Sunday, January 22, 2006

Explanatory examples: Darwin vs. Lamarck

Having distinguished two primary explanatory models, mechanism and agency, I want to try to apply that distinction in an example -- and the one I have in mind is one that excited much interest over a century and a half ago: the phenomenon of species change.

That the phenomenon existed at all was by no means evident, in the first place. Species appeared to be not just fixed, but admirably and ingeniously suited to their place in the world, which of course is just what one would expect from an Intelligent Designer. Thus, like contemporary ID advocates, the explanation for the remarkable "fitness" of species is modeled on the notion of the will or purpose of a supernatural agent, and that was all one could (or needed to) say. The fossils turned up by geologists (later, paleontologists), however, began to put this neat and static picture under increasing stress, and the distasteful but unavoidable conclusion seemed to be that species that once existed no longer do, and species that now exist once did not. What could account for this?

Jean-Baptiste Lamarck, as everyone knows, tried to answer that. He thought that individual organisms tried to perform certain actions in response to changing conditions, for example; that in doing so changed themselves, in the way that exercise develops muscle; and that these changes were then inherited by subsequent generations, which over time resulted in a changed species. Now, as everyone also knows, Lamarck was mistaken about the heritability of acquired characteristics, and this mistake is often taken as the defining error of Lamarckism. But I think that mistake was a relatively superficial one, and even somewhat understandable -- the deeper, more profound problem with Lamarckism has to do with how change comes about in the first place. It is true, of course, that some organs (mainly muscle) are developed through use, but many changes to species -- e.g., color of fur, structure of eye -- have nothing to do with anything that an organism is or could be "trying" to do. Moreover, even if an organism could try to change the color of its fur or the structure of its eyes, how would it know that that's the right thing to do -- that is, how would it know how to design itself so as to fit its environment? The primary problem with Lamarckism, in other words, isn't the notion of the heritability of acquired characteristics, but the fact that it lacks a mechanism for adaptational change -- instead, it simply located the agent of biological fitness within the organism itself, where it remained as mysterious as the ways of the supernatural Designer.

And then along came Darwin. By and large, he understood that acquired characteristics (which would be at least as likely to be damaging as beneficial) were not inherited, but that was not the essence of his surprisingly simple and remarkably general insight. What Darwin provided was precisely what Lamarck lacked -- a mechanism for adaptational change. It broke the process into two parts or aspects -- on the one hand, there was a blind but incessant generation of small random (i.e., not willed or agent-directed) changes; and on the other hand, there was an equally blind (again, no agent involvement), equally incessant process of culling those random changes. "Natural selection", as a phrase, emphasizes the agent-less nature of that culling process just by its contrast with the agent-based process that goes on in human or "domestic" selection. And with that two-part mechanism, what had seemed vague, obscure, and mysterious, suddenly became lucid and clear -- and perhaps all the more marvelous (and controversial) just because of that clarity.

This contrast between mechanism and agency is so important, I think, that it really has become implicit in the very meaning of evolutionary change. It's worth observing, after all, that the sort of willed or agent-directed change that Lamarck thought was at the basis of species change does indeed occur (even though he was wrong about its heritability), but is inherently limited -- we could think of it as exhibiting the potential of a species at a point in time. Evolutionary change, on the other hand, represents a change in that potential, and that is something that nature or the environment imposes, through the twin but independent mechanisms of natural variation and selection.

Friday, January 20, 2006

Explanatory models: what about "holism"?

The previous post distinguished two very general models for explaining a phenomenon: mechanism, defined by the primary role it gives to causation; and agency, defined by the parallel primary role it gives to will or purpose. What was absent was a term that a fair number of people would view as a possible alternative to, and improvement over, both -- "holism". In this note I just want to say why it was left out -- why, in other words, I think it's not really an alternative, much less an improvement, to either.

The short answer is that holism is not actually an explanatory model as such at all, but rather a set of explanatory guidelines at best, and an avoidance of explanation altogether at worst. Like many "isms", holism comes in a range of flavors or versions, from what's usually called "weak" at one end of the spectrum to "strong" at the other end. In all versions, I think, we can distinguish two broad aspects or themes to holism: that the whole is greater than the sum of the parts, and that context or environment is important in understanding a system. Let's briefly look at both:

That the whole is greater than the sum of the parts:
In weak or moderate versions of holism, this notion is certainly valid, but largely a truism -- any machine, after all, is greater than the simple sum of its parts in the obvious sense that the structure or relationship of the parts is a vital aspect of its functioning -- and understanding the mechanism of any phenomenon entails an understanding of how the parts interrelate. In its mild form, admittedly, this theme can sometimes provide useful advice when dealing with particularly subtle or complex phenomena, where the "whole" may not be immediately evident. But in its stronger versions, this idea is often taken to imply that the "whole" contains some mysterious extra quality or ingredient that is impervious to analysis, and this simply becomes a way of refusing or blocking explanation altogether.

