Monday, December 13, 2004

Rosenberg, Consciousness & Causality (Part 1)

Philosopher Gregg Rosenberg has written a book called A Place for Consciousness: Probing the Deep Structure of the Natural World. The book offers an ambitious new metaphysical proposal for understanding the natural world. It does this by exploring the deep connection between the philosophical problems of consciousness and causality, and then offering a thorough and detailed model for addressing both.

The outline of the book is as follows: first Rosenberg offers his take on the problem of consciousness in the context of contemporary philosophy of mind. Toward the end of this discussion he foreshadows how the issues which need to be addressed in this area connect to the challenges of understanding causality. He then shifts gears to critique past accounts of causality and present his own solution. Finally, he shows the connection between consciousness and causality and how to improve our understanding of both through a unified approach.

From the perspective of a lay reader, I would say that the more background reading you’ve done on these topics, the better you will understand the book. Between my first reading over the summer and my second try recently I read more about causality and this helped. But at the same time, I think there are so many good ideas in the book that I would recommend it to anyone, even if you end up skimming some parts.

It has been somewhat a revelation to me this year to realize the degree to which causality had still posed such a philosophical challenge. We are led to believe that the type of physical theories we have are also good objective causal explanations, but they are not. In showing how the challenges of understanding consciousness and causality are linked and making a proposal for a unified solution, Rosenberg’s book should make it extremely difficult for the reader to consider either topic in isolation from the other going forward.

Below I give a chapter by chapter summary derived from my notes on the book; please note that I can’t claim to be doing justice to the actual arguments here. I will follow this post up with another one containing some concluding thoughts and outstanding questions.
In an introductory chapter, Rosenberg outlines his agenda to provide a place for consciousness within nature. He introduces the term “liberal naturalism” to describe a perspective which looks to uncover some deeper aspects of nature which go beyond physicalism or materialism.

Chapter 2 presents a discussion of the arguments against physicalism. He presents his favorite critique, which builds on ideas which go back to Bertrand Russell and A.N. Whitehead. Physical theories describe quantitative differences and relations, but say nothing about the intrinsic, qualitative nature of being. Conscious experience is constituted by such qualities, and physicalism thus fails to explain conscious experience. As a thought experiment to support this argument, Rosenberg discusses the toy world of the 2-dimensional cellular automaton called the “game of life”. He shows how this toy world is built up from “bare differences” which do not reflect any intrinsic qualities. He then argues that the more complicated physical descriptions of our real world likewise share this shortcoming. I tend to like this argument, and agree with Rosenberg that it is stronger than the “knowledge” argument against physicalism (which, in turn, is stronger than the “conceivability” arguments).
He next spends a chapter detailing and responding to what he sees as physicalist challenges to this argument. Here, I found the discussion rather dense without much additional value-added, and I will be interested to look for responses to the arguments from other philosophers.

Chapter 4 sets out the boundary problem for consciousness. Here Rosenberg discusses something we take for granted – which is the unified and bounded nature of our field of conscious experience. He argues that experience is something which need not necessarily be this well contained. What do we make of the existence of multiple personality disorders? Do sub-components of the brain have any sort of experience? (BTW, see this recent post in Desert Landscapes). This argument prepares us for the discussion of panexperientialism in the next chapters (5&6). By starting with the boundary problem, Rosenberg also wisely introduces at the outset the challenge for panexperientialists (and more modest dual-aspect physicalists) which is usually termed the combination problem. If conscious experience pervades nature, why do certain organized systems have a coherent conscious field (like us) while others appear not to (rocks, galaxies)?

Next, Rosenberg argues the case for panexperientialism while distinguishing it from panpsychism -- the slogan is that experience “outruns” cognition in nature. How can one build consciousness out of parts which lack subjective experience? It is difficult to credit that subjective experience itself could be emergent (see my previous post), although we can more easily understand that certain cognitive capabilities probably are. Anyone who argues experience is limited only to humans (or to suitably cognitive systems) needs to explain why: Rosenberg reviews the candidate explanations for tying experience uniquely to our cognitive systems (arguments from “complexity”, functional arguments, and arguments from biology) and finds them lacking. Panexperientialism is the remaining solution standing.

Still, Rosenberg realizes that adopting a panexperientialist viewpoint leaves us with some key questions and problems (reviewed in chapter 7). Again, accounting for the binding of the perceptual field in space and time remains an issue. But an even bigger problem is that experience seems epiphenomenal. If we assume physics provides a full causal explanation of nature (the causal closure assumption), then if we also conclude that conscious experience exists we necessarily find that it accompanies the physical world, but doesn’t impact it in anyway. Now we know interactionist dualist models make no sense, but a model of the world which says experience and physics somehow move in parallel fashion but never affect each other is also deeply unsatisfying. It seems at this stage we are missing something. The time has come in Rosenberg’s book to shift over to a discussion of causality with the promise of returning to the problem of epiphenomenalism better equipped.

