Happy New Year to everyone out there in internet land. I’m currently feeling a bit awful, due to a combination of excessive merriment and a rather nasty cold I can’t seem to shake. I know I said I’d stop commenting on Graham’s posts, but as someone affiliated with the “Sellarsian scientistic wing of what used to be called speculative realism”, at least insofar as I work on metaphysics and am influenced both by Sellars, Ray Brassier, and his other philosophical descendants, I feel compelled to respond to what Graham has recently said about it (here) in the context of rebutting some of David Roden’s claims about his work (here). The relevant passage is a response to David’s claim that Graham’s position is a form of phenomenological idealism:-
2. “His famous reading of Heidegger’s tool analysis ups the metaphysical ante by presupposing that not being explicitly represented is a modality of things (or thinging, or whatever). If this isn’t good old phenomenological idealism, I don’t know what is!”
What is idealism is enemyindustry’s own next sentence: “In contrast, I hold that intentionality brings us into contact with the real with numbing regularity.”
This is idealism, because it holds that the real is convertible into the accessible. It gives no adequate account of the difference between the tree that grows and bears fruit and the tree that I encounter. No matter the level of “numbing regularity” with which I encounter a tree, that encounter is not the tree itself. Until you account for the difference between the two (as I do) then you are an idealist.
Ultimately, I think this is why Meillassoux remains in the Idealist camp, and the same holds even more for the Sellarsian scientistic wing of what used to be called speculative realism. They aren’t realists. They’re partisans of math and science.
Now, I agree with Graham that David’s characterisation of his position as idealism is incorrect, but I find the counter charge of idealism to be extremely thin. I’ve addressed some of these themes before (here, here, here and here), but I feel it’s worth restating the problems I have with this line of reasoning in a condensed form.
First, I’ll try and briefly reconstruct (and flesh out) the argument Graham is giving here. The overall argument goes something like this: i) realness implies ontological independence, ii) ontological independence implies epistemic inaccessibility, iii) realism demands that entities be real, therefore iv) realism demands that entities be epistemically inaccessible. However, the important step of the argument is (ii): demonstrating that ontological independence implies epistemic inaccessibility. This is where Graham’s claims about the difference between the tree itself and the tree I encounter are relevant. This argument goes something like this: v) epistemic accessibility implies the possibility of adequate knowledge, vi) adequate knowledge of an entity implies that the entity for-us and the entity in-itself are identical, vii) this contradicts ontological independence, therefore (ii).
Let’s put the question of whether this is a good argument to one side for the moment. If it is a good argument, then all positions that hold that the real is anything other than accessible are forms of idealism, because by holding that the real is some way they thereby commit themselves to its accessibility. This means that those who take the real to be composed of one-dimensional strings vibrating in 11 dimensions (or whatever your preferred fundamental physical theory holds) are idealists just as much as those who take it to be composed of water, fire, or a single self-causing substance (or whatever your preferred traditional metaphysical theory holds). In fact, it means that pretty much everyone who doesn’t occupy Graham’s position in the dialectical landscape is an idealist. It’s not just us partisans of science who should be worried.
Now, there’s a legitimate worry here that such broad applicability of the term ‘idealism’ voids it of its traditional substantive content. For instance, if it’s not useful for differentiating between Hume and Berkeley, or between Hegel and Russell, then there’s a good sense in which it’s just not doing the same job the traditional notion does. However, I’ll leave such semantics aside, because I think that the above argument isn’t a good one, and thus that most of us have nothing to worry about. There are two problems with the argument, corresponding to premises (i) and (vi), respectively.
