Monday, March 14, 2011

Russellian Monism and Dispositional/Categorical Properties

{Note: this is a draft of some work that I might develop further with add'l research at some point. Comments or suggestions are welcome.}

I’m gratified that the position in Philosophy of Mind known as Russellian Monism (also known as Russellian theory of mind and probably the best developed account of neutral monism) has gotten more attention in recent years. However, the terminology typically used to describe the position today is different from Bertrand Russell’s, as presented in his 1927 work, The Analysis of Matter. This post discusses some of the issues involved, and briefly looks at how some stances in contemporary debates would fit with the original account.

In a recent post on the Brains blog, Richard Brown (referencing an online discussion he had with David Chalmers) said: “RM [Russellian Monism] is the view that the dispositional properties talked about by physics have as their categorical base phenomenal or protophenomenal properties.” While descriptions vary, the reference to dispositional and categorical properties is common. In his book, Ignorance and Imagination, Daniel Stoljar says the position is a combination of two theses. First: “…that physical theory tells us only about dispositional properties.” And: “The second thesis we need to consider is that the dispositional properties of physical objects do require categorical grounds; that is, for all dispositional properties, there must be a non-dispositional property... (p.110)”

Now, Russell never uses discusses properties at all, and certainly not dispositional or categorical properties specifically! These are terms which have emerged in the more recent debates of analytic philosophy. So, how well is the intent of RM captured when using this terminology? (Please again note I’m only speaking of Russell’s work in The Analysis of Matter).

Brief summary of RM

Russell’s project is to show how careful analysis of the formulation of physical theory can reveal a common framework connecting what we think of as the physical and mental realms. First, he argues that the subject matter of physics can be interpreted as an abstract description of events and their linkage in causal relations. He then argues that the mental realm can likewise be described in terms of events, and that given a causal theory of perception, we can view perceptual events (“percepts”) as connecting with physical events. And a key point is that our knowledge of the physical events (which is inferred and ultimately derived from observations) includes nothing which is known to be inconsistent with the mental.

Russell speaks of events (or groups of events) as having “intrinsic qualities” or “intrinsic character”. These are things which are known to be an aspect of percepts, but are not part of physics, given its abstract structure. He doesn’t assert that physical events must have qualities like those of percepts – he is agnostic -- but he argues there’s no reason they couldn’t.

One thing I would note before discussing the recasting of RM in terms of properties. Russell is very insistent on an event ontology, rather than one of objects or substance. This distinction is usually ignored in contemporary discussions of RM. I’ll assume for present purposes that an event can be seen as something which either bears properties or is itself a bundle of properties, in the manner of objects, although there may be some issues with this assumption.

Dispositional and Categorical Properties

The topic of the nature and roles of dispositional properties (or dispositions -- also known as powers) and categorical properties is complex, and there is an extensive literature that concerns many aspects of metaphysics. But it seems fairly clear why philosophers have connected these concepts to RM. Dispositional properties (roughly) are those that when possessed, like the classic examples of fragility or solubility, direct its bearer toward some manifestation. A categorical property is (roughly) something possessed by an object which need not be connected to such manifestations. Dispositions seem to be closely associated with  physical interactions and causation, while categorical properties appear to offer an affinity to “intrinsic qualities”.

So, if physical theories only describe occurences and causal relations (as in Russell), and these are due to dispositional properties, then dispositional properties are the content of physics. If categorical properties are intrinsic qualities (as in Russell), then these are properties which are not captured by formal physical theory; they are, on the other hand, things which we still might be acquainted with (in some fashion) via perception. At a minimum, they comprise a part of the world which could play a constituting role for mental phenomena.  So this sort of thinking likely gives rise to the typical contemporary description of RM we began with above.

Does this description of RM reintroduce dualism?

However, there are a number of nuances and controversies regarding the status of categorical and dispositional properties, and there might be danger that certain stances on these might lead to interpretations inconsistent with RM.  For instance, do categorical properties play a supporting role in causation as well, or not?  If the latter, does this lead to epiphenomenalism with regard to the mental?  If the former, what is the basis for ascribing such a role?  Cannot dispositions fully account for causes?  I’m not able to delve into all of the issues in detail. But I have one especially pointed concern: given Russell’s goal of monism, do we run the risk of re-introducing a variety of dualism in the form of these two sorts of properties.  If dispositions do the "physical" work in the ontology and categorical properties do the "mental" work, isn't this just a "property dualism" in the context of Philosophy of Mind?  With this concern in mind, I want to conclude by briefly looking at contemporary stances which seek to eliminate one of these property types, and see how they fit with the goals of RM.

