PMS WIPS 007 - Joshua Knobe and Jesse Prinz - Intuitions about Consciousness: Experimental Studies

“Intuitions about Consciousness: Experimental Studies” by Joshua Knobe and Jesse Prinz, University of North Carolina – Chapel Hill

When people are trying to determine whether an entity is capable of having certain kinds of mental states, they can proceed either by thinking about the entity from a functional standpoint or by thinking about the entity from a physical standpoint. We conducted a series of studies to determine how each of these standpoints impact people’s mental state ascriptions.

The results point to a striking difference between two kinds of states — those that involve phenomenal consciousness and those that do not. Specifically, it appears that ascriptions of states that involve phenomenal consciousness show a special sort of sensitivity to purely physical factors.

[Link to full text of article]

[Link to further info on PMS WIPS]

9 Responses to “PMS WIPS 007 - Joshua Knobe and Jesse Prinz - Intuitions about Consciousness: Experimental Studies”

  1. I’m surprised no one has commented yet on this very interesting and readable paper! Well, it is the holidays….

    I find myself sympathetic with the view that some mental state ascriptions, such as “feeling pain”, strictly require a certain phenomenology, while others such as “believing” do not. And I’m also sympathetic with the idea that the division between these two types of mental states may correspond nicely to the division between those mental states we are not willing to ascribe to group entitites (those of the first type) and those we are willing to ascribe (those of the second).

    Yet I also wonder if the distinction you draw, Joshua and Jesse, between the functional and phenomenal states is a bit too sharp. It seems to me that belief — in human beings — is in some ways tied up, in folk intuition, with consciousness, with dispositions to feel surprise in certain cases or to utter things to oneself in inner speech, etc. If we imagined a behavioral twin of a human being without consciousness (a “zombie”), I suspect people would feel divided about whether it really, literally can have beliefs. Likewise, if asked straight out whether corporations can really, literally have beliefs, I suspect there would be similar ambivalence.

    Of course, we often speak of corporations and groups as believing, intending, trying, wanting, etc. — so that sort of talk doesn’t seem “weird”; nor do I want to say that it’s quite metaphorical; but maybe, for some of the terms, it’s somewhere between metaphor and the extension of a concept to cover less central cases. Compare: It doesn’t seem weird to say a computer “recognizes” a command (or that the immune system “recognizes” an invader) Maybe it’s not even, quite, a metaphor (anymore). But does it follow that computers (and immune systems) do have mentality — the mentality of literally recognizing things?

    Our ordinary folk psychology about such matters is, I suspect, a bit of a mosh, and answers to questions on these topics might be highly context-dependent — depending for example on what the contrast cases are taken to be.

    Consequently, I’m a bit nervous about some of the studies. As you point out, there are a number of potential confounds in Study 1, including simply number of words. Study 2 seems similarly confounded to me, actually. For example, the second group all employ the present progressive, and three of the five employ intensifiers (“great”, “excruciating”, “vividly”). The intensifiers, in particular, seem theoretically unnecessary and yet may skew the results a bit, since ascribing an intense mental state to a corporation may seem stranger than a more neutral ascription. The within-subjects design (if I read rightly) of Study 4 may also amplify the results a bit, if there’s a lot of context sensitivity since the one statement of the pair serves as a contrast to the other.

    All that said, I do think you’re on to something here. It’s interesting to point out the relative ease with which we ascribe intentions to corporations but not mental imagery; but perhaps mental state ascriptions align on a spectrum — some more strictly requiring a particular phenomenology, some more vaguely invoking or assuming it, others comfortably ascribed without the slightest trace — rather than dividing as you seem to suggest into two sharply disjoint classes.

  2. Joshua Knobe says:

    Hi Eric,

    I think you might really be on to something here. I wonder if it might even be possible to take your suggestion one step farther. You suggest that there might be a continuum such that different mental state ascriptions require phenomenality to different degrees. But might it also be that we regard different entities as possessing phenomenality to different degrees? If so, it might be that we only regard mental state ascriptions as acceptable when the degree to which the entity is taken to possess phenomenality is greater than or equal to the degree to which the ascription itself requires phenomenality.

    Does that approach sound right to you?