That context matters:
This theme, in moderate versions of holism, is actually quite important -- the behavior of many systems is quite dependent upon the environment in which they're situated, and may change drastically if that context is changed. It may still, of course, be a reasonable strategy to try to gain as good an understanding as possible of the system in isolation (a simpler problem), while realizing that a full or real-world understanding will need to include the larger "whole" of which the system is a part. That said, however, this theme too is frequently taken to untenable extremes in the "stronger" or more mystical versions of holism, where it's used to assert that "everything is connected" and hence nothing can be studied, understood or explained in itself.

So: "holism" in its more moderate flavors can add a helpful perspective to an explanatory project, but hardly qualifies as an explanatory model in itself; and in its stronger flavors, it tends to become a means of opposing explanation as such (sometimes, perhaps, in an effort to protect a favored belief, or sometimes just in the service of mystery itself).

Monday, January 16, 2006

Explanatory models: machine vs. agent

(Continuing reflections on the topic of "explanation"....)

The machine hasn't fared well as a cultural icon for the last while. From the image of the dehumanized, roboticized workers in the vast urban machine of Fritz Lang's Metropolis to the metallic death's-head horrors of the Terminator series, the mechanical in film and culture generally has been portrayed as something deeply antithetical to the human. No doubt the antipathy goes back at least as far as the coal-fed steam engines that powered Blake's "dark, satanic mills". It's as though there's an undercurrent of fear running throughout technological culture that we ourselves are in danger of turning into machines. Little wonder, then, given that sort of context, that there might be some serious resistance to the idea that, in fact, we already are machines.

But what exactly is a machine in the first place? Apart, that is, from a lifeless and soulless artifice, the very embodiment of the anti-human? In one sense, a machine is just a tool, as evident in that "simple machine", the lever. But there's a different, though related, sense, in which a machine is any system involving a number of parts, assembled in a particular manner, connected by determinate or causal processes, and capable of performing some function. As such, the idea of the machine or mechanism applies not just to artificial constructions but has become a general model of explanation for natural processes of any kind -- if we can provide an account for any phenomenon in terms of purely causal interactions between its parts or components, then we can say that we have provided a mechanism for that phenomenon. And with that we will have gained a metaphorical but nonetheless very effective "lever" for it, or a way of bringing it under our control. The idea of seeking out the "mechanisms" of natural phenomena lies at the heart of the technological or industrial civilization that has transformed the earth.

This idea wasn't always so prevalent, however. Just as the extension of the idea of the simple tool provided one explanatory model for natural phenomena, so an extension of a concept based upon our own social interactions provided another: this is the notion of "agency" as an explanatory model. At the basis of this notion is the idea of will or purpose, which plays an explanatory role parallel to that of cause in mechanism. Once, it was common to view such an animating will, instantiated in God or gods or spirits, as the explanation of virtually all natural phenomena of any significance, and we can still see a vestige of this in the tendency to speak of natural disasters as "acts of God", for example. Note that this notion of agency as an explanatory model offers nothing like the kind of effective control over phenomena that mechanism does -- instead, the forces behind events could only be flattered or appeased, with always uncertain results.

About four or five hundred years ago in Europe, however, this model began, slowly but implacably, in one field of phenomena after another, to be replaced by the mechanistic model. One of the last of such areas, and one which looked to be particularly resistant to the advance, was that of living organisms, which seemed, almost by definition, to exhibit the sort of will or purpose that was the essence of agency. But, of course, the last fifty years or so have witnessed perhaps the greatest success of the mechanical explanatory model yet, with the discoveries of the various mechanisms underlying the phenomenon of life itself*. (It's worth pointing out that living processes often depend upon randomness rather than strictly determinate causal sequences, but that such "stochastic" processes are nonetheless sufficiently reliable on a statistical basis to qualify as mechanisms.) So by now, the notion of agency as an explanatory model has been pushed back to the social and psychological arena in which it first arose, and is under some pressure there.

