A Summary of the Imperative Content Strategy

By Manolo Martinez

In my last post I was assuming familiarity with something not many people is familiar with. I apologise. Let me say a quick thing about the imperative content strategy to defend representationalism about pain.

Before that, representationalism: this is the idea that the phenomenal character of experiences (what it is like to undergo them, in a popular turn of phrase) supervenes on (or is identical to, depending on the flavour of representationalism) its representational content. So, for example, an experience as of red instantiates phenomenal reddishness -the characteristic  what’s-it-likeness of seeing red stuff- in virtue of having a content along the lines of “There is a red patch there”. This is an elegant way of combining realism about qualia and naturalism and, although I’m not sure I would marry the theory for good, I sure like it a lot.

One problem with representationalism is that it seems to be ill-suited to accounting for affective phenomenology, such as that of pains, pleasures or the phenomenal component of emotions such as fear or hope. For example, it looks as if representationalism goes hand in hand with the acceptance of illusory experiences (those with non-veridical content). But pains cannot be illusory: phantom-limb pain, it seems, is as genuine as real-limb pain. Also, it doesn’t seem as if pains or orgasms have representational content at all: experiences as of red seem to be about redness, but what’s an orgasm or a headache about?

Some have proposed -that I know of, Colin Klein’s ‘An Imperative Theory of Pain’ and my own ‘Imperative Content and the Painfulness of Pain’– that affective phenomenology should be accounted for in terms of imperative content. For example, in my version of the idea, the content of pain experiences is something like “Don’t have that bodily disturbance!” or, anyhow, an imperative with the satisfaction conditions that I don’t have that bodily disturbance anymore -where the bodily disturbance in question is, e. g., a cut or a burn. Note that these cuts or burns need not exist -witness phantom-limb pains.

Now: It’s no wonder that we cannot have illusory pains: the affective phenomenology of pain is neither true nor false. Also, it’s no wonder that the painfulness of pain does not look representational at all: this was assumed to mean that it doesn’t seem to have indicative representational content (which indeed it doesn’t), but our intuitions about painful phenomenology are compatible with (and indeed congenial with) the idea that it has imperative representational content, that painfulness compels us to move away from the disturbance in question, for it to be no more.

This kind of prposal is the one Tye criticises in the paper I discuss in my earlier post.


  1. Thanks for posting this. My first reaction is that we just need to expand our usual indicator-semantics to cover different states of the body such as damage, or low levels of blood glucose, etc.. Doesn’t Damasio argue that there is a unique population of relative activity in the interoceptors and autonomic activity that corresponds to various experiences like hunger, thirst, and even emotions?

  2. Colin Klein

    Nice summary, Manolo. I should point out that he and I disagree on precisely what the imperative content is—I think that it’s a negative imperative against using your body in certain ways. Manolo has a paper arguing against this, which I really need to come up with a good response to 🙂

    Also, if you want to see more of the flavor of it, a short response of mine to some objections just came out in the latest J/ Phil.

  3. @Eric, One problem with simply extending our semantics and saying that pains refer to damage is that it makes phantom-limb pains not real pains. The sufferer would be merely under the illusion of feeling pain. I don’t think such an error theory about pain is particularly attractive.
    The same goes for glucose levels, I think: the problem of affective phenomenology in general is that it does not tolerate being taken for an illusion -consider illusory thirst (because not accompanied with the correct level of liquid depletion, etc.).

    I should note that this not mean that you cannot be wrong in thinking you are thirsty when you are not. You can. It’s simply that, if undergoing a state with the right phenomenology (phenomenal thirst, let us say), it is not an option to say that you are not really thirsty.

    @Colin, I’m really looking forward to that response 🙂

  4. Eric Thomson

    If your concern worked for phantom limbs wouldn’t it work for visual and other hallucinations? Just extend the standard account of dreams/hallucinations to states of the body. I don’t see the problem. Just like you aren’t wrong when you say that you visually experienced such-and-such in a dream, you are not wrong when you say you had somatosensory experience such-and-such in your (nonexistent) arm. The brain’s representations of the relevant phenomena are activated without the corresponding event happening in the world/body.

    That said, even though it seems not too hard to extend standard indicator semantics to pain and such, the question is whether that is correct. These aren’t conceptual issues, after all, but empirical issues. Perhaps as empirical fact, you are right, that the content of pain is partly imperative.

