Epiphenomenalism is one idea I’ve struggled with for a long time: to my eyes, it doesn’t make any sense. But more importantly, when applied to philosophy of mind, it seems to me that epiphenomenalism does a great deal of damage. However, I can see that lots of people take the idea seriously, so my contribution is also a call for help: if, after reading my attack, you’ll think you can explain me why I’m wrong, I would really appreciate hearing from you.
I’ve never found the courage to write on epiphenomenalism because I nurture self-doubt, as an explicit epistemological method; in this context, I can see that many people who command my unconditional intellectual respect take the notion as seriously as possible (see for example Chapter 4 in Chalmers’ The Conscious Mind: In Search of a Fundamental Theory – 1996), giving me a strong indirect indication that I must be missing something crucial. I’ve finally decided to write my argument down as a consequence of a long discussion I’ve held on Conscious Entities. In the latter parts of my sparring with Jochen, I’ve realised that all my efforts on Consciousness (see also the Evolutionary Theory of Consciousness draft paper, ETC in short) rely on the assumption that epiphenomenalism is nonsensical. If you take away this assumption, ETC becomes a much weaker hypothesis, on a priori grounds.
In particular, Jochen noted that I’ve dismissed epiphenomenalism in a way which is “much (much!) too quick“. I think he’s right, and will try to make amends with what follows.
My argument is organised in two parts, first I explore the notion of epiphenomenon in general, along with why it has no place in the kind of scientific epistemology I’ve been pursuing here. In the second part I’ll look at how the concept is applied to the mind-body problem, or philosophy of mind in general. Overall, I’ll argue that the whole idea is irredeemably flawed and unquestionably harmful when applied to minds.
Epiphenomena in general.
According to the Merriam-Webster, the definition of epiphenomenon is (emphasis added):
A secondary phenomenon accompanying another and caused by it; specifically: a secondary mental phenomenon that is caused by and accompanies a physical phenomenon but has no causal influence itself.
My thesis in short. if something has no causal influence at all, none whatsoever, then:
- We have no way of knowing it exists,
- We have no interest in knowing if it exists,
However, the final conclusion is that declaring that something is an epiphenomenon (in the strong sense) is equivalent of declaring it non-existent, on epistemological grounds. I must add that there is a weak version of the concept: one where an epiphenomenon is a secondary phenomenon which has very weak causal powers (with respect to the powers of what causes it), or, in formal scientific subjects, it has causal effects only outside the field of interest. I have absolutely no problem with weak epiphenomenalism. I don’t think it’s a particularly useful concept, for comparing causal prowess looks a little pointless to me, but I certainly see no fundamental flaw in the general idea.
Before expanding on my position, it’s worth restating a few general considerations about what is amenable to scientific enquiry: science is in the business of identifying causal relations between observable phenomena. It usually relies on quantitative measures because very little in the world is binary (either present or absent, on and off, and so forth). The much trusted idea of falsifiability enters the picture because when a causal relation is hypothesised and quantified, it becomes possible to make predictions: put A and B together and they will do nothing at all, boil them up, or add this catalyst, and they will form X amount of C with a small Y residue of D. With such predictions, you can test if your hypothesised causal relation does hold, and if it does, you’ll be able to reliably produce substance C. If you don’t get the expected results and you are reasonably confident that there is no flaw in your experimental procedure, then your theory about the causal relation at hand isn’t good enough and needs to be either modified or ditched. Thus, science broadly constructed is concerned with the domain of all causal relations in the observable universe. Some may quibble that science is also interested in the causal relations between the observable universe and what remains unobservable, but this doesn’t make my point less relevant. The main message is that if some X can cause Y, and we can detect Y, then both X and Y can be studied scientifically. Thus, science encompasses all causal relations, no matter how subtle or difficult to detect or isolate.
Incidentally, this view of science is what justifies the core attitude of New Atheists: they claim that various religions describe how God causes stuff to happen in our universe, and therefore the causal relation between God and observable facts is by definition within the legitimate remit of science. I have no issue with this view: if something has causal relations with our world, it can and should be studied empirically. I’m remarking this because the implications for philosophy of mind are far fetching: one could be tempted to propose that the mental domain is inherently non-physical. That’s fair enough, but the moment we propose that mental phenomena have some influence on our behaviour, we are still admitting they can be studied scientifically, thus, we are saying that this non-physical domain should still be studied scientifically (in terms of causal relations), making the initial claim that the mental is not physical rather moot: it may not be, but we are still interested in understanding how it interacts with the physical world. If it does interact, it potentially pertains science.
As things stand, it is therefore no surprise that, outside sciences about mind/brain, the idea of epiphenomena is largely absent: when used, all the instances I could find entail a form of weak epiphenomenalism. For example, a biochemist studying how a given hormone is synthesised may note the effects that the hormone has on various organs, and treat these as an epiphenomenon for the purpose of the current study. In other words, in standard everyday science talking about something always assumes that this something can have some effect, however indirect, otherwise talking about it wouldn’t be worth the effort.
More importantly, the idea of epiphenomena as legitimate scientific concepts is so rare that I was unable to find any attempt to reject it outside the philosophy of mind field. It seems that Epiphenomenalism is a concept that is used only in one particular context, and implicitly avoided in the vast majority of scientific disciplines. In my quest, I was hoping to find some ready-made argument in support of my thesis, unfortunately (albeit this indirectly supports my claim that science and Epiphenomenalism don’t normally mix), the closest I could get is Swinburne (2011): the article does concern Epiphenomenalism as employed in philosophy of mind, but it does take a broad view on what justifies our belief in scientific theories. For example, Swinburne claims:
A justified belief in a scientific theory (which is not itself a consequence of any higher-level theory in which the believer has a justified belief) requires a justified belief that the theory makes true predictions.
Unsurprisingly, he then reaches conclusions analogous to mine: namely, that epiphenomena can have no place in a scientific theory, or, more precisely, that “(with [one] very small exception) no one can have a justified belief in epiphenomenalism [of the mind-philosophy kind]“.
This leads me back to my proposition 1. (above): if something has no causal powers (none whatsoever) then, whether it exists or not makes exactly zero difference in the world. Zero, nulla, nada, nothing! I could rest my case here: if something has no influence on the world, it is by definition impossible to detect it. Moreover, it should go without saying that if something has no influence on the world, whether it exists or not is entirely irrelevant, for all possible purposes, both practical and theoretical. Note here that we are talking about absolutes: a pure epiphenomenon is something that causes nothing by definition, therefore it concerns us neither in theory nor practice. In contrast, we could have a hypothesis that implies that particle such and such has no influence on what happens on the human scale, but at high energies it suddenly does acquire causal powers, and therefore it is theoretically detectable. Fine: this has nothing to do with epiphenomena, it may be that particle such and such is currently an entirely theoretical notion (was never detected and we don’t know how to detect it), but still, at least for theoretical reasons, whether it exists or not it does concern somebody. By contrast: a model of particle physics which included something which has by definition no causal powers, wouldn’t make sense, it would mean that taking this something out of the model will let the model behave exactly as before – otherwise this something would have casual powers, according to the model in question. This doesn’t negate the fact that plenty of current hypotheses of theoretical physics imply the existence of things we can’t detect, not even theoretically: in these theories such things still have causal powers, but exert them in realms which we can’t reach. Thus, such theories leave me rather cold, but I can’t and don’t wish to claim that they are nonsensical.
More on the point, I maintain that, For All Imaginable Purposes (FAIP), claiming that something has no causal powers, is equivalent to claiming it doesn’t exist. Why? Because whether such thing exists or not, according to the claim, makes no difference, therefore their existence is irrelevant FAIP. Thus, calling something an epiphenomenon is very different from claiming “this thing exists, but I don’t know what it does” (i.e. it’s causal powers are unknown). Such a claim implies that we already have detected something, or have some reasons to expect that this something is in principle detectable, and because of its detectability, we assume it does have at least some causal powers.
I could go on an on, but I would be running in circles: it doesn’t matter how I approach the subject, I always get to the same conclusion. If something can cause exactly nothing, it can’t be detected, it makes no difference whether it’s there or not (not even to itself), and thus FAIP it doesn’t exist. Epiphenomena are epistemically transparent: we can’t detect them, and hypothesising their existence allows us to explain nothing, as they allow to make exactly zero testable predictions. That’s it, from a scientific point of view, proposing the existence of something which makes no difference is a complete and undeniable waste of time and resources.
It remains to be seen whether the concept has some other use: if there is a categorical difference between science and philosophy, and since so much of philosophy of mind uses the concept, perhaps it is a useful concept in that particular domain? This is the subject of my next post.
Bibliography:
Chalmers, D. J. (1997). The conscious mind: In search of a fundamental theory. Oxford University Press.
Swinburne, R. (2011). Could anyone justifiably believe epiphenomenalism? Journal of Consciousness Studies, 18 (3-4), 196-216.
Writing my own user manual 
Hey Sergio!
Just a few quick comments (well, knowing myself, perhaps a few more): I wasn’t meaning to make an argument for epiphenomenalism, but merely trying to express my doubts about both the idea that ‘existence = having causal powers’, and that whatever parts of the mental a zombie supposedly lacks (I would say phenomenal experience, but you’ve made a distinction there that’s not entirely clear to me) may have causal powers even though one can subtract them and get a zombie world in which everything is physically equivalent.
In fact, I don’t believe in epiphenomenalism myself. In brief, my own take is that there’s simply (at least) two stories that one can tell about certain events: a third-personal, objective one, and a first-personal, subjective, qualitative one. Both stories are mutually exclusive, in fact incompatible, complete on their own terms, and nevertheless necessary. In both stories, there’s also causality—my experience of the apple’s sweetness causes my enjoyment of said apple, and hence, it must have causal powers, just as my low blood sugar causes me to grab the apple, bite into it, and start the whole digestion process.
This may sound strange at first, but such a situation also exists in physics, with the notion of complementarity, or contextuality—in general, the story one tells (i.e. the physical properties that can be said to have definite values) depends on the measurement context—one can tell, e.g., a wave-picture story if one doesn’t collect information on which slit the photon went through, and a particle-picture story if one does. In both cases, there’s a perfectly causal story that can be told; but both stories are incompatible, and nevertheless jointly necessary to account for the full range of phenomena. (I won’t go too much into the details here—for once.)
So if one takes this relationship as a model for the mind/body relationship, then likewise, in certain contexts, a story of physical causality is appropriate—say, in Leibniz’ mill, or when one examines a brain via fMRI techniques, etc—while in other contexts, involving introspection and first-person descriptions, a story of mental causation is more apropos.
In zombieland, now, there is only one sort of story, that of physical causation—a perfectly complete story, in which no details are left out, safe, of course, for those regarding phenomenal experience.
Of course, you also know that ultimately I don’t think that zombies are metaphysically possible; only our powers of reasoning, limited to the computable, make it appear so. In fact, the world only fractures into different contexts because it is not fully accountable in terms of computation, and hence, to human explanation—thus, it’s an effective, epistemic split, not an ontologically fundamental one.
Anyway, that much for a kind of sketch of a picture that I think meets your criteria for the existence of mental phenomena, while nevertheless allowing for the (epistemic) possibility of zombies; thus, even if “existence = having causal powers”, I think there’s still some work left to do for you.
Now, regarding that identification, I must confess that the more I think about it, the more doubtful it seems to me. First of all, you seem to have moderated your claim somewhat, now holding it to be rather epistemic—i.e. that we can’t know about what has no causal powers. But this epistemic claim doesn’t allow you to draw metaphysical conclusions, i.e. to infer the nonexistence of phenomenal experience. But it seems to me that it’s that latter claim that you need for your model to make sense; otherwise, the committed epiphenomenalist can simply accept your arguments, not being bothered by the idea that we can’t ultimately know mental phenomena at all.
Besides, it’s not at all clear that we can only know what has causal influence. It’s plausible for every ‘external’ thing, maybe, but to postulate that we only know our own experience by it having causal influence upon us puts a curious division between us and our experience—that is, it makes us some thing that’s external to experience, rather being influenced by it; and with that, we’re squarely into homunculus territory. So our very own experience seems to be something we know intimately, yet not through any causal mediation.
Furthermore, and somewhat contrary to what you’ve written above, I think there’s several cases in modern physics where entities that one might characterize as epiphenomenal are accepted as real. For one thing, every cosmological theory features some areas of the universe that are forever causally separated from us; that is, whether they exist, or not, makes not one wit of difference to anything that we could ever become privy to. So here’s a further objection to your views: it seems there’s at least one thing we have good grounds to believe we know about without causal mediation, and that’s exactly the thing you’re trying to get rid off.
Why are those things there? Well, because they’re straightforward predictions of our simplest theories. For one, to our best knowledge, our universe is spatially flat, that is, has no large-scale curvature. But if this is the case, then our theories predict that the universe is also infinite, and hence, contains infinitely many areas from where no causal influence will ever reach us.
Even more, it seems to be expanding, and the expansion appears to accelerate; thus, even if the universe should turn out finite (if it has some large-scale curvature merely undetected as of yet), parts of the universe drop out of causal contact with us. Eventually, galaxies reach a certain point from beyond which no signal can ever reach us, even in principle: expansion leads to galaxies receding from us faster than the speed of light (which is not a violation of special relativity). Does a galaxy that drops out of causal contact with us cease to exist? Couldn’t it be equally well be argued that, from the point of view of a hypothetical alien in such a galaxy, we would cease to exist? Yet I feel pretty existent, and presumably, so does the alien—even though we could in principle describe the universe as if that galaxy were absent.
There are other cases of similar phenomena—one might think of black holes, for example, but that’s a questionable case, since black holes eventually do evaporate, and are widely thought to release the information about everything that’s fallen into them in this evaporation. But in fact, that’s under hot debate (it’s not called the ‘firewall problem ‘for nothing!) at the moment, and some have suggested that whatever falls into a black hole is just gone forever. But falling into a black hole is nothing special at all: for a sufficiently large black hole, we could just have fallen into one, without noticing a thing (until we meet a violent end at the singularity, or whatever’s actually there, or some earlier time, when tidal forces tear us apart). However, you certainly cease to have any causal powers with respect to the rest of the universe once you go down that particular rabbit hole; but it would be strange to say that you cease to exist.
Also, some interpretations of quantum mechanics suggest the existence of properties—such as simultaneously exact position and momentum values—that are not even in principle jointly measurable, and hence, could not have any causal influences (otherwise, you could simply deduce them from those influences). Other interpretations posit parallel universes, between which no interaction is possible.
