Mikhail sent me the following post in email, giving me permission to post it if I so desire. I think it gets at a number of important differences and assumptions, so it might be of general interest to others. Following Mikhail’s post you will find my reply. I hope others interested in the realism/anti-realism debate and OOO take the time to read through the post as I think some key points are made here, as well as some arguments potentially central the epistemological grounds of OOO and why the “speculation” of OOO is not simply “making things up”. Basically I rehearse Roy Bhaskar’s argument for transcendental realism, trying to show why I think that epistemological questions can’t properly be resolved without robust realist ontological claims. However, there’s an important caveat here. While I’m strongly inclined to endorse the form of Bhaskar’s transcendental argument for ontological realism, I am more circumspect about the ontological claims he is making. In other words, it is possible to endorse much of the reasoning that leads Bhaskar to the conclusion that we can know something of mind-independent objects that exist regardless of whether anyone knows them, while rejecting the specifics of this ontology on the grounds that it is inadequate.

I think this particular exchange is not about SR/OOO/OOP or anything that has been discussed so far, it’s an old philosophical issue and this is why I think it is important to address as it seems to underlie
many of the disagreements. I’d like to begin with some very basic issues before going any further. You write:

“In my view this position undermines the possibility of any fallibilism so we’re left without the means of determining why we should choose one theory over another.”

This is important. Now just because a position undermines a certain possibility does not mean that it is wrong, just that it is inconvenient. I hope we agree on that. Therefore, say, if skepticism has a good argument, we cannot simply say that if we accept that argument we will be deprived of certain possibilities. I take your observation to mean more than just an expression of preference – if we cannot have an access to the world, we cannot have a true theory of it, because it’s neither true not false and cannot be shown to be
either true or false. I agree.

Now let’s slow down here a bit and see what’s going on. As you say, this is not a real point of disagreement, it’s just a statement and it has consequences. This is going to be very primitive not because I’m being condescending, but because I found of late that most of the disagreements seem to be about very small things we overlook because we think of ourselves (I mean myself primarily) as having long overcome these problems. It seems to me that you are affirming a kind of duality: there’s a level of the world and there’s a level of the mind (the theory of that world) – am I correct in reading you this way? An immanent “inside” and a transcendent “outside” – of course, as we both know from Descartes/Kant, we need a
“third” level, a point from which one can compare the two – the world and its theoretical description – and declare it to be adequate. Let’s reject Descartes’ solution and forget about God or anything that’s
truly “outside” and stick with Kantian types of solution that places that “third” on some transcendental level.

read on!

Actually, I don’t even want to name names here as it is bound to cause friction in terms of how we
read this or that philosopher. Let’s just say that there must be a third point from which we can compare the world and its theory – do we agree on this point? If we don’t, I’d like to hear your take on this.
If we do, then the obvious question is not that of realism/anti-realism, that is to say, not whether the world is out there and can/cannot be known, but whether if it is accessible, then what of it is accessible and how do we know that what we access is part of the world and not part of our own mind. That is to say, I’m
realist if realism is a simple postulation that the world is out there (barely anyone I know isn’t a realist in this sense). The issue at hand is the possibility of the only true theoretical representation of that world which I claim is impossible and you seem to claim is possible. Can we begin here for now?

Let’s me say a quick aside about epistemology/ontology. I think I see your point about collapsing the two, even though I don’t think I could ever be accused of doing it, but of course that’s my perspective. If I
understand the issue here, epistemology preceding ontology annoys you because of its certain counterfactual or counterintuitive move to ask questions of the conditions of possibility of something that seems to be very much there and possible. In a sense, it’s a stepping back from what we are doing while the task is precisely to stay in that what is being thought – stepping out, even if for a couple of epistemological
questions, ruins the effort. At this point, I’m willing to set the epistemological questions aside – that is to say, although my above observations seem to suggest that I’m interested in “knowing how you
know” if your theory corresponds to the real world, I’m asking the question not from a transcendental epistemological position (“let’s take a break and look into the knowing apparatus before we do
anything”), or some sort of methodological position, but from a simple procedural point of view. That is to say, I affirm that we never really are “outside” of anything, but are always “inside” both mind
and world, if you will (this, of course, makes any outside/inside distinction collapse, it’s all immanent).

