What is it about?

Yes, RL (Reinforcement Learning) is a machine learning technique BUT it core parameters are closely related to assumptions about human cognition. Those cognitive and behavioral researchers who do RL modeling may be missing out on some obvious sources of power. When to reward? -- at the end of a trial, end of a subtask? What to reward? -- what is the "objective function" you are trying to maximize? fast response times? accuracy? How much to reward? -- binary, categorical, or continuous values

Featured Image

Why is it important?

Different choices produce profoundly different learning paths and outcomes. Hence, these choices have theoretical as well as practical importance.

Perspectives

Unfortunately, there is little discussion in the literature of the effect of such choices. This absence is disappointing, as the choice of when, what, and how much needs to be made by the cognitive modeler for every learning model.

Professor Wayne D. Gray
Rensselaer Polytechnic Institute

Read the Original

This page is a summary of: When, What, and How Much to Reward in Reinforcement Learning-Based Models of Cognition, Cognitive Science, January 2012, Wiley,
DOI: 10.1111/j.1551-6709.2011.01222.x.
You can read the full text:

Read

Contributors

The following have contributed to this page