Anyway, point being: it’s at least a widespread belief that one should accept as real those entities that are indispensable for our best physical theories, and there plausibly are such entities that have no causal powers (since leaving them out would mean changing the theory).
So, why could the mind not be like that—behind a kind of ‘phenomenal event horizon’? Everything that crosses the threshold into conscious awareness ceases to exert any causal powers with respect to the rest of the universe; but nevertheless, it continues to exist.
Thus, I think the case that there are no causally inert entities that one wouldn’t nevertheless call real is, I think, a very difficult one to make. But the problems don’t stop here: what about, for instance, the abstract? Does pi have causal powers? And if not, does that mean that it doesn’t exist? Then what is that thing that keeps cropping up in my calculation? And other abstract concepts beyond that: take, for instance, causality itself—wouldn’t you need to say that causality exists for your account to make sense? But what causal powers does causality have? If A causes B, and nothing but A causes B, then there’s no further reified causality necessary to bring about B; causality causes nothing.
Speaking of which, shouldn’t we at least allow for the possibility of nothing existing? But nothing causes nothing—ex nihilo nihil fit. Your account of existence would seem to solve the old ‘why anything?’-problem; but the solution strikes me as somewhat cheap.
Alright, this has gotten long enough, I’ll leave it at that for the moment, and give you a chance to formulate an answer…
Whoops, something got garbled: the sentence starting with “So here’s a further objection to your views:…” was meant to be the last sentence of the paragraph above the one it now concludes. Sorry for the confusion!
Thanks Jochen!
I’m afraid you’ll get another meta-comment…
Well, I wasn’t exactly responding to your point either. Our discussion made me question my position in a wider sense, and yes, as you note, it made me apparently modify my position slightly: I’ve moderated my claim about ‘existence = having causal powers’ by making it explicitly a matter of epistemology. On reflection, I realise that I haven’t changed my position, I’m merely making my premises more explicit. In this view, before writing the above, I’ve considered the option of starting with a seemingly unrelated post about how ontology largely follows epistemology. However, I’ve decided not to, for various reasons:
1. Laziness: I had to do plenty of background checking to be reasonably confident I wasn’t going to make a fool of myself.
2. Opportunity: while I did have enough oomph to write the above (as a consequence of our discussions), it seemed silly to invest it on the digging of yet another, marginally related rabbit hole.
I’m glad I went this route, because I’ve realised that Swinburne’s paper is implicitly backing my position about ontology versus epistemology, so when I’ll write about it I’ll have one more piece of ammo.
Anyway, I’ll resist the temptation to answer in detail, for now. At the moment, the priority is to write the next post. If we start sparring on a point by point basis, I fear I’ll never get to it. FWIW, the current plan (which is likely to change!) is to write a “part 2” post about Epiphenomenalism in philosophy of mind, a third post explaining why ETC starts with (relies on?) the assumption that Epiphenomenalism is nonsense. Once this is done, I would have explained the scaffolding which is necessary to respond to the objections about ETC you’re raised on CE.
I do plan to reply here on the general points you make (about science), while the parts about zombieland and the like are better suited for the next post, I think.
[…] that I’m probably missing something important. This post is the second part of a series: in part one I’ve explained the concept in general, and explored my understanding of why it seems absent […]
Jochen,
Thanks again!
Now that the second post is out, I can tackle some of your actual points. To keep us disciplined, I will exclude plenty and concentrate on the questions you raise about epiphenomenalism and science in general.
Before doing so, a few words on what I’m going to skip and why:
1. “[the] two stories that one can tell about certain events”. This is interesting, and something that for me translates to the form of dual-aspect monism that I favour. I do hope to write on the subject explicitly, so perhaps it’s worth waiting?
2. Zombieland: I think the second post is a better fit for that discussion. BTW, the second post may already be an indirect reply to your prominent concerns (of course this doesn’t mean I expect it to be a convincing reply!).
Right, you raise plenty of general (i.e. outside the philosophy of mind topic) objections, and mainly worry that my claim “declaring that something is an epiphenomenon (in the strong sense) is equivalent of declaring it non-existent”. Some of these objections rely on actual scientific theories, one (the last) is more general and I’m not sure I can answer to it. However, as far as physics goes, I think you are actually making my point, because all the things you mention aren’t strong epiphenomena.
I’ll tackle one by one:
a. We presume that the universe contains infinite things with which we won’t ever be able to interact with in any way.
Yes, we do presume so, but we don’t presume that these things don’t have causal powers. We merely presume we won’t be able to learn about them. Thus, speculating about such things still makes theoretical sense. Besides, speculating about them might give us ideas about how to causally interact (learn about) some of the things that are currently beyond our reach, so it also makes practical sense. Of course, if the hypothesis about the infiniteness of the universe is correct, this would still leave infinite stuff out of reach, but would nevertheless make both theoretical and practical sense to try our inferring luck and make reasonable hypotheses.
So here the distinction is squarely epistemic: there are things that are outside our sphere of influence. Thus, we can’t know about them, we are sealed off. Do they exist? Well, we can’t know, can we? We can presume, and do so with a good amount of confidence, no more than that. But still, we are not presuming such things have no causal powers, in fact we presume the opposite, and add “shame that we’ll never learn about them”. In other words, we are most definitely not claiming that such things are epiphenomenal in the strong sense, and therefore the whole argument does not apply to my claim at all.
b. Black holes: same thing! The fact that we presume that some parts of the universe can’t interact with other parts is a very different claim from saying that “this part of the system may never interact with anything at all, not even in theory, and not even its internal parts”. No scientist in her right mind has ever proposed a model with such a part. In fact, we do assume have very scary causal powers, don’t we?
c. In QT, position and momentum values are not even in principle jointly measurable. Thus, either one or the other can have causal powers at any given time, but not both at the same time. It follows that neither are strong epiphenomena. That’s why we need to include both in our models: they both potentially have causal powers. But this example is even better: our theory says that once we’ve measured momentum, talking about position stops making sense, does it not? Could you stretch your imagination and say that measuring the momentum makes the position disappear?
e. Parallel universes: epistemic horizon again. We hypothesise that parallel universes are forever outside our reach, not that they have no causal powers. Having no causal powers on me doesn’t mean having no causal powers.
I think this is the full list. It follows that, even if you tried, you could not find one single example of a genuine epiphenomenon which is actively included in some scientific theory. Thanks, that’s my point: to make at least theoretical sense, hypothesising the existence of something entails assuming that this something has some causal powers. Whether we can measure those or not is irrelevant, for theoretical purposes.
Overall, you are reinforcing my hunch: declaring that something is an epiphenomenon is one and the same as declaring that we have no need to explain that something (but that we do have a different need, though).
First of all, your examples from physics and cosmology entail that there is plenty of stuff that has causal powers and we still can’t know about. This is very different from the claim that “we can’t know anything about what has no causal powers”. The two claims don’t contradict one another, can’t we accept both as true? They are epistemic, but neither equivalent nor mutually exclusive.
Second, yes, I do need to go one step further, but I think I can still remain on epistemic grounds. How about the following?
If we want to know about something, we may not know what this something is, but we do imply that this something isn’t epiphenomenal, because otherwise we’d have no reason to investigate.
I’ve tried to hint to this in the second post: if we want to know about something, and it turns out that this something doesn’t exist, we’d still want to know what caused our initial desire, wouldn’t we? Similarly, if we want to know about something and it turns out that this something is an epiphenomenon, we’d need to conclude:
-We can’t probe this something, because it has no causal powers.
-We can’t even figure out if/when it does exist, for the same reason.
-What caused us to want to learn more can’t be the epiphenomenon, so probably, we do want to understand the cause of our desire.
Once we explain this initial cause, and have in the mean time found out that the thing we wanted to explain is an epiphenomenon (a difference that doesn’t make a difference, i.e. something that is impossible to even notice, let alone measure), we are left with nothing else to explain.
I’ve included these remarks here because they don’t necessarily apply to minds only: is it ok for you to discuss the minds part below the other post? We can discuss everything in one place if that makes it easier, after all.
Finally, I am completely lost when it comes to your final points about abstract entities, causality itself and the existence of nothing. I do think that tackling the relationship between epistemology and ontology would be necessary to frame these problems in a way that I find intelligible. Do you think we can have a useful conversation while postponing the discussion on these particularly tricky cases? I can try to respond, if needed, but I won’t promise I’ll make much sense…
Thanks!
Hey Sergio,
upfront, I’m sorry if I don’t slice up my replies to neatly align with the topics of your posts, as you requested—I think these are very interconnected matters, and some clarity would be lost in doing so overly rigidly. I also haven’t read your second post yet, but wanted to react to your points first.
OK, so you only believe that what things that don’t have any causal powers per se, not those that have no causal powers with respect to us, don’t exist. I find this confusing right off the bat: we can’t know of either (provided we only know of things through causal mediation), in just the same way. So your criterion for having access to entities treats both strong and weak epiphenomena the same, yet you want to draw a distinction.
Personally, I must say I don’t really see the grounds on which to draw this distinction: you probably would hold that some A that causally interacts with B and C, none of which interact with our universe, is not a strong epiphenomenon—A has causal powers. However, how about the whole (causally closed) complex A, B, C? No lines of causal influence ever leave it, and hence, it has no causal powers, and would be a strong epiphenomenon. So, we have a strong epiphenomenon whose parts are not strong epiphenomena—which would, in your terms, imply that the collection of A, B, and C doesn’t exist, while each of A, B, and C does.
And again, while it may not have causal influence in the physical sphere, phenomenal experience may well be causally efficacious within the mind—with the pain of stubbing my toe causing my (subjective feeling of) being angry, and similar. So I think if you consider something that has fallen down a black hole a weak epiphenomenon, if you want to be consistent, you should consider phenomenal experience likewise.
Then there’s the matter of your stipulation that we can only know things via causal mediation. It seems to me that minds are the best counterexample to this: we certainly don’t know our own minds because they causally interact with us—there is no ‘us’ beyond our minds that they could interact with (and supposing otherwise incurs the homunculus regress). So ultimately, I’m not sure if this notion isn’t just a red herring.
Ultimately, there are three notions you’re concerned with: existence, having causal powers, and knowability. You hold that only that which has causal powers is knowable, that our theories ought to only concern the knowable, and that hence, we can treat the unknowable, that having no causal powers, as non-existing.
First of all, that argument, even if I agreed with all its steps, falls short of establishing what you need to establish, namely that the causally inert doesn’t exist. We may leave it out of our theories, perhaps, but there’s no metaphysical guarantee that only our theoretical entities exist. Indeed, in the case of consciousness, the causally inert thing is the theory’s very object, even if it has no consequences for the rest of the world—certainly, the world could in fact be that way: consciousness is real, but exists only as a cul-de-sac for lines of causality. But then, consciousness nevertheless must factor into a theory of experience, as explanandum. In such a world, you could run your argument exactly the way you do; but you’d be wrong. Hence, your argument fails to establish what you need to establish.
Furthermore, I think that basically every turn of the argument is at least questionable, if not fallacious. Consciousness is knowable, but not through causal mediation. Our theories are concerned with unobservables of various kinds—and I think at least the parallel universe case above isn’t refuted by your remarks: while there is causality within it, a given parallel universe has itself no causal powers, not even in principle (this is a necessary consequence of many worlds interpretation—universes are either identical, or completely causally separated).
Anyway, I’m going to go read your second post, to see if some of these concerns have been addressed there…
Jochen,
fast and short(ish) reply (by my standards)!
First, you are doing me a favour, no need to apologise for anything! You’re helping me in return of nothing but mild amusement, in the best case, so by all means, you’re welcome to proceed as you wish.
Second, by taking the side of the opposition, you are yourself making the assumption that gets us in the deadlock I’m trying to break:
In the above, you smuggle in the epiphenomenal quality of consciousness. I can’t figure out a way to reasonably accept that claim a priori. What is it that justifies it? The point of part two is all about that assumption: once you make it, you are declaring that consciousness is knowable in some mysterious a-causal way, and therefore:
1. Unexplainable, because how do you explain how we get to know it, if we know it without cause?
2. Irredeemably dualistic, because at least its knowability isn’t due to a mechanism.
3. Nonsensical: if we get to know about C but this happens without a cause, how do we get to know it? If you think something is a-causal, you should at least have the decency of not expecting an explanation ;-).
[See this essay by Brian Earp for another way of making my same point.]
So, in short, unless one accepts that your quote above needs to be vigorously questioned, the discussion can’t lead to anything: the epiphenomenalist can always flip the argument to its favour (which is what you’re doing, in part for the argument’s sake, I presume). As I’ve said: it’s an unassailable position, because its absurdity is well buried in its premises.
OTOH: now I think I should have written my post on epistemology Vs. ontology before writing about epiphenomenalism. As it stands, you are right at pointing out the confusions that arises between existence, having causal powers, and knowability. I promise I’ll put some order on this as soon as I can.
finally:
Hem, no. No homunculus regress, not really. You are the first one showing that it can be escaped, and ETC does it in a similar way. Otherwise you smuggle the same assumption again “we certainly don’t know our own minds because they causally interact with us” can be escaped in two ways: the most tempting is epiphenomenal (equivalent of positing that minds are unbreakable black-boxes), the other is by saying “we get to know something about our own minds because of causal interactions therein”. You then realise that this ability entails big limitations, and you end up proposing something like BBT, ETC, the Attention Schema, or the view proposed by Metzinger. All these share the same starting points and in practice they don’t venture very far from one another. The point however remains: if you start with the assumption that consciousness requires some a-causal element, you get stuck in some spooky extra-physical domain and can’t come back.
I don’t agree that holding that we get to know our conscious experience a-causally implies epiphenomenalism; but first of all, even if it did, I wouldn’t incur some question-begging problem: you are trying to argue that epiphenomenalism is incoherent, so all I have to do is to show an example where it is, in fact, coherent. For this, I can assume it nonfallaciously—this would only be a problem if I were to argue that epiphenomenalism is right. The burden of proof is not symmetrically distributed in this case.
The problem with the stance that all we know, we know via causal mediation, is that it does incur a regress: if for any A, for X to know that A, it must be the case that X causally influences A, then every time we use this relation to explain how X knows A, we must introduce an infinite chain of X’s. Consider our phenomenal experience, that causally influences some X: how does X know it has been influenced? If that is again supposed to happen via causal mediation, then we’re down to the regress; if not, then causal mediation doesn’t exhaust the ways we can get to know things. It’s the same with postulating God as uncaused first cause: either everything needs a cause—then God needs one, too. Or, some things don’t need causes—then we have no need of God.
