Phenomenology, Discourse and their Objects
Graham Harman recently responded (here) to my musings on his argument for his fourfold structure (here). That post was quite brief, but it suggested that the way to reject his approach is to reject the phenomenological standpoint that its based upon, loosely summed up in the idea that ontology must begin with experience. Given that Graham has picked up on the points I made, I feel that I should probably go into a bit more detail, although this will unfortunately fall short of presenting my alternative to phenomenology (fundamental deontology) in full. First though, I think it’s important to say a little bit more about what the core features of phenomenology are, before we try to provide possible reasons for rejecting them. I am of course no expert on Husserl, so my analysis will be faily crude, but hopefully fair nonetheless.
1. The Basic Problems with Phenomenology
Skipping over the methodological details of phenomenology (which are certainly most interesting), I think the two most important features to pick out are the theory of intentionality, and the correlative theory of meaning or content. The first encapsulates the real advance of phenomenology (via Brentano) over both the empiricist and Kantian accounts of experience. The advance over empiricism is threefold. First, phenomenology surpasses the empiricists’ indirect realism, holding that we do not experience ideas or representations of things, but that our experiences are directed at the things themselves. Secondly, experience is the encountering of objects as objects, not the encountering of bundles of sense data that must be actively united into objects. Thirdly, phenomenology takes the objects that our experiences are directed at to transcend our experiences of them. As Graham is fond of pointing out, this has the effect of opening up a distinction between the objects of experience and the qualities that they present to us.
The advance over Kant is more subtle, and more important for my purposes. For Kant, as for phenomenology, we encounter objects in experience, not patches of sense data (it is important to remember that, for Kant, experience and intuition are distinct). However, this is because experience has the structure of predicative judgment. The experience of a tree as a tree involves a synthesis of recognition in which the judgment ‘this is a tree’ is produced. By contrast, for phenomenology, experience is pre-predicative. The content of experience is not that of a judgment, but instead constitutes the evidence on the basis of which one would make, or acknowledge the truth of the judgment. Fundamentally, for phenomenology, judgment is grounded in experience, and not vice versa.
This leads quite naturally into the second feature of phenomenology I wanted to pick out: the phenomenological approach to meaning or content. Now, it’s important to point out that there are obvious internal differences between phenomenologists regarding how to understand meaning or content. I’m not going to address these here. I’m just going to make a fairly general claim which I think catches them all. In accordance with its prioritisation of experience over judgment, phenomenology makes the content of judgment, and the meaning of the words that express it, derivative upon the content of experience. Concepts or predicates get their sense from an originally pre-conceptual or pre-predicative level of phenomenal content. I think Merleau-Ponty describes this best in the preface to the phenomenology of perception, where he criticises the linguistic philosophy of the early 20th century for taking the meanings of words for granted, and failing to ground them back in the original experience of the phenomena.
Okay then, now that I’ve explained the phenomenological picture in very broad strokes, I can move on to providing some reasons why I think we should reject it. A good first point to make here is that although the three positions presented above (empiricism, Kantianism, and phenomenology) have quite radical differences, they all share one thing in common: they all take it that there is something like a content of experience prior to judgment. Putting it a different way, they all accept that there is something given in experience, and that this must somehow be translated into judgment. Of course, they all differ over what this content is, and thus also precisely how it is related to judgment. Empiricism is hard to pin down on this point, as its attempts to characterise the content of experience and its relation to judgment ranges from the crude theories of ideas and their association found in the original empiricists to the vastly more complicated accounts of the logical empiricists. Regardless, the most interesting contrast is still to be found between phenomenology and Kantianism.
