Neural correlates, computational correlates, and the prospects of computational explanations of consciousness by Wanja Wiese Abstract:The neural correlates of consciousness – as standardly conceived – face some fundamental challenges. In particular, neural correlates of consciousness do not enable inferences about the presence of consciousness in controversial cases, such as invertebrates, isolated neural structures (potential “islands of awareness”), or artificial systems. Computational correlates of consciousness – specified by computational models – may provide a solution to this problem, because computational correlates can be present in systems lacking the (neural) structures required to instantiate a neural correlate of consciousness. Still, it is doubtful whether computational correlates (that may be sufficient for consciousness in human beings) license the ascription of consciousness in controversial cases. In other words, one may wonder whether computational correlates should be regarded as sufficient or necessary conditions for consciousness (or neither). I shall argue that research on computational correlates should aim for necessary conditions for consciousness. This has at least three advantages over sufficient conditions: (i) If a necessary condition for consciousness is not fulfilled in a controversial case, this provides evidence for the absence of consciousness in that case. (ii) Regarding computational correlates as merely necessary conditions allows one to maintain a distinction between simulating a conscious system and instantiating consciousness. (iii) Necessary conditions for consciousness may furnish unification in the science of consciousness.