In my model, there is no need for such a regress—there is no external causal factor: due to it being a certain way, it produces a certain daughter pattern. There is an external causal factor that determines it being that way (a sensory stimulus, for example), but the stimulus does not cause the replication directly. Thus, the von Neumann mind meaning something to itself is not via a causal action of something upon it; it’s simply its being a certain way that suffices for its knowledge. (Also, but this is a minor point, I think you’re in places dangerously close towards equivocating between subjective experience and intentional content: my von Neumann minds have, arguably, the latter, but there is no reason to suppose they possess the former.)
Take again my example of a computation being ‘looked at’ through a screen (see my reply to part 2). For definiteness, let the computer perform some simulation, whose result you can either look at, or not. From the computation’s perspective, it matters not one whit whether there’s actually somebody looking at it. The ‘outside world’ is causally closed: there is no channel of input altering the details of the simulation. The user experience is thus a strong epiphenomenon with respect to the computation (although I’m still not sure how you regard complexes of things that may have causal interactions among one another, but which are causally closed in their totality, i.e. which have no ‘outgoing’ causal power).
This epiphenomenalism is wholly non-mysterious, explicable, and relevant regarding the question of whether there is experience. There is no causal back-action upon the simulation, and hence, one could describe it in its entirety without recourse to the question of whether there is a user looking in, or not. But concluding from there, that the zombie-world in which there are no users, and the world in which there are, are indistinguishable would be a mistake—in one, there is a user experience, in the other, there isn’t. If one thus believes your argumentation—the conclusion of which would be that it doesn’t matter whether we include the user experience (on the epistemic reading), or that the user experience doesn’t exist (on the metaphysical reading, where I’m still not quite clear which one you intend), one runs the risk of leaving out the most important part.
Jochen,
reading and re-reading you, I must conclude that you’ve lost me, because what you say doesn’t make much sense to me (deja vu?)… This reply took a long time to write (and re-write) because of this difficulty. In the end, I settled for the very direct and blunt approach, knowing we can adopt it without descending into fruitless confrontation. The result is also fragmented, that’s because, despite my attempts, I could not find a common thread to unify the various segments (excluding my inability of grasping your logic 😉 ).
Infinite regress of causality: as you point out, whether you start from “what causes knowledge?”, or a simpler “what causes X?”, the issue of infinite regress is always there. The way to escape it is to give up grand notions of explaining everything and pick more local versions and ask “with these basic ingredients (stuff we won’t try to explain), can we explain X?”. So for us, the question is “with mechanisms as our ingredients, can we explain PE?”. (My point in extreme synthesis: If you embrace epiphenomenalism, you have already answered “no”.)
I don’t see how you are going from the risk of infinite regress to the conclusion “we must be able to get some knowledge without anything causing that knowledge”. This to me may imply epiphenomenalism (I agree with your rebuttal, I was hasty in my previous reply) because positing that we can know about something without a cause of that knowledge directly allows to know about epiphenomena. Trouble is, following this route doesn’t really explain anything, we can only accept the story as is, because we assume there are no causes for our explananda. So this strategy looks to me as a self inflicted dead end: if nothing causes X, by definition we can’t isolate what we need to have in order to make X happen.
Second: there is the possibility of (somewhat) escaping infinite regresses, and it seems to me (apologies: I need to ask you lots of basic questions about the MM paper and/or find the opportunity to study the theory behind it on my own, didn’t find the time for either!) that the solution you propose can be generalised as a mechanism equipped with some form of circular reference.
A copies B, B copies A, with A that gets some (constrained) characteristics from the outside is one way, but you can add many passages and many I/O routes along the way and still remain circular . In fact, you say:
Which is the flip side of my:
(might be worth rehashing the implications of adopting a typical reductionist approach.)
In other words, I just don’t see where we are disagreeing on these details: we’re interested in the mechanisms within brains/minds. In trying to explain how brains/minds work we simply have no interest in treating them as unbreakable black-boxes. But still, you maintain that “Consciousness is knowable, but not through causal mediation” and/or that we can’t explain knowledge because of infinite regression (although you do propose one way of escaping it).
So, either you are selectively ignoring what’s convenient to keep disagreeing (sorry, I’m repeating a very unpleasant argument here) or I’m missing something macroscopic in your argument!
Now, more on the details (my emphasis):
Yes, I’m following so far. Give me one single example where a scientific theory includes something that is a genuine strong epiphenomenon and where the rest of the scientific community isn’t ROTFL, and I’ll understand where we disagree.
The examples you’ve produced so far are either not epiphenomena or merely weak ones, with the one exception I’ll discuss below. Weak ones are fine, but don’t pertain epiphenomenalism of mental events, because this latter idea rests on strong epiphenomenalism without qualifiers. In that view, “mental events(a)” are caused by the physical but cause nothing, that’s the assumption I wish to refute – the more I try, the more I marvel at the fact that there is even a need to.
But from definition (a), since we physically talk about something which we call “Mental events(b)” it directly follows that the (a)-stuff is not the same as (b): if we want to explain mental events, we actually want to explain at least the stuff that causes us to talk about them. But since we have no way of establishing if/where/when a strong epiphenomenon happens, there is, again by definition, no way of verifying the correctness of any explanation of what causes (a).
By contrast, we do have ways of investigating about (b), and these would be the same here or in zombieland. This is a very concrete remark: standard psychophysics can investigate what it takes to make a stimulus reach consciousness, what neural activity goes with it and so forth. However, to establish anything, it needs to rely on one or the other measurable physical event; thus, if you take the epiphenomenalist assumption, you are declaring that the whole approach is never ever going to be able to shed some light on the chosen explanandum (a), because all your measures relate to something that physically causes them (b), so you already assume they don’t relate to epiphenomenal consciousness.
Going back to your comment from Jan 17, which needed more than my first blunt reply (my emphasis):
What I’m saying is that the assumption (or hypothesis) that consciousness is (may be) causally inert is manifestly false. We don’t know what exactly we refer to with the word/idea of “consciousness”, but since we talk about it, and since we are positing it as our explanandum, whatever it is, it is manifestly the case that it isn’t causally inert – something must be the cause of our wish for an explanation. The moment you are proposing something is “to be explained” you are implying that it has causal relations with your world, but if you then propose it to be a strong epiphenomenon you make each and every explanation impossible (and accept contradictory hypotheses).
Your point on parallel universes (which are, by definition unreachable: if they are reachable, they don’t count as proper parallel universes because causal links do exist, so they would qualify as “almost parallel” universes) is quite handy, actually.
First, nobody came out with the aim of explaining all these spooky parallel universes that we know exist but have no idea of why they exist or even why we know they do. Instead, they “prop out” of various interpretations/extensions of current theories, by inference.
Thus, in various guises, theoretical physics hypothesises that infinite parallel universes might indeed exist and that we could never know for sure. Fine: I don’t have a strong problem with such inferences. You have a theory, try to figure out what it means and/or how to extend it, generalise it (or whatevs), and you get to hypothesise some logically inferred (and surprising) consequences. Good for you. I’d strongly object if someone told me “therefore, on the basis of inferences from our best theory so far, then infinite parallel universes, do exist and are organised in this way, for sure!”. As you know, it’s the “for sure” bit that would trip my alarm bells. As long as they remain a well substantiated hypothesis, it’s fine. But do we know anything about such universes? No, we just suppose they might exist.
All we have are some hypothetical ontological claims with no practical consequences, inferred from existing theories. Importantly, no-one started their quest trying to explain the existence of parallel universes. Secondarily, people are indeed asking whether it is still appropriate to consider such theoretical efforts as standard science. To me it’s a silly question (no doubt, it’s tied to very serious issues about funding!), as it assumes that science and philosophy have each their own distinct essence, which I believe is not the case.
But still:
(I) people do question if theoretical efforts which start as speculation over and above scientific theories and end positing the existence of epiphenomena are worthy of the the grand title of “science”. This reinforces my point: epiphenomena don’t mix well with science.
(II) such theoretical epiphenomena are nevertheless “side effects” of active inferences of pretty complex theories. No one is going around saying “ah, so now we’ve finally explained those spooky parallel universes”, because, since they are supposed to be strong epiphenomena, we never collected the tiniest piece of evidence that they exist.
(III) no one I know of is planning to demonstrate by experiment that parallel universes of the kind that is causally sealed off from us actually exist. That’s because you can’t, if they are indeed sealed off. You can test other parts of your extended theory, and then (falsely) claim that this demonstrates that the whole theory is right, but that’s just delusional and hubristic.
By contrast, with consciousness, we start collecting the evidence it exists very early on in each of our lives: since there is evidence, it’s not an epiphenomenon. QED.
So, as far as your best case goes, we have:
– Inference might in some case suggest the existence of strong epiphenomena.
– These can’t be investigated, not if the inferences are correct and they really are strong epiphenomena.
– It’s still not possible to start with an epiphenomenon as your explanandum, because if it’s an epiphenomenon you can’t have any evidence to account for. [Interestingly: you may instead try to disproof the epiphenomenal hypothesis, and show that your explananda do indeed have causal powers, refuting the underlining theory.]
– You can have a theory that predicts the existence of epiphenomena, but it would be a theoretical dead-end (remains theoretical) unless you’ll find out that the theory was wrong by stumbling on evidence about stuff you presumed was epiphenomenal.
In the case of parallel universes, we have evidence of non epiphenomenal stuff, and some theoretical interpretations of the theories that account for that evidence also allow to hypothesise that there may be stuff which is epiphenomenal to us. In the cases of consciousness, mental events and the PB in general, we have evidence that they exist (we’re discussing about them!) and want to build a theory to account for that evidence. In other words, we are going in the opposite direction: if there is evidence, our explanandum is not and can’t be an epiphenomenon. That’s where the incoherence lies.
In light of the parallel universes case, I’ll go back to your words:
Sure, “a” world like that could exist, but it can’t be indistinguishable from ours: in that world, no one will ever propose PE as a legitimate explanandum because it’s a cul-de-sac and therefore it can’t cause anyone to question what it is. For questions about X to happen pre-theoretically (i.e., for any observable X), X needs to have causal powers on the observable world. That’s where the analogy with the parallel universes case breaks down: in the latter quest, indeed no-one came out with the idea of explaining why parallel universes so obviously exist, on the contrary, someone came out with the hypothesis that they may exist, even if we can collect no evidence that they do. It’s a subtle difference, but very significant, to my eyes.
Finally, we can take the case of a computer doing its thing while being attached to a monitor or not, and being observed or not by someone. You say:
I can agree, but then “a strong epiphenomenon with respect to X” is at best a weak epiphenomenon or not an epiphenomenon at all. This means that I could come along with you if and only if what we wanted to explain was the computation itself. But you then remark: “This epiphenomenalism is wholly non-mysterious, explicable, and relevant regarding the question of whether there is experience”. So you explicitly set our field of interest as larger, we are interested on what the computation causes in possible observers. Indeed it is not mysterious, but if we’re interested in whether the computation produces an experience in the onlooker, then we say: the computation happens, whatever the onlooker may do, and whether the onlooker notices something depends on whether the screen is hooked up as well as other factors (such as the onlooker actually gazing toward the screen). The fact that the onlooker can’t stop or influence the computation is integrated by noting that there is no causal link in one particular direction, wholly not mysterious and relevant. What is either mysterious or irrelevant is why anyone would bother describing the situation as a special case of epiphenomenon, which can only be granted by acknowledging that the epiphenomenon label only holds in one direction between two specific identities (very different form: any direction from one specific identity). This definition is the same as saying the user experience isn’t an epiphenomenon, it just doesn’t have the ability to influence the computation. Now, if you reverse your focus, and are interested on what the computation is doing, you can observe that the user is irrelevant (epiphenomenal) and therefore you can (and should) take out the observer from your explanatory model of what’s happening in the computer. Why on earth would you include an irrelevant element?
All the way, we remain on my side: you don’t explain strong epiphenomena because there is nothing to explain.
In summary, I’m baffled. What is it that you disagree with? Perhaps you are mixing standard pro-epiphenomenal considerations with some of your own thoughts, and that generates a confusion because I can’t properly identify the boundaries?
[Meta: you are currently engaged on what must be a more important discussion (to you), so you really shouldn’t feel any pressure to answer quickly – actually, no pressure at all!]
Sergio, sorry for apparently confusing you. I think your troubles actually stem from a few misunderstandings, that should be relatively simple to clear up. So, starting with:
The problem here is that it doesn’t help, in any way, to escape the regress. The troublesome bit of this regress is that before we can know anything, it must be completed. Any theory of knowledge where such a regress lurks at the bottom flatly never explains anything, since something impossible must happen in order for us to know anything at all—the problem is right there at the bottom, and thus, we can’t escape it by just focusing on more localized areas. Wherever we look, the regress is already there.
To escape it, the only way (as far as I can see) is to hold that there is no further thing to be influenced by our experience, and so on, but rather, that we are just that bundle of experiences (shades of Hume, here), and so, know what we know because our experiences are what they are, and hence, we are the particular way we are (knowing some particular thing, for instance). Otherwise, you’ll always have the problem that before you know the first thing, an infinite chain of causal influences has to meet its end—which simply can’t occur.
And it need not be: it merely needs to be the case that the (b)-stuff is always lawlike connected with the occurrence of (a)-stuff. Take again the example of users of a simulation: whenever particular things are being computed [(b)-stuff], the user has a particular experience [(a)-stuff]. So, within the computation, what causes talk about ‘mental events’, i.e. user experiences and so on, is wholly determined by (b)-stuff; but the conclusion that hence, there is no (a)-stuff would be a grave mistake, since any theory proposed on this basis would leave out the phenomena we’re most interested in.
The (a)-things may perhaps not be the things that are causally responsible for our theory-building within the (b)-world of the computation, but it still would be a very different thing to explain only the computation—leaving out all the (a)-stuff—versus finding an explanation that includes it. The latter would be the explanation we were actually looking for, while the former would just be a mistake, perhaps brought on by erroneous theoretical presumptions, such as that only things with causal powers exist.
It might seem strange that our phenomenal experience is not the cause for our theoretical inquiry into phenomenal experience, but it is by no means incoherent. At least you now seem to agree that scientific theories can produce epiphenomena of the required sort; so why are you so hostile towards the possibility that a theory, prompted by the (b)-phenomena, produces likewise an explanation of epiphenomenal (a)-stuff, just like a theory proposed to explain, say, the absence of magnetic monopoles may end up requiring the existence of parallel universes?
