective was that it led participants to reach an impasse
more quickly.

Representational Change Theory

- Representational change theory: the theory that insight requires a changer in the way
participants represent the problem to themselves
- Constraint relaxation: an aspect of representational change theory: the removal of assumption
that are blocking problem solutions
- Chunk decomposition: separating the problem into the chunks that belong together and thinking
about them independently

Insight and the brain

- Recall that the that the anterior cingulate cortex (ACC) detects conflicting response tendencies
and facilitates the process whereby we become aware of such conflicts.
- Luo and Niki found evidence for ACC involvement in the insight process using both functional
magnetic resonance imaging (fMRI) and event-related potential (ERP) techniques.
- Luo and Niki (2003) also reported hippocampal involvement in the insight process.
- Luo and Niki (2003, p. 321) believe that this function of the hippocampus helps to
explain its role in insight.

Insight and sleep

- “Sleep on it” → excellent strategy
- Wagner, Gais, Haider, Verleger, and Born have shown that this may be an excellent
strategy
- To investigate the relationship between sleep and insight, they gave participants a number
reduction task
- This is a demanding task that requires close attention to detail. Participants are given a
string of eight numbers, all composed of only three numbers: 1, 4, and 9.
- They then have to generate a series of responses using two rules.
- The first, called the same rule, stipulates that if there is a sequence of two identical
numbers, then the response must be one of those numbers.
- Thus for the sequence 4 4 the response is 4.
- The second rule, called the different rule, stipulates that for a sequence of two different
numbers, the response is the third number.
- For example, the response to the sequence 1 4 would be 9. The response to the
first two numbers in a string is then compared with the third number in the string.
 Thus if the first two numbers are 4 and 4, for which the response is 4, this
response (4) is then compared to the next number in the string. If that next
number is 9, then the correct response is 1.
- Comparing 1 and 9, by the second rule, gives you 4 as the first response; then comparing
4 to 1 gives you 9 as the second response, by the same rule; 9 is then compared with 4 so
that 1 is the next response, again by the same rule; 1 compared with 9 gives 4, by the
same rule.

Functional Fixedness and the design of tools

- Defeyter demonstrated that young children may be less functionally fixed than older children.
- In one study, five-, six-, and seven-year-olds were divided into two groups at each age level and
presented with a task that required them to discover that a box they had been given could be used
not just as a container but as something to stand on.
- One group was presented with the pre-utilization condition, and the other was presented
with the no-pre-utilization condition.
- In the pre-utilization condition, the box was full of things, demonstrating its conventional
function as a container.
- In the no-pre-utilization condition, the box was empty
- five-year-olds were equally fast regardless of whether the boxes were full or empty.
- However, the six- and seven-year-olds performed much worse under the pre-utilization
condition.
- German and Defeyter (2000; Defeyter & German, 2003) interpreted these and similar results as
pointing to the development of a tendency to perceive the function of a tool in terms of the use for
which it was designed.
- By age six or thereabouts, children believe that the function of an object is the one for
which it was created, and have difficulty seeing any other use for it.

Is functional fixedness acquired solely in technologically advanced countries where

many objects are designed for a single purpose?
- German and Barrett (2005) examined functional fixedness in the Shuar of the Amazon region
- These people have been “exposed only to a small set of manufactured artifacts, and the set of
artifacts to which they are exposed tends to be ‘low tech’”
- The participants were adolescents and young adults, ranging in age from 12 to 25 years old.
- The task is similar as the one above
- The Shuar participants showed the same effect of pre-utilization as the older children in the
earlier experiment.
- it took them longer to solve the problem when the box’s function as a container had been shown
to them than when it had not.
- The researchers’ conclusion was that even in a technologically sparse culture, people will develop
the idea that an object’s function is the one for which it was designed.

