Carrie Figdor --- Verbs and Minds

[New Waves in Philosophy of Mind]

Verbs and Minds

Carrie Figdor

In this paper I distinguish two tasks involved in the project of naturalizing the mind and suggest solutions to them. The metaphysical task is to provide an adequate metaphysical framework for the purpose of accommodating the mind within the natural sciences. The semantic task is to provide a naturalistic framework for explaining representational and experiential content. I defend an adverbial metaphysics in which mental kinds are species of general activity kinds, and show how Kriegel’s anchoring-instance model of content individuation is one way to address the semantic task given that metaphysical framework.

Video of talk: http://www.youtube.com/watch?v=LTYCHGWA1R4

A PDF of the paper is ready to view and download in the attachment below. 

A direct link to the PDF: http://goo.gl/Sb2Ou

[Justin Fisher]

Hi Carrie,

Very interesting paper. You say without much explanation on pages 4-5 that activities can't be individuated functionally, and that functionalism was (at least implicitly) committed to viewing functional realizers as objects rather than activities. Both these claims seem false to me, so I wanted to ask you about them.

Let's take the activity of straightening the shaft of an arrow. I'd be inclined to say that the activity of my whittling off this piece of this arrow could realize that general activity. Of course that general activity could also be realized in other ways, e.g., by my sanding off part of an arrow, or by my steaming a crooked arrow and clamping the now-pliable wood into a straight position. Isn't this an example of (multiple) realization of activities by activities?

Similarly, I thought that one of the classic examples of multiple realization was that pain could be realized by c-fibers *firing* or by pneumatic chambers *expanding*, etc... So it seems to me that, from the beginning, functionalists were open to the idea that functional realizers might be a form of activity, again contrary to what you say on page 4.

I also think there can also be kinds of activities that are functionally individuated. E.g., my whittling a stick counts as an activity of arrow-straightening precisely because of the role that this activity plays alongside other activities (e.g., crafting an arrowhead and affixing it to the stick) in over-all arrow production. [Probably it would be better to instead construe these teleofunctionally, as it's very often better to move from functionalism to teleofunctionalism, but that's getting us into issues orthogonal to your question of whether activities can be typed functionally/teleofunctionally.]

So, anyway, I wanted to invite you to say more about why you thought activities couldn't realize functional roles, and why there couldn't be functional kinds of activities.

Thanks!

-Justin

[Carrie Figdor]

Hi Justin,

Thanks for your post, and I'm looking forward to your paper once I get out from under a stack of papers.

One of the problems is that realization talk is too loose and certainly does not when characterized more precisely fit the genera-species relationship. Cats don't realize mammalhood. Mammalhood isn't a role to be filled/realized. It's a general kind and cathood is a species of that kind. Similarly, crimson doesn't realize red; it is a determinate of the determinable red. It is unfortunate that the relation of roles and realizers is assimilated to genera-species or determinable-determinate. They're just metaphysically distinct.

As for why activities aren't functionally individuated, I think I'll use an example suggested to me recently by someone who also raised this question. compare a raindance with a mousetrap. the latter is any device/object that fills the role. but the only thing that can be a raindance is a dance (a raindance is a species of the genera dance) and dancing is ... well, its an activity kind. whatever causes it or is a consequence of it at best specifies a context for its performance. what is a dance? that individuative question isn't answered by saying its whatever brings about rain. unlike a device of any sort that kills mice, a motion of any sort is not a dance, and so dancing has to be individuated in some other way.

moreover, while some activities may have been selected by evolution (as so have 'teleofunctions' in that sense, yet another sense of 'function'), just as biological objects have been, it seems that they are selected because they are the kind of activity they are (e.g., rotating).

I think activities can play roles, I just don't think they are individuated essentially by their roles; at best, the roles lead us to individuate a species of a general kind of activity (raindancing), but it's important to see that the nature of raindancing depends on the nature of dancing, while the nature of a device that kills mice is precisely NOT dependent on the nature of the device ('what it's made of'). In short, the distinction made in functionalism between what something is made of and what it does collapses when we are talking about doings -- a doing just is what is done. what kind is that? and to talk of what it is 'made of' is to talk of its steps.

  

[Mark Sprevak]

Hi Carrie!

I confess that I had the same worry here as Justin.

Maybe the following could provide a different response to him?

