The Analytic Functionalists Were (Probably) Right!

The mind-body problem asks: How are mental states related to physical states of the brain, the body, and behavioral states more generally? Functionalists claim that mental states are identical with functional roles, defined as relations between environmental impingements, external behaviors, and other mental states.Analytic functionalists contend that these identities are imposed by folk psychological theory. Thus, analytic functionalism makes a specific empirical claim: that ordinary people conceptualize mental states in terms of their functional roles. Notoriously, this claim has met with intuition pump style challenges, such as Block’s Nation of China and Searle’s Chinese Room, which purport to demonstrate via our own intuitive assessments of cases that we do not ordinarily conceptualize mental states solely in terms of their functional roles. Discussion of these examples constituted a large chunk of philosophy of mind near the end of the 20thcentury, without reaching any very definite conclusions. But, as Lewis (1972) points out, analytic functionalism’s empirical claim “can be tested, in principle, in whatever way any hypothesis about the conventional meaning of our words can be tested.”


Recently, an experimental challenge to this empirical claim has arisen in the form of the embodiment hypothesis (discussed herehere, and here). According to the embodiment hypothesis, while information about behavior and other functional cues may be sufficient for ordinary attributions of intentional states, only entities with the right kind of biological body are typically thought to have phenomenally conscious experiential states. Proponents of the embodiment hypothesis have argued that empirical research on how people attribute mental states to groups, robots, and God support their view (here is a succinct overview). Based on this research, it may have seemed reasonable to conclude that analytic functionalism rested on a mistaken conception of how people talk and think about mental states. However, Wesley Buckwalter and I (along with others) have recently been uncovering a number of results that cast a more favorable light on analytic functionalism and seem to challenge the embodiment view. As discussed here, people are perfectly willing to attribute phenomenal states to robots when they are described in a way that calls to mind certain key functionalist assumptions. Mental state attributions to groups constitute evidence for the embodiment view only on the assumption that those attributions are to groups over-and-above their members—an assumption that Adam Arico, Shaun Nichols and I have argued against here. And Wesley and I have recently found that people are no less willing to attribute happiness and anger to a disembodied ghost than to his physical human counterpart, provided the ghost satisfies the appropriate causal roles. These findings led us to re-evaluate crucial studies taken to support disembodiment, and in many cases we found that those studies fail to control for function in important ways. As we discuss in the attached paper, this has led us via abductive argument to conclude that the analytic functionalists were (probably) right. It seems that ordinary people do conceive of mental states as functional relations between environmental impingements, external behaviors, and other mental states.




(Comments welcome as we revise for a special issue ofPhilosophical Topics.)

**Cross-Posted at the Experimental Philosophy Blog**

P.S.–Glad to be a part of this blog, and thanks John for including me!

07. December 2012 by Mark Phelan
Categories: analytical philosophy, Consciousness, Emotion, Experimental Philosophy, functionalism, functions, metaphysics of mind, naturalism, Philosophy of Mind, Psychology, reduction, representation, Robotics | 10 comments

Comments (10)

  1. Interesting stuff, thanks for posting.A couple of questions:

    1. Does this confuse ‘X is evidence for Y’ and ‘Y is constituted by x’?
    Even if people indeed tend to ascribe mental states based on criteria like causal role, is that really to say that they think such mental states are constituted by said causal role? Does grandma think her stomach hurts because it fills some abstract causal role, but has no intrinsic features, or does she think her stomach hurts because that is how it intrinsically feels to her? The latter seems right to me. And this is what makes things like qualia inversion type scenarios so tough for the hard-core functionalist. There is nothing intrinsic upon which to pin down the qualia of blue, say.

    2. Truth versus conceptual anthropology
    Given that the folk aren’t even good at physics, I am very doubtful that folk psychology is a useful avenue to the truth. To put it bluntly, why give such pre-scientific backwoods intuitions about extremely complex matters so much attention in the first place? Are philosophers really in the business because they want to understand uninformed people’s intuitions about extremely complex matters, about which they are almost certain to be wrong?

    On that vein, why not just say “Regardless of what the folk say, or what our polls suggest, I’m just advocating analytical functionalism as the best going hypothesis about the nature of mind.”

  2. Eric, on #2, it’s because then the thesis wouldn’t be *analytic* functionalism, since that modifier marks it off as a thesis about everyday meaning, and not (just) a metaphysical one.

  3. Thanks for your comments, Eric. John’s response handles 2. As for 1, we have a section in the paper discussing qualia. Our claim is that the folk psychological theory of qualia is representationalist. For example, people think of pain as physical damage and the experience of pain as a representation of physical damage. Same with colors. The folk psychological theory of color qualia is of a particular kind of experiential state (functionally defined, of course) that represents an external object as blue. (Of course, we don’t think people have explicit, verbalized theories that specify these accounts of qualia. But people do attribute qualia, and when they do so competently, we maintain, this is the tacit theory by which they do it.) The representationalist account of qualia identifies qualia with objects external to the mind. As such they can go unnoticed, be shared (insofar as a body part can be shared), and admit of an appearance-reality distinction. Our case for folk representationalism is simply that people think qualia exhibit all of these properties.

    This is all obviously quite collapsed…if you’d like to see more you can take a look at the last few pages of the paper. (Though, admittedly, the discussion of folk representationalism is much shorter than that of analytic functionalism.) In any case, thanks again for your comments!

  4. Thanks for the responses. On point number 2, fair enough.

    On point 1: I had actually looked over that final part of the paper on qualia, and that is precisely what triggered my concern.