And thus, we have a theory explaining the existence of something (parallel universes) even though that something is not causally responsible for our creation of the theory—which would seem to be exactly the situation you say can’t occur regarding the mind. The (b)-stuff can instigate a search for explanations that end up explaining the (a)-stuff, but not if we start with the presupposition that there is no (a)-stuff; hence, that presupposition should be avoided.
Not if the evidence itself is epiphenomenal. That is, if there is (a)-stuff, then every experience I have is evidence of the existence of (a)-stuff. However, this evidence is with respect to the ‘outside world’ itself causally inert; nevertheless, an explanation purely in terms of (b)-stuff would fail to account for it, leaving us instead forever in the dark.
In that world, people will talk about all the (b)-stuff in exactly the same way as they would in a world in which there only is (b)-stuff; so they would talk just as we do. But that world—which you agree is possible—also includes (a)-stuff, and hence, an explanation for the world in which there is only (b)-stuff is not an explanation for the world in which there likewise is (a)-stuff, even if the presence of (a)-stuff makes no difference for the (b)-stuff (which, ex hypothesi, it can’t).
Then I must again ask you for your definition of epiphenomenon. If A, B, and C causally interact, but form a causally closed whole, is that whole a strong epiphenomenon? Are A, B, and C only weak epiphenomena? If so, do they exist, while the whole they make up don’t?
Because that is exactly the situation we have with regard to the computation, where the user level may include causal interaction among its parts, but is completely causally closed itself. I took you to mean that anything that has no causal powers, no outgoing lines of causality, is an epiphenomenon; now you apparently turn around and say the opposite. So I need some clearer definitions, I’m afraid.
Otherwise, it seems to me that the user-experience case is an exact analogy for an epiphenomenalist mind-body relationship, where your approach would lead to eliminating the user experience—and thus, the phenomenon we care about the most—from consideration, producing instead merely an explanation of the computation, without, however, there being any lights on anywhere. Thus, you end up explaining only the zombie world, which, if we actually live in a world with (a)-stuff, will simply fail to account for the full phenomenology.
But isn’t that what you’re saying an epiphenomenon is—something without any ability to influence something else?
Anyway, the point of the example is that one can come up with a case in which there is one set of things that is wholly inert with regards to the third-personal description, but where that set of things nevertheless unambiguously exists, and thus, must be part of any complete explanation; whereas your stance would leave it out. Similar examples would be Plato’s cave, where the bound up people have no influence on the shadows on the wall, or Leibniz’ prestabilized harmony, where mental and physical, like two clocks set to the same time, evolve in unison without there being any causal influence between both.
Ultimately, the relationship between mind and body could be of that sort; the possibility is perfectly coherent. Again, this is a metaphysical possibility, and hence, immune to your epistemic argumentation; at best, you could hope to find an argument that if the world is built like this, we can never know; but that doesn’t tell us anything about whether it is, or isn’t.
Jochen,
Perhaps we are getting closer to isolating the disagreement, but I must confess that all your efforts are having the effect of convincing me more and more that I’m right. However, the more convinced I am, the less I can explain why anyone could ever think otherwise! So I can’t say that I feel we’re making progress…
I’ll cut some details and dive straight on what I think is the pivotal point:
Taking a charitable stance towards your position, the best I can say is that I can’t parse the expression “evidence is epiphenomenal”. It’s gibberish to me, like “water is grit”. Following through, I’ll agree to go in the particular case of experience, so:
“if there is (a)-stuff, then every experience I have is evidence of the existence of (a)-stuff”
No way. Every experience I have (the evidence) allows me to try to figure out how they come into existence and so forth, therefore it is manifestly not the case that my experiences are, with respect to the ‘outside world’ causally inert. Every experience I have is evidence that experience is not (a)-stuff. I can only repeat myself, the moment I’m asking “what is experience?” and what “causes experience?” I have already ruled out that experience might be a causal cul-de-sac – if it were, I wouldn’t have any reason to ask.
Perhaps the key to make you understand what I’m saying is to note that we don’t know what it is the thing we call “experience”. We have evidence to account for (at minimum, the undeniable existence of pain and pleasure, along with the evidence that what happens in the physical world influences both), and need to investigate the causes of the evidence. Since we have evidence, it’s impossible that the evidence causes nothing, because if it did, we wouldn’t have the evidence (in fact, we seek pleasure and avert pain, with exceptions, but still – the evidence is all around us). In the case of genuine strong epiphenomena we would be unable to detect the existence of something to be explained, exactly how we don’t have any evidence of the existence of parallel universes.
This brings me to:
I can’t find a way to accept your first sentence. We don’t don’t have theories explaining the existence of strong epiphenomena: we have theories proposing the existence of genuine strong epiphenomena (parallel universes). Such theories (with respect to parallel universes) don’t explain a single thing. They explain other stuff (evidence, things we measure, detect, fiddle with), and in doing so hypothesise the existence of undetectable stuff. There is an analogy with experience, in the sense that proposing that experiences are epiphenomenal is somewhat similar: we have evidence, try to produce a theory to explain it and end up proposing that our evidence is epiphenomenal. It would be OK, if it wasn’t self-defeating, such a theory is manifestly illogical, it can only be wrong: we do have evidence to account for, saying that the evidence we have is epiphenomenal is not even a theory, it just doesn’t add up – we end up with the idea that can be summarised as follows:
(1)Premise: something (unknown X) causes A.
(2)Hypothesis: A is a strong epiphenomenon.
(3)We know that A, but if (2) is true, nothing is causing our knowing that A. [your “the evidence is epiphenomenal” as well as your previous “the causally inert thing is the theory’s very object, even if it has no consequences for the rest of the world” and “we get to know our conscious experience a-causally”]
(4)Knowing that A, we want to figure out what X is.
You may keep telling me that the above isn’t incoherent, and I will keep telling you that sentence (3) is manifestly absurd. At the very best, it begs the question of how knowledge can spring out of nothing, but at the same time forbids every answer: you can’t explain something if nothing causes it. Every proposed explanatory X will look the same, because our knowledge of A isn’t influenced by anything, so we would be forever unable to produce falsifiable predictions – ex hypothesis, we don’t detect (measure, fiddle with) A, we just know.
Which sends me back to my conclusions: epiphenomenalism can only send us in explanatory dead-ends.
By contrast, parallel universes are not the evidence we are trying to account for: being epiphenomena, no one in her right mind would expect to find evidence they exist. In fact, finding evidence of the existence of “parallel universes” directly invalidates the “parallel” hypothesis: if they intersect with our universe (produce detectable evidence), they aren’t parallel. Once again, the theories don’t “explain the existence”, because to explain something you need to have some evidence to explain. These theories “propose the existence” as a side effect of explaining something else.
Disagreed again! Scientific theories can [!]propose[!] epiphenomena as a side effect of the effort of explaining some other evidence. Scientific theories most emphatically don’t ever offer “it’s a strong epiphenomenon” as the explanation of some evidence. They don’t because it’s incoherent: if you do, you are implying that the evidence you have could never be collected.
In other words, we have evidence as a consequence of the (b)stuff, we then call this evidence (our first person experiences) epiphenomenal (can’t be!) and thus make explanations vain: if the evidence causes nothing, we can’t even know when we do have the evidence, let alone verify predictions. [If nothing causes our knowledge of the evidence, we don’t have knowledge of the evidence: knowledge of something must depend on that something, if it doesn’t it isn’t knowledge.]
On to our users in front of screens: we want to explain the user experience. Good. Why? By using the example of people looking at screens you are making it very natural to assume that we can know that users exist, that we can detect how they react to what they see on the screen. I.e., we start with the assumption that the user’s experience of looking (or not) at the screen isn’t a strong epiphenomenon. By saying that you want to explain the user experience, you are saying that it is not an epiphenomenon. In other words, you have (pre-existing) evidence that the users exist, may or may not stare at the screen, and may have different (observable) reactions to what is shown on the screen. That’s why you have an explanandum. If, by contrast, we only had access to data from within the computer (can’t be influenced by the user), we wouldn’t know that people might be watching at the screen, and proposing the user experience as our explanandum would have no justification. It would be like me asking to explain why elves don’t like disco music. Totally arbitrary. Furthermore, we’d have no way to check if our explanations are correct. We always go back to the same point: if you have a legitimate explanandum, you already know/assume it isn’t a strong epiphenomenon.
Once again, it doesn’t follow. First of all, I am saying that a world where PE has zero causal effects can’t be indistinguishable from ours. It just can’t (I’m being consistent: PZs are incoherent). PE has generated the present discussion, therefore it has causal powers (it is not (a)-stuff): what other evidence would you need? So, I agree that a world where PE is a strong epiphenomenon could theoretically exist, but no, it can’t be indistinguishable from our world, because in that world nobody could ask questions about PE, as they wouldn’t know it’s there. It’s always the same point, no strong epiphenomenon can ever be adopted as an explanandum, because adopting something as an explanandum already implies that this something has some causal powers. If there is something to explain, there is, ahem, something detectable to explain. If something is detectable, it has causal powers. I am still unable to see how it could be otherwise.
Other approach, you say: “people will talk about all the (b)-stuff in exactly the same way as they would in a world in which there only is (b)-stuff”. Ok, so let’s accept the two worlds are physically indistinguishable. You are making my point: in our world, we are talking about PE, and it’s the (b)-stuff that causes this talk. Still with me? Thus, since our talk is about how to explain something, and since we don’t know precisely what this something is, it follows that what causes our talk is in fact what we want to explain. Therefore, we don’t want to explain the (a)-stuff, because, ex-hypothesis, whether it exists or not, it makes no difference. Instead, we want to explain the (b)-stuff. By proposing an explanandum, we imply our explanandum isn’t an epiphenomenon.
So, going back to the possible worlds: if the two worlds are indistinguishable, (a)-stuff makes no difference and doesn’t require nor admit any explanation (following my main point: can be regarded as non-existent). If, by contrast, there is a world where PE exists and causes nothing, then that world is distinguishable from our, because in that world nobody would go around asking what this undetectable (never detected) PE is.
In short, the counter examples you have provided don’t work. They don’t work because in the case of PE, the proposed explanatory theory says that the chosen explanandum is a strong epiphenomenon, which is incoherent. In the cases you’ve provided, the explananda are not strong epiphenomena, and for this reason they don’t mirror the case at hand.
Sure, your examples are not incoherent, but they don’t apply.
Sergio,
your entire line of reasoning rests on the idea that everything we know, we know via causal mediation. But of course, we know that that can’t be true, as it leads to infinite regress—if there is a ‘knower’ separate from my experience, who has to be informed of my experience via that experience having a causal influence on it, we just never bottom out.
Rather, the buck has to stop somewhere. Everything we need to know we have an experience is having that experience—the idea that we could have an experience, and then, without any causal mediation, would still not know about that experience is just incoherent. So the arguments you propose are predicated on a false premise.
Additionally, I’d still like to know whether you consider a collection of things that interact with one another, but that is in itself causally closed, to be an epiphenomenon; the way you now talk about parallel universes lets me suspect that you do, and I will consider that to be your stance pending disagreement.
Alright, now to some of your arguments in more detail.
An epiphenomenalist would hold that our investigations are due to the (b)-stuff; but that doesn’t mean that there is no (a)-stuff. Our talk about PE is due to (b)-stuff, our theorizing is due to (b)-stuff, and so on; but our experience is due to (a)-stuff. Compare: our cosmological theories are due to evidence within our universe, our talk about these things belongs to this universe, but the theory we end up formulating contains parallel universes—in which there may well be other intelligent beings drawing similar conclusions. These ‘other beings’ are (a)-stuff from our perspective, and their existence, their theorizing and so on are explained by our theory, which was prompted by (b)-stuff.
The only difference is that in the cosmological case, we have no access to the (a)-stuff at all, while within consciousness, that which is most accessible to us is the (a)-stuff. So, if the world were actually such that there are parallel universes, then a theory including only (b)-stuff would simply be wrong, because there actually are those other universes. Likewise, if epiphenomenalism is true, then a theory exclusively in terms of (b)-stuff would also be wrong, even if it accounted for all the facts prompting the formulation of the theory (i.e. all the (b)-facts). Because there simply are also (a)-facts as part of the world, and they would be left out.
Of course, the hope is that a theory that explains the (b)-facts also correctly explains the (a)-facts, the same way that our cosmological theories also account for the existence of things within parallel universes. But either way, assuming from the get-go that only (b)-stuff can be relevant will be misleading.
What a theory explains is not exhausted by the set of things that the theory was formulated to account for—every successful theory explains additional elements of reality, which, if they are observable, then may be used for theory assessment by way of falsification. But that doesn’t mean that a theory can’t also explain unobservables—if, e.g., eternal inflation is true, then how parallel universes come to be is straightforwardly explained; they are simply other domains in which the inflaton field has decayed. Of course, this prediction can’t be used for falsification, but it’s nevertheless a straightforward prediction; and ultimately, the world may be such that it is true that there are other bubble universes, in which case, a theory not accounting for them—not explaining their existence—would be wrong.
But that’s not how it works on an epiphenomenalist picture. We have (a)-stuff experience alright, but that experience is not the reason that we talk about phenomenal experience and so on; that reason is given by, say, the neural correlates of experience, i.e. (b)-stuff. So this conversation, and all similar to it, is causally accounted for by (b)-stuff, but that does not mean that (b)-stuff is all there is.
Rather, an epiphenomenalist’s blue-sky dream would be that our (b)-stuff caused theorizing leads to a picture of the world in which we have, besides that (b)-stuff, also (a)-stuff, which accounts for the fact that our experience feels a certain way to us—in the same way that our (b)-stuff inspired theorizing about the cosmos leads to the creation of a theory in which there is not only (b)-stuff, but also (a)-stuff, e.g. other bubble universes besides our own. Again, that’s just a possibility for the world to be: it could include only (b)-stuff, but there is no metaphysical necessity for it to do so, much less one granted by your epistemic unobservability arguments (deriving metaphysical consequences from epistemic arguments is a category error).
But the knowledge doesn’t spring out of nothing: it is by being, in a sense, made up from the experience we have that we know what experience we have. Now, again, neither that experience nor the knowledge of it does have consequences in the physical world; both is (a)-stuff. But that doesn’t mean it’s not there.
Let’s consider the user metaphor again. Actually, the user need not even be a conscious being: let’s say it’s just a second computer with a camera eye, that records what’s shown on the screen. Certainly, that second computer has knowledge—or rather, information—about what it saw, in the form of a data pattern stored on its hard drive. It could, for instance, play it back. There’s nothing (in the (b)-stuff world, that is, that of the computation it is watching) that this information ever causes, but that doesn’t make it any less real, or any less ‘known’ to the computer.