Kant takes there to be a content of intuition, a non-conceptual matter of sensation which must be related to concepts to produce the determinative judgment of recognition. The principal problem of the first critique is the attempt to work out how this non-conceptual content can be related to the conceptual in a systematic fashion. There is thus a relation between the non-conceptual given and the conceptually articulated content of judgment, but this is not a straightforward matter of derivation. For Kant, the conceptual has a life of its own beyond intuition, insofar as concepts are rules governing inference between the judgments they make up (e.g., if x is a dog then x is a mammal). In addition, Kantian intuition plays no justificatory role, i.e., it is not evidence for judgment. Phenomenology shares the Kantian concern with a pre-conceptual content of experience, but it takes this content to be much richer, so rich in fact that the conceptually articulated content of judgments is an abstraction from it. The content of a judgment is both derived from phenomenal content, and justified on the basis of it. Moreover, it is important to recognise that the derivative character of conceptual content is based on the role of phenomenal content as evidence. For phenomenology, a judgment is itself intentional, but its content is based on the intuition which would fulfil it, i.e., on the phenomenal content that would count as evidence for it.
Now, there has been much work done criticising ‘the myth of the given’ (inspired of course by Sellars). This is primarily aimed at the empiricist and Kantian conceptions of sense data. I’m not going to rehearse all of this here (in part because I’m not yet fully acquainted with Sellars, but get most of my take on him via Brandom). Instead, I’m going to take aim directly at phenomenology, and argue that the kind of pre-predicative phenomenal content it posits is redundant. As such, I’m going to do exactly what Graham hints at in his second counter point, and deny the existence of the phenomenological sphere entirely. I’m going to do this in two ways. First, I’m going to argue for a certain constraint on accounts of conceptual content which the phenomenological theory of judgment doesn’t take into account. Secondly, I’m going to argue that there are no good reasons for thinking that we need to posit anything like evidence for making a judgment, or taking it to be true. The first point shows that the pre-predicative phenomenal content is not sufficient for the derivation of conceptual content, and the second point shows that it is not necessary. Taken together, these demonstrate the redundancy of the notion of pre-predicative phenomenal content.
i) Concepts and Inference
First, as just noted, I’m going to argue for a specific constraint on accounts of conceptual content. I’m going to do this by examining what it is to grasp a concept, and thus also what it is to grasp the content of a judgment that is composed out of such concepts. The first important thing to point out here is the necessary relation between concepts and judgment. Although judgments are composed out of concepts, this does not mean that concepts can be understood independently of their role in composing judgments. In fact, the opposite is true – we must understand concepts in terms of the role they play in judgment. We will return to this point later.
Now, the ordinary way of understanding the grasp of a concept, and thus also the grasp of the content of judgments that it is involved in, is in terms of classification. The idea here is that one has a grasp of the concept if one can adequately (not necessarily infallibly) classify things into those that do and those that don’t fall under the concept. Phenomenology takes this rough approach insofar as it takes one’s grasp of the content of judgments such as ‘this is blue’, to be dependent upon one’s grasp of what phenomenal content would count as evidence for taking the judgment to be true. Our grasp of the conceptual content of judgments is directly linked to our ability to discriminate the salient features of the phenomena. I’m going to argue that, although there must be some role for such discriminative abilities in accounting for our grasp of concepts, on their own they are not sufficient to constitute a grasp of concepts and thus also the judgments they compose.
Here I’m going to turn to Brandom’s tried and tested parrot example. The idea here is that one could train a parrot to say ‘this is red’ whenever it was showed something red, and it would thus have a perfectly effective ability to discriminate between cases of redness and non-redness. However, we wouldn’t like to say that the parrot thereby understands the content of what it is saying. Indeed, we’d rather claim that it isn’t strictly saying anything at all, as much as making sound. Here the parrot functions in the same way that a red-detecting machine would. It has a reliable causal disposition to discriminate between the red and the non-red, but it does not thereby grasp the concept red. What the parrot (or the machine for that matter) is missing is a grasp of the consequences of the claim ‘this is red’. One doesn’t understand the content of ‘this is red’, unless one understands that this implies ‘this is not green’, or ‘this is coloured’, etc. and one doesn’t understand the concept ‘red’ unless one understands the systematic role it plays in constituting these relations (e.g., that ‘x is red’ implies ‘x is not green’, ‘x is coloured’, etc.). In essence, in order to have a grasp of any conceptual content, one needs to have a grasp of where it stands within what Sellars called the space of reasons. Now, this understanding need not be explicit and theoretical, but can (and for the most part does) simply consist in a practical ability to navigate the space of reasons within ordinary discourse. It consists in the practical abilities required to play the game of giving and asking for reasons that lies at the heart of discourse.
The constraint on accounts of conceptual content is thus the following: a grasp of the content of a judgment must involve a grasp of the inferential relations it stands in with other judgments, and a grasp of the content of a concept must involve a grasp of the systematic contribution it makes to the inferential relations of the judgments it is involved in. This spells out the intimate relation between concepts and judgments mentioned above. There are two qualifiers to this constraint. First, this constraint calls for some understanding of inferential relations, not a complete understanding. Precisely what constitutes a sufficient understanding in any given case is not something we need decide. All that we need defend is that it is not the case that one could grasp a conceptual content and have no grasp of its inferential role. Secondly, we must repeat that this doesn’t deny that classificatory or discriminative abilities play some role in the grasp of conceptual content, it simply holds that this is not sufficient on its own. The phenomenological account of the way in which the content of judgments is derivative upon phenomenal content does not meet this constraint, insofar as it says nothing about inferential role. It is thus isn’t sufficient as an account of the content of judgments.
ii) Evidence and Observation
I’m now going to argue for the redundancy of the phenomenological notion of evidence, and thereby complete my argument for the redundancy of pre-predicative phenomenal content as such. To do this its important to get clear about the precise role that the notion of evidence plays. As we have already noted, evidence is a justificatory notion. Among other things, it is meant to account for the contribution that perceptual observation (as well as the other non-sensary forms of intuition Husserl was interested in) makes to debate. On this account there are two different forms of justification: inferential and evidential. Inferential justification involves judgments (or claims) supplying warrant for other judgments (e.g., ‘It’s Sunday tomorrow, because it’s Saturday today’), whereas evidential justification involves appealing to something other than judgment for warrant, namely, the intuition which fulfils the judgment. Now, we all know that we need to give some account of the way that perceptual observation enters into discourse. It seems to be the case that some claims are justified on the basis of experience, rather than on the basis of other claims. The question is whether we need to posit anything like an evidential content distinct from the content of the judgment it grounds in order to do this.
What the phenomenological approach does is to effectively situate a mediating layer between the causal affection which produces sensation and the observation judgment which arises from it (causation – evidence – judgment). This has been described by others as a level that mediates between the space of causes and the space of reasons. This is problematic because it introduces a transition from evidence to judgment wherein the evidence functions like a reason, without actually being a reason. Specifically, evidence isn’t open to the same kinds of public challenges and criticism that genuine reasons are, i.e., it can’t contribute to the eminently social dimension of reasoning. Put simply, the evidence of my senses might have force analogous to that of a good reason for believing a particular claim for me, but it does not have that force for anyone else, whereas if reasons are good reasons, then they are so for everyone (at least in principle, there are some caveats about auxiliary commitments to be made here). Without this social dimension, wherein one’s reasons can be demanded and criticised it’s hard to see how evidence can count as anything like justification at all.
There are really two good arguments for thinking that there needs to be such a mediating level, both of which turn on appeals to what experience is like for us. First, it is noted that not only can we encounter something as being a certain way, without thereby explicitly judging that it is that way, but that we can in fact refuse to assent to a corresponding explicit judgment. I can encounter a pink elephant in an inebriated state and nonetheless refuse to judge that there is a pink elephant in front of me. This seems to indicate that there is something given prior to judgment, which can be, but need not be, translated into judgment. Secondly, it is noted that the phenomenal character of our experience as we experience it is indeed richer than any given judgment that it might count as evidence for. We genuinely do encounter objects as being certain kinds of things, and as being certain ways, but the content of the encounter is a unified experience that must be taken apart in order to isolate the specific features that any given judgment touches on. Moreover, there are features of this which are totally absent from judgment, such as the horizonal character of intentional experience. In order to show that the notion of evidence is redundant I need to tackle both of these arguments.