1. Are dispositions fundamental, or are they reducible?

Some philosophers argue that dispositions are not fundamental and can be reduced to categorical properties.  For instance, there has been an extended debate over whether the ability to describe dispositions in terms of conditional statements means they are amenable to reduction.  David Lewis argued for such a reduction as part of his metaphysical program (sometimes called Humean superveniece). However, Lewis' theory effectively eliminated causation itself as a fundamental aspect of reality (it just reflects regularities in a static mosaic of qualities). There are other avenues to support real causation without dispositions, but for present purposes I conclude that, since Russell was working with an assumption of causal realism, "disposing of dispositions" wouldn't be appropriate as part of a property-based description of RM.

2. Alternatively, can we do without categorical properties?

The key issue here is “intrinsic qualities”. RM relies on the assumption that events have intrinsic qualities.  If you eliminate categorical properties – perhaps by assuming that dispositions are identical with their bases, or don’t need bases at all -- you might risk losing these qualities, which are traditionally associated with the categorical side of things.  Many philosophers have argued that a world of "bare" dispositions isn't coherent, and that they need a categorical base for support.  This is debatable.  But for present purposes, the issue is simply that a world of “bare” dispositions lacking qualities (and adequately described by physical theory, at least in principle) would not be consistent with RM.

 However, there is another option available for a "disposition-only" ontology.  Some philosophers, notably John Heil and C.B.Martin, identify qualities with dispositions (they are another aspect of the same property). This move retains consistency with RM. In fact this view might be a particularly good fit when combined with the notion that we only encounter the qualitative side of causation when in direct causal connection with something.  Perception detects this aspect, at least in some fashion, while the abstracting process of physics has no place for it.  Note, in constrast, that if we maintained the dual property picture, where qualities are distinct from dispositions, then the reasons for their being “on display” when we are in causal contact are less obvious.

So, given the aims of Russellian monism, the stance in modern debates over the role and nature of dispositional and categorical properties which offers the best fit is the Heil/Martin view -- sometimes described as an ontology of "powerful qualities".


Jonathan Edwards said...

Hi Steve,
A friend was defending Russell last Friday and I began to think my own, modest, reservations on RM might be misjudged. I Googled and found your March post at the top. I share your gratification at an increased profile for RM if such is the case. I was even more gratified to find your helpful unpicking of interpretations of Russell.

I had come to the view that Russell was wrong to make qualitative features intrinsic because they ought to be relational. Conversely dispositional properties, like negative charge, are in a sense intrinsic, arguably independent of what they might relate to (repulsion on the other hand requires relation to another negative charge). I was also worried that experiential features should be ‘categorical’, in hypostasis to dispositions. Rather I see experience as reifying mathematical structure of dispositions by being what dispositions are (even if indirectly) to.

There is a lot of potential for getting tied in knots and I think you give some good reasons why traditional philosophical usages like ‘property’, ‘intrinsic’, categorical’ may be unhelpful. Moreover, you point out that it is critical whether ‘mental features’ are ascribed to substances or events and whether events are relational or unitary and indivisible. Whitehead seems clear on this but is Russell? He says he is talking of events yet he describes: ‘the material unit as a causal line, ie as a series of events connected with each other by an intrinsic differential causal law which determines first order changes, leaving second order changes to be determined by extrinsic causal laws.’ He suggests that a photon could be a good example. Photons may not be enduring but they seem more like substances than events. (Perhaps they are easy because they are easily seen as both.)

All this could be discussed at length but to cut to the chase I have come to think that experience must relate both to a ‘substance’ and to an event in an odd ‘dative’ way. The nice quality of a present given by Mary to John is dependent on the giving event and on both Mary’s giving dispositions and John’s receiving and appreciating dispositions (the former maybe like him being the same or opposite charge as the present). Moreover, QM may help flesh out detail, since an indivisible event with things compresent would seem to need to be at this level.

If our event is an x-ray photon going through, and being diffracted by, a protein crystal then we have a relation between one thing, the photon B, and a pattern of other things, the protein atoms Σ A(1,…n). The experience relates to the interaction B. Σ A(1,…n) but there are important assymetries. The domain of the event is defined by the life history of B, not that of the As. We could expect the pattern of As to form an experience for B, but experiences for As would be likely to relate to quite different, but overlapping, events A. Σ B(1,…n). Moreover, we would expect in the event B. Σ A(1,…n) B to be potential and indeterminate but each A to be actual and determinate – along the lines of determinacy or actuality occurring each time one Whiteheadian occasion is taken up into the next.

