Is the eliminative stance productive?

A number of recent conversations, some I’ve been in, and others witnessed, left me thinking about eliminative views like the strong illusionism of Keith Frankish and Daniel Dennett. This is the view that access consciousness, the availability of information for verbal report, reasoning, and behavior, exists. But phenomenal consciousness, the qualia, the what it’s like aspect of experience, doesn’t.

The problem with this view has always been clarifying what exactly is being denied. This seems complicated by the fact that terms like “phenomenal” and “qualia” have a number of different meanings. For example, many people use “qualia” to refer to something in the vicinity of the primary and secondary qualities discussed by early modern thinkers like Galileo and John Locke.

(Primary qualities include size, shape, duration, motion, etc. These are perceived properties understood to actually be out in the world. Secondary qualities include color, sweet, bitter, hot, cold, etc. These are argued to only exist in the mind, or at least only exist because of minds.)

These types of qualities definitely exist, and serve functional roles. Imagine a yellow elephant with green polka dots. Unless you’re aphantasic, I’m betting you had no trouble picturing it, even though I doubt you’ve ever seen a yellow elephant with green polka dots. (I was careful to make sure the featured image had different colors.) But, unless you’re blind, you have seen yellow things, green things, polka dotted patterns, and elephants before. You were able to combine these characteristics, these qualities, based on your familiarity with them.

Of course, the illusionists are denying a stronger claim. David Lewis, in asking whether materialist should believe in qualia, discussed the functional aspect I described above, a version he saw as compatible with materialism. But there’s another proposition regarding qualia that he discussed: the idea that we can know their full nature solely through self reflection.

I think it’s this assumption that causes the trouble. If we can introspect the full nature of qualia, then their seeming simplicity is irreducible simplicity, which implies they exist separate from the operations of the brain, allowing space for talk of inverted qualia and the absent qualia of zombies. And since no one can detect anything like that in the brain, they must be unobservable to anyone but the subject, who has special “direct” access, resulting in the intuitions behind Mary’s room.

This is the version Daniel Dennett attacked in his 1988 paper, “Quining Qualia.” But Dennett did more than just attack the concept, he attacked the term “qualia,” a standard other illusionists have followed. It’s not enough to attack the idea. The “tangled theoretical knot” of the terms themselves must go. Or at least that’s the argument.

But this causes a problem. There is widespread confusion about what exactly is being denied. For many people, terms like “qualia”, “phenomenal properties”, or “what it’s like” refer to the functional notion, the one we use to imagine weirdly colored animals. So when they see these terms attacked, it sounds like the basic concept is being denied.

The results over the years seem to have been endless conversations with the illusionists trying to clarify exactly what they mean. And yes, not all the confusion is genuine; some people use the conceptual confusion as a rhetorical weapon. But the very fact that it is such an effective weapon speaks to the confusion for anyone not familiar with the history.

Does this mean we should try to rehabilitate “qualia” and related terms? I personally stopped using them a few years ago, specifically due to the definitional confusion. For a long time I thought I was aligned with Pete Mandik’s qualia quietism, an idea I took to mean that these terms were best avoided due to the disparate definitions out there. There’s always other ways to talk about the perception of characteristics.

But qualia quietism seems to take a stronger stance against this language than I do. I don’t use the terms, but I’m not going to scold someone who does. For better or worse, they seem to have spread beyond obscure philosophical discussions. Instead I’ll typically try to figure out which sense they’re using them in, and deal with the concept they’re discussing. That said, qualia quietism remains the neo-Dennettian view I’m closest to.

But I’ve come to think being intolerant of terms like “qualia”, “phenomenal”, “what it’s like”, and similar labels, is drawing the battle lines in the wrong place, one that seems to sow confusion and produces a message that is easy to strawman. Perceptual qualities exist, at least in a representational and relational sense. This shouldn’t be a problematic admission for a physicalist.

Dennett noted in his 1988 paper (second endnote) that the difference between a reductive physicalist and an eliminative one is tactical, a difference in communication approaches. His goal was to confront people’s intuitions and try to force a reexamination. That seems to work well with some of us, who were already predisposed to agree with this ontology. But it seems to generate summary dismissal from everyone else.

Of course, a physicalist does need to deny the idea that we have introspective access to the full nature of our experience, that we’re perceiving something other than just the tip of the iceberg. Dennett compared these tips to the icons on a computer desktop, calling them a user illusion, but the actual software term seems less judgmental: user interface; experience is the brain’s user interface to its own operations. As Lewis argues, this is still eliminative, but look at how little is being eliminated.

All of which is why I prefer to just call myself a functionalist. It emphasizes more what I think is the case, causal roles, rather than what isn’t. Of course, with developments in AI, functionalism is becoming just as much a target. But in my experience it doesn’t generate the same visceral outrage.

What do you think? Am I overlooking benefits to the eliminative approach? Or missing vulnerabilities to just emphasizing functionality? Or worrying about something that doesn’t really make that much difference?

Can we in principle ever deduce the mental from the physical?

Christopher Devlin Brown and David Papineau have a new paper out in the Journal of Consciousness Studies titled: Illusionism and A Posteriori Physicalism; No Fact of the Matter. (Note: the link is to a free version.) As the title makes clear, the overall gist is that the difference between illusionism and a posteriori physicalism amounts to a definitional dispute.

