In this talk, we present a vector model of how consciousness, as depicted by Integrated Information Theory (IIT), can exert genuine causal powers on its physical substrate. IIT identifies consciousness with an irreducible structure of causal powers that a system of mechanisms in its current state specifies. So, IIT asserts that consciousness is “supremely causal” (Tononi 2012, p. 309). Moreover, IIT’s first axiom is that consciousness exists, and one of its key theoretical assumptions is that to exist is to have causal power. Therefore, consciousness must have causal power, on pain of internal incoherence. However, there is a tension between consciousness being causally powerful and the fact that, in IIT’s framework, the behavior of any conscious system seems fully determined by its non-phenomenal properties.To resolve this, we argue that IIT’s interventionist framework should be complemented with a “causal powers” metaphysical view of causation, and propose a mathematically formulated method to translate IIT’s irreducible causal structure (=consciousness) to a vector model where phenomenal powers interact and determine what state of activation the system is more disposed to be in the immediate future. In this way, we could advance our understanding of how consciousness as integrated information makes a causal difference.