The reactive programming paradigm enables programmers to express interactive programs as a transformation from input streams to output streams. Multiple strands of reactive programming have been proposed in the past. We argue that none of the existing classes of reactive programming sufficiently caters to interactive programs in which patterns of discrete events determine how the program should react. We propose a novel class of reactive programming, called logical reactive programming, in which patterns of events can be declaratively specified using logical rules. We analyze the differences between logical reactive programming languages and traditional reactive programming languages, and initiate discussion on its limitations.