Final Report Summary - UNIIND (A Unified Model of Induction)
The project offers a theoretical framework within which one can model case-based reasoning (analogies), rule-based reasoning (theories), and Bayesian reasoning (based on a prior probability that is updated by Bayes rule). Pre-existing literature has vast bodies of research on the latter two, but very little on the former type of reasoning in terms of formal, general modeling. Further, while the connections between Bayesian reasoning and rule-based reasoning have been discussed, there is, to the best of our knowledge no similar unifying framework for all types of reasoning.
As it is obvious that people use all types of reasoning in everyday life, when making predictions about economic, political, and financial events, to name a few domains of application, there is a need to have models that can incorporate all these modes of reasoning in order to provide a more complete picture of human reasoning. Such models are necessary if people switch between modes of reasoning, or if some use one mode and others – another. For example, it is possible that some people think of the stock market in terms of similarity to past cases, while others – in terms of general rules. Moreover, following critical events such as the 2007/8 crisis, people may switch between modes, say, believe in general theories less and try to reason by simple analogies more. These types of dynamics call for a unified model in order to understand economic and financial phenomena. Similarly, predictions regarding political developments also involve a mix of reasoning modes. People try to develop general theories, say, about the behavior of countries and regimes, the life cycle of empires, and the like, but they are also aware that cases can be unique and to judge them by similarity to other special cases more than by adherence to general theory. Again, the need for a general model derives from the fact that each person may have a mix of these modes of reasoning and switch among them depending on current events.
As it is obvious that people use all types of reasoning in everyday life, when making predictions about economic, political, and financial events, to name a few domains of application, there is a need to have models that can incorporate all these modes of reasoning in order to provide a more complete picture of human reasoning. Such models are necessary if people switch between modes of reasoning, or if some use one mode and others – another. For example, it is possible that some people think of the stock market in terms of similarity to past cases, while others – in terms of general rules. Moreover, following critical events such as the 2007/8 crisis, people may switch between modes, say, believe in general theories less and try to reason by simple analogies more. These types of dynamics call for a unified model in order to understand economic and financial phenomena. Similarly, predictions regarding political developments also involve a mix of reasoning modes. People try to develop general theories, say, about the behavior of countries and regimes, the life cycle of empires, and the like, but they are also aware that cases can be unique and to judge them by similarity to other special cases more than by adherence to general theory. Again, the need for a general model derives from the fact that each person may have a mix of these modes of reasoning and switch among them depending on current events.