The flexibility-rigidity dimension

- Luchins and Luchins water jar problems
- Imagine that you have three jars, labelled A, B, and C, and a supply of water
 - Your task is to use the jars to obtain a specific amount of water → For example,
suppose the three jars had capacities of 21, 127, and 3 liters, respectively.
- The solution to this problem involves four steps.
1. First, fill up the 127-litre jar (B).
2. Second, pour out 21 litres into jar A
3. Third, pour 3 litres into jar C
4. Fourth, dump out jar C and pour 3 litres into jar C again.
- You are left with 100 litres in jar B, and the problem is solved

- Einstellung effect (Luchins): the tendency to respond inflexibly to a particular type of problems;
also called a rigid set
- In the Luchinses’ experiments, participants were given a series of water jar problems just
like the two we have considered.
- All of them could be solved using the same formula: B minus A minus 2C.
- After solving five problems using that formula, participants had developed a rigid set, or
Einstellung effect.
- a set facilitates some responses while inhibiting others.
- A, B, and C have capacities of 23, 49, and 3 litres, respectively, and the required amount
is 20 litres.

- Woltz, Gardner, and Bell (2000) tested the generality of the Luchinses’ findings using a version
of the number reduction task
- Recall that number reduction problems have two rules, a same rule and a different rule.
- the three numbers were always 1, 2, and 3, and the given string was four numbers long
(e.g., 3213).
- Participants were trained on number sequences that required them to follow the same
sequence of rules over and over.
- For example, they might be asked to solve problems in which the rules had to be applied
in the sequence same–different–same or different–same–different.
- They were then tested on problems that required the use of new rule sequences, such as
same–different–different or different–same–same.
- Those participants who had the most practice during training showed the greatest
negative transfer when tested with problems requiring a new rule sequence.
- Negative transfer: the tendency to respond w/ previously learned rule sequence
even when they are inappropriate
- they kept responding with previously learned rule sequences and their
performance levels dropped as a result.

- Woltz relate their study to other research into common errors made in everyday life.
- The Einstellung effect resembles the kinds of errors we all make when we have
overlearned a particular routine and continue to follow it when we should do something
new.
- Strong but wrong: overlearned and response sequences that we follow even when we
intend to do something else
Flexibility-rigidity and the brain
- Recall that the prefrontal areas of the brain are thought to provide “a top-down bias that favours
the selection of task-relevant information.
- Recall that the left dorsolateral prefrontal cortex (DLPFC) has been singled out as playing a
particularly important role in selecting between alternative response tendencies.
- Recall on memory systems, we reviewed evidence suggesting that the left DLPFC should be seen
as an integral part of working memory, monitoring and controlling alternative courses of action.

- Colvin, Dunbar, and Grafman (2000) have extended this picture of the role of the left DLPFC in a
study of water jar problems done with patients with prefrontal lesions.
- Solving water jar problems requires a counterintuitive move—one that appears to take the
solver farther away from the goal rather than closer to it.
- Both frontal lobe lesion patients and normal controls were given water jar problems.
- The patients solved fewer problems than the controls, and also made fewer
counterintuitive moves.

Mindlessness
- Mindfulness vs mindlessness (langer): openness to alt possibilities vs the tendency to behave as
if the situation had only one possible interpretation
- People who are experiencing Einstellung effects behave mindlessly in assuming that
there is only one way to interpret a situation.
- to behave mindfully means to actively seek new possibilities.
- Langer and Piper reasoned that one way of preventing the development of mindlessness is to
encourage people to think about things in a tentative rather than absolute way.
- For example, describing objects in terms that allowed participants to see that they could have
alternative uses might encourage mindfulness, whereas describing them in terms of single uses
might not.

- Langer and Piper did an experiment in which participants were shown three objects: a dog’s
rubber chew toy, a polygraph pen, and a hairdryer attachment.
- For half the participants, the objects were described unconditionally as one thing only, as
in, “This is a dog’s chew toy.”
- For the other half of the participants, the objects were described conditionally, as in,
“This could be a dog’s chew toy.”
- The experimenter then pretended to need an eraser and asked participants what to do.
- A mindful response would have been to suggest that the chew toy could be used as an
eraser.
- If the chew toy had been described conditionally, then participants were much more
likely to make the mindful response than when it had been described unconditionally.

