Epiphenomenalism and the problem of psychophysical harmony. Thoughts?
Summary of the problem
Epiphenomenalism is the view that conscious experiences are produced by physical brain states but have no causal influence on the physical world. On this view, your sensations, thoughts, and feelings arise as byproducts of neural processes but do not themselves cause any actions, decisions, or changes in behavior. Consciousness is like steam rising from a traingenerated by the engine but doing no work of its own.
The problem arises when we observe that our conscious experiences are extraordinarily well-matched to our physical and behavioral needs. For example, we feel pain when injured, which motivates withdrawal from harmful stimuli. We feel pleasure when doing something rewarding or health-promoting. Our perceptual experiences generally track the external world in ways that are accurate and useful. This striking alignment between what we consciously experience and what would be biologically beneficial is whats referred to as psychophysical harmony.
But this harmony makes no sense under epiphenomenalism. If consciousness cannot influence behavior, then theres no reason for our experiences to be useful, well-calibrated, or even coherent. We could have evolved with conscious experiences that were completely disconnected from realitylike seeing a blue square all the time, or feeling pleasure when touching a flame. Worse still, we might have had no conscious experiences at all, and the physical behavior of our bodies would be exactly the same. Evolution could not select for good experiences because, by epiphenomenalist logic, those experiences dont do anything.
This leads to what philosophers call the "luck problem." The only way to explain psychophysical harmony under epiphenomenalism is to say we just got incredibly luckythat, out of the vast space of possible qualia, our consciousness just happens to perfectly mirror what is useful for our survival. But this level of coincidence strains belief. It would be like randomly pressing keys on a piano and composing a symphony. It suggests that something deeper is going on.
In sum, the argument from psychophysical harmony shows that if consciousness has no causal power, then its orderliness, usefulness, and alignment with behavior are utterly inexplicable. The fact that our experiences are not arbitrary or chaotic, but finely tuned to our lives and needs, suggests that consciousness must play a real role in how we act and evolve. This is a major challenge to epiphenomenalism and points toward views in which consciousness is causally efficacious, integrated into the functioning of physical systems rather than floating above them, inert and inexplicable.
My own view
If we assume epiphenomenalist dualism, then indeed I do think this is a problem. It seems, for me, a materialist, there are two options from here. Either we admit to some sort of mental causation that must comply with current laws of physics, or attempt to explain this issue through anthropic selection.
Let's take the first option. I, like many other materialists, believe consciousness to be a higher-order, emergent informational property of some kind. There is nothing particularly special about the matter that composes the brain; instead, what is special about it is how one part interacts and relates to another. It suggests that consciousness is not related to the actual substance in and of itself, but is instead an interactional/relational/informational property that is neutral to whatever substrate it happens to occupy. The only way I can see mental causation, in this case, happening without violating or massively changing our understanding of physics is via some sort of top-down, constraint-based causation.
In this view, mental states are not pushing particles around like little ghostly levers, but rather they emerge from and constrain the lower-level dynamics. Just as the macroscopic structure of a dam constrains the flow of water without being extra to the laws of hydrodynamics, so too might conscious informational states constrain the behavior of underlying physical systems without overriding physical laws. This allows for a kind of causal relevance without direct physical interventionmore like shaping and filtering whats already happening. Consciousness, then, would be a structural property with real organizational consequences, operating within physical law but not reducible to any single local interaction.
Alternatively, we could consider anthropic selection. Perhaps there are many possible physical-informational configurations in the universe, and only some give rise to conscious experiences. Of those, only a tiny subset might produce systems where consciousness is psychophysically harmoniouswhere experiences like pain and pleasure are meaningfully aligned with behavior. From this perspective, we happen to find ourselves in such a system precisely because only those systems would contain observers capable of reflecting on this harmony. But while this may explain *why* we observe harmony, it doesnt explain *how* such a configuration comes to be. It risks treating consciousness as an unexplained brute feature of certain arrangements rather than something that follows naturally from the structure of the system.
Ultimately, I lean toward the first path. While anthropic reasoning can play a supporting role, it feels more satisfyingmore scientifically fruitfulto investigate how consciousness could emerge as a causally integrated feature of complex physical systems. The top-down constraint view offers a promising way to make room for mental causation within a materialist framework, preserving both physical law and the apparent functional role of consciousness in behavior.