    Of couse indicator semantics doesn’t  just have to limit itself to inputs: we (neuroscientists) talk about the content of representations in M1 as being movements of the arm in a certain direction, after all, not the sensory inputs that ultimately evoked such movements. For some reason philosophers such as Dretske focus on the indication functions of primary sensory regions. My hunch is that this is because of their focus on perception and concepts. If Knowledge and the Flow of Information had been focused on motor control, perhaps we’d have a quite different book.Indeed, his discussion, and dismissal, of ‘consequentialist’ theories of content fixation (i.e., looking at effects not causes) suggests he missed this possibility entirely.

  5. That’s a possible move, yes: “having a pain” does not behave like “seeing” but rather like the (technical) “undergoing a visual experience”. The question, then, becomes this other one: why this other asymmetry? If having pains is just like seeing -that is, receiving information about thus and so goings-on-, why does “having a pain” not distinguish between veridical and non veridical instances and “seeing” does? That asymmetry stands in need of explanation, too.

  6. Eric Thomson

    “why does “having a pain” not distinguish between veridical and non veridical instances and “seeing” does”

    I don’t see the asymmetry. I’m frankly not even sure what you mean.

    It seems we do distinguish pains the way we do visual experiences. He took LSD and saw a monkey crawling up his leg. We produced the thermal grill stimulus and he experienced the pain of the thermal grill. His arm was cut off and he experienced pain in his arm.

  7. One last try 🙂 To “he saw a monkey during his LSD trip” a response such as “he didn’t really see it; he was allucinating” is appropriate. To “she was feeling pain in her phantom leg” the response “she was not really feeling pain, but merely hallucinating” is not.

  8. Eric Thomson

    Yes, there is a of ‘see’ that is a success verb or whatever, used in some epistemic/normative sense to endorse the match between percept and reality. But we are talking about phenomenal content, so that seems a red herring here as it is clear that ‘feels pain’ is not equivalent to ‘see’ when the latter is used in this epistemic sense. Since the word ‘see’ has ambiguities that seem to cloud the key issues, we should be using the expression ‘visual experience’ which is clearly the better semantic sibling of ‘feeling pain.’ (‘He had a visual experience of a flying pig’ to which the response ‘He was not really having that experience, but merely hallucinating’ makes less sense).

    Taking another route: you wouldn’t describe the visual experience during a dream as ‘Seeing a big rhino’ (not in the tendentious epistemic sense of ‘see’ anyway), so it seems clear that your ‘asymmetry’ is due to an incorrect characterization of vision versus pain. If we are really after the kinds of content you have in dreams, hallucinations, etc., there is no asymmetry between visual experience and pain. We have both in dreams, we hallucinate both.

  9. Leaving aside the stuff about affective phenomenology, let’s get back to the good old representationalism. I recognize that it is very difficult if not impossible to draw a distinct line between an experience’s phenomenal quality and its representational content. There is a funny interplay between the two, to the extent that it doesn’t take long to start suspecting that we may have a false dichotomy here. I think representationalism is an attempt to explore this, but by dismissing qualia in favor of the representational. I have never yet read a definition of “represent” that makes this plausible to me, or for that matter, any decent, crisp definition at all. I suspect that in order to get to the bottom of the aforementioned funny interplay in the mind, we are going to have to go the other way, and give due respect to the qualitative inherent within the representational. In doing so, we will likely redefine “represent” (at least within minds) in such a way that it ends up being a very different kind of thing than the “representing” that goes on in books, highway signs, and computers.

    -John Gregg


  10. Eric Thomson

    Yes, that seems a relatively uninteresting contingent fact about ordinary language, not something upon which to base a substantive theory of content. We could built a similar sense of pain or other interoceptive experiences (the degree to which the experience is internalized to the body seems the most relevant difference here). At any rate I really don’t like ordinary language focus. Even if such semantic distinctions didn’t exist in any culture what would that mean? Aren’t we trying to build an empirical, scientific, theory of consciousness, not doing linguistic anthropology?

    Just because there exists one particular verb associated with vision for which there is arguably no analogy seems to not be a very powerful reason to push in a completely new direction. Especially since we already have linguistic resources to describe vision in an obviously analogous way (i.e., ‘visual experience’). It just seems tendentious to focus on one verb about visual experience, the one that is clearly different in meaning, and make this special.

Comments are closed.

Back to Top
%d bloggers like this: