© 2006 Kotovsky and Cagan Sources of Insights in Engineering Design Kenneth Kotovsky, Dept. of...
-
Upload
victoria-glenn -
Category
Documents
-
view
218 -
download
0
Transcript of © 2006 Kotovsky and Cagan Sources of Insights in Engineering Design Kenneth Kotovsky, Dept. of...
© 2006 Kotovsky and Cagan
Sources of Insights in Engineering Design
Kenneth Kotovsky, Dept. of Psychology
and
Jonathan Cagan, Dept. of Mechanical Engineering Carnegie Mellon University
© 2006 Kotovsky and Cagan
Design Insights
• How do creative insights occur during the design process?
• Can their frequency and quality be enhanced?
• What are the underlying cognitive mechanisms?
© 2006 Kotovsky and Cagan
Overview
• A conceptual framework or model for design problem-solving
• Brief introduction of that framework
• Examples of some of the work that led to it
• Presentation of some current work within it
© 2006 Kotovsky and Cagan
A Multi-plane Model of Design
• Search is viewed as occurring within a representation depicted as a plane
• Multiple planes equiv. to different representations • Search for new representations/planes activated
by impasses often leading to insights• Search can involve both conscious and non-
conscious processes often operating in parallel or resulting in parallel search within multiple planes
© 2006 Kotovsky and Cagan
A Multi-plane Model of Design
*
impasse
solution
Mov
emen
t to
new
rep
rese
ntat
ions
© 2006 Kotovsky and Cagan
Prior work has demonstrated two distinct phases of problem solving; exploratory and insight-like final path
5 4 3 2 1
Ball Configuration
31
30
29
28
27
26
25
24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
StateProblem Space
31 Move Start
21 Move Start
Goal
Non-Conscious Insight Attainment
Isomorphs of Chinese Ring Puzzle
© 2006 Kotovsky and Cagan
Empirical Results
Kotovsky, K. & Simon, H.A. (1990) Why are Some Problems Really Hard: Explorations in the Problem Space of Difficulty. Cognitive Psychology, 22, 143-183
© 2006 Kotovsky and Cagan
Final Path: Non-Conscious Insight
• Now I want to ask you about the puzzle you just solved, how it worked, what you did.
• No idea.• No idea?• No idea. It was very painful, (laughs)• You did get it, right?• Yeah, but it was basically luck... that I got it.• You had no idea what you were doing?• Not really.• Suppose somebody else was going to do the puzzle who had never seen it
before and you had to give them some hints, tell them how to solve it.• Well, let's see ... I don't know what to say ... but, I guess, (garbled word) the
puzzle ...The good part was that there usually wasn't any more than like, one or two choices. I think there was one choice. Was there any more than one choice? I don't know. But I had (garbled word). Which is why I kept ending up back where I started from which was frustrating. I would tell them, I would tell them, good luck. That's all.
Reber, P. & Kotovsky, K. ((1997) Implicit Learning in Problem Solving: The Role of Working Memory Capacity. JEP, General. 126, 178-203
© 2006 Kotovsky and Cagan
Computational Model of Two-Stage Problem Solving
• Simulated annealing algorithm
• Works within the problem space with minimal assumptions about that space
• Generates objective function through experience
• Improves over multiple runs
Cagan, J. & Kotovsky, K. (1997) Simulated Annealing and he Generation of the Objective Function: A Model f Learning During Problem Solving. Computational Intelligence 13:4 534-581
© 2006 Kotovsky and Cagan
Computational Model of Two-Stage Problem Solving: Ball & Boxes
B&B - Misleading
move no.
problem state05
1015202530
050
100 150 200 250
B&B - 1st Ball Out
move no.
problem state0
10
20
30
40
050
100 150 200 250
Cagan, J. & Kotovsky, K. (1997) Simulated Annealing and he Generation of the Objective Function: A Model of Learning During Problem Solving. Computational Intelligence 13:4 534-581
© 2006 Kotovsky and Cagan
Summary: Non-Conscious Insight Attainment
• Dichotomous: Exploratory & Final Path
• Looks insightful
• Non-conscious insights– Also supported by work on
• Subliminally flashed hints (with Julia Lin)• Delayed realization of having reached an
insight/solved a problem (with Michelle Fahey)
© 2006 Kotovsky and Cagan
Multi-plane Model of Design
Some candidates for “Plane-jumping” or insightful representation change include:– Reaching an impasse
– Noticing relevant information
– Incubation
– Relaxing the most restrictive constraint.