You could draw a distinction between two claims about X: (i) X is individuated in terms of causes and effects and (ii) X is multiply realisable.

Functionalism asserts (i) is true of mental states as a way of securing (ii).

The claim in the paper only concerns (i): activities are not individuated in terms of their typical causes and effects. You don’t mention multiple realisation at all.

Perhaps a way of keeping Justin and others on board would be to not rule out that activities can be multiply realised (as Justin’s examples suggest). Instead, just say that activities are not individuated in terms of their typical causes and effects. On such a view, activities would be multiply realisable for different reasons: e.g. because they involve abstract dynamical properties that can occur in a number of different ways.

This seems to be true of your example of diffusion. A common way of understanding this is that a material in a medium diffuses iff it satisfies the diffusion differential equation. The diffusion equation is an abstract dynamical description, which many different physical substrates can satisfy in different ways.

Maybe such cases are not truly speaking cases of realisation. But I suspect that insisting on this may prove a distraction for you. After all, realisation is a term of art—perhaps it does serve a purpose to include these kinds of cases in its heading.

Might this provide an easier line to defend?

[Mark Sprevak]

[Repost: originally posted in the wrong place]

Hi Carrie!

Great paper! I found it very stimulating and helpful!

Just a quick question on how to get a grip on the overall positive view. I kept wondering: what are the entities?

Let’s suppose we switch to the activity-based view. However, it seems that, even on this view, the mind can’t be all activities—there have to be some objects on which the activities operate—activities involve something doing an activity. A cell isn’t entirely activities—it is composed of entities (ions, proteins, etc.) and activities (diffusion, binding, etc.).

On the act-object view, we have a story that involves both entities and activities: the entities are information-bearers (the mental representations), and the activities are the attitudes and inferences in which the information-bearers participate. On the adverbial view, I wasn’t sure what the relevant entities are.

I guess that one answer that an adverbialist could give is to say that the act-object view is, in one sense, right—there are information bearers, and activities that those information-bearers enter into. However, the act-object view is wrong in a second sense: if one looks more closely at those information-bearers, one can see that they are also activities themselves.

Then the question arises of what are the entities undergoing activities in the case of the information-bearers. Is the right thing to say here that the entities are neurological or physical entities, and they are undergoing some information-bearing or semantic activity?

But then it doesn’t seem so different from the old act-object view. That view also said that information-bearers are neurological entities undergoing distinctively semantic activity. So I was a bit puzzled on what the contrast was between the two views—whether the act-object view really is committed to mental representations being entities in a different sense from the adverbialist.

[Justin Fisher]

Hi Carrie,

I think I agree that not every instance of genus/species or determinable/determinate is an instance of realization.  But that's compatible with holding (as many people do) that every instance of realization is an instance of a determinable/determinate relation.  So just saying that what I've given are examples of determinable/determinate relations isn't enough to show that they aren't instances of realization too.

What sort of determinable/determinate relation is the role/realizer relation?  I'd be tempted to say that it needs to be one in which the determinable (i.e. the role) involves some sort of characterization in terms of inputs and outputs and/or in terms of patterns of interaction with other components of some larger system.  This criterion seems broad enough to encompass most examples that people have labeled "realization", and yet not so broad as to include, e.g., biological species as "realizing" their genera nor crimson as "realizing" red.  As for activities, it seems like at least some activities *could* easily be characterized in terms of inputs and outputs (e.g., arrow making is an activity that takes in raw materials and generates arrows) and there are many cases where components of a larger activity are characterized in terms of the relations they bear to the other activities in a larger system (e.g., nominations plausibly can be characterized only in terms of their relation to elections).  So I don't see any strong reason to think that activities couldn't be relata in what look to me a helluva lot like realization relations (e.g., my putting that slip of paper in that hat *realizes* a nomination activity, as it plays the role of a nomination in relation to the other activities of my club).

Also, you didn't respond to a point I made above, which was that the paradigm example of realization is that of pain by c-fibers *firing*, which I take to be an activity.  So it seems to me that it will be very tough for you to argue that people have been committed to thinking that traditional realization talk precludes activities as realizers, when the paradigm example of such talk employs an activity as realizer.

Cheers,

-Justin

[Mark Sprevak]

Justin raises some great points.