Artificial intelligence approaches to problem solving

- Artificial intelligence: the “intelligence” of computer programs designed to solve problems in
ways that resemble human approaches to problem-solving
 - Polya wrote a famous guide to problem-solving called How to Solve It (1945/1957) based on
heuristics: problem-solving procedures (typically rules of thumb or shortcuts) that can often be
useful but do not always work.
- Heuristic: a problem solving procedure typically a rule of thumb or shortcut → heuristic can
often be useful, but do not guarantee solutions
- Polya outlines his heuristic method as followed:
- First, to understand the problem you need to formulate it in a way that will allow
you to begin thinking about it → e.g you might draw a diagram
- Once you think you understand what the problem requires, then you can move on
to the second stage and devise a plan → For example, you might try to find a similar
problem that you know how to solve, and then see if the same method will work for
the present problem.
- The third stage is to carry out the plan, with careful attention to detail.
- Polya’s description of problem-solving has much in common with later artificial intelligence
techniques.
- Artificial intelligence requires as clear and precise a formulation of the problem as possible.
- Algorithm: an unambiguous solution procedure (e.g the rules governing long division)
- The rules governing long division are a common example: they are unambiguous, and a
computer can easily be programmed to follow them.
- Algorithms may be divided into two classes: systematic and non-systematic.
- “A systematic algorithm is guaranteed to find the solution if one exists . . . [but] non-
systematic algorithms . . . are not guaranteed to find a solution”

A simple example of artificial intelligence

- Computer programs that play games invented by humans are useful examples of artificially
intelligent systems.
- A tipping point in the history of artificial intelligence occurred in 1997 when the chess-
playing program called Deep Blue defeated world chess champion Garry Kasparov in a
six-game match
- Go-Moku is played on a lattice:

- One player tries to place five Xs in a line, while the other tries to place five Os in a line.
- the person playing Xs is in an unstoppable position → No matter where the person
playing Os moves, X will win on the following move → this situation is called an
open four
 - Meanwhile, the person playing Os has created a situation called an open three.
- If you can create two open threes, then you will be in a better position, because if your
opponent blocks one of them you can turn the other into an open four on your next move.
- There are various subgoals
- Subgoal: a goal drive from the original goal, the solution of which leads to the
solution of the problem as a whole
- If you were writing a computer program to play Go-Moku against an opponent, what
sorts of characteristics would your program need to have?
- First it would need a data structure and an evaluation function
- Evaluation function: the process whereby a plan is created, carried out, and
evaluated
- It consists of a representation of the playing board and the possible states of each
position on the board, whether X, O, or empty.
- The evaluation function handles all the elements that Polya referred to as
“creating a plan, carrying out the plan and evaluating the plan

The problem space

- Problem space: the representation of a problem, including the goal to be reached and the various
ways of transforming the given situation into the solution
- Go-Moku’s problem space is very simple, but more complex games such as chess have extremely
complicated problem spaces, and they must be analyzed at least two moves in advance.
- Search tree: all the possible moves from branching out from the initial state of the problem
- Its like a maze → you must get from the start to the goal

General Problem Solver (GPS) and the tower of Hanoi

- General problem solver: a computer program used to perform non-systematic searches
- To model problem-solving, it’s often useful to analyze the structure of toy problems.
- Toy problem” problems used to analyze the problem solving process
- They are useful because they have a known structure and interesting data can be collected
from participants as they try to solve them.
- the Tower of Hanoi
- In one version of this problem, three concentric rings (small, medium, and large) are
placed around one of three posts and the task is to move all the rings from the post
labeled
- The constraints are that only one ring may be moved at a time and no ring may be placed
on a ring smaller than itself.