In the second (and longer) part of the book, Rosenberg takes on causality. He starts (chapter 8) by reviewing critiques of the Humean view that what we see as causality is really just a pattern of regularities, not real causal connections based on dependency, constraint or production between events. These critiques include the fact that the Humean view undermines the existence of explanatory physical laws and the fact that the view must presuppose some kind of unity of the world, but it is the assumption of real causal closure which in practice provides the basis for the unity we assume. The Humean view also fails to objectively explain the temporal asymmetry we observe, which seems linked to “real” causality. Finally, Rosenberg asserts that the view has some epistemic problems which lead to extreme skepticism.

So, we need “real” causation (onto chapter 9). If we go back to physics, Rosenberg argues we will find part of the story, but not the whole thing. Physics underexplains causality. Rosenberg argues that association and correlation are shown by physical laws, but while we usually assume they explain causal connections, they really don’t. There is no causal dependency or production going from the state of the system at one point to the state at the next point. One can pretty easily take a Humean view of these laws, but we’ve discussed the problem with taking that route. (I would think that the fact that most laws are time-symmetric lends credence to this argument, but Rosenberg doesn’t use this point, and I infer he thinks the point would be valid even if laws weren’t time-symmetric).

Next, Rosenberg looks at philosophers’ efforts at non-Humean theories of causation, which he specifically classifies as theories of causal responsibility. He thinks these theories get off the wrong foot, because on examination they are not fully objective theories, but somehow fall back on intentional or interest-relative components. (I have thought it is interesting that the implied existence of a subjective point of view is so hard to remove from causation). To get more objective, Rosenberg moves to what he calls a theory of causal significance, which is simpler and has less baggage, compared to causal responsibility. Causal significance focuses on the constraints a thing places on how the world can be, rather than the effects it produces. He frames the difference as one between asking “Why is there something rather than everything?” instead of the usual “Why is there something rather than nothing?”

What types of properties must things possess for there to be causal significance? In a key move, Rosenberg proposes there must be two distinct (though interdependent) types of properties, called effective properties and receptive properties. Effective properties are the sort we usually picture as having the ability to impact other things (and ultimately on us as we investigate the world). It is argued that physics describes these effective properties. But effective properties alone cannot do the full job of real causation. A property of a thing can only be effective if some other thing is receptive to the property’s presence. You can’t have one without the other. And it is this necessary role for receptive properties that other theories have missed.

Further, receptivity is a connective property (and it can be non-local, such as in the case of entangled particles in a coherent quantum system). Importantly, in this model the existence of the receptive pole of causality seems to offer what I would call an extra “degree of freedom” which allows connections to exist at multiple levels of nature, which becomes very important in this overall system. Rosenberg argues that not all causation needs to be at the micro-level of nature, as physicalism assumes. (But the model is not in my view simply asserting some sort of “downward causation”; rather it is a more of a simultaneous multi-level causation). Receptivity can be a shared attribute among multiple systems. This leads to a higher-level causal nexus. In a sense, I interpret receptivity to be something which can provide qualities of organization or participation to a higher level system.

Rosenberg spends several pages detailing this model with notation and diagrams. A few important themes from the later part of the chapter:
When effective and receptive properties are bound together, it creates a natural individual in the world;
The ontology of this model can be considered an event ontology, where the actualization of the natural individual is the event (determination is an actualization);
Individuals seek completion (the filling of their receptive “slots” with effective properties) – this adds a bit of teleology to the model.

Chapter 10 goes over the model again with a toy universe where charge (+ or -) is the only effective property. Importantly, Rosenberg shows how at higher levels of the model, new kinds of laws emerge. He calls these strongly emergent laws, but I’m not sure why—I usually associate the modifier “strongly” with ontological emergence, which I don’t think is what is happening here.

Important ideas from the chapter:
The world as described by the causal significance model is a causal mesh. Causal significance is a system of constraint, and thus implies that an actual world is emerging from a possibility space which itself exists. It is a type of realism with two modes of existence, the possible and the actual; “…causation has no work to do unless there exist real alternatives to actuality.” This metaphysical perspective is compared to the world of quantum mechanics, and to Whitehead. An ingression is a (atemporal) movement from possible to actual.
Late in chapter 10, Rosenberg takes a stab at showing how space and time themselves could emerge from an underlying causal mesh which featured these ingressions. I plan to come back to this topic in my next post.