  3. Tad says:

    Are you familiar with Austen Clarke’s ‘Beliefs, Desires, Incorporated’? It’s in J. Phil. from 1994 I think. I think it’s relevant: he argues that belief ascriptions to corporate entities should be taken literally, that is, in the same sense as ascriptions to persons. From this he draws some interesting conclusions regarding our introspective access to the mental processes actually responsible for our behavior. (Just as belief ascription to corporate entites should not be taken as literally describing the processes causally responsible for their behavior, neither should such ascriptions to persons).

  4. Anibal says:

    It is not totally related to our folk intuitions concerning conscious states ascriptions to corporations but when a medical doctor failed in his diagnostics or err in therapy and due careness to a patient, sometimes corporativism happens, that is, all the medical staff in a hospital (group agent) cover up the failure.

    I think the great breakthrough open by this kind of studies which follows the rationale imposed by Knobe´s general philosophical program to asses to what extent our folk morality and our folk psychology intermix, settles the first stone of a good looking building which can answer what in law is called the “problem of many hands” (Mark Bovens 1998): how to distinguish the needdle in a haystack, or in other words, where our folk intuitions are prompted to, they are brought to blame the individual doctor or the entire medical staff (group agent)? Who is more “conscious” the individual or the corporation?

    Following the parallelism in the ascription of consciousness to corporations, can we say the temporal factors matters? if the group is recently form, is still view to posess virtual properties of consciousness or judged as if such were the case?, or the “similarity to humans” rule breaks when gropus are formed recently.

    If the temporal factor matters, another model for conscious ascriptions to corporations is valid, that devised by Pettit (2001) where he diferrentiates between “person aggregates” and “person intrincates” reading person as group and the later the only to be treated as genuinely person because it last enough time.

  5. J. Brendan Ritchie says:

    My comments here echo the same general worry (and possibly some specific ones) that Eric has: there is a real risk in all of these studies of significantly confounding variables. I think this in large part has to do with intuitions people have about the things you use in your examples, namely: feelings, corporations, magic, and people who “work a lot with fish”. Despite all this, I think you are right about physical makeup being important to attributions of consciousness, though I think you need to control for a lot more variables (accept for the moral dimension to our concept of phenomenal consciousness. I am not convinced you have shown that our FP concept of phenomenal consciousness has a moral dimension, rather than our concept of feelings).

    Study 1:

    There is an asymmetry in how Simone set up the phrases searched for with Google. For phenomenal consciousness it would be more accurate to search for, say “Microsoft feels” and “Microsoft experiences”. When I did the former I got several hits, though probably not nearly as many as for non-conscious states, but more than you report, possibly much more (there seems to be an especially large number of links to the phrase “Microsoft feels your pain” for instance).

    But, more importantly, as you have it your study is bias against sentences that might be found attributing phenomenal consciousness to Microsoft. It seems unreasonable to search for “Microsoft wants to eat cake”, because the desire expressed by this sentence, plausibly, is not something Microsoft, as an agent, is likely to want. Likewise, it seems wrong to be searching for “Microsoft feels is depressed”, because, if there are feelings Microsoft might feel, it is unlikely that depression is one of them.

    Finally, “feels” is a tricky term (this point I will be repeating a lot). There are more than one kind of conscious experience, and the term ‘feel’ seems to refer to a particular subset of conscious experience. What about “Microsoft sees” “Microsoft has heard” or, again “Microsoft experiences.” It is too quick to generalize based on using almost exclusively the term ‘feel’. At minimum, to prove your case, you are going to have to use other words in its place.

    Study 2:

    I have essentially the same issue as before. Consider two of your sentences:

    1. Acme Corp. intends to release a new product in January

    Compare to:

    2. Acme Corp. is getting depressed.

    The choice of language here should be more impartial. What about:

    2.* Acme Corp is depressed about low profit margins.

    Similarly, you might see what happens with the following:

    1.* Acme Corp. intends to eat cake.

    My hunch is that we would see a statistically significant change. My running worry here is that what is driving the attributions of mental states and phenomenal consciousness is the content of the states that are attributed, NOT the kind of states themselves. Until you control for the sorts of cases I have just suggested, it does not seem your data has ruled out this hypothesis.