Epiphenomenalism is the view that conscious experiences are produced by physical brain states but have no causal influence on the physical world. On this view, your sensations, thoughts, and feelings arise as byproducts of neural processes but do not themselves cause any actions, decisions, or changes in behavior. Consciousness is like steam rising from a traingenerated by the engine but doing no work of its own.
The problem arises when we observe that our conscious experiences are extraordinarily well-matched to our physical and behavioral needs. For example, we feel pain when injured, which motivates withdrawal from harmful stimuli. We feel pleasure when doing something rewarding or health-promoting. Our perceptual experiences generally track the external world in ways that are accurate and useful. This striking alignment between what we consciously experience and what would be biologically beneficial is whats referred to as psychophysical harmony.
But this harmony makes no sense under epiphenomenalism. If consciousness cannot influence behavior, then theres no reason for our experiences to be useful, well-calibrated, or even coherent. We could have evolved with conscious experiences that were completely disconnected from realitylike seeing a blue square all the time, or feeling pleasure when touching a flame. Worse still, we might have had no conscious experiences at all, and the physical behavior of our bodies would be exactly the same. Evolution could not select for good experiences because, by epiphenomenalist logic, those experiences dont do anything.
This leads to what philosophers call the "luck problem." The only way to explain psychophysical harmony under epiphenomenalism is to say we just got incredibly luckythat, out of the vast space of possible qualia, our consciousness just happens to perfectly mirror what is useful for our survival. But this level of coincidence strains belief. It would be like randomly pressing keys on a piano and composing a symphony. It suggests that something deeper is going on.
In sum, the argument from psychophysical harmony shows that if consciousness has no causal power, then its orderliness, usefulness, and alignment with behavior are utterly inexplicable. The fact that our experiences are not arbitrary or chaotic, but finely tuned to our lives and needs, suggests that consciousness must play a real role in how we act and evolve. This is a major challenge to epiphenomenalism and points toward views in which consciousness is causally efficacious, integrated into the functioning of physical systems rather than floating above them, inert and inexplicable.
My own view
If we assume epiphenomenalist dualism, then indeed I do think this is a problem. It seems, for me, a materialist, there are two options from here. Either we admit to some sort of mental causation that must comply with current laws of physics, or attempt to explain this issue through anthropic selection.
Let's take the first option. I, like many other materialists, believe consciousness to be a higher-order, emergent informational property of some kind. There is nothing particularly special about the matter that composes the brain; instead, what is special about it is how one part interacts and relates to another. It suggests that consciousness is not related to the actual substance in and of itself, but is instead an interactional/relational/informational property that is neutral to whatever substrate it happens to occupy. The only way I can see mental causation, in this case, happening without violating or massively changing our understanding of physics is via some sort of top-down, constraint-based causation.
In this view, mental states are not pushing particles around like little ghostly levers, but rather they emerge from and constrain the lower-level dynamics. Just as the macroscopic structure of a dam constrains the flow of water without being extra to the laws of hydrodynamics, so too might conscious informational states constrain the behavior of underlying physical systems without overriding physical laws. This allows for a kind of causal relevance without direct physical interventionmore like shaping and filtering whats already happening. Consciousness, then, would be a structural property with real organizational consequences, operating within physical law but not reducible to any single local interaction.
Alternatively, we could consider anthropic selection. Perhaps there are many possible physical-informational configurations in the universe, and only some give rise to conscious experiences. Of those, only a tiny subset might produce systems where consciousness is psychophysically harmoniouswhere experiences like pain and pleasure are meaningfully aligned with behavior. From this perspective, we happen to find ourselves in such a system precisely because only those systems would contain observers capable of reflecting on this harmony. But while this may explain *why* we observe harmony, it doesnt explain *how* such a configuration comes to be. It risks treating consciousness as an unexplained brute feature of certain arrangements rather than something that follows naturally from the structure of the system.