    To repeat my concern in a different way: functional ascriptions are likely necessary, but not sufficient, for qualia ascription. For grandma, just having the experience is enough to ascribe the experience of pain (e.g., phantom limb pain, as her arm was removed in the war). Sure, she will say it is unpleasant, that it is an experience that is usually caused by tissue damage to her arm. However, the essential feature is that she has this pain, and all these other extrinsic facts are secondary.

    The brief discussion, at the end of the paper, of the folk analysis of ‘pain’ (not the experience of pain, which is a very important distinction) didn’t assuage these concerns. There is quite a bit of research on folk-concepts about experiences, for instance, that seems to contradict the main conclusions. For instance there is clearly a sense in which people don’t think their dreams are public or can be shared (I can track down the papers on this if you are interested but Google can probably reveal them). Perhaps there is a sense of ‘pain’ in which it is public, can be shared, but clearly that is not the sense relevant here, for the quali-freaks.

    On the other hand, there could simply be two sets of conditions, each sufficient to ascribe experiences of pain. One is purely functional, one purely experiential. In that case, you would need to show that the two are actually equated among the folk.

    To get at this you could look at folk responses to inversion thought experiments, in which an experience of a red sunset is swapped for an experience of pain in my toe (or something). E.g., ask subjects if this ‘experience of a red sunset’ caused them to jump about holding their toe, would it become an experience of pain in the toe? Something like that. Now of course, if you were to ask, which is more similar to a pain in the toe–an experience of a red sunset enjoyed at the beach, or an experience of a red sunset that causes you to jump around screaming about your toe–they will say the latter. But nothing will take away their conceptual ability to discriminate between the two experiences, regardless of any functional connections.

    That’s my hypothesis, anyway.

  5. Eric, let me repeat myself. The proposal is that people identify qualitative states with representational states of a certain type (where the relevant type is functionally specified). Given that proposal, we would not expect them to identify an experience of (a representation of) a red sunset that causes you to jump around screaming about your toe with an experience of (a representation of) a pain in your toe. Qualitative states are not simply functional states on our proposal. What they represent matters.

    The stuff on dreams may or may not be relevant. As described it strikes me as irrelevant. We obviously think mental states are proprietary, and dreams are paradigmatic mental states.

    That said, the discussion of representationalism at the end of the paper is, as you say, brief, and not worked out sufficiently to constitute a full proposal. It is a suggestion at the end of a paper whose main point is to defend analytic functionalism and argue against the embodiment view. We hope to write another paper on representationalism.

  6. We obviously think mental states are proprietary

    It wasn’t obvious to me, due to claims like pains are “a property of objects external to the mind…If pains are properties of bodies, they can go unnoticed (or un-experienced) as can other properties of our bodies. Pains can be shared according to this view.”  So pains can be shared, but dreams cannot be shared? I agree they shouldn’t be shared, but I still don’t see in your proposals what would preclude this.

    I think painting folk theories of consciousness as proto-representational is promising, but if you parse representational views in purely functional terms then it could run into trouble. However, you do say “Qualitative states are not simply functional states on our proposal. What they represent matters.” But what they represent is also defined functionally, no, in terms of relations to inputs, not in terms of their neural or other intrinsic properties. Which leads back to grandma explicating her pain in terms of how it feels, not functionally.

  7. Well there’s an ambiguity to the term “pain”. It’s probably easier if we distinguish pains and pain experiences. Pain is physical damage, as we say in the paper, so is not a mental states. Pain experiences are mental states–that is, experiences of pains are mental states. Pain experiences are proprietary. It may be thought that they cannot be shared. I’m not sure saying their proprietary requires any particular position on that, but I expect what is thought to be true of dreams in this respect is thought to be true of pain experiences, beliefs, etc.

    I don’t want to parse representation in purely functional terms (it wouldn’t be much of an addition to functionalism in that case!), but nor do I have a specific theory of representation. I’m inclined to adopt some version of externalism.

  8. Very interesting. Here are some thoughts/ questions that could be addressed in future work:

    First, the distinction between “individuals” and “groups” seems to be relative to a level of description: On a microscopic scale, our brains are vast groups of neurons, and while they interact and are closely related, so do group entities like corporations, though admittedly to a lesser extent. One question is then whether the folk, on being given a detailed explanation of how some part of the brain works, or of neurons, dendrites and synapses in general, would feel an intuitive pull towards a dualistic view: they might think such an ensemble couldn’t be conscious any more than a corporation could, and posit a more unified agent like a soul (BTW, this resembles Scholastic arguments for the simplicity of the soul). Another question is, if the folk would still think that brains could be the subjects of intentional and phenomenal states but NOT that corporations could, whether the difference would arise because the members of a corporation have thoughts and experiences of their own, while neurons are thought not to have them. If so, that would seem to show that people fail to ascribe thoughts and experiences to corporations because they think that if the members of a group have thoughts and experiences the group cannot–the mental states of the members “crowd out” any potential mental states of the group–not just because they are a group. On this view groups could still be conscious if their members are not.

  9. This all seems reasonable, which is why I wrote a couple of comments ago:
    “The brief discussion, at the end of the paper, of the folk analysis of
    ‘pain’ (not the experience of pain, which is a very important
    distinction) didn’t assuage these concerns [blah blah blah concerns] On the other hand, there could simply be two sets of conditions, each
    sufficient to ascribe experiences of pain. One is purely functional, one
    purely experiential. In that case, you would need to show that the two
    are actually equated among the folk. ”

  10. Thanks Eric and Jason for your comments, they will help with this and future work. Sorry for my delay in responding. Jason, I make some slightly more detailed comments where you posted on the Experimental Philosophy blog.