Taking the first argument first then, Sellars provides us with a very neat counter point. The pink elephant case is one in which we might say something like “It seems like there is a pink elephant in front of me”, rather than “There is a pink elephant in front of me”. The standard Cartesian approach inherited by Husserl takes such claims about seeming or appearance to be incorrigible. I might be able to be mistaken about whether or not there is in fact a pink elephant in front of me, but I can’t be mistaken about whether it seems to me to be the case. The Cartesian strategy is then to try and give an account of how such incorrigible (or immanent) knowledge can be used as the basis for genuine empirical knowledge. So, from a first personal perspective, experiential evidence functions as the basis for making a particular claim, but from the perspective of the public activity of giving and asking for reasons for the claims we make ‘seems’ statements serve to make this warrant explicit, even if we recognise that things can seem to be one way and actually be another. Sellars’ insight is that we needn’t see ‘seems’ claims as providing anything like justification for proper claims. Instead, Sellars takes it that proper claims (or assertions) and ‘seems’ claims share the same content, whereas the former assert that this content is true, the latter withdraw such assent. What is withdrawn is the responsibility to justify a claim that one undertakes when one makes an ordinary assertion. As such, ‘seems’ claims are not claims about some special inner dimension which are automatically justified, but are rather a privative form of assertion for which the question of justification does not arise.
This is of course only part of the story. We also need an account of how it is that perceptual observation does enter into the game of giving and asking for reasons, of how there can be something like a non-inferential input into discourse, that does not appeal to this mediating level of phenomenal content. Sellars also provides us with the basics of this (and Brandom expands on it, although, there is a massive debate between him and McDowell here that I’m not an expert on). There are a couple ideas that need to be elaborated to explain this. First, we need to recognise that there is indeed an additional level to the process of justification. However, this is not that of evidence, but rather that of deference. As already noted, when we make a claim we undertake a responsibility to justify that claim, or to demonstrate our entitlement to it. There are two ways to do this: to provide reasons for it, or to defer to another’s authority. Deferring to another’s authority means we pass the buck on to them to justify the claim. This is unproblematic, because the only way I can do this is if they have made the claim themselves, and if they have done so, then they have already undertaken a responsibility to so justify it. So, making a claim not only means that I undertake a responsibility to demonstrate my entitlement to it, it also involves licensing others to make that claim, and thereby to defer to me when they are called on to justify it. Secondly, we need to recognise the default and challenge structure of discourse. The idea here is that assertions are entitled by default. This default entitlement can then be challenged, but such challenges are not free, and must themselves be warranted. In essence, claims stand unless there is a good reason to question them. This is what ultimately stymies the pyrrhonian skeptic’s regress of reasons.
The Sellarsian solution leverages these two insights in order to provide a second person account of how observation functions as a non-inferential input into discourse. If I’m sitting in Coventry, discussing with my girlfriend what the weather is like in Sunderland, I can call up my brother and ask him whether he can see snow outside his window. If he says ‘yes’, then I can confidently assert that ‘It’s snowing in Sunderland’. Here I have deferred to my brother’s authority, but I haven’t deferred to him as someone who can provide further reasons for his claim, or as someone who is himself deferring to someone else. Instead, I am deferring to him as a reliable reporter of snow. In order to challenge my claim that it is snowing in Sunderland, my girlfriend then needs to provide a good reason to think that my brother is not a reliable reporter of snow, or at least that he is not a reliable reporter in the given circumstances. She might know that he’s currently tripping on acid for instance, which would potentially put his ability to reliably classify the weather as snow in question. In essence, when we endorse an observation claim made by someone else, we endorse an implicit reliability inference, ‘x says a is F’, ‘x is a reliable reporter of F’s’, therefore ‘a is F’, and challenges to the claim have to provide reasons for taking this inference to be faulty.