Ultimately, I lean toward the first path. While anthropic reasoning can play a supporting role, it feels more satisfyingmore scientifically fruitfulto investigate how consciousness could emerge as a causally integrated feature of complex physical systems. The top-down constraint view offers a promising way to make room for mental causation within a materialist framework, preserving both physical law and the apparent functional role of consciousness in behavior.
Comments (13)
Great post. I have brought up this issue many times. One neat historical point here is that Plato's last (and best) argument in the Phaedo against the Pythagorean view of soul as simply being akin to a tuning on a lyre (i.e. an emergent epiphenomena) closely mirrors a lot of modern discussions here. This isn't totally surprising since the Pythaogrean view that being somehow [I]is[/I] mathematics is quite popular today (e.g. ontic structural realism, Max Tegmark's "Mathematical Universe Hypothesis, etc.).
So would a carefully constructed neural network made from pipes and water wheels that is set up to process inputs and outputs like a human brain be conscious? Could we carefully set up toilet paper rolls to be conscious?
My take is that, while I think the computational theory of mind view gets [I] something[/I] right, I am not sure it gets everything right. For one thing, the reduction of thought to computation (essentially discursive [I]ratio[/I] without [I] intellectus/noesis[/I] in medieval thought) seems to open up a host of epistemic challenges that undermine our very faith in reason or science itself.
Also, if one adopts the popular view in physics that the universe itself is essentially a computer (endorsed to varying degrees by a veritable whose who of physicists: Tegmark, Lloyd, Davies, Landauer, Vedral, etc.) then the brain's "being a computer" is nothing special and cannot explain its uniqueness: But it actually turns out to be very hard to define computation in physical systems. Tight definitions are hard to justify and loose ones make it so that anything with enough informational complexity can be said to be computing anything else.
Perhaps one easy way out here is to say that contemporary neuroscience simply makes too many simplifying assumptions. Steam pipes cannot become conscious by being set up as neural networks because human bodies do way more than our neural networks. Perhaps all the very small actions of cellular metabolism, glial cells, quantum scale behaviors, etc. all play a role such that substrate [I]is[/I] important because pipes and toilet paper rolls cannot actually do what the human body does. That quantum behavior has been found in phenomena like photosynthesis seems to me to indicate that it would be more surprising than not if life didn't take advantage of it in some ways in the nervous system.
Are you familiar with Terrance Deacon's "Incomplete Nature" or his other work?
You might be interested in this introduction . His theory works similar to this but brings in semiotics to help, although I recall that it didn't seem to totally lean into the triadic semiotic view (as opposed to dyadic mechanism, which is so dominant). John Deely is another interesting guy here. Deacon also ties his constraint-based absential influence back to Aristotle's notion of formal causality.
As Deacon notes, Jaegwon Kim has some very strong arguments against any sort of emergence from the perspective of a substance metaphysics of supervenience (i.e. one where "things are what they are made of," a building block/bundle ontology). However, process metaphysics avoids this issue (Mark Bickhard has a good article on this, although it simplifies a bit too much). So, again, Aristotle is a nice example of a process metaphysics that doesn't run into the problem of collapsing all being into a single monoprocess and making all predication accidental (a particular vice of process metaphysics that is sort of the mirror image of the excesses of reductionism, a sort of "bigism," e.g. "only quantum fields exist, and they are unified, so only the field of fieldsjust one thingexists"). One nice thing about process metaphysics is that it also seemingly incorporates information theory better in some ways (instead of having it reduced to mechanism).
Actually, I think Aristotle can be more useful here through the later development of his thought in Neoplatonism, Islamic thought, and Scholasticism, but those are sort of a dark zone in contemporary thought.
Although, I'm also partial to the idea, advanced by David Bentley Hart in "All Things are Full of Gods," D.C. Schindler, and others, that the problem also one of framing. Mechanistic philosophy is essentially a giant inversion, orienting higher levels towards the lower, act to potency, form to matter, etc. Historically, one can see how this shift was motivated by a number of theological and political concerns, as a sort of reaction against the existing model (e.g. describing nature in terms of "laws and obedience" isn't any less anthropomorphic than speaking of "desires and inclinations," it's just motivated by a particular sort of theology). Smallism and reductionism grow out of this moment and they seem to make a number of phenomena impossible to explain.
