June 20, 2010

Notes to Why "Why Transhumanism Won't Work" Won't Work

The article is now up on the IEET website, HERE. (A PDF version can be found here.)


[1] Cognitive enhancements have the potential to significantly augment the cognitive capacities of the individual, thus (possibly, to some extent) closing the epistemic gap between what the collective whole and the individual knows. At some point in the future, then, it may be that each individual knows as much as the entire group.

[2] See Jaynes, E.T. and G.L. Bretthorst. 2003. Probability Theory: The Logic of Science. Cambridge: Cambridge University Press.

[3] Gubrud writes: “Since these multiple criteria, not all clearly defined, may sometimes conflict, or may tend to different formulations of ‘identity’ and its rules, philosophers have here a rich field in which to play. ‘Progress’ in this field will then consist of an endless proliferation of terms, distinctions, cases, arguments, counterarguments, papers, books, conferences and chairs, until all tenured positions and library shelves are filled, the pages yellowed, new issues come into vogue, and the cycle starts over. I come as a visitor to this field, and while I must admire the intricate Antonio Gaudi architecture of castles that have been raised from its sands, twisting skyward and tumbling over one another, my impulse is bulldoze [sic] it (see first sentence of this essay), flatten it out and start from scratch, laying a simple structure with thick walls no more than ankle-high above the ground, as follows.”

[4] Consider the opening sentence of Gubrud’s “Balloon” paper: “Physical objects exist, consisting of matter and energy, or any other physical substance that may exist, but please note that ‘information’ is not one; neither is ‘identity’.” Where to begin? I cannot think of a single philosopher – or, for that matter, scientist – who would argue that information doesn’t exist or is non-physical in nature. (Indeed, information theory is a part of physics.) Most philosophers today are ardent physicalists who see information as perfectly compatible with their metaphysical monism (which asserts that “everything is physical”).

Gubrud’s reflex here is, no doubt, to think: “Yeah, well, that doesn’t make sense to me. How could information really be physical? I mean, you can’t reach out and touch information…” I would encourage Gubrud not to leap to any conclusions; first try to understand why philosophers today take information to be physical (a crucial first step that Gubrud repeatedly fails to make). Then you can proceed to critique the thesis, if you’d like, once you know what that thesis is. (Perusing this article on physicalism would be a good start – but only a start.)

[5] Gubrud writes: “For transhumanism itself is uploading writ large.”

[6] Take note that philosophers typically distinguish between the qualitative and non-qualitative aspects of mentality; in Ned Block’s phraseology, the former is “phenomenal” consciousness and the latter “access” consciousness. Chalmers (1996) also emphasizes an exactly parallel distinction between "psychological" (or "functional") and "phenomenal" conceptions of the mind.

[7] Note that this computation may be of numerous different kinds; again, see this article.

[8] To be clear, functionalism takes mental states to be “ontologically neutral.” That is, while purely physical (e.g., neural) systems could indeed instantiate a given mental state, so could, in principle, an immaterial substance of some sort. All that’s relevant, according to the functionalist view, is the substrate's causal-functional properties.

[9] As Howard Robinson writes: “Predicate dualism is the theory that psychological or mentalistic predicates are (a) essential for a full description of the world and (b) are not reducible to physicalistic predicates. For a mental predicate to be reducible, there would be bridging laws connecting types of psychological states to types of physical ones in such a way that the use of the mental predicate carried no information that could not be expressed without it. An example of what we believe to be a true type reduction outside psychology is the case of water, where water is always H2O: something is water if and only if it is H2O. If one were to replace the word ‘water’ by ‘H2O’, it is plausible to say that one could convey all the same information. But the terms in many of the special sciences (that is, any science except physics itself) are not reducible in this way. Not every hurricane or every infectious disease, let alone every devaluation of the currency or every coup d'etat has the same constitutive structure. These states are defined more by what they do than by their composition or structure. Their names are classified as functional terms rather than natural kind terms. It goes with this that such kinds of state are multiply realizable; that is, they may be constituted by different kinds of physical structures under different circumstances. Because of this, unlike in the case of water and H2O, one could not replace these terms by some more basic physical description and still convey the same information. There is no particular description, using the language of physics or chemistry, that would do the work of the word ‘hurricane’, in the way that ‘H2O’ would do the work of ‘water’. It is widely agreed that many, if not all, psychological states are similarly irreducible, and so psychological predicates are not reducible to physical descriptions and one has predicate [or descriptive] dualism.”