But a theory in terms of (b)-stuff would simply fail to account for this information, and thus, as a theory of that world, would be false. Hence, considering all non-(b)-stuff as irrelevant, or worse, non-existing, at the outset may make it impossible to account for the world as it actually is.
The epiphenomena can never be the evidence that the theory set out to explain, that’s true. But a being in a parallel universe would have tons of evidence that, to us, is epiphenomenal, and a theory including such epiphenomena would include an accounting of that evidence. Likewise, while it’s not the reason for formulating a theory, a mind has tons of evidence of the subjective what-it’s-likeness of the experience it has; and a theory formulated without producing an accounting for this evidence would, simply, be wrong. Again, that evidence isn’t why the theory is formulated, but that is just a matter of what is epistemically accessible, not what is metaphysically true.
Consider the case of a far-future astronomer. We know we live in an expanding universe; so, in a few trillion years, all galaxies beyond our own will have disappeared beyond the universal horizon. Thus, the far-future astronomer will have no evidence from which to deduce the expansion of the universe. Hence, theories they produce will account for the evidence available to them, but will, if they don’t feature universal expansion, be wrong—they would, for instance, contain a small, static, and eternal universe, while the universe is actually much bigger (possibly infinite), expanding, and originated in a Big Bang. Their picture of the world will be wrong, because they don’t have the same epistemic reach we do—some evidence is just not available to them.
Now, consider a rogue astronomer who comes up with a theory of the Big Bang—say, for entirely unrelated reasons. Perhaps they just can’t stomach the idea of an eternal universe, or they’re just being willfully contrarian, or whatever. Nevertheless, there’s a matter of fact, even if nobody can detect it, as regards to who is right—simply because there is a way the world actually is, and it is actually the way the rogue astronomer supposes it to be, and not the eternal, static reality orthodoxy presupposes.
Certainly, this would be lamentable—there would not be an experimentum crucis that could decide between both theories, and hence, the future astronomers conceivably could never get at the truth—not in a scientific way, at least. But the world is under no obligation to make itself easily explicable to us. It could be the case that there are truths about the world that we can’t get at via the available evidence—that is, that evidence that can make a causal difference in the world.
The same situation obtains for the anti-epiphenomenalist in an epiphenomenal world. They would propose their theories, perhaps accounting for all the available evidence; but they still would be wrong: in fact, there is epiphenomenal experience in the world, that the theory of the anti-epiphenomenalist simply fails to account for.
That all of those facts that prompt theory-building are in fact all the facts that need to be accounted for is certainly something one can hope—and it’s something that I do believe. But it’s not something one can, as you do, require of the world. The world flatly won’t care.
This is again talking as if there is some external knower, that knows about both the user and the computation. But again, this talk is incoherent. Knowledge and facts about the world are not co-extensive: something can be true without anybody knowing this. (I’m talking, here, about knowledge in the computation.) So that there is a user, which either looks at the screen, or doesn’t, which is something that does not have any causal effect within the simulation; so, every theory that only takes causally efficacious elements into consideration will fail to account for it,
Thus, taking that theory, and building a world from it, will lead to a different world than the one the example stipulates: namely, it will lack any user experience. It will, concretely, lack the pattern of ones and zeros that contains a record of what the camera-eye saw when looking at the screen, since those ones and zeros, or their arrangement, have no causal influence within the computation. It’s only in the sense that this pattern must be present that there need be any ‘knowledge’ of the user experience; we need no further meta-user (and consequently, meta-meta-user, meta-meta-meta-user, and so on) to know about the user experience and the computation.
That’s only true if you already assume that epiphenomenalism isn’t right. To an epiphenomenalist, this discussion has not been caused by PE; it’s been caused, say, by the neural correlates of experience. But that doesn’t mean that there is no PE over and above these neural correlates.
Saying that a world in which PE is a strong epiphenomenon could exist, but that that world would be (physically) different from ours, is in fact inconsistent—because then, there is a physical fact that one can point to which differs if there is epiphenomenal PE, which hence would mean that this PE isn’t epiphenomenal.
But if the (a)-stuff exists, then a theory that we could come up with in which it doesn’t figure would be wrong. Take a god’s eye perspective: you could create a world in which there is only (b)-stuff—in this world, nobody ever would have felt a pain, would have enjoyed a meal, and so on. Nevertheless, people would talk the same way we talk right now—that’s what it means to have a zombie world.
Rather, if you want to create a world in which there is genuine enjoyment, pain, and so on, you have to fix additional facts—the physical facts alone don’t suffice. So there is a genuine difference between those worlds—the question of whether anybody has ever felt like anything. But it’s not a fact that’s fixed by the physical facts, and hence, if only physical facts figure into theory building, it’s not a fact that can be responsible for theory building; but it well might be the case that a theory can be found that does account for those additional facts, which would then be the right one—just as the rogue astronomer’s Big Bang theory is actually the right one, despite not having more empirical support than their colleague’s static universe.
Explananda are facts about the world—thus, if there actually are parallel universes, these are explananda for any correct theory of the world (those theories that leave them out, are wrong). We might not be able to get at them, but to expect that due to this, they don’t exist, is again to expect metaphysical conclusions from epistemic arguments—which just won’t happen.
So if the world is such that there is (a)-stuff phenomenal experience, then that is an explanandum in any theory that aims at being correct, as opposed to merely accounting for observed phenomena. That phenomenal experience, and here I agree with you, cannot be the cause of formulating a theory, anymore than parallel universes can be the cause of formulating a theory in which they exist. But whether the theory correctly accounts for the facts of the world is a question which does depend on whether there is (a)-stuff. You mustn’t confuse those two, or you end up reasoning in circles.
It occurs to me that you may be getting tripped up by thinking that, since they have no effects in the physical world and hence, leave no evidence, we can’t sensibly know, theorize, or even talk about epiphenomena. This is, however, not the case—after all, we do quite sensibly talk and theorize about parallel universes, without (I presume) anybody of us ever having been to one. Moreover, we can even talk about, and have knowledge about, non-existent entities—I can know, for example, that Sherlock Holmes lived in Baker Street 221B, and even know that he didn’t exist; both of which are true statements about Sherlock Holmes. So leaving evidence clearly isn’t a precondition of knowledge.
Indeed, it can’t be the case that all knowledge is only due to evidence, is empirical—that proposition itself—that ‘all knowledge is empirical’—is certainly not one we can know via empirical evidence. Knowledge can also be arrived at through reasoning—I know that there are infinitely many prime numbers, without there being any causal connection between me and prime numbers. Indeed, if anything, I would say that I am more certain of this knowledge than of any that I arrived at empirically—after all, there is always the possibility of Cartesian hyperbolic doubt, such that I might be merely a brain in a vat. And whether that is the case is not something I can ever know, so every empiricist conception has unknowables at its very core.
Thus, while I can’t come to the position of epiphenomenalism via the evidence left by the epiphenomena, that doesn’t mean I can’t rationally arrive at it—I could, for example, believe that some particular theory of nonreductive physicalism gives the correct account of intentionality (or some other feature of the mind), and that theory might also imply that qualia are epiphenomenal—such that then, I would be as justified in believing in epiphenomena as I am in believing that particular theory. Alternatively, I could have a priori grounds for this belief: say that there is only one consistent way for the world to be, and that happens to be one in which there are epiphenomena; then, a sufficiently great intellect could deduce the existence of epiphenomena simply from the knowledge that the world exists (the rationalist pipe-dream).
Indeed, if one is truly of the opinion that things we can gather no evidence for should not figure into the conception of the world, it seems the position one is driven to is one of idealism, not of mechanism—because mechanism already implies the assumption of the existence of a material world, which we, however, have no direct access to—everything we actually observe are mental phenomena; everything else is subject to the above-mentioned sceptical hypotheses. But we can’t know via evidence that we’re not brains in vats. So there’s ultimately no evidence for the physical world, and hence, it should be rejected.
So ultimately, it’s perfectly well possible that one can speak intelligibly of epiphenomena, and theorize about them. On an epiphenomenal conception, whenever I utter something like ‘I am in pain’, I actually am in pain, even if pain is epiphenomenal—it’s just that the same physical cause that makes me say that—say, the firing of c-fibers—also makes me experience pain. Moreover, that experience of pain can, e.g., cause me to be angry, or even cause me to want to know what this pain is. The physical correlates of my anger—elevated heart rate, high blood pressure, the uttering of choice swear words—will of course be caused by the physical cause of my pain, but that doesn’t mean that my experience of the anger is not due to my experience of pain; likewise, my physical behaviour of looking for an explanation of this pain is due to its respective physical causes, but that does not mean that my subjective experience of wanting to find out the pain’s cause is not due to pain. All you have to assume for this is some mild psychophysical parallelism—that whenever I show the behaviour appropriate to some subjective experience, I am actually having that subjective experience.
Hence, I can quite coherently say that ‘I want to figure out the causes of my epiphenomenal experiences’, since if I do say that (and epiphenomenalism is true), it is actually the case that I have the subjective experience of wanting to find out those causes. Of course, I can be wrong about this sort of thing—in a zombie world, I would be. But likewise could I be wrong about the existence of the natural world—in a brain in a vat-world, I would be. The question is simply what hypotheses we are willing to entertain, a priori.
So we don’t need to be causally connected to the things we investigate. Nobody ever had a causal connection to pi, or to the infinite set of all prime numbers. Nobody ever had a causal connection to parallel universes. On first blush, it might seem somewhat strange that the epiphenomena can’t be the (causal) reason for our investigating them; but this strangeness is far from sufficient grounds to decide that they can’t exist. Our belief in the external world is just as much an a priori conviction that we ultimately can’t justify by pure appeal to what is in causal contact with us, or what we have evidence of. So if you consider yourself reasonable in basing your inquiries on the hypothesis that the external world exists, you don’t really have grounds on which to call a belief in epiphenomena unreasonable.
Jochen,
thanks! The unrelenting pressure you’ve been applying is finally bearing some fruits (I hope!). I think I have a clear idea of why we’ve been talking past each other for most of our discussion here.
Your last comment in particular highlights plenty of things I can agree with, while the previous one made me worry that we might never even understand each other (i.e. at least agree about what we’re disagreeing on): it was hard to read, I had to interrupt reading it a handful of times before reaching the end, more than 24h after receiving it (a very remarkable effect in itself!). I’m not making a value judgement, just communicating a thing that happened ;-).
Anyway, over the week-end, life permitting, I plan to write a “rehash” post, to re-propose my argument in a way that you’ll hopefully find more intelligible, in the mean time, if you can indulge and haven’t already, you may want to check my previous post on essentialism, and in particular the last bit, under the “exceptions” header.
Actually, no, scrap that: the whole post is relevant as it really is foundational to my whole approach. Writing it has clarified my own position in an irreversible way (frightening!).
The key for us here is the implication about truth-finding, which I see as a major obstacle in our communications: to my eyes, your way of thinking sort of springs from the assumption that there are matters of fact about the outside world that can be shown to be true or false. Mine is the specular opposite: facts are more factual the more you move towards pure concepts (and away from the physical world). Facts about external reality are ultimately mind-tricks. Yes, I am fully aware of the contradiction, it scares me appropriately, but I don’t think it’s avoidable.
[For example: in my framework, regarding something as “non existent” can be more or less appropriate or correct, but true or false enter the picture as shades of grey (falseness) not as a binary distinction. If everything you know indicates that a particular something can have no effect, you’d be justified to regard it as “non existent”. Whether that something really exists or not is irrelevant, because all our ontological claims (I should use the word “beliefs”, really) can only be a function of our local epistemological situation. The theorist work in my view is to find/develop theories that are fit for purpose, the fitter they are, the “true-er” they are (and the more purposes they fulfil, the better), but absolute correctness is confined to maths and the like.
The moment you understand objective truths as legitimate but unreachable aims, and embrace local, relative truths instead, remarks like “if X exists, excluding it from your theory because X is a strong epiphenomenon would be (objectively) wrong” become nothing more than indulgent mind-games: all our theories are objectively wrong, all we care is whether they help making the world more or less predictable.
That’s why I don’t know how to answer your question about existence of stuff which is epiphenomenal only to us: in my position, the question is unanswerable and not worth asking (I will write more on this, it’s a promise!).]
Sergio, just some brief comments:
I do believe there are matters of fact about the outside world; however, they may never be demonstrated with certainty. That’s sort of the whole point of my last post.
Then I think you should all the more care about the possibly epiphenomenal nature of mental states, since they’re all you have, in such a case. Additionally, on this conception, I don’t understand your claim about basing existence in causality anymore—causal relations are just the sort of facts you now seem to be claiming are beyond our reach. Furthermore, I think it’s grossly misleading to talk about existence etc. at all on such a conception.
I think that’s an entirely valid stance. The only problem I (personally) have with it is that it entails giving up on explanation—every theory is then just a bookkeeping device. A spectacularly practical thing to have, certainly, but in the end, just useful fiction. To me (again, personally) that’s just not the kind of thing I want from a theory; indeed, if things are this way, I find I loose all interest in theorizing.
Jochen,
now we’re talking! After all we’re not that distant. I’ll reorder your last statements while trying to better qualify where I stand.
Well, mental states are all I have, indeed. That’s why I am trying to take the hypothetical epiphenomenal nature of mental states as seriously as I can. A very difficult thing for me to do, since the more I try, the more I find it nonsensical. Thanks to you, I’ve built a much clearer picture of why I can’t accept it, which will be the subject of my next post. I may not be able to finish it this week-end (I didn’t take into account the impact of the 6 nations championship: if I’m at home, not watching the matches is physically impossible for me 😉 ).
I’m not completely sure I understand what you mean with this sentence, but it may be pointing to the reason why I had to leave your question(s) about closed causal systems unanswered (for now). Indeed, in my “world” talking about existence is very tricky. All along, I’ve tried to be careful with my language, adding qualifiers such as “regarding X as non-existent” and the like, precisely for this reason. If you take my stance, ontological claims are always hypothetical, are never blessed with full certainty and always dependent on what it is that you are trying to achieve. So, depending on context, sometimes it might be appropriate to regard the same X as existent, sometimes not. This does make me dizzy, and it probably should.
Now I’m reaching the juicy part:
OK, the first thing to clarify is my statement that “Facts about external reality are ultimately mind-tricks”. What do I mean with “mind-tricks”?