The first problem I’ve talked about quite a bit elsewhere (here and in the TR essay). The fundamental idea that underlies the notion of realness is the possibility of error, i.e., the possibility that our beliefs/claims about a real entity could be wrong. This can be articulated in two ways: in terms of mind-independence, or in terms of attitude-independence. The former is a matter of the ontological independence of the objects of thought from the thoughts themselves; at minimum, that they are numerically distinct. The latter is a matter of the epistemic independence of the way things are from the way we take them to be. I take it that Graham’s argument takes the former route, whereas I prefer the latter. It is important to recognise that there is an alternative way of cashing out the notion of realness, not because this lets us advocate for a real that is ontologically dependent upon our thought, but because it allow us to articulate the relation between the real and thought in non-metaphysical terms. This opens up the possibility of treating thought in an entirely non-metaphysical fashion. There is more that could be said here, but I’ll leave it at that for now.
The second problem I’ve also talked about elsewhere (here), but I’ll try to be more concise here. Premise (iv) – the claim that the adequate knowledge of an entity implies that the entity for-us and the entity in-itself are identical – is already dependent upon Graham’s metaphysics of objects. This is because it depends upon us treating the for-us and the in-itself as two different types of entity, rather than as different modes of apprehending one and the same thing. This is the basis of Graham’s famous distinction between sensuous and real objects. It is only given the assumption that we must analyse knowledge as a metaphysical relation between two types of entity that (iv) makes sense, because it is only under this assumption that relations of identity and distinctness could be at all relevant. There are two independent objections to this assumption. First, one might challenge that knowledge should be understood in metaphysical terms at all (as hinted at above). That it must be requires some further justification that I think Graham has yet to provide (see here), especially when there are seemingly viable non-metaphysical approaches (such as my own, I would hope). Second, even if we admit that knowledge should be understood in metaphysical terms, there are reasons to reject the specific metaphysical analysis Graham provides.
The most prominent reason to reject the analysis is that it implies that what it is to represent (e.g., believe/claim) something as having a certain property to stand in a relation to a special kind of object that possesses that very property. To believe that the real tree is an elm is to stand in a certain relation to a sensuous tree that is an elm. This means that in order to believe that a piece of coal is combustable, I must be related to a sensuous piece of coal that is combustable in the very same sense. The argument to inaccessibility would not work if the properties in question were proxies for their real counterparts. A sensuous object that had all the right proxy properties would not be identical with the corresponding real object. This account of representation seems to lead to absurdity, because we have no idea what it would be for a sensuous object to combust, and thereby what it would be for it to be combustable.
Now, Graham’s position isn’t quite as absurd as this, because he takes it that the properties (or qualities) that sensuous objects possess are entirely different than those that real objects possess (this distinction between real and sensuous qualities is the other aspect of his fourfold). This effectively downgrades the status of those properties we ascribe to things. Combustability is a phenomenal matter, rather than a noumenal one. Science (and most other activities) only ever deals in appearances, never with the things themselves. However, this distinction is motivated on the basis of epistemic inaccessibility, which (on this reading) is itself motivated by this account of representation. In essence, the distinction between types of qualities is motivated by the absurdity that the distinction between types of objects leads us to. Although Graham’s approach is certainly a viable way out of the absurdity (albeit at great expense), it is just as easy (if not easier) to reject the account of representation and the distinction between types of object it is based upon. Again, what’s needed is a stronger justification of the initial split between types of objects. I suspect that this is where Graham would fall back on his interpretation of the tool-analysis. It seems that all roads lead to the tool-analysis, and if you don’t find that convincing (which I don’t), then too bad. Anyway, I’ll leave explaining my worries about that till a later date.
In conclusion, there are plenty of alternative accounts of knowledge and representation (both metaphysical and non-metaphysical) which don’t support Graham’s argument. Given that those of us of a Sellarsian bent advocate such accounts, I think it’s safe to say that we’re immunised against this particular charge of idealism.
DEONTOLOGISTICS 
I’m broadly in agreement with you about the epistemological costs of the OOO approach and Graham’s ontology of proxy objects. I’ve addressed this in a response here http://enemyindustry.net/blog/?p=569
in terms of the concept of semantic information.
If I have time, I’ll expand on my critique of Graham’s account of the tool analysis some time; but it’s not a big concern at the moment.
Happy new year and take it easy!
Hi David,
I think we’re mostly on the same page here, though I’m not as into accounts of semantic information as you. We should trade notes on the tool-analysis at some point though.
Best to you too!
Yes, that would be good. it’s entirely possible that I’ve missed something essential in GH’s reading. Heidegger is a bit of a blind spot for me!
D
Hi Pete, I’d made these notes on the tool analysis earlier. Hopefully, these explain my position better. I’m quite happy to concede that it is incorrect if it is demonstrably so.
The key argument for Harman’s ontology is to be found in his revisionary reading of Heidegger’s tool analysis.
Heidegger notes that we encounter objects ‘present to hand’ (Vorhanden) as explicitly represented, bearing certain properties, some of which may be perceptual (like colour or shape) others which may be representable abstractly (as when we use the diagrams of lines of force to represent the variation across space of a magnetic field).
1) According to the standard pragmatist reading of Heidegger’s tool analysis this explicit relationship to objects presupposes a network of everyday practices and coping strategies which presuppose – but do not represent – ‘a range of functions or effects that are relied upon’ (Harman PhD thesis, 4). We encounter things as present to hand only as disturbances within the network of coping practices which we use and identify things in our environment.
As phenomenology this seems plausible. One might object – coming from the direction of computationalist theories of mind – that the fact that we are not conscious of representing objects explicitly does not entail that we don’t. The ready-to-hand hammer might be represented unconsciously in one’s language of thought (we know that not all perceptual states are conscious states). But we can waive this objection for now.
2) However, Harman claims that the tool-beings on which these practices continuously rely, are not merely unrepresented when subjects of everyday coping strategies but unrepresentable in principle. Harman writes: ‘This is the true scenario of the tool in Heidegger’s description: equipment as an agent thoroughly deployed in reality, as an impact irreducible to any list of properties that might be tabulated by an observer.’ (Ibid., 7)
Now how do we get from 1 to 2? When Harman says ‘irreducible to any list of properties’ I take him at his word to mean any list whatsoever. Even the God of infinite data retrieval could not represent the ‘hidden execution’ of an object. The point is not that the object has too many properties, but that its tool-being is something of a different order from anything that could be represented:
‘For Heidegger there is certainly no simple objective world to which the mind can “adequate” with relative ease, but neither is there a subjective sphere for which the outside is irrelevant to the point of being non-existent. Instead of these options, he gives us a world of visible broken equipment, which in its visibility is utterly incommensurable with the real effect of equipment in action:
‘ “Realism” will probably always be a loaded term that awakens dozens of misconception; for this reason, it should be avoided whenever possible. but if there were ever a philosopher who respected the force of a reality absolutely distinct from its conditions of being perceived, Martin Heidegger is that philosopher. At least classical realism believed that the things themselves could be adequately copied by human knowledge. But for Heidegger, no such adequation is possible, since the tools themselves forever elude any attempt to represent them in the flesh: the tool analysis has no other result than this’ (Harman 2002, p. 120)
At this point, I admit to be being beguiled but also puzzled since, assuming that there is no crucial premise that I have ignored, nothing obviously warrants the inference from 1 to 2.
What is it about the efficacy about objects that is so ineffable? Why can’t we describe this efficacy in terms of an objects relations, dispositions and affordances say. For example, a bridge by virtue of linking two sides of a river, affords certain opportunities to ambulatory creatures which it doesn’t afford to stones or plants. There just doesn’t seem to be anything very difficult about that and certainly nothing that contradicts the phenomenology of equipment.
3) From 2 – the non-representability of the Zuhanden – we seem to get to the final step in Harman’s metaphysics. Since tool-beings are unrepresentable,
insofar as we seem to represent them we must really be representing something else (the famous sensual objects). Meanwhile, tool-beings subsist in a night beyond representation, etc.
Now, I can accept that something along the lines of 3 might follow from 2, but no independent grounds are offered for 2.
