Papers review: Kepecs’ framework for decision confidence Papers Adam Kepecs et al., “Neural Correlates, Computation and Behavioural Impact of Decision Confidence,” Nature 455, no. 7210 (September 2008): 227–31, https://doi.org/10.1038/nature07200. Adam Kepecs and Zachary F. Mainen, “A Computational Framework for the Study of Confidence in Humans and Animals,” Philosophical Transactions of the Royal Society B: Biological Sciences 367, no. 1594 (May 19, 2012): 1322–37, https://doi.org/10.1098/rstb.2012.0037. Scope Humans and other animals must often make decisions on the basis of imperfect evidence.

Confidence as hyperparameter tuning for sequential decision-making Metacognition Sequential decision-making Speed/Accuracy Tradeoff. Canonical model: integration of noisy information until a threshold is reached. Many refinements: multi-alternative choice, impact of learning, change of mind, etc. Confidence for decision-making Quantifies the degree of certainty associated to a decision. Canonical model: post-decisional computation based on accumulated info. Can be used to alter the subsequent decisions. Proposed architecture: confidence as hyperparameter tuning for sequential decision-making Current & next steps Article for ESANN 2025.

Confidence in decision-making Terminology Confidence: general definition Everyone knows intuitively what confidence is about, yet it is seldom defined explicitely. In the broadest sense, confidence quantifies a degree of belief in something or someone [Meyniel et al., 2015]. It is fundamentally linked to its object: a thought, a choice, an external actor, etc. Belief A belief is a feeling of certainty about a proposition (i.e. a statement or a decision).