For reasons that I've already touched on -- having to do with an inherent indeterminacy in cultural self-knowledge -- I think agency is secure in this arena, finally, against further encroachment. In this more limited field, agency really serves more significantly as a moral concept than as an explanatory one in any case (though of course it can also play a different kind of explanatory role to some degree as well). In this, it addresses directly much of our fear of the machine, since what really arouses concern, in this contest between models, is not our humanity per se (except in a broad or metaphoric sense) but rather our status as moral agents. At some point in our future, after all, an expanded cultural ecumene might well include actual machines (e.g., robots, androids, etc.), not to mention aliens, speaking animals, or what have you, and they too would be simply incorporated as fellow agents into the moral realm that communication engenders. What repels us, quite rightly, about the image of ourselves as machines is the thought that it seems to turn us into mere tools or instruments (in the hands, implicitly, of another will or agent) -- but that is something that mechanistic explanation in itself is simply unable to do (though some ideologies have thought they could, and tried, with horrifying results). It's not the machine that we should fear, in other words, nor mechanism as an explanatory model, both of which have brought immense benefits and opportunities, and have, if anything, on the whole enhanced rather than diminished our humanity. It's, as always, the tragically mistaken uses to which they're put, but to which we can, and must, refuse. Understanding that can make the enquiry into the mechanisms even of consciousness and culture not just a useful, but an entirely human undertaking.

*Including, in just the last twenty or so years, the mechanisms behind the enormously complex processes of developmental biology -- this is a plug for a fascinating book I'm in the middle of, Endless Forms Most Beautiful by Sean B. Carroll.

Saturday, January 07, 2006

The "explanatory gap" series: a summary and a Q-and-A

(The end of the series, but likely not the topic. Earlier posts are:
In this series of posts, I've tried to look more closely at the widespread thesis that any attempt to explain conscious experience in purely physical terms -- i.e., any attempt to "reduce" experience to a physical level -- is fraught with a very fundamental problem, termed, by Chalmers (after Levine), an "explanatory gap" or "the hard problem of consciousness". Neither Chalmers nor Nagel, the two authors of seminal papers in support of this thesis that I've quoted from here, want to say that this means such a project is necessarily doomed -- in fact, towards the end of their papers, both make useful suggestions for making some headway with it -- but the thrust of both their suggestions and their texts as wholes is that, for any purely physical explanation, the "hard problem" isn't going to go away.

In this, as I've indicated, both philosophers are tapping into a very deep and very long-standing intuition -- which is that experience is an inherently non-physical phenomenon, or, in other words, that there is an absolute and fundamental gulf between the mental, whatever that may be, and the physical (whatever that may be). Such a gulf, of course, creates all kinds of problems concerning the interaction of such radically disparate realms, which is no doubt among the reasons that neither of the two wants to assert that the gap is unbridgeable in principle. But in general when we try to get down to details about mental-physical interaction, we're often met with mystification or ad-hocery or both -- e.g., the mental (aka "experience") is a new fundamental entity in the world; the mental operates in mysterious synch with the physical; the mental is just some mysterious by-product of the physical; the mental is some quantum thingy, etc. What I've wanted to propose is that, instead of such evasions, we question that deep-seated intuition of a gulf between mental and physical in the first place -- it wouldn't be the first time that our intuitions concerning ourselves and our place in things have lead us astray.

The preceding posts in this series (see the list above) have all just been suggestions toward this end. The last one (before this) in particular puts forth an alternative to the ontological divide, and an explanation of the intuition itself, in the form of a difference in orientation toward, or perspective on, experience -- a difference between viewing it as a phenomenon, on the one hand, and being a part of the phenomenon, on the other hand. But, after all, even if we're willing to accept that the intuition of a fundamental gulf between "mental" and "physical" realms may be mistaken (with the scare quotes indicating that these terms themselves may be part of the problem), we're still left with the considerable difficulty of coming up with an actual physical explanation of the phenomenon of conscious experience. That's properly a scientific, not philosophic or speculative, task, but I thought it would be worthwhile, after a series of often critical posts, to try to answer a few of the more obvious questions about this approach in a more positive vein:

If experience is, as you say, just another phenomenon among phenomena, then why is it, after all, that we can't observe it as we can any other phenomenon?
The key point about conscious experience as we experience it is that it's actually a function of two things: a particular kind of behavior-control mechanism, on the one hand, and our situation as a component of such a mechanism, on the other hand. So we can't "observe" this experience directly in any other entity simply because the "we" component isn't in those other entities (see this post). If we could unplug the "we" from our own brains and plug it into the analogous slot in another entity, then we could indeed know what it's like to be a bat, for example. Short of that, all we can do instead is observe the effects of conscious experience in the behavior of other organisms, and infer that the underlying mode of behavior control is consciousness.