The only grounds that would seem to support an inference from 1 (phenomenology of equipment) to 2) (unrepresentability of objects) is that not-being explicitly represented is a modality of being that already distinguishes one kind of entity (executant objects) from representable things (sensual objects).
This would account for Harman treating the business of representation of things as on a par with mugging. But if so, OOO, far from being a realist doctrine, makes the the ontological category of an entity depend on its accessibility or inaccessibility to other entities.
All this is complicated further by the fact that many Object Oriented Philosophers move frictionlessly between an ontological realism of vacuum-sealed objects and an ontic realm of natural objects. Thus when Levi Bryant gave me a righteous telling off for misconstruing OOO as a common sense metaphysics he pointed out that we are composed of cells and atoms and the like, none of which correspond to middle sized dry goods. That’s fine, but I can’t see how Harman or the others have the right to help themselves to this vocabulary since – ex hypothesi – it only applies to the sensual realm.
Hi David,
I understand your frustration. I re-read the opening of tool-being last night and was once more confounded by it, thoroughly vexed in my attempt to divine the argument. Anyway, I’ll do my best to build on your points.
1. Graham’s reading of Heidegger is openly hostile to anything resembling the pragmatist reading of the tool-analysis, and of the existential analytic as such. This means that he’s not trying to move from (1) to (2), because he wouldn’t accept (1) as you characterise it. I think this makes it a dubious reading of Heidegger, as the pragmatist reading you gloss is largely correct (even if their are devils in the details).
2. The tricky question is precisely what his alternative account of the ready-to-hand is, as it largely consists in a number of suggestive phrases that aren’t entirely cashed out: “function”, “effect”, “act”, “capacity to effect”, “impact”, “execution”, etc. There are at least two problematic tensions involved in these descriptions: between an emphasis on possibility (“capacity”) and an emphasis on actuality (“act”, “execution”), and between an emphasis on causal efficacy (“capacity to effect”, “efficacy”) and functionality (“function”, “in-order-to”). These leave me unsure of precisely what Graham thinks the ready-to-hand is. I’ll do my best to reconstruct it as we go.
3. The two major features of the ready-to-hand that are supposed to follow from these characterisations are invisibility and totality. The former is Graham’s gloss on the phenomenological idea that objects recede from view in use, and the latter is his gloss on the idea of referential totality. I’m really unsure the extent to which the latter is supposed to shore up the former, because unless it is then there seems to be little real argument for invisibility. Nonetheless, I’ll try and say something about each independently.
4. There seem to be two ideas underlying Graham’s conception of the ‘absolute’ invisibility of the tool:-
i) The first is that our encounters with the tool should not be understood as us ‘using’ the tool, but as us ‘relying’ on the tool. This is how he initially extends the notion of ‘encounter’ beyond anything like human awareness, and it’s what completely separates him from pragmatist readings of Heidegger. This means that I ‘encounter’ the earth’s gravitational field, and my own internal organs, insofar as I am in some sense dependent upon them from moment to moment. One can intuitively accept that I can depend upon such things without having any awareness of them, and thus that they are in a certain sense ‘invisible’ or ‘fade into the background’.
What’s important to note is that the kind of invisibility he’s appealing to in the latter examples is simply the *absence* of awareness or understanding, whereas the kind of invisibility Heidegger was concerned with characterised a *special kind* of awareness or understanding, i.e., circumspective practical understanding. This highlights the importance of Graham’s move from ‘use’ to ‘dependence’, because it is only the former that can underlie anything like circumspective understanding: an implicit practical grasp of something as part of a larger goal-oriented activity, in which one’s attention is not directed at the thing grasped. Heidegger’s notion of invisibility is fundamentally one of *understanding without attention*, whereas Graham’s is one of *interaction without understanding*.