The important fact to recognise here is that this makes the argument about whether or not the claim is warranted an argument about the individual’s discriminative capacities, but it treats those capacities in purely causal manner. In challenging observational claims (leaving issues of sincerity and such aside), we are interested in the causal factors that affect the individuals disposition to make certain assertions on the basis of certain stimuli. At no point do we need to say anything about an internal dimension of phenomenal content that mediates between the causal sensory stimuli and the judgment or assertion. This account can then be transposed to the first personal perspective. In making an observation report I am implicitly endorsing a reliability inference. Although from my first person perspective it might seem as if I am presented with evidence on the basis of which I make a certain claim, the form of the inference that I endorse by making an observational claim within discourse is only concerned with my causal dispositions.
This leads us nicely into the second argument, which stresses the complex structure of our experience in excess of this seemingly crude account of causal dispositions. To counter this argument it is important to recognise what phenomenology aims at in general, namely, a universal account of the structures of consciousness qua consciousness. This account is meant to be independent of any account of the particular causal mechanisms that such a consciousness might be manifest in, indeed, it’s meant to be indifferent to discussions of causal mechanisms entirely. Now, the Brandomian/Sellarsian account of observation statements as non-inferential inputs into discourse presented above is universal, but not indifferent. It does imply that anything that could count as an observer have causal dispositions to make reports, but it doesn’t present any specific limitations on what the specific mechanisms underlying such dispositions would have to be. That being said, it doesn’t preclude an analysis of the specifics of those mechanisms, in fact, it encourages it. Importantly, this doesn’t just involve an analysis of the mechanisms underlying our discriminatory capacities, but also an analysis of the mechanisms underlying our capacities to engage in discourse. This means we’re open to an account of how our perceptual mechanisms are already bound up with the mechanisms which constitute those abilities in which our grasp of concepts consists (which I think is what McDowell’s whole discussion of conceptual capacities is about). This kind of empirical analysis promises to account for the richness of experience in its difference from (yet connection with) the conceptual content of judgment, but it promises to do it in a way which is specific to the actual mechanisms in which the given ‘consciousness’ is manifest, without appealing to anything like a universal form of content shared by all such ‘consciousnesses’.
I’ve put scare quotes around the word ‘consciousness’ in the last line because I ultimately feel that the word ‘consciousness’ has lost any usefulness it might have had in philosophical debate. It has way too much baggage. Some take it to be an empirical matter, some take it to be a non-empirical matter, and the vast majority end up somewhere in between, and such quasi-empirical notions either need to be cleaned up or jettisoned. Personally, I vote for the latter. In essence, what I’ve been arguing for here is not just the redundancy of the notion of phenomenal content (which I think I’ve provided some good reasons for), but also for the redundancy of the idea of a universal structure of consciousness. We don’t need something like consciousness mediating between the particular causal structures within which discursively capable beings are manifest and the universal normative structures of discourse within which they move. We don’t need to posit a self-sufficient phenomenological sphere, no matter how much its self-sufficiency is apparent to us.
2. A Small Note On Causation
Having got my reasons for rejecting phenomenology off of my chest, I think there is an additional interesting point that can be made about Graham’s work in light of the above considerations. Although I characterised the phenomenological level as mediating between the causal and the discursive, phenomenologists obviously wouldn’t accept this analysis straightforwardly, precisely because of the indifference to causality I mentioned above. Graham is an even more interesting case though. He effectively does the opposite of what I’ve done above, by trying to show that causation needs to be understood in terms of the phenomenological account of phenomenal content. This means he actually needs to maintain phenomenology’s indifference to causal mechanisms in an even more radical way than most, because he has to secure the universal character of the structure of intentionality for all objects, independently of their particular causal make up, in order then to base his account of causality on this structure.