One can understand that experience, as a form of information, might have causal effects -- that is, that it makes a difference -- but why does experience have content (like red or ringing or sour)? What does such qualitative content add?
The short answer is because information, unlike the smile of the Cheshire Cat, has to have a carrier. The content itself is largely arbitrary and unimportant (in principle -- there may be technical advantages to certain content in practice) -- what carries the information, and therefore does the work, of experience is simply the difference between one token-like quale and another. Which is the reason that "inverted qualia" arguments -- sometimes used as arguments demonstrating the ineffectiveness of qualia in general -- are irrelevant here.

But why is the content not just voltages or magnetized regions or some such, rather than red, sour, etc.?
Because if voltages, magnetic regions, etc. are interpreted simply as causal factors (which is what we mean by invoking them), then this violates the principle of "loose connectivity" that defines consciousness as a behavior control mechanism. Regardless of how qualia are caused or instantiated themselves, they must be detected solely as distinct tokens of information -- at this stage, there is no notion of quantity as opposed to quality, and certainly no notion of "voltage", etc.; there is only arbitrary content and difference.

But then notice the difference between this desiccated language of "tokens" and what David Chalmers rightly calls our "rich inner life" -- doesn't that alone suggest that there's more to experience than mere information tokens?
Well, there's certainly a linguistic or conceptual difference, but that has to do with the different aims or objectives of phenomenological description, on the one hand, and mechanistic explanation, on the other -- and those different objectives derive, in turn, from the distinct orientations toward experience spoken of already. But it's interesting that Nagel, toward the end of his paper "What is it like to be a bat", suggested the possibility of developing what he called an "objective phenomenology": "... its goal would be to describe, at least in part, the subjective character of experiences in a form comprehensible to beings incapable of having those experiences". Examples of concepts for such a phenomenology of qualia, abstracted across different sensory channels, might be the notion of a spectrum (linear or circular?), intensity, locality, definition, motivation, duration, etc. With such as this, though we could never match the richness of experience itself in objective terms, we might begin to approach it, from the outside as it were, by deliberately trying to avoid terms dependent upon a particular kind of experience for their meaning.

In summary and conclusion:
  • The "mental" and the "physical" (in reference to consciousness) are not two different realms, nor two different aspects of things, but are just two different ways of speaking about the same thing, their difference a consequence of the simple fact that we the speakers are at the center of the thing we're speaking about.
  • In physical terms, "we" are not agents lurking in the machine, but are complex components of the machine -- a component specialized to receive standardized (tokenized) signals, and integrate such tokens into the processes of behavioral decision-making. (Among other things, this implies that the notion of "feeling" does indeed apply to certain mechanical structures.)
  • Far from being a pointless and/or mysterious after-effect of physical processes, then, phenomenal experience or feeling is just information in the form of distinct tokens, and is the linchpin in the physical explanation of consciousness -- it provides the essential, loose (i.e., informational as opposed to directly causal) connection between the two key components of consciousness as a uniquely flexible behavior-control system.

And to anyone who makes it through the whole of this post, my apologies for its length.

Thursday, January 05, 2006

Is experience a paradox?

(This is another post in a series, starting with this and continuing here and here, on the so-called "explanatory gap" in any current, or perhaps any possible, theory of consciousness and the reality of conscious experience.)

It might seem so, going by at least some of the statements made in Chalmers' paper "Facing up to the problem of consciousness" "Why," he asks, "should physical processing give rise to a rich inner life at all? It seems objectively unreasonable that it should, and yet it does." And later: "We know that conscious experience does arise when these functions [such as visual discrimination] are performed, but the very fact that it arises is the central mystery." Similarly, Thomas Nagel, in "What is it like to be a bat?", his famous earlier paper that focused attention on the problem of phenomenal experience, frames the issue more sharply:
We appear to be faced with a general difficulty about psychophysical reduction. In other areas the process of reduction is a move in the direction of greater objectivity, toward a more accurate view of the real nature of things. This is accomplished by reducing our dependence on individual or species-specific points
of view toward the object of investigation....

