Four-dimensional computed tomography (4DCT) offers an extra dimension of 'time' on the three-dimensional patient model with which we can incorporate target motion in radiation treatment (RT) planning and delivery in various ways such as in the concept of internal target volume, in gated treatment or in target tracking. However, for all these methodologies, different phases are essentially considered as non-interconnected independent phases for the purpose of optimization, in other words, the 'time' dimension has yet to be incorporated explicitly in the optimization algorithm and fully exploited. In this note, we have formulated a new 4D inverse planning technique that treats all the phases in the 4DCT as one single entity in the optimization. The optimization is formulated as a quadratic problem for disciplined convex programming that enables the problem to be analyzed and solved efficiently. In the proof-of-principle examples illustrated, we show that the temporal information of the spatial relation of the target and organs at risk could be 'exchanged' amongst different phases so that an appropriate weighting of dose deposition could be allocated to each phase, thus enabling a treatment with a tight target margin and a full duty cycle otherwise not achievable by either of the aforementioned methodologies. Yet there are practical issues to be solved in the 4D RT planning and delivery. The 4D concept in the optimization we have formulated here does provide insight on how the 'time' dimension can be exploited in the 4D optimization process.