This indicates the real problem with Graham’s extension of the notion of ‘encounter’. He’s abandoned the notions of awareness of understanding in favour of *causal dependence* (or perhaps *functional dependence*, but we’ll return to this). However, in doing so, he’s already abandoned phenomenology for metaphysics. Any possible defence of his account as performing phenomenological description rather than metaphysical argumentation is thus untenable.
ii) The second idea is that it is the fact that the entity is in some sense “in action” or a pure “execution” that renders it invisible. The extension of the notion of ‘encounter’ simply opens up the possibility of extreme cases in which there is no awareness or grasp of the things encountered. The appeal to execution seems to be the crux of the argument insofar as it’s supposed to show that even in ordinary ‘encounters’ there is no real awareness or grasp. However, how this is supposed to work isn’t elaborated. I have no idea why our encounters are limited to pure activity or why if this was the case it would underwrite a notion of absolute invisibility. This whole thing is problematised by the first tension noted above, i.e., the fact that despite characterising tools as pure actual execution he also frequently appeals to their possibilities or capacities.
5. The best way of making sense of Graham’s account of the referential totality is by paying attention to his descriptions of it as a “world-effect”, “world-system” or even better “world-machine”. The idea seems to be that each individual entity can only be understood in terms of the specific functional role that it plays within its local context, and that this local context must itself be understood in terms of the functional role it plays within a larger whole, and that this bottoms out in a single system of which every entity is a part. The real difficulty in understanding this is that it deploys a very unusual (and extremely unclear) notion of functionality. I hate to appeal to my favourite watch-word, but what distinguishes between a functional role and a causal capacity is that the former is a normative notion, the entity not only *can* perform in a certain way, but there is some sense in which it *should* do it. This ties to the notion of the ‘in-order-to’, insofar as it is the fact that there is some end that the performance aims at which underwrites this *should*.
The principle thing that distinguishes Graham from Heidegger here, is that Heidegger thinks that these kinds of purposive relations between things are established by Dasein and its community, i.e., through the specification of general functional roles by the latter and by their involvement in the practical projects of the former. Graham thinks this is too anthropocentric. He thinks that entities must be enmeshed in these kinds of purposive relations on their own. This is in no way a phenomenological claim, but is already a really serious metaphysical one. It either amounts to endorsing something like Aristotelian teleology, or something that a Spinozistic immanent teleology. This kind of metaphysical claim requires serious justification, as most people would deny this kind of position.
The best way I can reconstruct his position is by looking at the way the normative notion of “function” gets blended with the idea of *actual* causal dependence, i.e., with the idea of the relations of causal dependence an entity is engage in in its current activity or *execution*. This is most apparent in the explicitly mereological examples Graham gives of reference relations (e.g., the bolts referring to the bridge they’re part of), wherein he identifies the whole that the parts compose as the end they aim at. In these examples, the causal capacities that are actualised in composition get transformed into functions because they are normatively underwritten by the whole they *actually* compose. However, this extends beyond *mereological* dependence relations, to include things like *environmental* dependence (e.g., my dependence upon gravity and oxygen), and even goes so far as to incorporate *negative* dependence relations (e.g., my dependence on a meteorite *not* falling from space into me). In essence, whatever relations of *causal dependence* things stand in in order for a thing to be the way it *actually is* are to be treated as relations of *functional dependence*. This means that every other entity in the world gets assigned some functional role in virtue of the dependence relations I stand in to it, and that there is such an assigned for every other entity in the world. If this is right, it puts Graham into much greater proximity to Spinoza than Aristotle (for whom functions are intrinsic).
Anyway, the idea is that these different perspectives from which purposive relations are assigned all enmesh insofar as they are transitive, meaning that, because, if x depends on y, and y depends on z, then x depends on z, then (reversing direction) x is the telos of z. Everything stands (either directly or indirectly) in dependence relations with everything else, and therefore everything stands in purposive relations with everything else. This is supposed to imply that the world forms some single system of ends, tantamount to an all consuming world-machine in which everything has its proper place. I’m not entirely sure about that inference, but I’ll let it slide.
