To a certain extent I agree with the definition of interactivity: a cyclic process in which two actors alternatively listen, think and speak. There should definitely be threads of actions that take input (listen), process it (think) and speak (output), however I believe that it does not have to limit to two ‘actors’, nor does it have to be in strictly alternative order. For instance, when someone interacts with a website, it may seem like there are two actors, but if you divide the website entity to two parts, the frontend and the backend, the interaction would be between three entities. In this sense, interactivity could contain multiple entities. Furthermore, I believe that there needs to be some sort of correlation between the threads. In other words, the output of one actor must be directly or indirectly be taken as an input for the other actor. Otherwise, there would be no relation between the two actors.
Another aspect that piqued my interest was the sentence “The refrigerator does interact with a user but it does so at a low level” . When it comes to measuring the level of interactivity, besides the three requirements, the time it takes for the interactivity to occur could be an important parameter to measure. There could be many micro-interactions simultaneously happening in a short time, which could be argued as a high level interaction. While on the other hand, there could be complex interactivity but the time it takes spans a long period, therefore lowering its level of interactivity. The word interactivity spans large area in our society, and depending on how it is used in various contexts, the definition changes.