    For your follow up study: ‘experience’ and ‘feel’ seem to have different connotations, as I have already suggested, so the fact that you got such low results for feels does not necessarily support your hypothesis. Also, plausibly, an urge is the same thing as a want or desire, in which case, it would count as a non-conscious mental state, and would count against your hypothesis. At the very least, it seems like the following phrase:

    Acme Corp is experiencing a sudden urge to pursue internet advertising.

    Is problematic. You should be able to substitute it with other phrases. Also, the term ‘experience’ does not necessarily refer to phenomenal consciousness. Many things, for instance “experience change over time”; Often people refer to the DOW “experiencing” things. While for philosophers the term ‘experience’ refers to phenomenal consciousness it is at least not obvious that laypeople use it in the same way.

    Study 3:

    The introduction of the sorceress is almost certainly a confounding variable for these results. I would like to see what Edouard’s results are. Another test would be whether people think a sorceress could magically make someone a philosophical zombie: suppose the sorceress puts a spell on some unassuming graduate student such that he no longer forms complex thoughts, plans to do things or anything (so the story would ony explicitly state he lacks certain non-conscious mental states). It would be interesting to see whether people attribute phenomenal consciousness to said graduate zombie (what would they answer to the question: “does the grad student feel happy or sad?”) My hunch would be that subjects are willing to deny that the student still has consciousness, in which case, it seems magic overrules function and constitution. In which case, I think you are going to have to appeal to something other than magic to prove your case. At minimum, this is another test to be done.

    Study 4:

    Again, when you keep the amount of “upsetness” fixed, you might try with other words than ‘feel’. Perhaps people are willing to attribute some conscious states to corporations, but not others. So, corporations cannot feel anything, but they might be able to see and hear things, they might also be able to be “aware” of things. Of course, sight, audition and awareness are not necessarily conscious, but then, it is also controversial whether or not there can be unconscious feelings ( e.g. unconscious pain).

    Also: you might run the sentence pairs you have here with your enchanted chair. I wonder what the difference would be. My hunch: a similar discrepancy would arise when you add and remove the term ‘feel’.

    Part V:

    While some have explicitly claimed that folk psychology is functional it is not clear that it follows from the theory-theory that one is committed to the same claim. So, two hypothesizes: A. folk psychology is functional (that is: FP states are specified by the functional role they play in predicting and explaining behaviour), B. folk psychology is a kind of theory. Your results do not seem to cast any doubt on the second hypothesis. While the grand vision might be to understand folk psychology as a way of predicting and explaining behaviour for the most part, this does not entail that said vision is incommensurable with the idea that not all folk psychological states are characterized in functionalist terms for this purpose.

    In fact, I am not sure why anyone would claim that folk psychology only plays a role in predicting and explaining behaviour, though I must profess my own ignorance here. While certainly it seems that many claim that FP states are specified by functional roles I cannot think of anyone off hand (maybe Jackson and Pettit 1990) who has claimed that ALL FP states are specified by functional roles that contribute to explaining and predicting behaviour.

    Actually, I can think of one reason why they would not just in virtue of the theory-theory. Typically theories of FP have focused on unobervables, which are posited for explanatory and predictive purposes. But it seems consciousness has more in common with observables, like limbs, than with unobservables (even if it is only “observed”, so to speak, in the first-person). This is because consciousness is something that stands in need of being explained. So just as scientific theories are concerned with both unobservable and observable phenomena FP has similar breadth.

    In short, I think you are right, but I see no reason why someone committed to the theory-theory has to disagree with you, if one takes the supposed analogy between science and folk theory seriously. Though perhaps it was only the functionalist thesis, and not the theory-theory itself, you were targeting?

    Study 5:

    Two worries: I am worried here that corporations might be problematic, since people often have the perception that corporations are bad (Enron, World Com, etc.). In which case, you might worry that they think it is not bad to kill the corporation because people think corporations are bad. The story could be about a club, or something more neutral. My second worry: the parts of the chair are not moral agents, where as the parts of the corporation are. Suppose that the sorceress enchants the parts of the chair, in order for them to make the chair, and afterwards, once it is destroyed, uses them to build other chairs. My hunch is people would find it less wrong if the parts of the chair were given more significance. My worry is that people see the chair more as a whole than the corporation because they see the individuals of the corporation as the moral agents, as well as the corporation itself: so rather than people showing less concern for the corporation than the chair because they posit the chair as having phenomenal states it is the fact that they see the parts of the corporation as moral agents who are not being wronged. This contrasts the chair case, where the parts of the chair do not constitute agents, and so any moral considerations apply solely to the chair.