Recall where I start in my post on essentialism: our brains build models, and to do that they classify entities and events that exist and happen in the real world. To do so, the first unavoidable step is to build distinctions: the apple is a separate entity from the tree. The trouble with this operation is that there is always an arbitrary element in doing so. I’ll fast forward for a moment and consider causality: if A always causes B and B is always caused by A alone, without exception, what justifies considering A separated from B? Nothing that I can see. For example, apples (B) are always created by trees (A), and indeed, when the apple is on the tree, there is no real objective distinction between the two (ultimately, deciding where the tree ends and the apple starts is a matter of convention). However, apples are edible, trees much less so, thus, we consider the apple a separate “thing”, which neatly accounts for the fact that we can pick and eat the apple while leaving the tree there. Thus, what justifies the cognitive operation is ultimately our own, intrinsic aim. You may find this perspective intriguing, because at the bottom it leads to the necessity of having self-references (without infinite regress!), and your own account of intentionality (the way I understand it) fits in this scheme rather well. So, overall, our idea that (some specific) trees “cause” apples, isn’t strictly true in objective terms, because the distinction tree versus apple is arbitrary.
However the distinction is necessary because:
(a) cognition is limited, it can’t encompass the whole universe in one sweep, so needs to make distinctions and simplifications (build models).
(b) cognition is justified by our biological needs, in this case, eating apples.
From this I derive that knowing something is, even if we may not immediately see why or how, always dependent on some of our biological needs. Those needs are the reason why we think about stuff, and intrinsically define what distinctions are justified. These distinctions are always somewhat approximate and somewhat “arbitrary”, in the sense that it’s not the world out there that makes them “objectively true”, their ultimate justification is self-referential as it depends on our own needs/aims.
Thus, the mind trick is about believing that objectively true facts about the world are made true by what the external world is, while it would be more accurate to acknowledge that they also strongly depend on how our minds/brain work.
From there, I derive that knowledge is aim-dependent.
In terms of the essentialist fallacy: the essence of the apple is a consequence of how we are, it doesn’t depend exclusively on what the apple is, so attributing an essence to the apple isn’t “objectively” justified, it’s only relatively justified.
On the other hand, when I look at a road-sign and interpret it as a road-sign that tells me to slow down, there is no error there, because the sign is a designed artefact, built on the basis of pre-existing idea of its essence, so overlooking the physical detail and just picking up the essence of the sign is somewhat more justified (albeit still not objectively justified in the sense that still depends on how humans work).
At the extreme of this route (from entirely physical, to fully “essential”) you get pure logic and maths, where 2+2=4 without any arbitrariness (and therefore no uncertainty) because we are dealing with pure concepts: we’ve already abstracted away all sources of undecidability.
In summary: if we define “objectively true” as what we can know for sure, without any doubt, the only objectively true statements that we can find are required to be 100% abstract. The closer we get to physical reality, the less “objectivity” we can find. A pretty surprising inversion, if you ask me.
However, this doesn’t mean that the reality out there isn’t real (your “there are matters of fact about the outside world”), it just means that our knowledge of what is real is always unavoidably coloured by what we are (the self-referential side, again).
This brings me to causality: the outlook above brings a very refreshing (to me) view on causality. In this framework, the world out there just is, and the very idea of causal relations is the consequence of our need of slicing the world into discrete entities. If you are still with me, it immediately follows that you can find alternative and perfectly valid ways of describing the world: depending on how you slice it up at the start, different causal relations will result. Certain classifications will lead to more precise, or more generalisable, or more reliable models, but because at the bottom there is always an arbitrary (self-referential) step, we don’t even need to assume that alternative accounts are necessarily mutually exclusive.
In the context of philosophy of mind, this last little addendum is of astonishing importance:
If we start by slicing the world up by defining each person as a separate identity (an individual self), an operation that is justified by our social needs, we end up constructing a vision which includes (classical) intentionality, meaning and the whole problematic bunch. In this outlook, people do things because of pain and pleasure (desirable and undesirable states) and mental events are causally efficacious.
On the other hand, if we slice up reality in terms of fundamental physics, we get mechanisms, neural correlates and the prospect of creating a causal picture that fully accounts for what humans (and living things in general) do in these different terms.
The two accounts seem mutually exclusive because in both cases we forget the arbitrariness of the distinct starting points. But from where I stand, both approaches are legitimate and their difference is a consequence of the different epistemological approach, not necessarily a consequence of how the world is. Thus both stances can be somewhat true and because our ontological conclusions depend on the epistemological approach, they might (I’m not saying they are) both contain comparable amounts of “truth”. That’s roughly why I regard myself as a dual-aspect monist: the underlying reality is the same, but the two approaches necessarily illuminate different aspects of it (ontological claims always depend on epistemological premises). The differences are ultimately caused by the fact that we have finite brainpower, and therefore need to start analysing things by slicing up the world in one way or the other.
Now, going back to the theorist’s job: one legitimate aim is to improve theories by making them as generalisable as possible. Having a very precise, very reliable and very local theory is quite good, but if you can expand it so that it becomes less local, without eroding precision and reliability, it would be better. Thus, if we can find a unifying account that can show how the two aspects (mental and mechanical) relate to one another, we would be hitting the jackpot. Importantly, on my account this equates getting significantly closer to what is really true out there, even if we are forever bound by our intrinsic limits.
With this long explanation, perhaps you can now grasp how profound is my commitment (Yes, I find being profoundly committed genuinely alarming):
Useful fiction, but unavoidable. Furthermore, it is also something that does ultimately capture snippets of truth about the world out there. The limitation is only about absolute truths, but there is no hard upper bound. We can get our theories to become ever more accurate, precise, reliable and generalisable, so it’s still a game that’s worth playing.
Furthermore, this account also helps understanding what an explanation is. Essentially (yes, being conceptual, explanations do have a legitimate essence!) explanations are predictive devices: they make our world intelligible by allowing us to “know” what is likely to happen when.
[Incidentally: that’s why I don’t buy your claim that some theories *explain* the existence of parallel universes (they predict it, which is different). In these theories, the part about parallel universes explains nothing because that part doesn’t in itself make our universe more predictable (i.e. accounts for no evidence).
Also: note that I’m not giving up to explanations. On the contrary, I’m gaining some understanding of what explanations are (assuming I am capturing some “truth”)!]
Overall, I think my stance accounts for cognition, for the reason why science is so fantastically efficacious, for the intrinsic uncertainty of knowledge, for the existence of our dualist inclinations, and much, much more. Thus, because of the extreme generalisably of this approach I find myself unable to let it go.
Finally:
I can really see why! I am myself very surprised of where I find myself, after all my very first pair of premise and aim was that:
“I believe that reality exists and is unique” and “[the] process of constant refinement and occasional re-invention of reality representations (models) is inherently good and it is what makes my own life worth living”.
As a result, finding myself in my current predicament is surprising, but hasn’t led me to lose interest. One thing I’m only starting to realise is how deeply humanistic my stance is: it places what we are and what we yearn for at the very centre of my efforts. At the same time it highlights why it makes sense to feed our inexhaustible hunger for knowledge and why I (we?) invest so much time and effort in this endeavour. I’m saying this because, if you’re still with me, you may get a glimpse of why what at first sight seems a fairly bleak stance is in fact full of hope and excitement.
Sergio,
I think you’re trying to have your cake and eat it, too, here. You’re sort of muddling down the middle between two options that I think are mutually incompatible, but each of which entails a commitment you’re trying to avoid:
1) You can hold that theories, in some approximate way, can at least get closer and closer to truths about the world. But then, you’ll have to take seriously the arguments I posed above regarding epiphenomena—in particular, you can’t use the argument that there are no epiphenomena, and hence, there must be some explanation of (evidently real) PE in mechanical terms, since there could be epiphenomena even though they are not capable of making themselves causally known to us. So a theory predicated on such an assumption runs the risk of failing to account for the phenomena actually present—in particular, a world build according to this theory would lack any form of phenomenal experience, i.e. be a zombie world.
2) You can hold that theories are just engines for prediction, for producing successful behavior in the world. But then, you can’t say that ‘that’s how phenomenal experience works’—at best, you can say that such-and-such a belief produces adequate behavior. You loose touch with how the world is, and instead you only concern yourself with what we can say about the world. In this case, you don’t have to worry about epiphenomena; but you also don’t have to worry about PE, or even about whether some things exist or not, because in this case, scientific theorizing simply does not concern questions about what’s really out there, but is a mere instrumentalist construct to account for the clicks of detectors.
Either would be fine—but you try to have the benefits of both, without the drawbacks of either: a theory that approximates truth in some relevant way, yet allows you to disregard entities you consider inconvenient. This, I think, gets you in both a methodological and metaphysical muddle, which you have to find a way out of by facing the problems that confront you before going any further in your inquiries. Otherwise, all you construct on top of this will be built on sand.
In particular, I still think that there’s a confusion between epistemological an metaphysical matters that underlies your stance: you say that our knowledge of the world is always just partial, fallible, open to revision, and bundled up with concepts whose roots are more in the human condition than in the world (putting in—or on?—some Kantian shades here). This, I think, is mostly reasonable.
But in order to reach your desired conclusion, that we can neglect epiphenomena, you need to make a metaphysical commitment—otherwise, all that follows is that while epiphenomena exist, we can never get at them, and that’s that (provided your argument goes through as intended, which I still don’t think it does). Hence, in particular, you can’t conclude that some mechanistic explanation of PE ought to be possible. Thus, you want to conclude from our inability of getting at the epiphenomena to their non-existence—but this is no longer a claim about knowledge, but a claim about ontology. And this claim is not warranted by your arguments, since the world is not constrained by our ability to gather knowledge about it.
So you can either hold theories to be solely instrumental—but then, you loose all pretense of them approximating truths about the world, or anything like that. Then, you genuinely need not worry about epiphenomena. Or, you can claim that theories come to approximate truths about the world—but then, you can’t neglect epiphenomena, because it might be true of the world that they exist (even if they do not make themselves known to us directly).
Jochen,
thanks again. It seems that we’ve managed to get over a major obstacle as I’m not having big difficulties in following your objections now (I used to, a couple of exchanges ago). So far, it’s all good, but of course I still disagree, and in part, that’s my fault, as I still haven’t managed to clarify how I get to categorically discount epiphenomenalism. Please bare with me, while I finish writing the next post which will be tackling that topic in particular.
For now, with my previous comment and the current one, I took a step back: one of the issues at hand have to do with the relation between epistemological stance/method and ontological claims. I do plan to write a separate post about this latter subject, but never felt ready to put things in writing, as I couldn’t find a way to make my argument clear enough (I write by the principle: if you can’t explain it in simple terms, you haven’t understood it).
Our discussion gave me the chance of testing my ability of writing it down without too much commitment: I regard comments as “work in progress”, it feels safer to test ideas within an ongoing discussion.
Furthermore: in my last comment I was also trying to show why taking a very weak epistemological stance, while also inverting the typical understanding of “objective facts”, we can still be passionately committed to theorising.
Bottom line: (if you can/wish) please try to let epiphenomenalism aside for a little longer, and let’s see if I can show that my position is indeed worryingly complicated, but I still don’t think is self-defeating. Moreover, I am trying to eat and have my cake, and do so at the cost of never eating the whole cake!
Your two points, depicting two apparently incompatible stances, offer a very good point of attack, once we remove the (currently) confounding consequences about epiphenomenalism. To get started, I’ll rephrase you:
Assuming you are happy with the way I’ve synthesised the two approaches, my aim here is to show that they are not alternative. Instead, I believe that accepting the second directly entails the first.
So, let’s see: we start from the weakest possible stance (your second). Theories are “just” instrumental, we don’t have access to truth and just care about successful predictions. We should all be able to agree that this stance is pretty solid in its positive form (defining what can be done): our theories do indeed allow to make somewhat reliable predictions, and we are able to produce better theories as we go along.
If that’s the case, one can ask: why? The straightforward answer is that there are lots of regularities in the world, and that therefore we can proceed by identifying (empirical) and formalising (theoretical) them. I assume you can overlook the practical details (we frequently run ahead and formalise stuff before having any empirical data to match) and that we can agree with the statement above.
If that’s the case, that’s all I need: it follows that the regularities we are undoubtedly able to identify and exploit instrumentally are actually real (really real: they do exist outside there). But this entails the first stance: our theories get to capture some aspects of the real world, and can progress into capturing more and more truth.
To exemplify, I’ll hijack the Feng Shui example you’ve made recently. Let’s limit ourselves to the instrumental case (second approach) and claim that Feng Shui is a theory used to design architectures which maximise well-being. Let’s also assume it does work: building houses following Feng Shui practices does produce comfortable homes (in measurable ways, Feng Shui compliant houses are nicer than other ones).
Does this mean that mystical Qi force (or dragons) exist? Not really, however, we still have to admit that the theory must be capturing some truth about the world. Sure enough, it does start from ontological claims that we feel compelled to reject (existence of Qi, dragons and the like), but its efficaciousness tells us that it does get “something” right. The problem is that discerning what is right is prohibitively difficult, getting something right doesn’t automatically help to isolate what you got wrong.
I wouldn’t have the slightest idea of how to “correct” the ontological claims of Feng Shui, to the point that I have to admit that it may be impossible. Thus, we end up admitting that Feng Shui captures some truths (under the assumption that it is efficacious), but also that we can’t tell precisely what they are.
Theorising therefore is inherently hard: one can get some things spectacularly right, follow that route, and eventually get stuck in a dead end. Since there is no simple method to keep what’s right and discard/revise the rest, many theories dry up when pushed to the limit.
To me this view is quite convincing as it explains some of the puzzling qualities of scientific progress: from example, the controversial idea of “paradigm shifts” becomes immediately intelligible. Some ways of slicing up the world are inherently better because they allow to illuminate a wider range of phenomena. So far, mechanistic/reductionist approaches have proven to be very versatile, which doesn’t guarantee that they can be used to explain everything (in fact, we expect they will always fail to explain something), but does imply that the approach is particularly powerful (particularly in the domains of precision and reliability, while versatility lags a little behind). From here you derive the expectation that fundamental physics is able to slice-up reality in a way that captures truth to an unprecedented extent.
But you can even go one step beyond: this same outlook also suggests where the unprecedented efficacy of reductionism comes from. You get there by noting that it’s a method that revolves around systematically evaluating the efficaciousness of its foundational ontological claims. As a method, it is the one that has a direct feedback on its own way of “slicing up reality”. Thus, it is inherently powerful as it directly addresses the main difficulty that comes with building efficacious theories.
This doesn’t make (physical or otherwise) theories infallible, because we still start from the second stance, and it remains the case that the way we decided to slice-up the world will be able to illuminate some but not all aspects of reality.
