How many parameters to model states of mind ?

TL;DR

This paper discusses the trade-offs between complex and simple computational models of mental states, emphasizing that simpler models with fewer parameters are more practical for understanding internal states.

Contribution

It argues that models with fewer parameters are more suitable for interpreting internal states due to computational constraints, contrasting detailed complex models with simpler approaches.

Findings

Complex models with many parameters are computationally less feasible.

Simpler models provide qualitative insights into internal states.

Rich models' assumptions often remain unchecked.

Abstract

A series of examples of computational models is provided, where the model aim is to interpret numerical results in terms of internal states of agents minds. Two opposite strategies or research can be distinguished in the literature. First is to reproduce the richness and complexity of real world as faithfully as possible, second is to apply simple assumptions and check the results in depth. As a rule, the results of the latter method agree only qualitatively with some stylized facts. The price we pay for more detailed predictions within the former method is that consequences of the rich set of underlying assumptions remain unchecked. Here we argue that for computational reasons, complex models with many parameters are less suitable.