Noisy coding of time and reward discounting

I present a model generating delay-discounting from noisy mental representations of time delays. The optimal combination of noisy signals about time delays with prior information results in a stochastic model predicting discounting to exhibit present-bias, but to be stationary and delaydependent once up-front delays are introduced. The derivation from an optimal encoding-decoding process sidesteps arbitrariness concerns voiced about earlier models. Data collected in an experiment support the need for separate but interacting parameters to capture present-bias and delaydependence. The account explains why non-trivial discounting is routinely observed in experiments using monetary rewards instead of consumption.