If epistemological questions are preparatory questions, I’m willing to set them aside and affirm
that since we are already in the world and already have all sorts of theories about it, and we cannot leave it to ask epistemological questions (although we don’t necessarily have to, at least according
to Kant), we must proceed with questions of ontology. At least this is how I’m reading your complaints about epistemology/ontology. This is just as aside, if we disagree here as well, it might slow down the
discussion even more which is fine with me, I’m on Thanksgiving break.

I appreciate the effort Mikhail made in this email. I’m extremely tired as I only got two hours of sleep last night, so hopefully I’ll make some sense in my response.

I do not feel that I’ve ignored the basic issues as Mikhail suggests, but rather have carefully worked through these issues. Above Mikhail proposes a model of mind-world relations drawing on the spatial figures of the inside and the outside. His premise is that our minds are characterized by interiority or “inside-ness” and the world is outside. In my terminology, the mind is here treated as immediate, while our relationship to the world is mediated. Put differently, the relationship of mind to itself is here treated as immanent, while the world is treated as transcendent. From this it follows that we can only have access and certainty about our minds, never about the world. The question of knowledge becomes that of how we can get out of our “bubble”.

Much of my book, Difference and Givenness, is devoted to contesting this model of mind, though not in the way one might expect. This is particularly the case with chapter 7. My thesis is not that we have immediate access to the world, but rather that our access to ourselves is itself mediated. In other words, we no more have a direct relationship to our own minds than we have to the objects of the world. Put differently yet again, we are transcendent to ourselves. I won’t here rehearse all the arguments for this claim, but merely reiterate it to point out that I do not share the first premise Mikhail proposes.

If it is true that our relationship to ourselves is mediated or that we don’t have direct or incorrigible access to our own minds, it follows as a consequence that there is no reason to claim privileged access to our relationship to our own minds over our access to objects. In other words, the epistemological problem of access is not restricted to that of how we can have access to the world, but arises in exactly the same way with respect to the question of how we can have access to ourselves.

Mikhail seems to hold the view that I reject epistemology or do not raise epistemological questions. I do not think this is true. My thesis is rather different: we cannot have an adequate epistemology without making ontological considerations. Here my argument, following Bhaskar, is transcendental and is driven by an examination of the actual practices involved in knowing the world. Following good Kantian transcendental methodology, the question here is “what must the world be like for knowledge to be possible?” The question here is a transcendental question insofar as it seeks after conditions that would render our knowledge praxis intelligible. However, it differs from Kant’s transcendental idealism in that where Kant asks “what must our minds be like for synthetic a priori propositions to be possible”, this position asks what the world must be like for knowledge to be possible. In other words, it is a question about ontology, not our minds. The thesis is that an adequate epistemology requires an answer to certain ontological questions.

Like Kant’s transcendental arguments, it begins from the premise that we do have knowledge of certain things in the world and independent of mind. For Kant it was an uncontroversial fact that mathematics and Newtonian physics are instances of knowledge. The question then became that of determining what mind must be like for this form of knowledge to be possible. It differs from Kant in that it asks what the world must be like for this knowledge to be possible. In the domain of the sciences, a question closely related to this is the question of why the practice of experimentation in the sciences is necessary.

Now the move I am making, following Bhaskar (wish I could claim credit for these arguments), is controversial so it requires some justification. We are entitled to ask “Why is it necessary to make this move through ontology to answer the knowledge question of epistemology? Doesn’t this simply beg the question by claiming to know before we’ve answered the epistemological question?”

