Week 2/3- Reasoning the fast and frugal way

This is my first entry and this is the first time that I have created a blog, so I am quite uncertain on how to compose a blog. Nevertheless, I will give it a go. I am going to write about week 2, after reading Reasoning the Fast and Frugal Way: Models of Bounded Rationality by Gigerenzer and Goldstein (1996), of which I found quite interesting.

Simon's (1956,1982) models of bounded rationality attempts to understand how humans who have little time and knowledge behave. He argued that information-processing systems typically need to satisfice rather than optimise. Unlike the classical view, whereby the laws of human inference are the laws of probability and statistics, Simon argued that we would choose the first object that satisfies our aspiration levels, instead of calculating all possible alternatives, estimating probabilities and utilities for possible outcomes. In accordance with Herbert Simon (1956,1982) I believe that us humans are limited information processors and that our minds are shaped by the environment.

To demonstrate Simon's notion of satisfing, various simple algorithms were tested and by computer simulation Gigerenzer and Goldstein (1996) held a competition between the satisfing "Take the Best" algorithm and several "rational" inference procedures. In this article, satisfing algorithms use limited knowledge and performed, to my surprise, extremely well. One example of an inferential task, where participants must choose between two alternative choices by only using knowledge retrieved from memory only is.." Which city has a larger population? (a) Hamberg (b) Cologne". This is a great example to test participants under limited time and knowledge (obviously the participants were not German). Surprisingly 90% of US students gave the correct answer (Hamberg) and vise versa with German students when asked to choose between two American cities.

Many studies seems to support this idea that under limited time and knowledge we make correct inferences, maybe this could have arisen through evolution. For example, in an emergency (a lion approaches you) we go in a flight or fight mode and we have to make a decision under limited time and knowledge to survive, we would not have the time and knowledge to search for all relevant information, as the classical view suggests, otherwise by the time we would have gathered all information we would have been eaten (This has not been written anywhere, nor based on evidence I am just hypothesising).

In this example of which two cities is more populated, Gigerenzer explained that we try to find a cue, for example Hamberg has a football stadium or airport, and therefore participants choose one city over the other. The Take The Best algorithm is the first satisfing algorithm, it is called take the best as its policy is "take the best, ignore the best". Initially I thought that this algorithm was quite simplistic, until I learnt how accurate it made inferences about a real-world environment. The Take the Best algorithm parallels Simon's notion of satisfing, as the algorithm stops search after the first discriminating cue is found, just like Simon's satisfing algorithm stops search when the option meets an aspiration level.

To conclude, before I write another 1000 useless words, I agree with Simon's concept of satisfing and I am impressed that the Take the Best algorithm performed so well with limited information and did not need to compute weighted sums of cue values.