Risk should be objectively defined: reply to Zentall and Smith.

Zentall and Smith (2014) have published a comment on Pelé and Sueur (Anim Cogn 16:543-556, 2013) in which they raise two issues, one about the definition of risk and a second concerning the optimality of decisions. When making a choice, subjects are faced not only with several possible alternatives but also with the risk of opting for an unsuitable choice which depends on several variables (context, internal state, knowledge and perception). Although it is true that animals might learn about their environment and adapt their decisions to the context and to their experience, strong constraints make some behavioural traits stable over individual lifetime and even generations. We therefore consider that experience has limited impact on the variability of temporal discounting. These behavioural traits make the difference between perceived and actual risk. If the perceived risk strongly differs from the actual risk, a decision should be considered as suboptimal. If we want to lead individual and collective cognition to a common decision science, it is crucial to use the same definitions for terms implied in decision-making.