Here is the reason why I believe this move is justified. In outlining certain points about Kant and Hume I am not trying to patronize Mikhail or suggest he is unfamiliar with these things, but to trace out the line of thought. It is necessary to backtrack a bit through the history of philosophy to see the problem. Hume tried to reduce questions of being to questions of sensation or experience. Everything had to be traced back to experience. For Hume causality is just a constant conjunction of sensations. Kant shared Hume’s premise that there is no knowledge apart from sensibility, but noted that sensibility alone could never give us the idea of necessity. Paraphrasing Kant’s famous statement, “while it is true that all knowledge begins with experience, and that knowledge is impossible apart from experience, it does not follow that all knowledge arises from experience.” If we want to understand why experience or sensation alone is inadequate for grounding the relations of necessity asserted in causal judgments, we need only look at the logic of the lower portion of Aristotle’s square of opposition. Quite frankly I’m shocked that it took nearly two thousand years after Aristotle developed the square of opposition for philosophy to arrive at Hume’s skeptical conclusions.

From the observation that sensation alone cannot ground relations of necessity in causality, and that experience is all we have to go on, he inferred that the idea of necessity comes not from sensibility but is contributed by mind. Like Hume, Kant holds that causality is a constant conjunction of sensations. Unlike Hume, he argues that minds contributes the category of causal necessity that links these sensations. The thesis that judgments of causality are judgments about the constant conjunction of sensations is positivism.

Here’s the problem: The concept of causality and knowledge producing practices become incoherent if causality is understood as a constant conjunction of sensations. This for three main reasons:

First, outside of astronomy we very seldom encounter a constant conjunction of sense-events, yet we still hold that causal relations are functioning in the world unobserved. Sex, for example, doesn’t inevitably lead to conception, but without sex (setting aside artificial insemination) conception cannot take place. Antibiotics don’t inevitably get rid of an infection, but we still hold that antibiotics have the power to kill infections. If we take seriously the thesis that causal claims are constant conjunctions of sense-events we would be forced to reject the thesis that sex, antibiotics, and many things besides have causal powers.

Second, many events are constantly conjoined in experience, but we hold that they do not possess a causal relation to one another. I get up before the sun rises and make a cup of coffee for myself. The sun then rises. Why am I not led to the conclusion that making coffee causes the sun to rise? It might be argued that I do not assert a causal relation here “because there have been occasions where you haven’t made coffee before the sun rises, yet the sun still rose.” However, returning to our first problem with positivism or the thesis that causality is a constant conjunction of sense-events, we see this doesn’t work. Why? Because the failure of a consequent to occur after the antecedent occurs is not grounds for rejecting a causal relation. The antibiotic is taken (antecedent) and the sickness does not go away. Yet we do not, on these grounds, arrive at the conclusion that there is no relation between the antecedent and the consequent. In short, the positivist theory of causal relations doesn’t allow us to distinguish genuinely related sense-events from unrelated conjunctions of sense-events. This is true even under the Kantian model.

Third, the thesis that causal statements are simply constant conjunctions of sensations does not explain why scientific experimentation is necessary. If causality is a constant conjunction of sensations then the idea of engaging in experiments in a controlled and isolated setting makes little or no sense. What would be gained from such a strange activity?

Bhaskar’s thesis is that the problem with the empiricist thesis is that it conflates causality with sense-experience, when in fact, the two are very different things. Recall the first argument: it is possible for antecedents to occur without the consequent occurring, yet for there to still be a causal relationship between the two terms. What is being said here? In our practice we are saying that the causal mechanism is independent of 1) whether or not we experience its consequent, and 2) whether or not the consequent takes place as an event in nature.

In other words, we have three terms: The causal mechanism, natural events that may or may not be experienced by anyone, and experiences. The sense-data empiricist tries to collapse the first two into the third. The problem is that if we do this we are unable to explain 1) how it is possible for something to exercise its causal powers without producing the accompanying natural event, and 2) how it is possible for something to exercise its causal powers without us experiencing it. In other words, this model, according to Bhaskar, fails to distinguish the real, the actual, and the empirical, instead trying to collapse the other two into the third.

