Ned Block (2009), “Comparing the major theories of consciousness”, in M. S. Gazzaniga (ed.), The Cognitive Neurosciences, 4th edition, Cambridge, MA: MIT Press, chapter 77, pp. 1111–1122.
Key takeaways
- Three theories. Higher-order thought (HOT): an experience is phenomenally conscious in virtue of a thought that one has it. Global workspace: consciousness is global broadcasting of a winning coalition to frontal consumers; it abstracts from neural details and is a form of functionalism. Biological: consciousness is a biological state of the brain, illustrated by a recurrent MT+ representation of motion plus a background such as the “thalamic switch”.
- The explanatory gap as the test. For HOT, why a thought about an unconscious representation should make it conscious is a puzzle of HOT’s own making, like a “prime order thought” theory’s puzzle about primes, and not the real gap. The same holds for the workspace: why broadcasting is conscious would arise equally for an electronic workspace. Only the biological theory states the gap as it is.
- Gap without dualism. Following Nagel and Loar, subjectivity and objectivity are properties of concepts, so Mary acquires a new concept of a fact she knew. The gap is a failure to see how a subjective and an objective concept could pick out the same thing.
- Against ambitious HOT. Its conditions are stipulated to avoid counterexamples (“the advantages of theft over honest toil”); it faces the mismatch problem between a HOT and the sensory state it is about; and our case for experience in infants and autistic children exceeds our case for higher-order thought in them.
- Machines. The workspace and HOT views are friendly to machine consciousness. On the biological view it may matter that neural coding passes from electrical to chemical and back, so only machines with the right biology could be conscious.
Standing in the record
Filed on 2026-10-03 at the owner’s request, with six other papers by Block on consciousness. No anthology topic holds a comparison of theories of consciousness. The section on machines is an argument about which theories allow machine consciousness, not an instruction for machine-learning practice.
It is the “Block 2009” that Chalmers’s meta-problem paper (read in NOTE-354) names as the source of “biological theories”, against which it argues that non-conscious silicon duplicates would share our intuitions. It carries forward the concept-dualist reply to the gap of Block (2002), LIT-639, in a shorter form, and leans on Block (2007), LIT-646, for the claim that the biological theory allows experience a subject does not know about.
Bearing on THEORY-023. That theory holds that architectural indicators presuppose computational functionalism. This chapter states why the global workspace, the source of four of Butlin et al.’s indicators (LIT-056), is a functionalist theory, whose architecture “can just as easily be realized in silicon-based computers as in protoplasm”, and states the biological alternative on which those indicators would not suffice. It is a clear statement of the rival premise that theory says the evidence cannot yet decide.
On integrated information. The chapter calls the integrated information theory functionalist and criticises it as unable to tell consciousness from intelligence, since a seizure lowers both. It discusses the 1998 Tononi–Edelman version, not IIT 3.0 (LIT-445, read in NOTE-365), and offers integration as an adjunct to a biological theory. No relation is declared.