Carnegie School of Thought - Bounded Rationality

Discuss Herb Simon some and Organizations

When I began my PhD, I was introduced early on to a particular strain of work on management called the Carnegie School of thought. This work was primarily done in the 50s and 60s by researchers at Carnegie Institute of Technology (now called the Carnegie Mellon University). Herbert Simon and Jim March were the primary individuals involved, March continuing the work with Cyert.

Herb Simon was a very complex, analytical man. Though Simon began his life as a political scientist, publishing his dissertation in a book called Administrative Behavior, he became more interested later in artificial intelligence. My first introduction to him was in a cognitive psychology course. The instructor was describing how Simon began the first lecture of the Fall semester of one of his courses by asking his students what they accomplished. After all of the students had described their summers, Simon said that, over the summer he had designed a computer that could think like a person. It was an early version of artificial intelligence that based its decision making on the same ways in which people make decisions. Simon was an extremely influential person in computer science, artificial intelligence, cognitive psychology, and management. His influence in management, for the most part, is due to his encouragement and collaboration with James March.

The ideas within the Carnegie School are quite diverse and for this post I will focus on a concept called bounded rationality (also known as satisficing). Within Economics, it is assumed that actors make the best choice in any given decision. Satisficing, proposes that there are increased costs for some decisions or that the outcome is not as important to the actor, leading to the actor to willingly make a suboptimal choice. An example that Simon used to give was about lunch [I have modified the story from the original but the idea is the same]. If an actor is in their office and needs to get lunch, they could have multiple values that they desire to maximize: timeliness, cost, health, etc. An actor could determine the relative weight of those characteristics and make the optimal choice. But as Herb said "I would instead just always go to [the student center]. For those who have been there, it is obviously a non-optimal choice." The humorous example does have certain limitations but provides a good overall example of the concept. Satisficing proposes that the act of making a choice is costly and ones own desires are not always clear, leading to a "good enough" choice being much easier to determine than the best.

Though this concept, while somewhat of a refinement of the economic theory of optimization, was a revelation to the academic world. It is not without its detractors. A comment that I have heard from several proponents of bounded rationality is that it is not testable meaning it is not a proper theory. The reason for this is that people may actually be making optimal choices but they are optimizing on unknown or unmeasured criteria. I personally think that satisficing is a very useful concept though it does have a subcurrent of nondeterminism that also arises in March's concept of the Garbage Can Model of Organizational Choice. This concept is a bit unsettling, but still interesting to me.