In chapter 11, Rosenberg returns to the theme that physics underexplains causality. He uses the example of quantum entanglement to show evidence of receptive binding: the system cannot be captured simply from the particles’ physical (effective) properties of mass, spin, charge. He also revisits the game of life world to make the point that physics designates the nature and the regular behaviors of effective properties, but the receptive structure of the world does not ontologically supervene on the facts about low-level effective properties. It is shown by the fact that different sorts of receptive connections could dovetail with the same physical facts. The physical facts suggest the structure of receptive facts, but don’t explain them.

Now it is time to start connecting back to consciousness. In Ch. 12, Rosenberg takes the next key step in his story by introducing the “carrier theory of causation”. Returning to themes from his anti-physicalism arguments in chapter 2, he discusses how physical systems need to be instantiated in the real world; an example is the instantiation of the game of life on a checkerboard or computer. The qualities of the checkerboard or computer “outrun” or are “extrinsic” to the system. If you look at physics, the same thing is true. Physical concepts are circular- they are difference relations which don’t “sit on” anything else. There must be a wider system of properties on which these differences and relations are instantiated. These properties are called carriers. He then argues that phenomenal properties are perfect candidates to be carriers. Phenomenal properties are differentiated yet qualitative and are extrinsic within a system (their nature is not exhausted by the difference relations in the system). So, it is postulated that phenomenal properties are the carriers of the effective properties described by physics.

So what carries receptive properties? An experiential property. Experiencing carries receptivity. Putting it together, a natural individual is one which experiences phenomenal properties. Each event is an individual experience of phenomenal properties. This is a panexperientialist model, where each event in the world is proto-conscious, by virtue of having some sort of experience.

Finally, how do we build up from this to explain human consciousness? Well, each consciousness is a cognitively structured high-level individual with an experiential receptive field.

In Ch. 13, this model is tested for how it addresses problems from recent work in the philosophy of mind. The status of knowledge is discussed: here it is argued that the binding of causal connections provides a kind of direct acquaintance with the world. It is argued that the consciousness model addresses both the binding and epiphenomenon issues from chapter 7.

Other applications of the model to commonly discussed problems follow in chapter 14, including additional discussion of emergence, and the relationship between this model and functionalism. While most forms of functionalism are false, there is a consistency with a more modest sort of non-reductive functionalism. Additional sections address the subjective flow of time and possible contributions to cognitive neuroscience. These sections are brief pointers toward possible future work.

But the real substance of the book is metaphysics, and this is where it makes such a valuable contribution. Chapter 15 is a brief summary of what the theory means for how we should view the natural world and the individuals in it.


Giordano Sagredo said...

Thanks for summarizing it. It is helpful.

So when a billiard ball hits another billiard ball such that the hit ball moves into the corner pocket, there is a phenomenal experience associated with that event?

I like the idea that a purely relational physics is anemic, i.e., leaves out intrinsic properties, and I could easily be convinced this is connected with the problem of causality. As for connecting this with consciousness, I am very suspicious: it seems superficial to say, A has intrinsic properties, B does too, maybe A and B are deeply connected. (kind of like saying consciousness is mysterious, so is quantum mechanics: maybe there is a deep connection between the two).

Also, I don't see how he gets outside epiphenomenalism. Couldn't we imagine that his causal story is implemented, generating identical dynamics, but with no experience accompanying it? What does adding 'and the carrier properties are experience' really add to the dynamical/causal story?

Steve said...

That's a good question. In Chapter 12, Rosenberg discusses a list of reasons why he thinks phenomenal and experiential properties are structurally "a good match" for effective and receptive properties respectively. These include on the phenomenal/effective side: relations of inclusion, exclusion, compatibility and incompatibility; scalar relations; status as "determinables". On the experiential/receptive side: being constrained by the effective/phenomenal properties; having an inherent openness to the content.

It's hard to do justice to the argument here since it depends on the progressive build up of the discussion through the book.

Giordano Sagredo said...

It's hard to do justice to the argument here since it depends on the progressive build up of the discussion through the book.

Hmm. That is rarely a good sign for a philosopher's theory.

If it is not clear that he isn't subject to epiphenomenalism, then it isn't clear what we get out of his theory. An isomorophism between phenomenal properties and his causal story brings with it all the problems with the isomorphism we will likely find between neuronal states and conscious states. Why should a physical system with these causal carriers and the like, with these physical dynamics, have an experience accompanying it? It is a nomological dangler. Unless, that is, he is up front that this is simply a postulate, like Chalmers' initial information-qualia connection. But in that case his theory of causality should stand or fall independently of its connection with consciousness. This I take to be a good thing, because the extension to consciousness seems forced (like something you'd throw into a PhD thesis to graduate, become wedded to, but not realize it just isn't very good).

But I am a physicalist, and most days think the qualiaphiles are just wrong.At any rate, I am glad you read it and not me. Seems too far out there.