As computing becomes ubiquitous and intelligent, it is possible for systems to sense their use of context and adapt their behavior accordingly. Utilizing an appropriate context model that shows representative contexts and behavior in computing can provide convenience and efficiency to users for their ubiquitous access. Human activities and preferences tend to be diverse and dynamically changing, depending upon material and social circumstances. Current context models usually capture the factors that significantly affect human decisions or behavior, but hardly offer cognitive properties that are essential to human activities and decisions.In order to support human-centric adaptation without frustrating and disorienting users, we propose a cognitive context modeling framework for capturing and analyzing end-user's cognition of context information. In this framework, diverse human preferences are represented with context views and dynamic activities are elaborated with relevant contexts. A case study of computer power saving schedule is conducted to demonstrate the capability of this framework for capturing and representing cognitive context information.