Experience itself however, does not seem to fit the pattern. The idea of moving from appearance to reality seems to make no sense here. What is the analogue in this case to pursuing a more objective understanding of the same phenomena by abandoning the initial subjective viewpoint toward them in favor of another that is more objective but concerns the same thing? Certainly it appears unlikely that we will get closer to the real nature of human experience by leaving behind the particularity of our human point of view and striving for a description in terms accessible to beings that could not imagine what it was like to be us. If the subjective character of experience is fully comprehensible only from one point of view, then any shift to greater objectivity—that is, less attachment to a specific viewpoint—does not take us nearer to the real nature of the phenomenon: it takes us farther away from it.
So the paradox seems clear and stark (though practical rather than logical): on the one hand, "subjectivity" is an inherent source of error while "objectivity" is the path to reality; but on the other hand, subjectivity is here the reality to be explained -- how could it ever be possible to be "objective" about the phenomenon of the subjective? As soon as we try to grasp such an illusive phenomenon in objective terms, in other words, we seem to lose the very quality that defines it.

Now, when you come up against a paradox in an explanatory project, you have a number of options (not counting the one of just moving on to another project): you can paste a big label "Mystery" over the whole thing and store it away (perhaps bringing it out occasionally to inspire awe); you can invent new entities and/or ontologies and endow those with just the right features or qualities that you hope will make the paradox go away; or you can take the appearance of the paradox as an indication that there's a flaw or problem buried somewhere in your assumptions. The first option might have its uses, but is clearly an expression of explanatory failure. The second might work, but has an ad hoc or arbitrary aspect to it that rubs off on the concocted entities/ontologies. The third option is the one I'd say has the most promise (and may help clear up some other quandaries as well).

In this case, for example, the assumptions I would question are those lurking behind the quasi-realist epistemology that seems implicit in these approaches. Among the most basic of those assumptions, as we can see from the Nagel quote above, are two: first, that "appearance" is to be distinguished from reality, and second, that explanations can and should approximate that reality (in the process, necessarily shunning appearance). Which obviously leads directly into dilemma and paradox when trying to cope with "appearance" itself. So, first, it might help to look again at that "epistemological inversion" suggested previously, in which appearance isn't a veil but is bedrock, and in which explanations don't aim at approximating a non-phenomenal reality (an impossibility), but rather at constructing more effective and comprehensive structures built out of appearance. Such "efficacious myths", as Quine called them, would still be characterized by ever greater abstraction, in which concrete experiential content is increasingly reduced, but they could no longer be seen as structures inherently alien to appearance or what we've been calling experience.

And that in turn might open the way to a more efficacious understanding of experience within a framework of two distinct perspectives or orientations. Earlier in this series, I'd distinguished those two perspectives as the view from within the phenomenon, as it were, and the view from without (in which experience as such appears only presumptively, based largely on observed behavior). As I said in the previous post, however, this doesn't quite capture the situation, and might even suggest that there is the possibility of a view (of anything) from outside of experience itself, when of course such a notion doesn't even make sense. It might help to expand a little on these different perspectives by pointing out that, on the one hand, experience is all that we are aware of, and is the ground and building material for all concepts, abstractions, and explanations of anything -- this is the first perspective. And, on the other hand, (presumptive) experience (as when we encounter other conscious entities) is but one phenomenon among others within the world that is built out of experience -- this is the second perspective. The second perspective, in other words, is contained within the first -- the situation suggests an Escher-like painting, in which the world is wrapped around and condensed into a localized object within itself. Any explanation -- any attempt at explanation -- of such an object will always be contained within our experiential situation, and can never contain that situation. And that, I think, is a sufficient explanation of the wrongly-named "explanatory gap" -- it would be better to think of it as a situational gap. By its nature, that gap is ineradicable and unbridgeable, since it simply refers to two distinct perspectives (and lies, I believe, at the origin of the many forms and varieties of dualism that have haunted cultures since long before Descartes). But, once we abandon the realist assumption that there must be a single, "true" perspective on the phenomenon, and once we disentangle the two perspectives, it presents no obstacle in itself to the provision of a coherent, effective, and physical explanation of experience as a phenomenon within experience.

UPDATE: Here's the conclusion to this series.