Reflections on “The Scientist as Problem Solver”

Abstract

H. A. Simon has described how the scientific research, he has been conducting is very similar to problem solving in other form for example a game of chess. The process of scientific discovery according to him is a process of recognition. He describes the whole scientific discovery process in the following parts:
• Formulating Problems
• Laws from Data
• Representations
• Finding an Explanatory Model
• Designing Good Experiments
• Problem Isomorphs
• Experiments without Independent Variables
• The Scientist as Satisficer

Formulating Problems

The first and foremost part of any discovery process is identification and formulation of the problem. However, it’s the case that a person actually finds the solution of another problem while trying to solve one problem. Pasteur’s famous dictum as “Accidents happens to the prepared mind” has been cited to justify this. In explaining the failure of neo-classical economics theory in a described situation Simon could figure out how bounded rationality of human nature can overlook the rational global solution to the problem.

Laws from Data

Many a times, scientific theories are found embedded in data rather than an established theory proving the same. For example, when Kepler provided information on earth motion it was purely data driven. Newton postulated laws gravitation and proved that his laws can explain Kepler’s observed data. Similarly, Lotka’s assumptions could be solved lot later by logical arguments provided by the author. More over similar research has been found to be conducted by other researchers in parallel fields without being aware of each other’s work.

Representations

Representations are very important for any scientific experiment. Although, words can describe the problem a clearer understanding comes from imagery describing the problem at hand. Einstein and Hadamard agreed in their exchanges that words do not stimulate the process of thought significantly. Simon has shown that although mathematical models help but a physical mental picture like the room and cell example helps getting a clearer thought process.

Finding an Explanatory Model

Finding an explanatory model for a phenomenon is quite important and useful in coming up with a solution. One way is to systematic thinking in the direction as has been done Lotka’s problem case. The other approach is identifying a similar problem which has been solved. The example the author uses is of defining theory of problem solving by using digital computers as a model.

Designing Good Experiments

Designing a good experiment is very critical in a scientific problem solving exercise. Many a times experiments bring out facts that were never captured in the model thus bringing in surprises. The author tries to find out an experiment on Chinese language on STM which is a natural experiment rather than creating an artificial system.

Problem Isomorphs

Problem isomorphs are identical in task domains and legal domains yet have been described by different words. He describes two scenarios of tower of Hanoi and missionary and cannibal problem both being isomorphs may take a human being different times to solve. The idea that the solution of a problem depends
only on the size of task domain is not a valid assumption to make.

Experiments without Independent Variables

Many a times in experiments it’s hard to define exact independent variables. However, redesigning the experiment in such a way that the results are unintuitive can be a good means of coming up with results. The impossible problem formulation has provided the author existence two distinct human cognitive system in addressing the problems. Simon wants to portray that observation is far superior in solving the problem than hypothesizing and proving based on the hypothesis.

The Scientist as Satisficer

Scientist is ultimately a satisficer as well. When she is in search of something she postulates certain possibilities that may be occurring in the nature. Ultimately she provides a model of sorts and tries to validate that with the natural happenings. When the model matches with the natural occurrence within limits she concludes the completion of the experiment. However, it’s not an endless quest for finding the right answer but a relationship of matching a model definition with the expected natural occurrence.

References

[1] H. A. Simon. Models of My Life, chapter The Scientist as Problem Solver, pages 368–387. The MIT Press, October 1996.