Right, this seems very unpalatable. It's almost a non-answer. It seems similar to some responses to the Fine Tuning Problem. These sorts of answers are only accepted so long as better answers don't exist.
I'm not making fun of your terminology at all. I have none better to offer. It's very hard to use our current concepts to construct a simple picture of what's going on here. And yet it's the simple picture that we need: Am I along for the ride, like steam from the train, or do my decisions really cause anything to happen? And if these binaries are misguided, how should we understand that?
Excellent OP by @tom111! I'll reply directly to them later on.
Thanks for the Deacon reference. I had read The Symbolic Species circa year 2000 just before I began being seriously interested in philosophy and was much impressed, but I haven't read his more recent work. What remained with me from this reading was the powerful idea of the co-evolution of human thought and language (at the cultural level) with our biological evolution, and the Baldwin effect accounting for the influence of the cultural on the biological.
I just now perused what Deacon has to say about emergence, downward-causation and Kim's causal exclusion argument, in Incomplete Nature.
I find it interesting that Deacon stresses the analysis of diachronic (and dynamical) emergence over the analysis of synchronic (inter-level) emergence as the key to understanding strong emergence, while my own proclivity is to stress the latter. The way in which Deacon purports to circumvent Kim's argument is by questioning mereological assumption regarding the nature of matter and hence also rejecting the thesis of the causal closure of the physical (or micro-physical) domain. This reminds me of a similar move by another scientist advocate of strong emergenceGeorge Elliswho, in his defence of the reality of non-reducible downward-causation, also argues for the need for there being "room at the bottom" (i.e. at the micro-physical level) for high-level processes or entities to select from. Ellis, unfortunately, merely gestures at the indeterminacy of quantum mechanics without providing hints as to the nature of this selection process. Deacon is a bit more explicit and seemingly makes a similar moves when he points out that:
"The scientific problem is that there arent ultimate particles or simple atoms devoid of lower-level compositional organization on which to ground unambiguous higher-level distinctions of causal power. Quantum theory has dissolved this base at the bottom, and the dissolution of this foundation ramifies upward to undermine any simple bottom-up view of the causal power." (Incomplete Nature, from my unfortunately unpaginated epub copy of the book)
My own view is quite different, stemming from my reflections on the problem of free will, determinism and responsibility (and mental causation), since I think Kim's argument fails even while granting to him the thesis of the causal closure of the physical.
The reason why mental states aren't epiphenomenal is because antecedent low-level physical properties of embodied living agents don't causally determine their subsequent behaviors and mental acts but only the material supervenience basis of the latter. Where Kim's causal exclusion argument fails in discounting antecedent mental states (or acts or events) as causally efficacious (and non-redundant with physical causes) is in misconstruing the cause of the specific low-level physical configuration of the antecedent mental state that merely accidentally happens to realize the subsequent mental state (or intentional behavior) as its cause. But a successful causal explanation of a mental event ought also to explain why it is an event of that specific (high-level) type, that happened non-accidentally to be realized, and the fact that the antecedent physical state was realizing the specific antecedent mental state that rationalized the subsequent one constitutes an indispensable part of this causal explanation. What Kim seeks to portray as the real (low-level, physical) cause of mental events is actually causally irrelevant to what it is that rationalising explanations of behavior disclose as the genuinely informative cause.
(On edit: My last paragraph is rather dense, so I asked Claude 4 Opus if it understood the structure of the argument and its relevance to the thesis of the causal closure of the physical. It understood both and rephrased those ideas elegantly.)
I also liked your discussion of the Pythagorean view of being as mathematics. Personally, I dont resonate all that well with this idea though Ill admit I havent explored it deeply. Im a nominalist, in the sense that I see mathematics as an incredibly effective tool for describing and predicting the behavior of systems that exhibit regularity. The universe, as it appears to us, behaves in highly structured and regular ways, which is likely why mathematics is so successful. But I dont think that means mathematics is the essence of being.
As a physicist myself, Id argue that physics can only ever tell us about the relations between things, not their intrinsic nature. This is because physical theories work by specifying how systems change relative to one anotherhow one thing influences another over time and space. Whether were talking about force fields, wavefunctions, or geometric structures in spacetime, all of it is defined relationally. We never get a glimpse of what a thing is in itself, only how it behaves in a web of other things.