The last remark is important to philosophy of mind: by acknowledging the limits of cognition (needs to start from somewhat arbitrary classifications – and therefore knowledge starts from your point 2), we can make the meta-prediction that, for the same underlying reality, any theory will be able to approximate some truths while necessarily neglecting some others. In this context, minimising the blind spots is very useful. From this I derive that the apparent impossibility of bringing together purely mechanical and purely mental descriptions (brain mechanisms versus classic intentional explanations) isn’t mysterious: it’s expected. Also: trying to build a theory which brings the two together might be possible, and would be a theory which is better, because it has fewer/smaller blind-spots (generalises better than the current ones). In this way we get to describe our common aim (linking physical and mental), have a clear explanation of why it’s desirable and at the same time have an explanation (note that it takes the form of a prediction!) of why it’s so freaking difficult. In other words, this approach makes the challenge posed by dualism entirely predictable, which is a promising start (for me). At the same time, it puts the onus on the theorist: the challenge is to find a way of “slicing up the world” which may bring together mechanisms and mental events.
So far, I’m still with ‘classic’ Chalmers: his attempt of adding new fundamental ingredients to “explain” consciousness is 100% compatible with the interpretation I’m offering.
However (I come back “on topic”), I am more and more convinced that the following step goes wrong, and yes, that’s the introduction of epiphenomenalism (not discussed in this comment!).
In short: I think you’re asking me to choose between two alternative approaches. I refuse to do so, because I don’t see the two approaches as alternative: one entails the other. Furthermore, accepting this connection immediately accounts for a lot of disparate pieces of evidence, from the apparent jumps and U-turns of scientific progress, to the possible efficaciousness of theories built on outlandish claims. It even accounts for why classic understanding of mechanism and the mental look so hopelessly irreconcilable, and in doing so, allows to tentatively reject dualism. As far as meta-theories go, it’s consistent and accounts for a lot of evidence. For these reasons, I find it convincing.
Alternative view: in your description of the second stance, you say “But then you loose touch with how the world is, and instead you only concern yourself with what we can say about the world”. That’s where I disagree: we never had touch with how the world is (we can’t loose what we never had), we can only start from what we can say about the world. It may be symptomatic that you placed “finding some truths” before “making some predictions”: I think you got the order wrong. We start with making predictions, and from there we get to approximate more truths by getting ever better at predicting.
Sergio:
Then I’m afraid you’ve misunderstood it. The point of this approach is that it doesn’t entail anything about the world; hence, the inference you’re trying to make—something like the familiar ‘no miracles’-argument—simply has no grounds to start from. In order for that argument to work, you already need to have a commitment to (some part of) the first stance, namely, that the regularities we discover are regularities of the world; otherwise, the success of our theories says precisely nothing about the world. If there are no ultimate, fixed truths about the world, then it’s simply not the case that the success of our theories tells us anything about the world. (If, say, all truths are socially constructed, then so will the truth of our theories’ success be; hence, that success does not allow us to infer anything about the world beyond its socially-constructed characteristics.)
Rather, to make that inference, you need to already assume that ‘the truth is out there’, so to speak; i.e. endorse option one. Then, the observed success of our theories allows you to conclude that they approach certain mind-independent truths. But then, you don’t get out of the muddle.
Jochen,
First obvious question would be, “misunderstood what?” What’s the “it”? Considering the following sentence, I think you probably mean your second approach. Since you’ve spelled it out, it’s possible that I misunderstood it, but at the same time, I think you’re misunderstanding me.
To try getting over double-crossed misunderstandings: I’ll put down my approach in a rather scholastic way:
Hypothesis: (H1) there are regularities in the world.
Hypothesis: (H2) we don’t have direct access to what’s out there. What we can do is make predictions and see if they work out. Thus, we can build predictive models.
These two are my starting points, and I’ve (perhaps mistakenly) attributed them to your second approach. Whether I’ve misinterpreted you or not, it doesn’t really matter to me, what I care about is whether the case I am making is solid or not. In your reading, it’s true that H1 already links to your first approach, that’s what I was trying to say (will come back to this later).
On H1: you simply need it to do any epistemological work. If there are no regularities in the world, everything is truly random, and:
(C1) no knowledge is possible.
(C2) no structures exist, as they would produce some regularities.
Thus, not accepting H1 is a definitive non starter.
Furthermore, the negation of H1, “there are *No* regularities in the world” is just so manifestly false that I seriously doubt you find it an acceptable. In fact, while describing option 2, you say: “such-and-such a belief produces adequate behavior”, and how can that possibly be if the world was completely devoid of any regularity? So, unless you have a rather spectacular trick up your sleeve or I’m barking at the wrong tree, your own definition of the second approach already implied some commitment to H1.
For these reasons, I never even considered the idea that your second approach was negating H1: it wouldn’t make sense. Even if we take the (not uncommon enough) view that the only regularities are socially constructed, we still have to accept that it’s nonsense, without even leaving the armchair: after all, the armchair itself displays plenty of regularities, many of those would hold even when “tested” by some non-human, non-social agent: even a roomba, a pet or a spider can bump into an armchair. Do I really need to make this point??
Instead, my take on your second stance, interpreting in the way that makes it sensible, is pivoting on H2. The key point being that we don’t have direct access to the world out there, all we have to work on are our own mental states. This interpretation follows our previous exchange:
“[Mental states] are all you have” is your own assessment of where I start from, after all. From there, you note that I’m trying to get the best of two approaches, and I answer: true, I am, because (any defensible version of) the second approach does lead to a defensible version of the first. The two approaches are not only compatible, one is the consequence of the other, but you are looking at them in the wrong order, the second precedes the first.
It is because we don’t have direct access that we can’t be 100% sure of what really is out there, and this does create a problem: if Feng Shui does work (assumption, made for the sake of the argument), does it mean that Qi really exists? I say: probably not. I also say: finding out what really real facts the theory captures is freaking difficult, perhaps impossible. And therefore I am watering down the first approach: we can get closer to truth, but for any given theory, identifying what’s true is probably impossible. This goes back to modelling: models get off the ground by assuming the distinctions they posit are “really real”, but that’s always somewhat arbitrary.
Instead of responding to these points, you have switched to what I think is an untenable version of the second approach, and told me “See? it’s untenable, so you you’re not making sense”. Sorry, but this won’t do: you’ll have to address my position to get further.
Once again: espousing option one (light version) doesn’t mean that we know what, in any given theory, is right and what is wrong. If we did, we would systematically and automatically discard the wrong parts. So yes, I am committing to some part of option one, the part that makes sense. As a result, the full argument is:
Hypothesis: (H1) there are regularities in the world.
Hypothesis: (H2) we don’t have direct access to what’s out there. What we can do is make predictions and see if they work out. Thus, we build predictive models.
First conclusion: (C1) the performance of predictive models can be assessed.
Second: (C2) for a given purpose, from C1, you can rank alternative models, based on how well they perform.
Third: (C3) you can keep going, tinker with your models, assess them again, see how they perform for additional purposes, find the variations that work best.
(C4) by doing so you get to capture more and more regularities about the world. You do capture some truth.
(C5) however, you don’t ever capture only the truth (this comes from noting that models require discrete elements, but there always is an arbitrary element in the distinctions you make).
(C6) (as per Feng Shui) for any given theory and purpose, separating what’s true from what isn’t is prohibitively difficult (perhaps impossible?).
(Corollary): this incidentally shows why reductionism works so well, it’s a method that inherently allows some degree of iterative self-correction (by making ever more precise, finely grained distinctions).
By C6, I’ve espoused a watered down, not-entirely positivist version of your option 1. This does allow to get out of a hell of a lot of philosophical trouble (causality, the problem of induction, dualism, ineffability, you name it!), which to me is a bonus. (It does feel cheap even to me, especially to me, but if it works, cheaper is better, isn’t it?)
Does this put me somewhere in the middle between two mainstream stances? I hope so! Would mean that I am producing some original ideas, yay! Does this mean I’m categorically wrong (I’m capturing less truth than both alternatives)? No. Having an unusual stance doesn’t automatically make it wrong. To understand where I’m getting it wrong, I’ll need you (or some other generous soul) to attack my schema above (the labelled statements). Going after mainstream theories, or forcing my view to adhere to them will not help, it will make us talk past one another again. I know you don’t want to do this, so I’m feeling free to ruthlessly point to the danger.
Trouble is, I find my labelled statements unassailable, but you should take this for granted: if I could spot the flaws myself, I would have fixed them already…
Other (shorter) approach:
No. I do need to commit to something that points to the first stance, that’s agreed, but I do not need to accept that “the regularities we discover are regularities of the world”. We discover regularities that, in one way or the other are co-caused by the regularities of the world, this is because we don’t have direct access. Having only indirect access, the regularities we find are indirect (always also a function of what we are), that’s what compels the conclusion that some uncertainty always survives (no theory about the outside world is 100% correct).
In short, “my” versions of options 2 and 1 are compatible, they depend on one-another and lead to conclude that we can get to more and more truths, but can never get to “only the truth and nothing but the truth” (someone should tell the judge! 😉 ).
Sergio,
Yes; I quoted your bit about ‘the second approach’, hence my ‘it’ referring to that. This approach was my attempt at spelling out what you would need in order to deal with the problem of epiphenomena in the way you seem to want to—namely, a position of there being no ultimate facts of the world (as opposed to our knowledge of these facts being limited of necessity).
Such a position is perfectly consistent, despite your last post’s rather sweeping denials, and has been formulated many times by many different philosophers—pragmatists, constructivists, poststructuralists, and so on. Consider science as modeled by the game 20 questions: at a party, one player leaves the room, while the others decide on some kind of object, or a person, or a concept, or even a story. Now, it’s the task of the other player to interrogate the ones that stayed in the room, by means of yes/no-questions, to identify whatever they have agreed upon. This is science, as usually conceived: provided the right questions are asked, the possibilities get ever more narrow, until they eventually converge upon some matter of fact.
Now, however, consider that the assembled guests rather decide on the following strategy: whenever the question ends with a letter in the first half of the alphabet, they answer with ‘yes’; whenever the question ends with a letter in the second half, they answer consequently with ‘no’. Additionally, and overriding this, contradictions are not permitted. Thus, when the interrogator re-enters, no fact has actually been fixed for him to uncover; yet, invariably, his asking will lead to the creation of some form of ‘truths’, and may lead to the wildest stories unfolding. In this case, there are no facts about the world; whatever ‘facts’ there are, are solely dependent how we ask. One may even be quite successful in such a setting, i.e. receive desirable answers—but not due to the fact that one is uncovering any kind of pre-fixed truth at all.
This is of course nothing but a hasty sketch of a position that has been elaborated in much greater detail; but I think it should suffice to show that this sort of thing is not as rashly dismissed as you attempted to, but deserves a much more careful discussion. You’re sometimes a bit quick to consider matters that seem obvious to you to be essentially settled (‘existence means having causal powers’, ‘there are objective regularities in the world’, etc.), while in reality, they’re subject to much subtle and careful debate, and dismissing these things out of hand hurts your argument.
Anyway, in the rest of your comment, you essentially seem to agree that there are objective facts about the world. Good. This is all that’s needed for my arguments on why you can’t dismiss epiphenomenalism so easily—simply, because the world could in fact be such that epiphenomena existed; that they’re objective facts of the world that any theory aspiring to yield an as complete description as possible must account for. Moreover, this means that any theory that is built on the assumption that epiphenomena can’t exist can be right, at best, by accident—you are, after all, trying to argue that mental phenomena are actually real, and must yield to explanation in terms of chains of cause-and-effect, since epiphenomena can’t exist; but if epiphenomena in fact can exist, then this means that we have no grounds for placing any credence in the idea that mental phenomena can be explained in the terms you avail yourself of.
Nice one Jochen,
(Do I detect some signs of simmering irritation?)
First, the general remarks: yes, I know that I do get carried away, more often than I would like. I don’t think I’ll be able to avoid repeating the same mistake, however: for a start, I kind of allow it to happen, as, especially during a discussion, one has to cut some corners – we can’t produce a full philosophical tractatus at each iteration. When heuristically/semi-consciously picking shortcuts, I naturally tend to choose the ones that correspond to the things I find more obvious and/or easier to “fill-in” by some basic charitable reading. This approach can misfire (frequently with you in particular), so apologies for that. Problem is, given the time and energy constraints (energy, mostly) that drag me down, I haven’t been able to find another cost-effective way to keep the conversation going, so I guess I’ll be doing the same mistake again, sooner or later :-/.
Re the specific “rash dismissing” I may have committed in the last comment, I can try to fill-in the blanks (briefly!): I started from “there are regularities in the world” (H1), which doesn’t say anything on our access to them. Also, you mention that claim as an example of something that I’ve thrown-in with too much haste, but it seems to me that you may be committing the same mistake (haste) in the way you read me. For example, in your second 20 questions scenarios, it remains true that there are regularities in the world (answers still reliably depend on the questions), so I don’t think introducing H1 was rash or unjustified. Furthermore, I wasn’t even trying to dismiss much, I was trying to apply a commonly accepted background rule of philosophical debate: you start by spelling out your best, strongest, more convincing interpretation of your interlocutor’s position (I refuse to refer to you as the opposition!). It’s fine to correct me when my interpretation is wrong, of course!
With that in mind, I was trying to look at your “2) You can hold that theories are just engines for prediction, for producing successful behavior in the world.” option and build an interpretation that doesn’t clash with my (and I assume, pretty much everyone’s) experience of the world (we encounter regularities all the time, that’s how we manage to remain alive). My aim wasn’t to summarily dismiss the possibility “2)”, I was explaining how I understand that position before responding to your objections on my possible “confusion between epistemological an metaphysical matters”. The way I understood that objection was resting on the incompatibility of the two epistemological stances you described. Thus, what I was trying to do is to show why my (I believe reasonable) epistemological stance can be seen as a mixture of the two options you’ve laid down (see also below). If you want, you can see my point as an attempt to show that metaphysics and epistemology are inextricably intertwined – attempts to separate them neatly will introduce their own new errors.
More in general, please note that what follows is “one way” of positing my claim (a way suggested by the points you are making: I’m trying to raise to the specific challenges you are posing): if, as I claim, different models can be built by changing our ontological assumptions (or less pompously: by changing how we “slice up the world” to start with), it follows that two alternative, seemingly incompatible explanations can both capture some truth about the world, sometimes even the same truths (or mainly overlapping truths).[Incidentally: that’s the kind of consideration that allows me to embrace a form of dual-aspect monism. The apparent divide of mental and mechanical is due to how human brains work (can and do change how they slice up reality depending on context/aim).]