Thinking more about it, if you wanted to go down the line I suggested, the claim (i) X is individuated in terms of causes and effects seems implausible to deny for every activity. The examples of activities individuated in terms of causes and effects that Justin gives pretty much establish this.

But maybe (i) can still be denied for some activities, and those are the kinds of activities you have in mind for naturalising the mind? 

For example, (i) seems false for diffusion. What makes something diffusion isn't the typical inputs and outputs of activity, but that the material in the physical system satisfies an abstract, high-level, description.

On the determinable/determinate issues, I think that the question of whether the the determinable/determinate relation is or isn't an instance realisation it's a bit of a distraction from your main point and just something that you're likely to just get flak on—I think you should remain neutral on it!

Hope that this helps!

[Carrie Figdor]

Hi Justin and Mark,

Give me a day or so to respond -- I'm also trying to do my duties to the others here, including Justin.

These are good and helpful. I think I'll end up converting you both though. ;-)  

Carrie

[Phil Woodward]

Carrie--

I have three questions, all of which may be entirely clarificatory.

(1) On the one hand, you say that there is some sense in which activities wear their natures on their sleeves. I think your point is that entities at lots of different metaphysical levels can be said, quite literally, to perform activities of the same general type. (Certainly there are some descriptions under which activities have a hidden essence; you use the example of waving/hailing a taxi--but isn't it possible to recognize an instance of waving without knowing that it is an instance of waving a taxi? This looks to be an example of an posteriori identification, i.e. an identification of an activity with its nature exposed, on the one hand, with an activity with its nature hidden, on the other.) But later you say, "we cannot take for granted that the ways we use verbs to pick out mental activities captures what, if anything, they have in common," which suggests that their natures may very well be hidden--at least, it may be hidden whether their natures constitutively involve external relations. I'm not understanding how these claims hold together.

(2) I'm not sure I understand what your ambitions are in this paper with respect to defending adverbialism's naturalistic prospects. This is my best attempt at reconstructing the dialectic: Kriegel takes it that adverbialism is straightforwardly unnaturalizable because adverbial modifications are not external relations, and any naturalistic semantic theory will appeal to external relations. You respond that adverbialism is actually compatible with some activities' constitutively involving external relations, so Kriegel's judgment is too hasty. So far, so good, but at points in the paper I wondered if you were trying to go further, viz. to provide a strategy for naturalizing content-modified activities. But isn't Kriegel correct in assuming that the only naturalistic semantic theories are those that make central use of external relations? If so, then adverbialism is only naturalizable if paradigmatic contentful mental states do constitutively involve external relations. If the relevant content-modified activities are of a piece with rotating, rather than with hitting, then abverbialism's naturalistic prospects are shot--or so it seems to me, and I can't tell if you would concur or not.

(3) Related, I think, to Mark's question about what distinguishes your view from act/object functionalism, I found myself wondering what it means to "manipulate activities" (per p. 6) if not to manipulate entities that perform semantic activities. (I think I know what it means to manipulate a rotation, but that's to change what sort of determinate activity it is; but I'm not sure how to fold such a notion into a broader story about information-processing.)

Phil Woodward

[Carrie Figdor]

Hi Mark and Justin,

Reading your comments makes me realize how far I have traveled in thinking about these issues to get to where I am, and how much I need to do to bring people to where I am now. So these are helpful reminders. That said, much of your concerns here are really not concerns of this paper at all, but of another paper which is prior to this one where I discuss functional individuation at length. This paper is the next step in the larger project. So basically you are responding to the prior paper. I'm tempted to just send you the other paper, but I'll try to answer your questions anyway without it. And I'll be happy to send the other paper and point you to the relevant pages -- I'd be happy to get comments on that one too.