[10] More generally, Gubrud seems especially susceptible to confusing terms with the entities signified by those terms. That is, Gubrud reasons that since there are two (or three, etc.) different terms in the discussion, then there must be two (or three, etc.) different referents. Consider, for example, the following passage from his Futurisms article:

"Thus Moravec advances a theory of
pattern-identity ... [which] defines the essence of a person, say myself, as the pattern and the process going on in my head and body, not the machinery supporting that process. If the process is preserved, I am preserved. The rest is mere jelly.
Not only has Moravec introduced 'pattern' as a stand-in for 'soul', but in order to define it he has referred to another stand-in, 'the essence of a person'. But he seems aware of the inadequacy of 'pattern', and tries to cover it up with another word, 'process'. So now we have a pattern and a process, separable from the 'mere jelly'. Is this some kind of trinity?"

See the first "rule for avoiding sciolism" mentioned in my article.

(Additional note: ontological dualism seems to imply descriptive dualism, but descriptive dualism does not necessarily imply ontological dualism.)

[11] Chalmers’ view is called “property dualism.” It holds that certain particulars have physical and non-physical properties. In contrast, Cartesian substance dualism posits that those particulars themselves are non-physical in nature. My own tentative view is that this is probably wrong, but that we (unenhanced humans) are simply "cognitively closed" to the correct answer. (This is McGinn's "transcendental naturalism.")

[12] As Georges Rey puts it, "consider some theory, H, about houses (which might state generatlizations about the kinds of houses to be found in different places). The ontology of this theory is presumably a subset of the ontology of a complete physical theory, P: every house, after all, is some or other physical thing. But the sets of physical things picked out by the ideology of H -- for example, by the predicate "x is a house" -- may not be a set picked out by any of the usual predicates in the ideology of P. After all, different houses may be made out of arbitrarily different physical substances (straw, wood, bricks, ice, ...), obeying different physical laws. Houses, that is, are multiply realizable. To appreciate the generalizations of theory H it will be essential to think of those sundry physical things as captured by the ideology of H, not P. But, of course, one can do this without deny that houses are, indeed, just physical things."

[13] The answer to Why? here, on Chalmer's view, is that consciousness is simply a brute fact about the world in which we live. Psychophysical laws connecting matter and conscious states are fundamental laws, just like the laws of thermodynamics, or motion. They are, as it were, the ultimate "unexplained explainers."

[14] As Schneider points out, patternism is thus a computationalist version of the "psychological continuity theory" of personal identity.

[15] This is, in my opinion, a rather interesting thought: the uploaded mind would indeed by psychologically continuous with me. Mind clones seem, I suppose, more intimately related than genetic clones (such as identical twins).

Back to the article -->


  1. Philippe,

    Classical information theory can be regarded as pure mathematics, motivated by certain engineering and physics scenarios. Like other bits of math, which tell us how to compute the result of some infinite series or the solution to some equation, classical information theory can be regarded as implying results in physics.

    Quantum information theory, now regarded as a generalization of classical information theory, and the "true theory" of physical states considered as "information-bearing", is clearly inspired not by pure mathematics (although it is now available for further development by pure mathematicians) but by physics, i.e. quantum mechanics.

    None of this implies that "information" (classical or quantum) is a "second stuff" or dual of matter and energy. Rather, "quantum information" is just another way of looking at the same physics of matter and energy. You could say the "quantum information" is just as real - unlike classical information, it can't be destroyed or cloned - but at the same time, it is still the fact that there is no quantum information without matter/energy, and that quantum information theory only develops the physics of matter and energy in a different form, rather than adding something new.

    I argued further, in the paper you are ostensibly critiquing, that all mathematics can be interpreted in purely physical terms, so there is no need for Platonic ideals or other dualistic notions. As must be so, unless our brains, in performing mathematical calculations, somehow violate known physics.


