Additively Combining Utilities and Beliefs: Research Gaps and Algorithmic Developments.

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Behavioral Economics

Background:

• Complex environments present challenges for optimal decision-making due to uncertain reward probabilities.
• Normative frameworks may not fully capture effective learning and behavioral flexibility under neural constraints.

Purpose of the Study:

• To review a suboptimal strategy in value-based decision-making: additively combining reward magnitude and probability.
• To present the computational details of the MIX model, an algorithmic implementation of this additive strategy.
• To discuss the opportunities and challenges associated with the MIX model.

Main Methods:

• Review of existing literature on value-based decision-making strategies.
• Computational modeling of a specific additive strategy (MIX model) for sequential decision-making with two options.

Main Results:

• The MIX model algorithmically implements an additive strategy for combining reward magnitude and probability.
• The additive strategy may offer advantages in learning and behavioral flexibility despite being suboptimal by normative standards.

Conclusions:

• The MIX model provides a framework for understanding a specific suboptimal decision-making strategy.
• Further research is needed to explore the model's potential and refine it as a general model of value-based choice.