Let me jump in right away and elaborate on a specific but rather central point Haugeland makes in Pattern and Being. To perceive an object in the world is to get hold of a "pattern", and such a pattern is objectively perceived inasmuch as it, together with other objects in the same empirical domain, is governed by constitutive standards (e.g. governing rules, norms, and/or enabling material/ecological/social conditions) in such a way that it (e.g. the purported object or phenomenon perceived) could conceivably fail to accord with some of those standards, and hence signal a case of misperception of the object, or misunderstanding of the purported empirical domain. Hence the objective perception of an object involves some degree of understanding of the laws (or norms) that objects of this sort obey (or are held up to, in the case of objects that belong to socially instituted domains and practices). And perceptual acts can't count as objective -- in the sense Haugeland seeks to explain -- unless the observer also has, at the very least, a tacit understanding of, and commitment to, the constitutive standards governing the empirical domain this object belongs to.

One paradigmatic example of such an empirical domain is the domain of "intentional" (in Brentano's sense) mental "states" such as belief, desires and intentions. Ascribing such states to persons only is possible, and only makes sense, when effected against the background of a larger pattern of rational behaviors (including speech acts) that justifies a large set of broadly consistent such ascriptions. This constitutive standard of rationality can be assimilated to Davidson's constitutive ideal of rationality (making possible the "radical interpretation" of the linguistic performances of an individual), and to Dennett's insistance that the intentional stance towards intentional systems only can succeed trough treating such systems as suitably "rational".

In this post (and the following one), I am going to take for granted Haugeland's accounts of objects, or "objecthood", and of the objectivity of judgment/perception. Other participants are free, of course, to question them. I rather want to explore an interesting consequence of Haugeland's thesis, defended in Pattern and Being, that objects can be identified with patterns that are objective (i.e. that are real objects, though not necessarily material "things") only inasmuch as they are constituted by standards governing broader empirical domains those objects are "sub-patterns" of. This thesis might by regarded as the ontological counterpart of the epistemological thesis of the theory-ladenness of experience/observation. There doesn't exist "raw data" of either sensory experience, or scientific observation, uncontaminated by theory and, likewise, there aren't any "raw objects", as it were, that would be the objects that they are quite appart from the "laws" governing them. The laws that govern the objects found in an empirical domain make up broad patterns -- patterns of conformity with coherent sets of constitutive standards -- within which alone those objets are, not only discernible by us, but can so much as exist as the sorts of objects that they are.

I now come to the consequence of this thesis that I wanted to highlight. Conway's Game of Life, as well as the idea of a universe governed by universal laws, such that the "temporal-evolution" of such a universe could be simulated on a digital computer, suggests the possibility -- seemingly ruled out by Haugeland -- that the fundamental entities making up our universe have ontological primacy in the sense that they exist in themselves appart from the super-ordinate patterns that they make up, or (equivalently) appart from the laws (deterministic or not) that govern their interactions and "temporal" evolution. We can thus imagine that the "inhabitant/scientists" who might have evolved into, and "inhabit", a "universe" thus simulated could, by means of "observation" of the empirical phenomena that occur in their "world" (and that we can also discern in the printout or display of the computer simulation), together with suitable theorizing, come to disclose the "laws" that govern their own simulation. Those "laws" just would be the rules of the Game of Life, or of whatever set of fundamental physical laws govern the temporal-evolution of universe we have chosen to simulate and in which those "intelligent creatures" "evolved".

This is just a thought experiment, and it may be the case that the laws of quantum field theory (or whatever fundamental physical theory can be held to subsumes them) preclude our own universe being simulated in any kind of digital computer imaginable, even with the provision of infinite resources. This will not affect my main point since I am merely using the thought experiment as a loose metaphor to highlight two distinct conceptions of natural laws. There are, first, laws* as Haugeland understands them, which fall under the generalized concept of constitutive standards of an empirical domain. Those laws can, indeed, be disclosed empirically and/or imposed conventionally, in the case of socially instituted domains such as the the domains of chess, or economic markets and currencies, etc.