A quick primer. Illusionism is the stance that consciousness exists, but only in the sense of functional capabilities such as modeling the self in its environment, attention, learning, episodic memory, self monitoring, etc. What’s thought to be illusory is phenomenal consciousness, the “what it’s like” nature of subjective experience, but particularly in the strong sense as something distinct from functional capabilities, and with properties, such as fundamental subjectivity, that imply it’s non-physical.

Reductive physicalism is the stance that the mental can be reduced to physics. However, there are different views on exactly what can be understood in that reduction. In one, we can find correlations between conscious states and physical ones that imply an identity relationship, but one that can only be discovered and understood empirically, not justified in a logical sense. This is the a posteriori physicalism of the type the authors discuss in their paper.

The other view is a priori physicalism. It argues that we can go further than just brute identities, and understand the logical relationships, in a way where, in principle, we could deduce the mental from the physical. A common example of this view is analytic functionalism, which describes mental states in functional terms, such as the experience of pain being a negative reaction to a perceived state that motivates a system to try to avoid or ameliorate it.

It’s long been acknowledged among illusionists that the distinction between illusionism and functionalism is definitional. Functionalists generally target functional capabilities for their explanation. If they speak about phenomenal consciousness, it’s usually in a weaker sense of being the inner perspective of a functional system without the non-physical attributes. (As a functionalist myself, this is certainly the sense I use it in older posts on this blog.)

This weaker sense is one that the authors seem to call for in their paper. They point out that it’s always a judgment call whether to eliminate or reconstruct the concept when it turns out not to have all the attributes we assumed in our pre-scientific understanding. (David Chalmers has a similar discussion in his book, Reality+, which I discussed a while back.) For example, we eliminated the concepts of ghosts and witches from our ontology after scientific investigation revealed too many of their properties didn’t exist. However, we retained planets and stars, holding reconstructed understandings very different from the medieval ones.

But I think this is the first time I’ve seen an argument that the differences between illusionism and a posteriori physicalism are definitional.

There is some resonance between illusionism and the phenomenal concept strategy, an argument often made by a posteriori identity theorists about why we tend to think phenomenal properties are distinct from physical ones. In short, our phenomenal concepts are thought to be isolated from our physical ones, making the relationship one we can’t bridge, leading to an epistemic gap, the notorious “hard problem”. This is similar to possibilities explored by some illusionists, such as François Kammerer, who see the illusion as deeply enmeshed in our cognition, something we can’t avoid, and so no explanation of consciousness, including of the illusion itself, will ever feel right.

But it seems like there are differences. For one, the phenomenal concepts strategy is often described as recognizing the conceivability of functional zombies, entities that are behaviorally indistinguishable from a conscious being, but aren’t actually conscious. Most illusionists I’ve read see zombies as an unproductive concept.

And many illusionists take the illusion to be more of a theory error, a failure in philosophical reasoning more than something universally embedded. That’s the feel I get from Daniel Dennett’s writing, although in reality I suspect he would have rejected the distinction.

Still, most illusionists seem in the a priori camp, rejecting any notion of an unbridgeable divide. The phenomenal concept strategy, and a posteriori physicalism overall, seem to skirt mysterianism, a view generally rejected by the a priori camp. To be sure, most of this camp see empirical investigation necessary for progress in any practical sense, but the idea that we can’t have a theory explaining the identity relations is rejected.

Of course, a lot depends on just how much work we’re asking these identities to do. Often the identity relationship between H2O and water, genes and DNA sequences, or heat and molecular motion are given as examples of identities that, once established, we don’t need to explain any further. But these identities have 1:1 relationships, and the reduced concept can in principle be used anywhere the higher level version can in descriptions, making the concepts causally equivalent.

Much depends on what we mean by a conscious concept like “pain”. Is pain a relatively simple primitive like water above? As a phenomenal property, the painfulness of pain is often assumed to be that kind of primitive, which is how many end up thinking of it as something separate from the functionality.

Or is pain more a complex collection of processes, in a way similar to the concept of “democracy”? In principle we could find the physical identity relationship between the concept and a physical occurrence of democracy, although it would be extremely complex. But more broadly, democracy as a type encompasses too many physical instantiations with too many variations for this kind of identity primitive to be useful. We need intermediate abstraction layers, such the role people play in governance. Such roles are multi-realizable, which puts us in functionalist territory, where I think most illusionists live.

Ultimately the difference between the views seems to remain, although it doesn’t seem vast. I suppose it could come down to what is expected of an explanation. If it doesn’t feel right, does that mean we’ve failed to bridge the gap? Given scientific theories like general relativity and quantum mechanics, it doesn’t seem like we have any right to expect an explanation of mental states to necessarily feel right, but that’s a view from someone firmly in the functionalist camp.

I do think the authors are right that “consciousness” is a semantically indeterminate concept. Its meaning has varied too much over the centuries for anyone to claim a particular version is the one true definition. It can mean introspection, perception of the outside world, attention, sentience, imagination, a non-physical ineffable essence, and a host of other notions. Which means these definitional disputes are probably unavoidable.

What do you think? Are these views more similar than I’m seeing? Are all physicalists basically illusionists, even if only implicitly? Or does the ambiguity of the word “consciousness” render these kinds of distinction a hopeless muddle?

Featured image credit

https://selfawarepatterns.com/2024/08/24/illusionism-and-types-of-physicalism/

#Consciousness #eliminativeMaterialism #illusionism #materialism #Philosophy #PhilosophyOfMind #physicalism