Curiosity is a notion that is increasingly receiving special attention, particularly in the context of the emerging fields of developmental robotics. In the first part of the paper we give a brief critical overview of the research in motivational systems in intelligent robotics. The overall impression is that the prevailing understanding of curiosity is rather one dimensional and reductionist in spirit. We argue that this is a result of rather simple agent’s representations of the environment that are usually adopted. In the second part of the paper we put forward some arguments towards modeling of curiosity in context of other cognitive phenomena like feeling of understanding, analogy-making and expectations, integrated in a more general cognitive architecture.