Reward Function Complexity and Goals in Exploration-Exploitation Tasks
- Brian Montambault, Tufts University, Medford, Massachusetts, United States
- Chris Lucas, University of Edinburgh, Edinburgh, United Kingdom
AbstractPeople are often faced with choices where there is a conflict between seeking reward and gathering information. In many of these cases there exists a functional relationship between the features associated with actions and their corresponding rewards. Accounts of how people make decisions in these circumstances have not considered how peoples' strategies depend on the complexity of this function, as well as the person's goal. In a sequential decision making task we found that people chose between a number of different exploration strategies, but that strategy selection did not necessarily align with goal or account for function complexity.
Return to previous page