  2. Dear Philippe,

    To your note #4 again, physicists get to rule on this. "Information is physical" only in the sense that it is an aspect of the physics of matter/energy. No matter or energy, no information. Attempts to derive "It from bit" or the physics of matter and energy from pure information theory have always failed. Perhaps this is related to the fact that classical information is not even conserved. Quantum information is conserved, but it, too is scrambled (decohered) and lost to the noise pool. This may be related to the statement that quantum information theory is just a redevelopment of quantum mechanics, which always refers to states having energy/mass.

    Information is not "physical" in the sense of being a substance. A thing cannot exist because it is made of pure information. It must be made of matter/energy. The "information" is the pattern of matter/energy, which we can describe in many ways. It is not a second substance supervening on the first. It is just aspects of the first and only substance.

    Best regards,

  3. To your #8: By "immaterial substance of some sort" I take it you mean things other than physical quantities such as matter/energy, spacetime, the wavefunction, etc. - the things we know to exist. Those other things are things which are ontologically null, which cannot instantiate anything, and which have no causal-functional properties. This is so because they do not appear in the theories that describe the physics of those things that we know do exist. Therefore, any effect of them would be a violation of those theories, and should be detectable by experiment. I do not expect that to occur.

  4. To your #10: You sound so pedantic, judging that I reason that "since there are two (or three, etc.) different terms in the discussion, then there must be two (or three, etc.) different referents."

    Didn't you notice that my whole argument was that all the terms refer to the same object, same as "the soul" or "the true identity"?

    I am of course accusing transhumanists of either being very confused themselves or trying to confuse others with these many euphemisms for the soul, that elusive object of true identity.

    To make it sound plausible that I am not just joking about a "trinity," you cut off the quote at that point. The rest was,

    "Or is the “mere jelly,” once appropriately patterned and undergoing the processes of life, what real human beings are made of — that and nothing else that is known to science?"

    Shame on you, Philippe.

  5. To #12: Your ideology of houses is your ideology of houses, and would be nothing if no one had it. The reality of houses is written in the book of P. Do you think you are an ideology, Philippe? Or are you a particular house?

  6. First of all, I would encourage you to read my article before commenting on my footnotes. Although I am at fault for posting them prematurely (I honestly didn't expect you to be checking my blogpost hourly), the fact that you posted 5 comments is, I believe, right in line with my criticism of your paper: you are *very quick* to state, often in strong language, your opinion, without taking the time to *understand* what is being said.

    So, for example, with respect to #12, do I think that *I* am an ideology? No, no, no. NO! Quine (and others following him) made a distinction between the ontology *of* a theory and the ideology *of* a theory. (Forget whatever connotations "ideology" may have -- it's being used here in a technical sense.) Note also a difference between talking about "fundamental ontology" and "a theory T's ontology." This being said, a real estate theory has an ontology that consists of houses (since houses must exist for the theory to be true), and indeed such a theory would (centrally) include the predicate "is a house" (where this predicate would pick out objects having the property of being a house).

    Presumably, houses are physical things(!), which means that the the ontology of this real estate theory would be a *subset* of the ontology of physicalism. But there need not be (and very likely wouldn't be!) any predicate in a fundamental theory of physics that corresponds to "is a house." Indeed, the real estate theory and the physical theory would radically cross-classify objects, since houses are *multiply realizable* entities.

    What does this mean? Well, it means that one cannot reduce *talk* of houses to the language provided by the fundamental physical theory -- again, this is because what makes a house a house is not its material substrate (what it is, materially), but its function (what it does, roughly speaking). (Consider the different between water and poison: what makes water water is that it has a certain chemical structure; but what makes a poison a poison has nothing to do with its chemical structure. Rather, it has to do with its *function*. The distinction here is between natural kinds and functional kinds.)

    The same idea applies to mentality. Mental states are, if you're a functionalist (and just about every single cognitive scientist and philosopher today is), functional rather than natural kinds. They are like houses, which means that a theory of mentality may share the ontology of physicalism while holding that talk of mental states is *not reducible* to talk of neural (or electronic) states. Thus, just as one can "transfer" (or, call it "upload") a house from one substrate to another (by, e.g., gradually replacing red bricks with cinder blocks, or whatever), so too can one transfer a mind from one substrate to another.

