Computational modeling builds mathematical models of cognitive phenomena to simulate patterns of perception, decision-making, and belief updating. These models mathematically represent the information processing by combining an anterior probability distribution, a likelihood function and a set of parameters and hyperparameters. Their use popularized the conception of a nervous system functioning as a predictive machine, or "bayesian brain". Applied to psychiatry, these models seek to explain how psychiatric dysfunction may emerge mechanistically. Despite the significance of emotions for cognitive phenomena and for psychiatric disorders, few computational models offer mathematical representations of emotion or incorporate emotional factors into their modeling parameters. We present here some computational hypotheses for the modeling of affective parameters, and we suggest that computational psychiatry would benefit from these modeling parameters.