Here we can finally return to the question “why is scientific experimentation necessary?” Bhaskar’s answer to this question spins on a distinction between open and closed systems. According to Bhaskar, if it is possible for 1) an antecedent to be triggered without being actualized in a natural event, or 2) an antecedent to be actualized in experience without producing the consequent event, then this is because most causal mechanisms function in open systems where other causal factors intervene, overdetermining the event. As a result, the other causal factors prevent the causal mechanism from actualizing itself in a natural event or an experience for an observer. This is the reason, contends Bhaskar, that it is necessary to engage in experimental activity. Experiment creates a closed, artificial system allowing the inquirer to trigger the causal mechanism to determine what consequent it produces without the intervention of other causal mechanisms.

So returning to Bhaskar’s question “what must the world be like for knowledge to be possible?” we now have a thumbnail sketch of an answer to this question. First, it is necessary to distinguish between the real, the actual, and the sensed. Second, it is necessary to distinguish between causal mechanisms, natural events, and sense-events. Third it is necessary to distinguish between open and closed systems. The ontological dimension of Bhaskar’s epistemological problem lies in the transcendental claim that the condition under which science is possible is the existence of causal mechanisms that can function or act without producing a corresponding natural event or consequent. It is only on these grounds, argues Bhaskar, that 1) our engagement in scientific experimentation, and 2) our claim that certain entities like antibiotics have causal powers even when they fail to successfully cure illness are intelligible or coherent. Experiment creates a controlled and isolated environment in which causal mechanisms can be triggered without the interference of other causal mechanisms. When or if these mechanisms are found they are then accorded the status of “transfactuality”, which is to say they are treated as functioning in the ordinary world of open systems even when they do not actualize themselves in an event.

These are specifically ontological claims about the nature of the world, not epistemological claims about how we come to know the world. Without these ontological claims, argues Bhaskar, we can’t render our epistemology intelligible.

Before wrapping this up there are two crucial points to be made:

First, the ontological claim that causal mechanisms exist and that it is possible for causal mechanisms to act without being actualized in a natural event makes no claim as to what causal mechanisms exist. In addition to the existence of these causal mechanisms, Bhaskar argues that these mechanisms must be differentiated, structured, and stratefied (more on that another time). Finally, he argues (and this gives Harman fits because he’s an actualist, I’m not sure where I shake down on this issue though I tend towards potentialism) that these causal mechanisms must have powers (capacities, “able-to-do’s”) that can go unactualized. The discovery of what causal mechanisms exist, their differentiation, their structure, and their stratification is the responsibility not of philosophy, but of actual experimental inquiry. All the transcendental argument purports to demonstrate is that causal mechanisms exist (because knowledge exists) and that these causal mechanisms are mind-independent and continue to function regardless of whether any human perceives them.

Second, within the domain of experimental inquiry the claim that such and such an entity is a causal mechanism is not infallible. It can turn out that subsequent inquiry shows that such and such a claim that “x is a causal mechanism” was, in fact, mistaken. All the transcendental realist is committed to is that causal mechanisms exist and can function without being actualized in natural events. He is not committed to the claim that specific knowledge-claims about causal mechanisms are infallible or that we have direct access to these mechanisms. Getting at the causal mechanisms is, for Bhaskar, hard work. It requires the laborious construction of closed systems that allow for the causal mechanism to be triggered, producing both a natural event and a sensible event independent of intervening causal mechanisms in open systems where the contribution of the causal mechanism being sought is ordinary disguised or mute in its functioning.

Here I think we get at one of the central failings of traditional epistemologies. For whatever reason they begin from the premise of a passive observer that simply has sensations that it links in some way or another. What is missing in this intellectualist model of knowledge acquisition is the work it takes to produce salient experience. Basically, what is missed is the crucial role that the fact that we are embodied (and therefore ourselves causal agents), that we act on the world, that we use instruments to trigger these mechanisms, and that we carefully build closed systems to trigger events. But here’s the central point: We strive to create closed systems that allow us to trigger the causal mechanisms we’re searching for, but no system is entirely closed and subsequent experimental inquiry might reveal that there were intervening causal mechanisms that led us to misinterpret the triggers. In other words, Bhaskar’s position is fallibilist, allowing for the possibility of error.

Alright, that’s enough for now. What do you all think? I’m still working through Bhaskar’s arguments myself, but I confess I find them deeply appealing and convincing.