Traditionally robots are autonomous systems that plan and execute actions without taking into account any external feedback. This paper studies to what extent a robot is able to alter its behavior based on observing a series of positive or negative human reactions. We describe the result of introducing these human reactions in a plan of actions by means of fuzzy logic analysis of the reactions. In this way, the election of the actions to perform is based on a combination of the initial plan of actions and of a fuzzy logic weighting of the utility of previously executed actions.