1. When I talk about functionalism I mean as a theory about how mental activities are individuated. If you look at the Stanford Encyclopedia of Philosophy you will see a standard definition of functionalism that captures the core idea of functionalist (functional role, not teleological-selection) individuation. Functionalist individuation is not a mind-body theory, however. It's a theory of how certain kinds are individuated, period. It's compatible with the identity theory. And I'm saying that this theory of individuation is inapt for activities, in particular for the purposes of mechanistic explanation in science. It presupposes the individuation of activity kinds in some other way (since functional roles depend essentially on prior activity-kind individuation) and it asserts the individuative priority of the activity kinds in the roles over the activity kind to which a token occurrent already belongs (just as what an object is "made of" does not matter or at best recedes to secondary status when individuating objects by functional role -- that's what the famous distinction between what a thing is and what it does amounts to for objects. For occurrents, it is the distinction between its intrinsic activity kind and the role it plays -- that is, between doings. In other words, functionalism claims that mental types are networks of doings, so if mental tokens are activities then what fills the roles are doings, and what an occurrent does intrinsically (what it is "made of") is not essential to what it is; instead what it does extrinsically (the activities that comprise the role) are essential to its being the kind of activity it is. If functionalism is the right theory for mental types, then to say a mental token is an occurrent that realizes or fills a particular functional role is to say that the kind to which it belongs intrinsically is metaphysically subordinate to the network of occurrent kinds that comprise the functional role. When it comes to doings, the token both is and is not individuated by what it does without contradiction because individuating an occurrent by what it does is ambiguous, and functionalism resolves the ambiguity via a claim of individuative priority for the activity kinds that are in the role rather than the activity kind to which the occurrent belongs intrinsically. But how are all these doings individuated? How can functionalism possibly help answer this question? This is the critical question we are motivated to face when mental phenomena are ontologically classified as activities and we take that claim seriously. (See Machamer, Darden and Craver on this too; I defend their view in my prior paper.) Of course a token of (say) transmitting can play a role in the brain (neurotransmission); but had a token of some other kind of activity filled the role, it would be a different activity (neuro-diffusion, say) -- this is ruled out by functionalism, however. It doesn't work for activity kinds; their intrinsic kind never disappears. Instead, the role (as well as the kinds of the entities that may perform the activity) may provide reasons for individuating species of the same general activity type. 

In this paper, I try to move forward a little bit on how to assimilate this metaphysics with the structure of mental representation-talk that permeates philosophy of mind and cognitive science. I don't give that up; I'm trying to reformulate it into a consistent naturalistic metaphysics.

2. The best evidence I have at the moment that mental representations are currently conceived in object terms is circumstantial, but collectively strong. The metaphors that are employed (putting items in belief boxes, homunculi) involve objects. Functional individuation works nicely for (artifact) objects, which are the invariable examples. And, lastly, if the mind were really, genuinely conceived of in activity terms (not just as lip service or 'why couldn't it be?", to which I say, it could and should!), there would be a huge literature by now about how to individuate activity kinds (e.g. what is it to diffuse? To encode? To rotate? etc. etc.), with mental activity kinds as a special case. But there isn't. The absence of work here is overwhelming. I of course am planning much more work on that (in the next step, not this one, where I only gesture). A common response I get to my attempts to motivate this work is that it is "interesting and neglected". The simplest reason for the neglect is that we don't really fundamentally think of mental phenomena as occurrents/activities/doings.

3. I don't mention multiple realization anywhere in the paper. It's just orthogonal to my concerns, along with the vast literature on multiple realization. Mark draws a distinction between functional role individuation and multiple realization, and that's fine and his general perspective seems to capture what I'm trying to say. But the simple response is just that MR has no role to play in anything I'm defending. Functional individuation is compatible with unique realization, if that matters, but it doesn't. Denying that mental kinds are functional kinds (unless one means "activity kinds", but that would be to surreptitiously adopt my view) does not rule out that there can be various complex activity kinds that can be composed of distinct kinds. Dual process theories (processes!) in general claim that some complex activity can be performed in various ways. This is true however you think activities are individuated, and whether you think mental kinds are a special case of activity kinds.

4. Just to beat a dead horse (albeit one that seems to have an infinite number of lives), c-fibers firing includes "firing" as an activity-kind that is filling a role of pain -- it is part of a mechanistic explanation of what goes on inside our brains when we are detecting harmful stimuli. The biology of pain is pretty complex so it's difficult to take the example seriously. But here's the point: what is firing? Firing isn't what it is because it is filling a role; it is because it is firing that it can fill the role, and if a token of another kind of activity filled the role it would be that kind of activity. The role doesn't individuate the firing token (which is what functionalism claims). Now if firing were a token of a mental activity kind, the same response would apply. Now you might say: yes, but the mental kind is the role! Firing is a physical activity kind! Well, this is a little progress. How is that activity kind individuated? Good question. (See above.) But now what is the argument for saying that mental activity kinds aren't on a par -- that they have tokens just as the physical activity kind firing has individual firings as its tokens? I know, I know: we have a history of Davidson and "Mental Events" to deal with. Which I do in my prior paper. The basic idea is that Davidson was right about quantifying over (token) events (he said nothing about how event kinds are individuated, he took all that for granted) but wrong to suggest that mental events were just properties of physical events.