6. Now, let’s examine the possibility I mentioned above, that this idea of totality somehow underwrites the idea of absolute invisibility. The idea seems to be that, despite the fact that the account of totality (and the expanded notion of ‘encounter’) is premised upon the idea of specific isolatable causal capacities and the relations of dependence they underwrite, the being of any entity as an ‘execution’ must be its active role within the whole system of ends, its execution within the world-machine (he actually goes further than this, to say that there is only the world-machine, but I’ll leave that). I simply don’t see how this follows. It only works if we insist on collapsing the notions of causal capacity and functional role into this unitary notion of ‘execution’, in which they can’t be separated out. However, as just noted, we’ve only been able to make sense of any of this by isolating out these different parts. The only thing that could motivate us to treat ‘execution’ as such a black box would be an independent argument for invisibility, which as noted at (4 (ii)), we have yet to see.
Heidegger’s argument that there is no such thing as ‘an equipment’ was an argument to the effect that one couldn’t grasp one kind of equipment without being able to grasp a whole bunch of other kinds to which they are related, because it is these relations which constitute our understanding of the practical activities we can engage in. Graham’s argument here bears no resemblance to Heidegger’s, and it’s hard to see how, even if we grant him the very odd metaphysical postulates which are supposed to shore up his argument, they license anything like absolute invisibility.
7. Really, it all comes back to the earlier point (4 (i)) that Graham is not simply performing phenomenological analysis, but that he’s already engaged in metaphysics. The corollary of this claim is that he doesn’t abandon phenonemology for metaphysics, but instead sits in an uneasy place between them. I think this is what you’re getting at when you say:-
“OOO, far from being a realist doctrine, makes the the ontological category of an entity depend on its accessibility or inaccessibility to other entities.”
When Graham comes up with his extended definition of ‘encounter’, he doesn’t just move from talking about the phenomenological topic of awareness/understanding to talking about the metaphysical topic relations of causal interaction and dependence, but still takes himself to be talking about the former in talking about the latter. The ordinary form of invisibility, i.e., interaction without understanding, is extended and taken to herald some important metaphysical fact about the relations between objects and understanding, which is ultimately built into an important metaphysical insight about relation between objects in general.
I’m not going to go into his reading of the as-structure, broken tools, or how this leads into his later stuff in any further detail, as I’ve done enough already, but I think it’s fair to say that what all of this reveals is a profound methodological mess. The conceit that we’re somehow starting from a descriptive phenomenological analysis has the effect of disguising some pretty sketchy metaphysical assumptions and glossing over the absence of some well needed inferential connections.
As ever, my interpretation may be wrong, I may have just missed the point entirely, but I assure you that it is not for want of trying.
Thanks Pete – this is very useful. There are many philosophers who would argue that individuals acquire new capacities or powers as part of complex; that such arrangements alter the identities and powers of components. I also rather like the idea of technics as a system with its own emergent, dynamic properties. I can buy into this much emergentism but the idea that the operations of a complex whole are not be ontological decomposible into the operations of notionally independent parts does not seem to be what is at issue here. If it is, then execution seems to be just an emergent causal power. But this hardly warrants Harman’s drastic ontological claims, as you say. Oh well.
D
Sorry this emerged as word salad. I’ll try again:
Many philosophers would argue that individuals acquire new capacities or powers as part of complex wholes; that such arrangements alter the identities and powers of components. I also rather like the idea of technics as a system with its own emergent, dynamic properties. I can buy into this much emergentism but the idea that the operations of a complex whole are not decomposible into the operations of notionally independent parts does not seem to be what is at issue here. If it is, then execution seems to be just an emergent causal power. But this hardly warrants Harman’s drastic ontological claims, as you say.
Oh well.
D
What is idealism is enemyindustry’s own next sentence: “In contrast, I hold that intentionality brings us into contact with the real with numbing regularity.”
This is idealism, because it holds that the real is convertible into the accessible. It gives no adequate account of the difference between the tree that grows and bears fruit and the tree that I encounter. No matter the level of “numbing regularity” with which I encounter a tree, that encounter is not the tree itself. Until you account for the difference between the two (as I do) then you are an idealist.