– Using and tracking the use of analogy to generate other search spaces or planes
© 2006 Kotovsky and Cagan
Incubation Impasses & Open Goals
(with Jarrod Moss)
• Open goals related to prepared mind view of insight
• Affect behavior on a range of tasks, starting with Zeigarnik Effect (1927) & long line of work since (ex. Patalino & Seifert (1994, 1997)
• Attempt to demonstrate open goal effect and explore its functioning in impasses
Moss, J. (2006)The Role of Open Goals in Acquiring Problem Relevant Information. Ph.D. Dissertation, Carnegie Mellon University
© 2006 Kotovsky and Cagan
Study 1a• 39 Participants• 30 seconds to work on Remote Associate
Test (RAT) problems (Mednick, 1962)– Short insight-like problems (fox peep man) hole– Fixation on distractors (Smith & Blankenship, 1991)
• Answers for unsolved problems presented in lexical decision task (hint)
RATProblems
(20)
RATRepeated
(20)
Lexicaldecision task
(50)
© 2006 Kotovsky and Cagan
Lexical Decision
550.00
560.00
570.00
580.00
590.00
600.00
610.00
620.00
630.00
640.00
Solved Unsolved
RAT Result
RTControl
RAT
© 2006 Kotovsky and Cagan
Unsolved Problems Solved on Second Attempt
0.00
0.05
0.10
0.15
0.20
0.25
0.30
Previously Unsolved Problems
Proportion Solved
No Hint
Hint
© 2006 Kotovsky and Cagan
Study 1b
• 30 Participants
• Method similar to Study 1a– Hints after unsolved problems
– Hints before new problems
RAT Problems RAT Problems
Repeated+ New
Lexical decision task
© 2006 Kotovsky and Cagan
Effect of hint for new and unsolved problems
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
New Previously Unsolved
Type of problem
Proportion solved
No Hint
Hint
22% relative improvement
50% relative improvement
© 2006 Kotovsky and Cagan
Effect of hint on solution time (solved problems)
0.00
5.00
10.00
15.00
20.00
25.00
New Previously Unsolved
Type of problem
Solution time in seconds
No Hint
Hint
© 2006 Kotovsky and Cagan
Results
• Hint more effective with open goals– 50% relative increase vs. 22%
– Improvement in solution time
– Majority of participants unaware of any relationship between the two tasks (> 90%)
– Also did study showing sizable no. of open goals could be maintained
© 2006 Kotovsky and Cagan
Study 2 Impasses: Protocol Study • Hint vs. no-hint• Concurrent Verbal protocols
– 30 participants
– Assessment of candidate generation & fixation
– Measured Impasses (10 sec. or throwing up hands)
• Varying amount of time spent on a problem– 15, 30, 60 seconds 1st attempt, 30 seconds 2nd attempt
RATProblems
(30)
RATRepeated
(30)
LexicalDecision task
(50)
© 2006 Kotovsky and Cagan
Fixation on Distractors
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Unsolved Improved
Result of Second Attempt
Proportion of Candidates that are Repeats
No Hint
Hint
© 2006 Kotovsky and Cagan
Impasses
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Impasse No Impasse
Type of Impasse
Proportion Correct
No hint
Hint
© 2006 Kotovsky and Cagan
Impasses: Exploration and Fixation
15 s, Hint
30 s, Hint
60 s, Hint
No Hint
Interactions:
(1) Hint xTime Limit
(2) Hint x Candidates
Time limit for second attempt
always 30 s
© 2006 Kotovsky and Cagan
Model• Probability of retrieval
– Activation of word– Activation of distractors leading to fixation
• Fixation– Retrieving a word increases base level activation– Multiple retrievals result in greater increase– Interference as more words are retrieved
• Implemented in ACT-R• Shows that activation and fixation alone can’t
account for results, need open goal mechanism
© 2006 Kotovsky and Cagan
Results & Conclusions• Hints are effective in presence of open goals that
act by sensitizing us to information from the environment– Less effective with more fixation– Without hint, fixation continues into 2nd solution attempt
• We can assimilate hints without awareness• Redefining impasse as a more continuous process
due to interplay of– Fixation– Exploration
• We’ve seen some of the same impasse related fixation on engineering design problems
• Compatible with other work showing non-conscious problem solving leading to multi-plane search
© 2006 Kotovsky and Cagan
Future Directions• Map findings onto the design of design
activities so as to optimize: – Timing via tracking exploration & fixation
– Exposure to potentially useful information to increase range and frequency of creative insights
• This research thrust is itself a multi-plane approach as we try to both enhance our understanding of cognitive mechanisms and apply them to engineering design activities and computational implementations.
© 2006 Kotovsky and Cagan
Bibliography• Mednick, S. A. (1962). The asssociative basis of the creative
process. Psychological Review, 69, 220-232• Patalano, A. L., & Seifert, C. M. (1994) Memory for impasses
during problem solving. Memory & Cognition 22(2), 234-242• Patalano, A. L., & Seifert, C. M. (1997) Opportunistic planning:
Being reminded of pending goals. Cognitive Psychology 34(1), 1-36.
• Smith S. M., & Blankenship, S. E. (1991) Incubation and the persistence of fixation in problem solving. American Journal of Psychology, 104(1), 61-87.
• Zeigarnik, B. (1938). On finished and unfinished tasks. In W. D. Ellis (Ed.), A source book of gestalt psychology (pp 300-314). New York: Harcourt-Brace. (Reprinted & condensed from Psychologische Forschung, 9, 1-85, by B. Zeigarnik, 1927.
© 2006 Kotovsky and Cagan
Acknowledgements• Jarrod Moss • Julia Lin• Michelle Fahey• Paul Reber• Matt Campbell• Herbert Simon• J.R. Hayes
• AFOSR under grant # FA9620-04-1-0201
The U.S. Government is authorized to reproduce and distribute reprints for governmental purposes notwithstanding any copyright annotation thereon. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of AFOSR or the U.S. Government.