The formation, adaption and learning mechanism of risk preference within the neural noisy coding framework of decision making

Recently, Khaw et al. (2021) proposed a seminal theoretical framework building on the dominant paradigm in neuroscience, which provides a micro-foundation for small stake risk aversion, as well as accounts for stochastic choice patterns. While this modeling approach has the potential to unify individual decision making and perhaps behavioral economics at large, I aim to theoretically expand and empirically test the noisy coding framework of Khaw et al. (2021) across three work packages. In the first work package, I aim to augment the models of Khaw et al. (2021) and Vieider (2021a) by introducing a neural notion of loss attitude, which attributes loss attitude to asymmetric information processing mechanism. In the second work package, I propose to empirically explore how the priors are updated following large income and consumption shocks. This promises to yield new insights on how the risk preference is formed and adapts to environmental changes—an issue on which little if anything is known to date. In the third work package, I aim to augment the static noisy coding model with a dynamic process by which the model parameters themselves are learned from a sequence of stimuli. While some recent papers have discussed how encoding noise could systematically adapt to the prior in an efficient coding setup, the prior itself has so far not been discussed in the literature.