    There's nothing mysterious here at all: any functional kind whatsoever can be so "uploaded" from one substrate to another. Thus, if the mind *is* a functional kind, then it too will be transferable from one material substrate to another.

    The self is a separate issue. Please take a look at my article (later today, I believe) before commenting any more. And, furthermore, I would encourage you to let the ideas simmer for a day or two before responding. People spend years and years studying these issues for a reason -- they are complicated!

  7. Finally, regarding your comment:

    "To your #10: You sound so pedantic, judging that I reason that "since there are two (or three, etc.) different terms in the discussion, then there must be two (or three, etc.) different referents."

    Didn't you notice that my whole argument was that all the terms refer to the same object, same as "the soul" or "the true identity"?"

    Yes, absolutely. This is precisely my point: you assume that terms like "pattern" refer not to physical entities, but to some immaterial thing. *That* is precisely the problem, and why I argue in my article that you confuse what's sometimes called "predicate dualism" (which one finds in just about all of the special sciences) with "substance dualism." Talk of minds is no more dualistic than talk of houses, in the above example, even though both types of descriptions are non-reducible to the language of lower-level physical theories.

  8. Philippe,

    Your house is a particularly bad example, because a house is definitely, at any one time, a particular set of atoms, with very little exchange. If anyone would say a house stays the same despite complete change of materials, I will say this proves nothing at all, except that people view houses this way. Likewise, people view people as staying the same people through life even as they change, by appeal ultimately to the numerical identity criterion imposed by the single organismal history.

    At any time, one is a particular structure made of a particular set of atoms. If such an object has a fate, then its nanodisassembly is a fate that is not altered by any subsequent assembly of any kind of copy, biological or more generally functional. Nor by two copies, nor by N, nor by the fates of those copies, nor by any other fact in the future. You should not volunteer to have your brain disassembled for any reason, unless you are willingly giving up your life for some higher purpose.

  9. There's a distinct asymmetry in this discussion. One party is able to express both (and various) views on this topic, thus demonstrating the more encompassing understanding. The other other party is merely certain that he is right.

    A blunt example: Children often "know" that they are "right", despite being unable to express the views of their more experienced parents.

    I've found no direct solution to such problems, since the necessary context can not be conveyed but must be constructed.

  10. Mark:

    Again, if you're still at the University of Maryland, I would *strongly urge* you to print out the IEET article, plus the comments (if you'd like), and take them over to the philosophy department. There are several quite brilliant and well-known philosophers working on the philosophy of mind who could help explain these issues to you. At this point, I'm really not sure what else to say, except that I have the overwhelming support of philosophers, cognitive scientists, and everyone else who specializes on these issues. If the house example doesn't make sense to you, then *ask someone more articulate than me to explain it to you*. Because I assure you, it *does* make sense.

  11. Mark: "At any time, one is a particular structure made of a particular set of atoms. If such an object has a fate, then its nanodisassembly is a fate that is not altered by any subsequent assembly of any kind of copy, biological or more generally functional. Nor by two copies, nor by N, nor by the fates of those copies, nor by any other fact in the future. You should not volunteer to have your brain disassembled for any reason, unless you are willingly giving up your life for some higher purpose."

    Fate is a human way of looking at things. The universe doesn't look at things like that at all. It just has atoms which do things in response to other atoms. If you are interested in recording the specific history of particular atoms, that's fine, but it does not say anything about the survival of human beings, which happen to be patterns of a particular sort imposed on interchangable sets of atoms. Human fates (for the meaningful definition of human) do not map to particular sets of atoms, otherwise biology would be destroying and replacing us quite frequently.

  12. Luke,

    If your brain is cut up into little bits, you're dead. There is no way to put the little bits back together and nothing else that happens changes what happened to you, the last thing that happened to you. I am talking about the you that counts, the true you, the you that wants to be saved from death. That thing is certainly destroyed when the brain is cut up into little bits. The position of any (approximate) copies made from information obtained in cutting the brain up into little bits is another matter to be considered. It happens after your death, so why do you care?