The end result is the slightly surprising idea that an experience is a feature (best not say property) of an event B. Σ A(1,…n) and is an experience for B but is an experience of As. Is the ‘quality’ of B or of As? The traditional conceptual framework seems to fail totally here. The new formulation, however, is close to a common sense microcosm of how we normally see the world.

The importance of all this for me is that it is relevant to trying to work out what event in the brain will host experience. Russell says that it will have to be very local, to ensure compresence, with which I strongly agree, but does not seem to distinguish the total event B. Σ A(1,…n) from the protagonist As and B. As I see it we can only bypass the combination problem if we can find a domain that sits naturally on a particular 'biography of B' in a brain.

Jonathan Edwards

Steve Esser said...

Hi Jonathan. Thanks for reading the post and sharing some of your thoughts. In addition to thinking about where qualities reside, you've been working on the further difficult problem of where to locate and how to characterize experience when there's an interaction. I will think about what you say here.

To make one comment, I've been thinking experience might have something to do with composition of a more complex "substance" or entity -- so the photon (assuming it's simple and not complex) would carry quality but might not be a subject of experience.

Jonathan Edwards said...

Hi Steve,

I guess Russell and Whitehead started from Leibniz and it is easiest to do the same. So we want complexity within unity, or Russellian ‘compresence’. ‘Composition’ of a substance seems to me to run into the problem of aggregates and of course Russell holds that the units are ‘events’ rather than substances. All the physical features we have to define these events are dispositional and relational, and to instantiate relational dispositions I think you need interactions. So for me the natural solution is the one Leibniz takes – that complexity within unity is not based on complex internal composition of a substance so much as on complexity of interaction with the universe. We are looking for one-to-everything else complex interactional events. QM tells us that these get more complex as you get smaller, and are as complex for a photon as for anything. Moreover, it gives us rules for indivisibility (of interaction), as Leibniz wanted.

That said, I agree that for a human subject we tend to think we need something with more complex structural parameters than a minimal complex harmonic oscillation. I tend to think in terms of more exotic biological examples of the sort of complex quantised modes of energy that inhabit things like ice crystals – with complex symmetries but no actual parts (even if the crystal itself seems to have parts). Nevertheless, for the purposes of the essential abstract arguments, I am not sure that a photon would not do. It can have a hugely complex event as its autobiography. The layman’s sense that we ought to be involving fairly large networks of brain cells does not have any justification to my mind.

However, the quote from Russell I gave last time seems a bit of a muddle and I think this is where the trouble starts. He talks of the unit being a causal line connected by an instance of a causal law that determines ‘first order changes’ with extrinsic laws doing the second order stuff. It is as if he wants the 'event' (maybe photon) to do its ‘eventing’ on its own, with just a nod to the universe, keeping its qualities to itself like a child jealous of her packet of sweets. However, physical, and especially QM, laws are about relational dispositions. The photon has no autobiography, indeed no life, other than the unfolding of its disposition to interact with the world. And if the substance is more complex in structure surely its experience, to be causally relevant, must be similarly about the interactions that are the operation of its dispositions rather than just the abstract mathematical structure of its dispositional parameters.

In short, I think Russell wants to have his event cake and still eat his substance because he has convinced himself to reject the idea of a human subject as a receptive ‘ego’. For me, Leibniz gave a better account. We just need to update some of his stipulations about the parameters of monads. And I think Leibniz, contra Whitehead, is right to deny that one can solve the combination problem. The only strategy for me is one that invokes no combination – but I think complexity of autobiography, or for Leibniz apperception, delivers the goods well enough.

Best wishes
Jonathan Edwards

Steve Esser said...

Thanks for the thoughts and the monad advocacy in particular. That's good to revisit.

Jonathan Edwards said...

P.S. I note I have made dispositions both intrinsic and relational and I think maybe the trip-up just reflects the fact that for monads there is no real distinction. The relation of supposed hypostatic categoric properties to dispositions like fragility may just be the relation of the monads to the aggregate? The vase is fragile because its monads are working away relating all the time according to micro-dispositions that are equally manifest in non-breakage as in breakage.

I will let you think in peace.