Now, I suspect Graham will respond to the above by suggesting that it takes the nature of causality for granted. To defend myself against this in advance, the above doesn’t commit itself to any specific metaphysics of causality. If it is incompatible with Graham’s account of causality, it is not because it makes some particular assumptions about the nature of causality as much as because it rejects the phenomenological grounds on which Graham’s account is built. Now, at this point one might maintain by reductio ad absurdum that my account is false, because Graham’s phenomenological approach is necessary to solve a problem with regard to the metaphysics of causality, namely, the problem of how it is possible that objects can interact. My response to this is that I don’t really see the problem Graham finds here. This isn’t because I am committed to a naive scientism, and just take causality for granted. I think that there are a lot of very interesting questions with regard to the metaphysical status of causation. I genuinely believe a metaphysical theory of causation is required, and that it needs to involve such things as an account of the relation between causation and law, the status of statistical causation, and the relation between causation and mereology. However, I don’t see why it is any more problematic to claim that there can be direct interactions between entities than it is to claim that there can be interactions between real objects and sensual objects.
Now, Graham has given quite a detailed historical background to this problem. He sees this problem as being dealt with both in the classical metaphysical occasionalism of Islamic philosophy and Malebranch, and in what he has identified as the more epistemologically inflected occasionalism of Hume and Kant. Graham criticises both of these trends as global forms of occasionalism, which require all causal relations to be mediated by the same thing, be it God in the former or the mind in the latter, and proposes, along with Latour, a local occasionalism. However, although I can just about see the unity of the occasionalist tradition as he presents it, I don’t see a unified philosophical problem that stands outside the various sub-traditions that make it up. For instance, classical metaphysical occasionalism did not only provide a theological solution, but was motivated by a theological problem about the power of God. If we no longer take metaphysics to be bound by theology, then this problem has no hold on us. Similarly, Hume and Kant are motivated by empiricist epistemological concerns that we also may now reject (and I do). This means we need a good reason to accept the problematic status of direct causal relations independent of these traditions. Now, I think Graham does provide such a reason, but this reason is based upon his account of how real objects withdraw from our experience of them, and also from each other. This means that within Graham’s system, this is a genuine problem which needs a solution. However, this problem can’t thereby motivate us to take up the rest of Graham’s position. Graham either needs to motivate the phenomenological standpoint he adopts independently of the problem of causality, or to provide a non-phenomenological account of the problem.
My original intention in this post was to go a bit more into what I’d promised to say about a discursive account of the for-us/in-itself distinction. However, this post is already a lot longer than I’d intended it to be, and I need to get on with some thesis work. I will get back to talking about this in more detail at some point hopefully. I’d also intended to say something more about the object/qualities distinction, as Graham said I ignored it in my previous post. I didn’t exactly ignore it, but I didn’t entirely suggest where I stand on it. I suppose I’m still slightly confused as to whether Graham means that an object has essential qualities which are in excess of its accidental qualities, or whether an object is in excess of all its qualities. I can just about see reasons for why you’d think either from a discursive position (though I’ll have to leave these to another time too). However, I’d like to close by addressing Graham’s third point, which I’ll quote in full:-
“Moreover, if our understanding of reality is not indexed to some notion of experience, then we need not think of the real properties of entities in terms of any model of experiential qualities, be it an empiricist model or the Husserlian model Harman advocates.” This gets me slightly wrong. I don’t think that the real properties of entities can be experienced. For me there is an absolute gap between real properties and experienced properties. It’s not I, but rather the cognitive science wing of S.R. that insists on the correspondence theory of truth (which is not an essential part of realism, incidentally; I’d even say the opposite– if you think the real can be exhausted by knowledge of it, then you aren’t a genuine realist).
I wasn’t suggesting that Harman thinks that the real properties of an object can be experienced. What I was suggesting was that the second axis in Harman’s model is derived from the Husserlian account of experience, and that he extends this from the sensual into the real. It’s my conviction that we shouldn’t base our account of real properties on anything like experience, even if we take them to be impossible to experience. Properties like mass, hyperconductivity, and economic stability shouldn’t be understood even on analogy with experiencible qualities.
Right, back to Heidegger for me…