Hurdles for computationalism: The many difficulties facing a computationalist theory of phenomenal consciousness

There are several things that might be called ‘computationalism’. The one most central to the philosophy of mind is a theory that attempts to answer what Chalmers calls the ‘hard problem’ of consciousness: how can it be that brute material processes can give rise to phenomenal experience? The existing literature has addressed metaphysical challenges for computationalism piecemeal. In this thesis, I begin the work of assessing them as an interconnected whole. I analyse six key problems faced by the computationalist, showing how they mutually complicate each other in ways the existing literature does not appreciate. In mapping this problem space, I draw on and expand upon the work of thinkers including John Searle, Hilary Putnam, Tim Maudlin, B. Jack Copeland, and Colin Klein, refining their insights and showing how the problems they highlight interlock. The interactions between these problems, I argue, undermine the solutions these thinkers offer. In the first chapter, I take a look at the mathematical structure of implementation – the relation between abstract computation and concrete physical system in virtue of which the latter ‘does’ the former. In the literature, the usual approach is to treat isomorphism between the abstract and the physical systems as the only natural option; in a step forward for the field, Matthias Scheutz introduces the notion of bisimilarity as an alternate relation. But as I show, there is actually a wide space of possible implementation relations, including at least a plethora of relations intermediate between isomorphism and bisimilarity, with no obvious best choice among them. The possible answers to this problem go on to shape, and be shaped by, the problems I investigate in the subsequent chapters. The problem laid out in the second chapter is ontological. In order to avoid the problem of trivial implementations – the conclusion that all kinds of everyday physical systems are carrying out all possible computations up to some high level of complexity (as Lycan and Searle argued) and thus that such systems are undergoing phenomenal experiences – the computationalist needs there to be a natural ontology that somehow privileges ‘real’ parts or states of systems over arbitrary disjunctions of their micro-level parts or states. This, as I argue, is only the first of a number of onerous ontological commitments the computationalist will have to make. The next, discussed in the third chapter, is if anything more onerous still. In applying the neat categories of abstract computation to messy concrete reality, I argue, computationalists must be extremely intolerant of vagueness – only allowing edge cases to come up under certain very restrictive conditions. Arguing by cases, I demonstrate this is true regardless of how the computationalist conceptualises phenomenal similarity and distinctness. The fourth and fifth chapters map the various obstacles facing computationalists in their attempt to make use of counterfactuals – which they must if they are to escape the problem of trivial implementations. In addition to iterations of the previous chapters’ problems, I argue, a forest of novel problems arise. In the fourth chapter, I discuss those that come with the attempt to use counterfactuals per se; in the fifth, those the computationlist faces if they instead attempt to use counterfactuals as indicators of underlying categorical conditions. Finally, in the sixth chapter, I lay out the conditions the computationalist needs their theory of causation to fulfil, in light of all the obstacles so far encountered. These conditions not only place forbidding obstacles before the two most common kinds of theory in the literature – counterfactual and transmission theories – but threaten to contradict one another. Computationalists, I conclude, face a stiff uphill struggle if they are to put forward even a minimally plausible theory