    Study 6:

    Again, I am not sure about using the word ‘feeling’. It is easy to see people as associating this with pain, in which case, it seems obvious they would answer in the way they do. This is because it seems feelings, qua feelings, have a moral dimension to them: but it does not follow that FP conceptions of consciousness have a moral dimension. This is because feelings are not the only kind of phenomenal consciousness. What about the person wondering if fish can reason about things, or imagine things? Perhaps you will get the same results, perhaps not, but in order to claim that it is our FP concept of phenomenal consciousness with the moral dimension you cannot generalize from feelings alone. Also, I would be interested whether there was a difference if you were less vague about the person’s relationship to the fish. Suppose you asked responders to guess what his occupation was: I bet a large number would say he was a fisherman. If so, then it might be this implicit assumption that is guiding their answers about why he wants to know if the fish feel something. To test this, you might make the person a marine biologist, or just some person, who does not necessarily work with fish a lot.

  6. Thanks for the reply Joshua! I must say that I also agree with some of Brendan’s criticisms. It seems very important to me to balance the language in some way to avoid confounds.

    On the point in your response to me: I’m actually somewhat leery of thinking that entities have phenomenology to different degrees, that having phenomemenology is a vague matter like baldness or being tall — I just don’t feel I understand what the gray zone would be. (This could easily just be a lacuna in my understanding, rather than a sharp line in the world, though.) Maybe this is a good way to think about it: Different entities have different *types* of experiences (e.g., maybe fish feel pain but they don’t have visual imagination?), and responses ought to be sensitive to that.

    But the original thought in my remarks was more that the terms are fuzzy, and align along a continuum of phenomenality, and less that entities exist on a continuum in this way.

  7. Joshua Knobe says:


    You make a number of very cogent crticisms of our experiments here. I wonder now whether you might have any suggestions about how we might more effectively figure out whether people are willing to ascribe phenomenal states to group agents. Is there, for example, a sentence ascribing phenomenal states to Microsoft that people really might find acceptable? If so, I’d love to have a chance to try it out — either on human subjects or just by entering it into Google.


    I’ve been thinking more about your suggestion and recently tried taking a look at the original data to see whether it might provide any illumination here. The data actually show a striking pattern. Within each question, the distribution of responses is bimodal. In other words, very few subjects thnk that any particular mental state ascription has an ‘in between’ status. Instead, each subject tends to think that each ascription is either right or wrong — but there are some ascriptions that almost everyone regards as right, some that almost everyone regards as wrong, and some about which there is more disagreement.

    So I think you might really be right that some ascriptions sound more phenomenal than others, but perhaps any given subject treats any given ascription either as phenomenal or as non-phenomenal. Does that sound plausible?


    An interesting point! It’s quite striking that we are quick to blame corporations for their activities but that we do not blame people for the harms they do to corporations. Perhaps one could say, then, that corporations have ‘moral agency’ but lack ‘moral patiency.’ Then it seems that moral agency is associated with intentional states (beliefs, desires, etc.) And that moral patiency is associated with phenomenal states (emotions, sensations, etc.). So maybe the fact that people are willing to ascribe intentional states to corporations has something to do with the fact that we regard them as moral agents…


    Unfortunately, I haven’t yet read that paper, but I’m grateful for the tip!

  8. Thanks for your response, again, Joshua! My hunch is that the responses are bimodal not because of the intrinsic nature of the phenomena but because your questions were chosen so as strongly to suggest phenomenality or not to suggest it. Or at least maybe that’s the case. What about adding an intensifier to a belief attribution: Acme Corp “deeply believes”? Other kinds of mental state attributions that might generate intermediate results (some piloting might be necessary) are “Acme Corp has a hunch” or “Acme Corp is seeing the traffic on the freeways” or “Acme Corp is imagining what it would be like to take over Beta Corp”.

  9. Let me change the last to “Acme Corp is thinking about what it would be like to take over Beta Corp”.