I find this from Harman utterly perplexing. First he says that idealism is in the offing when the real is taken to be “accessible” (he says convertible to, but I would gloss that also as “assimilable to”)–so any view which holds that what is real is “accessible” to some subject is idealism. Second, it is idealism when you cannot account for the difference between the tree that grows and bears fruit and the tree “you encounter.”
For the first, I really think that this is, at some level, (at least, an invitation to) definition-mongering. The real is that which is “inaccessible” (say, paradigmatically, to perception) and to treat the real as if it were accessible is idealism? The only way to save that line of thought is to impute to enemyindustry the thesis that *what it is to be real* is to be accessible. That sounds like something it make sense to apply the concept of idealism to. But, notice the quote–it has nothing to do with characterizing what it is to be real in terms of intentionality–if anything, the opposite, it characterizes intentionality in terms of its target: real things.
Which perhaps accounts for the second criterion for idealism–if you can’t explain the difference between the tree that grows, etc. (*the real tree*) and the tree you encounter, then you’re an idealist. But a direct realist will see no reason to try to make that distinction because she will hold that the tree you encounter is the real tree. (That may be an incorrect view, but its a bad move to use Harman’s second criterion for idealism to characterize direct realism as an idealist view.)
I’m probably missing something.
Slocum: Some good thoughts, and I don’t think you’re missing much. There’s a few issues here.
First of all, Graham’s notion of ‘inaccessibility’ is not supposed to be limited to perception, but to all modes of knowledge. However, that it is phrased in terms of ‘accessibility’ in the first place indicates that he understands other forms of knowledge in relation to the paradigm case of perception. I for one think that this privileging of perception causes problems.
Second, I think you’re right that Graham is bordering on accusing David (and others) of defining the real as that which is accessible, in direct opposition to his defining it as that which is inaccessible. This seems to be in part because he thinks that one has to understand accessibility/inaccessibility in metaphysical terms. This is understandable if we a) model all knowledge on perception, b) understand perception as causal interaction, and c) take the nature of such interaction to be a metaphysical issue. However, it’s quite easy to reject any and all of these assumptions. If one understands knowledge principally in terms of assertion and inference (with observation claims as one type of assertion subject to a special form of inferential challenge and justification), then one can thereby understand knowledge, and more importantly *knowability* in non-metaphysical terms. This means that one can take oneself to have metaphysical knowledge of entities without taking the knowability of those entities to be a part of that metaphysical knowledge, i.e., one’s metaphysical theories need not say anything about ‘accessibility’ in Harman’s terms.
Thanks for the response!
Regarding your last point, that’s why I find some kind of inferentialist understanding of the validity and content of practical and theoretical cognition appealing. Inferences are things we do (or are open to us to do): take up, defend, scrutinize, retract, and so on… so, we leverage and refine the vocabulary we use ordinarily, and we come to have no *special* metaphysical problems about cognition, beyond those we might have with understanding actions (understood broadly) and the vocabulary associated with them.
Of course, there is the interface of the organism with its environment and how we want to understand both that interface and the environment… do we adopt a Quinean-style view about stimuli, a McDowellian “openness” at the perceptual level, look to a kind of Merleau-Ponty-ean embodiedment etc.–Is the environment to be conceived physicalistically (as only physical actions that stimulate nerve-endings), as propositionally-structured states of affairs, as co-original with the embodied subject, etc… But those issues run quite orthogonal to broad categories like “idealism”, especially as Harman seems to use it, have “metaphysical”* aspects, but are open to input from the behavioral and biological sciences, and, why not, literature and art. Assimilating this concern to a very abstract category like “accessibility” doesn’t do justice to the details of the arguments and varied considerations that can be brought to bear.
*I sometimes start to lose my grip on this word, given, among other issues, how often it is used as an accusation or vindication by we philosophers.