You raised this thought experiment:
Quoting Count Timothy von Icarus
Very good point. If we take informational or structural accounts of consciousness seriously, then in principle, any system that implements the relevant patterns should be consciouseven ones made from absurd materials.
That said, I think we can sort these possibilities into three broad camps:
Personally, I lean toward either the first or second view. I find the third possible, but it would likely require extensions or reinterpretations of physics. The second view seems like the best middle ground: consciousness is still relational or informational, but subject to real physical constraints. In other words, not all informational structures are physically realizableonly certain complex, fine-tuned systems (like living bodies) can support the right kinds of interactions.
On Terrence Deacon I havent read his work yet, but Ill definitely check it out. The idea of using semiotics and constraint-based models to rethink causation sounds right up my alley. I also completely agree that substance-based metaphysics struggles to handle emergence. If we view things as nothing more than what theyre made of, then so-called emergent phenomenawaves, organisms, mindsend up being treated as mere large-scale approximations of the "real" underlying stuff. That feels deeply unsatisfying. It undermines the explanatory power and causal relevance of the very patterns that define higher-level reality.
Thats one reason Im drawn to process metaphysics. Id argue that much of modern physics is already (mostly) process-based. While we still refer to entities like quantum fields as kind of substances, (although, again, we don't really explain what they are - only their role in an enormous web of interactions with other things) what physics mostly does is track interactions, transformations, and constraints over time. It describes evolving relationships, not timeless chunks of being.
In my own fieldcomplex matter organizationIve noticed that many systems seem to operate in a way thats semi-teleological. We regularly work with attractor states, self-organizing processes, and feedback loops that appear goal-directed - a shift I've seen in recent years within physics. This contrasts with more mechanistic branches of physics where behavior is rigidly determined by initial conditions. But it seems unlikely that we live in a reality where some systems operate teleologically and others mechanically. That would imply a strangely split ontology. Whatever causal model we settle onmechanistic, teleological, or something elseit likely applies universally.
Thanks again for such a thoughtful comment Ive really enjoyed reflecting on all of this.
While I disagree with much of your thinking, I agree with your thought about epiphenomenalism being wrong [Url=https://emersongreenblog.wordpress.com/2020/06/12/why-epiphenomenalism-is-almost-certainly-false/]Here's an article I like along those lines.[/url]
To start, I'll echo what many other's here have said - really good OP. Clear, well-written, and interesting.
One problem with almost all discussions of consciousness here on the forum is the failure to define just what is meant by the word "consciousness." Different people mean different things, which almost always leads to confusion. Since you haven't really done so, I'll offer this one from "Feeling and Knowing: Making Minds Conscious" by Antonio Damasio.
I think this highlights one of the most important aspects of consciousness. It's not a little guy sitting up in our brains pulling the levers. It's a complex interaction among mental processes and it doesn't just do one thing. It participates in the entire system of mental processes. There isn't a one to one correspondence between a specific conscious experience and a specific behavior. In Damasio's view, consciousness is meant to give the affected organism ownership of it's mind. That makes sense to me.
I'm also wondering about "meant". That sounds like it was the plan, which I don't assume you meant?
I would say that the steam has the potential to do work if it were to come into contact with something else. At the very least the steam would merge with the water vapor in the air and become part of the air we breath. Everything is a causal process, including the mind. The relationship between causes and their effects is information.
Quoting tom111
Exactly. Is consciousness like eye color in that it is just a by-product of accumulated mutations that have no beneficial or detrimental effects on survival? Eye color could play a roll in sexual selection as some might prefer a certain eye color in a mate. It seems to me that many of us select mates that match, or add to our mental lives as well. One might add that we also select mates based on their mental states as well.