[Carrie Figdor]

Phil,

Thank you immensely for your comments -- requests for clarification are perfect for me right now with this project. I'll do the best I can and let me know if this helps or not (and what other issues arise ....):

1. Activities 'wearing their natures on their sleeves": I do want to distinguish the metaphysics of activities from what we recognize (or know, or how we learn, etc.) -- i.e. the metaphysics of activities from the epistemology of how we know them. I think in many case we know them basically by acquaintance, from exemplars. I also think we can observe an activity without recognizing the activities which it generates in the right context (i.e. which other descriptions of it are true). My remark about mental activities reflects the following: I want to advocate for the view that activity kinds that we identify as mental are special cases of activity kinds that need not be mental -- recognizing is something an antibody can literally do even if it is not recognizing that P, and even if our understanding of recognizing starts from our acquaintance with exemplars of recognizing by people (including ourselves). It doesn't follow that recognizing has a hidden nature -- human instances of recognizing may capture its nature just as we can observe what makes a cat a mammal without supposing that all mammals are cats. Recognizing may have parts (other activities) which comprise it (e.g., encoding?) but some of those parts may be of the same activity kind (just as adding a large sum will include adding as steps. This is metaphysically and epistemologically very interesting. I should add that I'm not against hidden natures for activities, and it wouldn't matter to me if there were some. I'm trying to explore these differences and how they affect the project of mechanistically explaining the mind (e.g. naturalizing the mind). It does strike me as interesting that rotating nuclei will be parts of rotating tops -- the top is composed of atoms (partly composed of nuclei) and their activities and relations, and the rotating occurs and is explanatory at different material levels of composition. That tells me that whatever rotating is, it doesn't have to be composed of some other kind of activity. In fact, it seems incorrect to even classify it as complex or simple, unless we classify it as such relative to the complexity or simplicity (mereologically speaking) of the entities performing it. 

2. Naturalizing: yes, these are my first attempts to wrestle with the issue of content and how my framework accommodates it. My basic line is that while it's true that naturalistic theories (well, causal-informational ones) rely on external relations, it's also true that they start hand-waving with all the cases of content that do not involve existing external objects. So the demand that naturalistic theories of content must involve external relations is rather strange: since these theories are clearly (at best) incomplete, it's hard to say why we are forced to accept that they tell the whole story and the only story. But even if we just say, OK, let's set aside all content that can't be comfortably explained via external relations (!!!!). Can my framework accommodate at least those? And my thought is, why not? It is open to the individuation of some activity kinds partly in terms of external relations (e.g. hitting as an intuitive crutch here), and for all I know all propositional attitudes (special cases of activity kinds that are not essentially mental, see above) require individuation in terms of external relations. I haven't dealt with issues of compositionality and many other content-related problems. But the point is that the adverbial framework does not preclude or require any particular way of individuating activities -- it's neutral, or more precisely, that's not something to be determined a priori. 

3. Manipulating activities: when you adjust a faucet you are manipulating an activity; you are changing the flowing of the water. It flows quickly or slowly. Manipulating activities is just fancy talk for acting (or doing) in various ways, including acting on other actings; I use the fancy talk because it's how philosophers of mind routinely talk about the mind: manipulating mental representations. This is more fodder, come to think of it, in my response to Mark and Justin above about what distinguishes my view -- if you can't think naturally of what it means to manipulate activities when we talk of manipulating mental representations, that indicates another way in which we don't really take seriously the idea of the mind as (complex) activity, and that we implicitly think of mental representations as objects (the sorts of things we can put in belief boxes, for example).

4. the entities: I think I forgot this one above. The entities get all the attention, so I really haven't bothered talking about them. I'm for the underdog here. So, what entities are involved? Well, neurons, glia, Hebbian synapses or neural networks if they are reified, cells, etc. -- I'm happy to adopt whatever entities the various sciences posit. 

I hope that's helpful and I appreciate your close reading of the paper. Perhaps we can continue the conversation (Mark and Justin too).