For “gradual behaviours”, whether a specific instance of that behaviour is blame- or praiseworthy depends on how much of the behaviour is done. For instance, for a behaviour like “replying quickly to emails”, whether a specific reply is blame or praiseworthy will depend on the timeliness of that reply. Such behaviours lie on a continuum in which part of the continuum is praiseworthy (replying quickly) and another part of the continuum is blameworthy (replying late). As praise shifts towards blame along such behavioural continua, the resulting blame-praise curve must have a specific shape. What determines the shape of that curve? Moral psychologists have developed models to explain when people “blame” others (e.g. Malle et al., 2014), but these do not address how blame and praise function for gradual behaviours. In the current project we aim to develop a novel framework to solve this problem by using non-linear regression to estimate “blame-praise curves” that directly link the amount of praise and blame to an underlying behavioural continuum for a wide variety of behaviours. Doing so, allows us to estimate the parameters that define the shape of these curves and investigate what has an impact on these parameters. Within this project we investigate how blame-praise curves vary as a function of a) RL1: the characteristics of the behaviour in question, b) RL2: the characteristics of the individual doing the judging and c) RL3: the characteristics of the individual being judged.