Re your two “20 questions” scenarios. I guess I could rephrase my stance on the basis of the sketch you’re offering: it may work quite well.
My two statements “(C6) (as per Feng Shui) for any given theory and purpose, separating what’s true from what isn’t is prohibitively difficult (perhaps impossible?)” and “models get off the ground by assuming the distinctions they posit are ‘really real’, but that’s always somewhat arbitrary”, can be linked to the two situations you propose, and lead to what looks like a hard to escape undecidability conundrum.
It’s hard for me to see how one can start an empirical investigation without making some a-priori ontological assumptions, the consequence is that we have to introduce some arbitrary elements from the start (at the risk of getting observer-dependent results) can be managed, minimised and re-iteratively pushed back, but not eliminated. My view is that the situation above also applies to cognition in general, where our biology makes us take for granted the legitimacy of classifications, when in fact, it’s only a necessary step to bootstrap cognition. But since cognition is all we have, realising that this is the case isn’t easy, and realising what the consequences are is even more difficult, perhaps impossible.
What I’m trying to say is that yes, there are (observer-independent) regularities in the world (or at least: there seem to be, although we can’t be sure), and that however, our observations are inescapably somewhat observer-dependent. The observer-dependent side means that we don’t know how much (or exactly how) the answers we are getting are a function of how we are posing them. We know (or better: I believe) that we can never get rid of this sort of undesired effect, so our theories end up being somewhat tainted for the start. The problem is, as shown in the 20 questions example you mention: the tools we have don’t allow to isolate what isn’t quite right.
Thus, for me the default assumption is that in practice we end up in a situation that is like a generalised 20 questions strategy:
the people in the room decide to use the first strategy (observer-independent) for X out of 20 questions (with 0 <= X <= 20), and the second strategy (observer-dependent) for the remaining 20-X questions. When we enter the room, we know the rules, we know the players will use this last strategy, but we don't know X (or the order they will use to pick the strategy to use for each question). They could be playing as per your first scenario (X = 20), your second (X=0) or anywhere in between. Would we be able to estimate X? Could we at least spot the cases where X=0 or X=20? I wouldn't know where to start, and that's basically my epistemological stance.
I know (I expect that) our theories are somewhat distorted by the fact that how we ask our questions does, to an unknown degree, influence the answers. Thus, despite the fact that there are (probably) plenty of regularities in the world out there, I'm still never able to take anything for granted.
If the above sounds reasonable to you (I guess not!) it means we've made some progress, and I'll feel free to re-hash my "against epiphenomenalism" argument in light of what I've learned in this discussion. The question is not whether my view is acceptable from within all the existing philosophical traditions. Of course it isn't. The questions are: "Is my view reasonable enough to be contemplated?" if it is, "does it help solving a useful amount of explanatory riddles?". If you can answer a provisional "yes" to the last two question, then we can move on.
Otherwise, I'm afraid I'd prefer to sort out the epistemological/metaphysical muddle first and keep epiphenomenalism out for a little longer.
Hey Sergio,
Yes, I spoke somewhat harshly in my last reply; my apologies. If you’ll permit me a brief attempt at explaining the cause of my doing so, it’s that there’s a kind of evasiveness to the way you’ve responded so far. So let’s chart the way this conversation went (at least, according to my perception of it, which is undoubtedly different from yours).
So, first, you make two claims: one, that you can show epiphenomenalism to be false; and two, that you’ve solved the riddle of existence, and that existence is equivalent to ‘having causal powers’. Now, these are, if you’ll excuse my language, both big f’n deals; they need absolutely airtight justification in order for your claims to be taken seriously by anyone. So, I try and probe your justification (after all, the ultimate end of all of this is to get your ideas into a shape where they have some chance to stand up to critical scrutiny by your peers).
And, as I believe to have demonstrated, it’s possible to construct scenarios where your justification doesn’t hold up—the parallel universes, the lone astronomer who can’t detect the universe’s expansion, and the observers in front of their screens. In all of these cases, theories built upon your assumptions would simply turn out wrong: they would fail to account for all the facts of the world. Hence, it’s conceivable that, likewise, in the case of mental epiphenomena, theories thus built would turn out wrong.
Then, confronted with these scenarios, you suddenly make (what seems to me) a 180°-turn, and claim that, well, none of this ever was about facts of the world, anyway. Now, if this has indeed been your position from the start, then I don’t understand why you made any claims of existence in the first place; rather, you should have been talking about our knowledge about what exists, which is a very different issue (epistemological rather than metaphysical).
Furthermore, you come in with yet more big f’n deal claims: that your stance accounts for ‘cognition, for the reason why science is so fantastically efficacious, for the intrinsic uncertainty of knowledge, for the existence of our dualist inclinations, and much, much more’. Now, any position able to do that—plus account for existence (or maybe not), plus prove epiphenomenalism false—would basically be a holy grail for philosophy, so what you’re effectively saying is that you’ve figured out, all on your lonesome, an answer to questions that have kept the best minds occupied for these past 2500 years.
This, again, needs some very, very strong justification. I mean, you must pause to consider the effect of such claims on somebody at least familiar in outline with the problems you’re claiming to address—it’s a bit like somebody claiming to just have proved the Riemann hypothesis, the twin prime conjecture, and found Fermat’s original proof of his last theorem with elementary algebra. Absent very strong indications to the contrary, such a claim does not carry high prior credence (which should not be taken as any slight against your—obvious—intelligence).
So again, I probe. My best charitable reconstruction of what you’ve been saying is that you’re making one of two (I still believe) mutually exclusive claims: one, theories are just instruments for behaviour—in which case, you do get out of the muddle of epiphenomena, but you can also make no pretence at ‘explaining consciousness’, since on such a reading, explaining just isn’t what theories do. Two, theories are about, or at least approximate, what is the case in the world, independent of our knowledge thereof. This is the stance upon which your comments regarding existence and being able to show epiphenomena inconsistent make sense (on the other reading, you simply wouldn’t have to bother with either concept, and that’s that).
But you (quite understandably) want both: the ability to make explanatory statements, and to be rid of epiphenomena (whose existence otherwise would threaten the theory of consciousness you cherish). So, you come back with yet another big f’n deal claim: you say that you can show that there are regularities in the world, which underwrite our experience, and you say that such a thing actually follows from taking a stance in which theories are merely engines for behaviour production.
This, again, I find impossible to accept without very strong justification. In particular, scenarios can be thought up in which it’s not the case that there are underlying regularities, yet we can theorize; it’s just that the theories don’t tell us anything about the world, but merely about the way we’ve been asking questions. If this is the case, and I hope you’ll agree with me here, then it does not follow that you can get to the theories-as-reflecting-truths view from the theories-as-instruments view.
To illustrate this, I gave the example of the 20 questions game. Admittedly, the example I used was not the best—I wanted to show that the answers the world throws at us may be unconnected to the content of the questions asked, hence I just went with the rule Daniel Dennett used in Consciousness Explained. I could, and should, have just as well proposed that the answers are given at random. Indeed, even the consistency meta-rule is unnecessary: one, without some bedrock of facts to gauge consistency against, lapses in consistency are not noticeable—think about dreams, for instance, which are often rampantly inconsistent, which however the dreamer rarely notices; entire changes of scenery, plot, and even identity are just taken for granted. There’s nothing that says the world is not like that, as in such a case, our subjective sense of consistency is just as questionable as any other facts about the world we believe we know. Using the scientific method in a dream doesn’t serve to uncover anything about the world; but there’s no way to prove you’re not dreaming now.
Two, for any two inconsistent answers, there’s always a hypothesis that can be postulated to make them consistent, by enlarging the context. So, the apparent contradiction between receiving the answer ‘yes’ to both ‘Is there a girl in the story?’ and ‘Are there no females in the story?’ may be resolved by receiving (by mere accident) a positive answer to the question ‘Is the girl transgender (or transsexual; I honestly don’t know the preferred nomenclature)?’, or any of a number of other consistency-restoring hypotheses; at some point, you will receive an affirmative answer to one of them, and your sense of consistency will be restored—but not due to some underlying fact of the world you’ve discovered.
So, that’s where my frustration stemmed from—whenever I point out a problem for your stance, you resort to some new, and, to me, out of left field proposition to save your story, every time introducing new, often even greater implausibilities. We thus get further and further removed from the original issue, and honestly, to me, your case looks weaker and weaker. By now, your argument against epiphenomenalism not merely rests on your claim that existence = having causal powers, but on your ability to prove that there are regularities in the world, which entails that theorizing is necessarily in contact with the facts about the world, even though (somehow) not all facts play a role, which allows you to ignore the possible facts about epiphenomena, which then, finally, allows you to make your claim that phenomenal experience must be mechanistically explainable (though, as noted, there are other problems with that claim besides epiphenomena).
It’s a bit like when I point out that you’re doing a difficult tightrope walking act, you retort by noting that you don’t do any balancing, you’ll be comfortably seated on a chair—standing on a stack of books on top of a pencil that’s standing on its tip on the rope. You’re heaping implausibility upon implausibility, and resort to more implausibilities to prop up those of the previous step. At some point, I think, you need to ask yourself whether you want to continue this balancing act, or if it’s not better to try and face up to the problems of your approach, and hope to sort them out—and if that doesn’t work, change your approach.
Now, for another, final (for this post) matter:
I’m not saying that you aren’t allowed to make assumptions. You’re perfectly free to make any assumptions you wish—although, with any assumption, there will be those disagreeing with it. Anybody makes assumptions; as long as they’re clearly labelled as such, that’s not a problem. But what you’ve proclaimed yourself to be able to do is not making assumptions, but settling matters that have been enormously controversial over the whole history of philosophy—thus, with one quick stroke of a pen, erasing thousands of pages of spilled ink, and entire lifetimes’ worth of thought.
So, just for ease of reference, my argument so far:
1) You can take theories to be about, in an at least approximate way, facts of the world; to provide explanation, make predictions, etc.
1a) If you do so, you’ll have to face the scenarios in which there are epiphenomenal facts that nevertheless are necessary for a given theory to give a correct accounting of the way the world is.
2) You can alternatively take theories to be mere bookkeeping devices, instruments to make sense of the world, removed from the facts about the world.
2a) If you do so, you indeed get out of having to account for epiphenomena, but not, as was your original claim, by showing that they don’t exist, but merely because they don’t play a role—as is the case with any other putative facts about the world.
3) The two options do not reduce to one another; you can’t have 2a)’s immunity against the epiphenomenal threat while simultaneously harvesting 1)’s explanatory power—if you have that power, you also have to deal with the possibility of epiphenomena (or at least, your arguments so far do not banish that possibility). This is demonstrated by the possibility of scenarios—such as doing science in a dream—in which theories can be formulated, which however don’t have any connection to facts about the world.
4) Hence, you either need to find a new argument against epiphenomena, or accept their possibility, if you simultaneously want to uphold your explanatory claims.
Before you now go on to try to defend your position against the ‘science in dreams’ scenario and its ilk, I would strongly urge you to take a step back, and look at the edifice of argument you have erected: it teeters, it totters, it creaks and cracks. A competent philosopher—not a dilettante like myself—will, I’m pretty certain, not be much impressed with it. If this again sounds harsh, I apologize; but after all, I still believe that there is some value to your ideas relating to ETC, but your insistence on erecting some grand epistemo-metaphysical architecture incurs so many debts of commitment that I can’t see anybody willing to meet them, before even getting at the substantive claims your theory makes.
Thus, you’re raising a neigh-impossible wall of prior commitments anybody interested in your idea would have to scale first, and hence, you’re simply asking for too high a price of admission, which will leave your theory beyond the reach of most you want to get interested in it.
So, my advice would be: KISS (Keep It Simple,…Sergio 😉 ). Work out what you really need to claim; assume what you need to assume for that, and clearly mark what is being assumed as an assumption—but don’t go around claiming you’ve solved the ten greatest riddles of philosophy before breakfast. (Even for Wittgenstein, that didn’t work out too well.) Your ideas don’t support that, and more, they don’t need it, either—you can keep the price of entry much lower, and thus, increase your potential audience.
Jochen,
quick reply, without substance. I don’t want to leave this unanswered for days, not without saying immediately what needs to be said.
First thing is “Thank you!” You do not have anything to apologise for: I’ve been asking to never sacrifice clarity in favour of courtesy. It’s hard enough to understand one another anyway, so thanks for not sugar-coating your words.
If apologies are due, clearly I owe you some, as I haven’t managed to make myself clear. I’ve re-read my post above here, and I can see my intention of making it very clear that my point is epistemological (intent: “declaring that something is an epiphenomenon (in the strong sense) is equivalent of declaring it non-existent, on epistemological grounds”. Further down “if something has no influence on the world, whether it exists or not is entirely irrelevant, for all possible purposes, both practical and theoretical” and “whether such thing exists or not, according to the claim, makes no difference, therefore their existence is irrelevant FAIP”), but then I was all too eager to follow your bait and get onto metaphysical grounds. Perhaps I should have kept the helm straighter, but I’m not sure what would have worked best. Perhaps this (undoubtedly frustrating) back and forth is actually useful and necessary, I really don’t know.
Thank you also for providing a description of the external view, I can only suppose how I look from the outside, and my suppositions feel weak and inherently biased (no sh*t?). Your words don’t really come as a surprise (I’m sure you are right in describing how I probably appear to a professional philosopher – I wish I could say that I don’t care!), but they help by making things more concrete, otherwise I have only cold inference to guide me and that’s ineffective whenever I get excited.
Thank you more than anything for not giving up, I really don’t have a theory(!) of what your motivation might be, but I do know that I’m accumulating a debt of gratitude. If you’ll ever have half an idea on how you may collect your debt, please do let me know. Reciprocating would make me feel better.
On the contingent stuff: my immediate reaction makes me think that I will proceed as follows (might still change my mind).
1. Finally write/publish and Epiphenomenalism Pt.3 post, using your well advised KISS strategy.
2. Answer here on the meta-stuff about approach: I think I owe you an explanation on how it feels from my side (yes, of course it’s different).
3. After that, we’ll see. I don’t know if I’m smart enough to find a way to apply KISS to the whole thing, I don’t think I am, but that’s another story.
[Thank you also for introducing some edulcorated bad language! I was hoping you would 😉 ]
[…] follows is a direct reaction to the discussion I’ve had with Jochen below my first epiphenomenalism post. The discussion meandered into scary places, and Jochen ended up recommending me to keep it simple […]