If we were to explore what makes consciousness useful I would point to learning. Whenever we learn something we are fully conscious of what it is we are doing. When learning to walk or ride a bike, you are fully conscious of every movement of your feet, legs, balance, etc. Your attention is focused on these things and the effect it has on walking or riding a bike successfully and efficiently. You observe how others do it and try to duplicate the action and then observe the effect and repeat until you eventually get it right. After that the task of walking and riding a bike is handled subconsciously, You no longer have to focus your attention on the task of walking or riding a bike, which is why you can focus on other things while doing these things. It's as if consciousness is training the brain and muscles what to do so they can handle the task on their own in the future so that the conscious mind can tend to other, more important things.
I would add that the notion of what it means to be "physical" is a mental abstraction. The brain processes sensory information at a certain rate relative to the rate of external processes that it observes. This will have an effect on how we perceived certain processes compared to other processes. Slow processes would appear as static "objects" and what we tend to think as physical. Faster processes will be perceived as actual processes and even faster processes might not be perceived at all, or as blurs of motion. The point is that a physicalist is confusing the map with the territory and trying to reconcile the "physical", static, solid object of the brain they perceive with the "non-physical" (dynamic and unbounded) aspect of the mind. The brain is not a physical object. It is a process and consciousness is a sub-process of the brain.
Here's how I think about it. Keep in mind that I don't claim any scientific truth to this. I don't know what goes on in a baby's mind.
Babies when they are a few months old seem to be fascinated by their feet. Sitting in a stroller they keep reaching out and holding them. At that age, their hands seem to be the primary way in which they interact with the world, so they recognize them as part of themselves. In my imagination, one day, the baby touches his feet and realizes that the feelings he has are coming from the feet. That leads them to the amazing realization that those feet are actually part of him.
That's how I see ownership - recognition that something is part of your inside world, you, and not the outside world.
Quoting Patterner
I guess my wording is imprecise. The subject on the table is whether consciousness actually does something or if it's just along for the ride - an epiphenomenon. When I said "consciousness is meant to give the affected organism ownership of it's mind" I only meant that is the value, or at least one value, of consciousness to the organism. That doesn't address whether consciousness actually provides the motive drive for action. I suspect it doesn't, but I can't give you any specific evidence for that.
Thank you.
:up:
Funny enough, I think the rise of information theoretic/complexity studies approaches to the physical sciences make a good case for a sort of "immanent realism" as opposed to a thoroughgoing nominalism (which opens up a number of epistemic challenges).
There is, a sort of definiteness to something like an ant. That is, ants were around as organic wholes organized around aims (life being "goal-directed") long before there were men to say "this counts as an ant." The ant is not an arbitrary ensemble; at least that's my position.
I also think this goes with Aristotle's idea that it is organism who are most properly beings (things like rocks or puddles being more nexuses of external causes). But of course, unity, self-organization, and self-governance occur on a sliding scale. Being a "whole" is not a binary distinction. Unity and multiplicity are contrary opposites, not contradictory. Some things are more unified as wholes than others, organisms being the best example (but dissipative systems with "life cycles" are interesting cases too).
To my mind at least, the opens up a via media for explaining why there are such things as trees, men, ants, etc. in a non-arbitrary way, instead of "just universal quantum fields" or "just a sea of particles," which we then "give names to." Afterall, if we give things names because it is "useful" to make such distinctions, we will next have to ask: "but why are these distinction useful and not others? Why did disparate linguistic groups all come to speak of ants, trees, etc., and have the idea of species?" Some prior cause seems necessary to explain the distinction. To my mind, the most obvious cause is that ants and the like already represented organic wholes.
But then the informational pattern by which something is an ant, even though it is defined by a relation to the entire cosmic order (i.e. not wholly intelligible in itself) is, nonetheless, abstractable, and so has a sort of unchanging being in abstraction.
I think one of the difficulties here is that the obsession with logic in late medieval/early modern thought led to a sort of calcification of realism, such that it became a real life strawman of itself. "Ant" had to be an eternal, unchanging univocal form, rather than being a certain sort of principle instantiated in all ants. Evolution is fatal to the calcification, but not really to more nuanced formulations that existed earlier.
There is at least an interesting etiology here, when you think about "information" defining thought (even in steam pipes perhaps!). Because the term "information" comes from the old "form," eidos (from whence we also get idea). Well, if computational theory of mind or integrated information theory are correct in some key respects, then there definitely is a link between "idea" and "(in)form(action)."