- Production rules: a production rule consists of a conditional and an action ( C → A)

- Means-end analysis: the procedure used by general problem solver to reduce differences
between current and goal states
- In order for the problem-solving process to advance, subgoals may have to be substituted
for the original goal.
 - For example, if there is no action that would follow from the current condition and lead
directly to the goal, there may be a subgoal that can be reached directly from the current
state.

- Goal stack: the final goal to be reached is on the bottom of the stack, with the subgoals piled on
top of it in the reverse of the order in which they are to be attained

Thinking aloud as a method for studying human problem-solving

- Thinking aloud: concurrent verbalization → the verbalization of info as the participant is
attending to it
- Ericsson and Simon referred to the thinking-aloud method as concurrent verbalization: the
verbalization of information as the participant is attending to it.

- By contrast, in retrospective verbalization the participant is asked about cognitive processes that
occurred at an earlier point in time.
- Concurrent verbalization relies on short-term memory, whereas retrospective verbalization relies
on long-term memory.

- When participants think aloud, they put into words a process that normally takes place non-
verbally.
- This verbal description of the solution process is called a protocol.

Programming Insight
- Kaplan and Simon (1990) showed that even a very difficult insight problem can be analyzed in
terms compatible with an artificial intelligence approach.
- The problem they focused on is called the mutilated checkerboard problem
- There is a standard 8-by-8 checkerboard, from which two corners have been removed.

- Participants are asked to imagine placing dominoes on the board.

- Each domino covers two vertical or two horizontal squares. Dominoes cannot be placed
diagonally. There are 62 squares. Can 31 dominoes be placed to cover the 62 squares
exactly? If not, why is it not possible for 31 dominoes to cover the 62 squares? Spend a
few minutes thinking about this problem.
- This means that there are 32 black squares and 30 white ones.
- Each domino must cover one black and one white square, and there are 31 dominoes.
After covering 30 black and 30 white squares with 30 dominoes, there will still be two
black squares left over. Remember that a domino must cover one black and one white
square.
 - It is not possible for 31 dominoes to cover the 62 squares.

- historical accounts, observation of ongoing scientific investigations, laboratory studies, and

computational models: diff methods for studying problem-solving in science

Historical accounts
- Cognitive history of science: the study of historically important scientific discoveries in a
framework provided by cognitive science
- A landmark in cognitive historical studies was Gruber’s reconstruction of the process by which
Darwin arrived at his theory of evolution through natural selection.
- Scientists are often meticulous record-keepers, and this was especially true of Darwin. His
notebooks trace the development of his ideas over more than 50 years. Gruber observed that
Darwin’s enterprise illustrated the strength of the Zeigarnik effect: the “quasi-need” to finish
incomplete tasks

Observation of ongoing scientific investigations/laboratory studies

- In vivo/in vitro method: in the case of scientific problem solving, in vivo research involves the
observation of ongoing scientific investigations, while in vitro research involves lab studies of
scientific problem solving
- In vivo means “in the living” and in vitro means “in glass.”
- In the case of scientific problem-solving, in vivo research involves the observation of
ongoing scientific investigations, while in vitro research involves laboratory studies of
scientific problem-solving.
- To set up an in vivo study, the investigator needs to find research settings in which successful
scientific problem-solving is likely to occur.

Unexpected findings
- Unexpected findings: although scientists may initially resist info that disconfirms favoured
hypotheses, successful
- Although scientists may initially resist information that disconfirms favoured hypotheses,
successful problem-solvers don’t persist in trying to confirm their original hypotheses: instead,
they set themselves a new goal of explaining the unexpected findings.

Distributed reasoning
- Distributed reasoning: reasoning done by more than one person

Computational models
- BACON: A computer program that has been able to “discover” several well known scientific
laws
- For example, it searches for patterns in the relationships between two variables, such as
whether they increase together, or one increases while the other decreases. BACON has
“discovered” several well-known scientific laws.
- Among them was Kepler’s third law of planetary motion, which expresses the relationship
between the length of time it takes a planet to orbit the sun and the distance of that planet from
the sun.

- Face validity: methods that clearly measure what they are supposed to measure are said to be
“face valid”