(Let me note, as my previous "and/or" suggests, that Haugeland doesn't hold to a dichotomy here between laws of nature and laws of men (or customs or whatever). It can be a matter of empirical discovery for a field anthropologist to find out that members of some foreign tribe are playing a game such as chess in an unfamiliar medium. Instituted domains also obey internal constraints, may have a higher degree of contingency than the domains of the natural sciences, but still be suitable objects of empirical, and objective, discovery. The patterns at issue may be natural patterns, social patterns, mathematical patterns, etc., though, maybe, the applicability of the idea of empirical discovery could be quibbled with in the third case.)

And then let me come to a second conception of "laws", let me label them 'laws**', that describe patterns of subsumability such as the "rules" that govern the evolution of the "life plane" in Conway's Game of Life. This second conception of the "laws of nature" dovetails with "metaphysical realism" (as Hilary Punam uses the term critically in Representation and Reality), and with the idea that our world (or "universe") is as it is, at a fundamental level, and temporally evolves as it does (in accordance with laws**), independently of whatever our contingent interests might be. The common belief among physicists, possibly shared with some "realist" philosophers (though by all means not all, since Haugeland himself was an avowed realist), is that such "fundamental" laws**, just like Haugeland's own laws* (e.g. the constitutive standards of specific empirical domains) can be discovered empirically. I would like to question this. And indeed, I would like to suggest that, inasmuch as laws** can't be disclosed empirically at all (as I purport to show), and can't make any difference to the nature and structure of our experiences, they are unintelligible.

I'll make my case, which I take to be continuous with Haugeland's thinking, in a followup to this post.

I'll make my case, which I take to be continuous with Haugeland's thinking, in a followup to this post. — Pierre-Normand

I've got an area I'd like to explore but will wait until Pierre-Normand has made his further case. Very rewarding essay to bounce off.

I've got an area I'd like to explore but will wait until Pierre-Normand has made his further case. Very rewarding essay to bounce off. — mcdoodle

Please go on! I'm rather busy at the moment but will try to complete my followup later today. My argument also is harder to articulate than it had seemed to me when I rehearsed the argument mentally. In any case, I don't mind this being a peripheral sub-thread.

This is a followup to this post in which I was intending to show that laws** purportedly governing "things-in-themselves" are unintelligible, unlike the laws* that govern the objects ("patterns") that show up in empirical domains constituted by those laws* (i.e. normative standard) as they are potentially understood by us when we commit to them. The demonstration of this thesis, and the explanation of what it means, though, is a bit more difficult than I had anticipated. I had conjectured that the fundamental laws that govern Conway's Game of Life, which are the rules that determine from one 'time step' to the next one the state (on/off) of each individual pixels as a function of the states of the 8 neighboring pixels, would be unintelligible and unknowable to (eventual) intelligent 'inhabitants' of such a Life Plane. That conclusion likely is too strong and not quite what I needed in order to show (or, at least, suggest) that objects can't exist apart from the normative standards that we hold them answerable to.

My initial idea was that if we were to run a computer simulation of such a 'world' (i.e. an artificial life simulation) then the temporal evolution of the simulation would be quite divorced from the 'time' 'experienced' by the simulated 'creatures' who 'inhabit' the simulated 'world'. That's because such an experienced 'time' is part of the content (or, more properly, the form) of their intentional states and those states are disclosed by our adopting an intentional stance towards their 'behaviors'. (I am using single-scare-quotes to refer to all the phenomena that are, from our own perspective, merely simulated and a matter of our interpretation of the regular features -- the patterns -- that emerge in our interpretation of the computer output).

If we were to interrupt the simulation and resume it at a later time, then, this obviously would have no impact whatsoever on the 'experiences' of the simulated 'creatures'. This only shows, though, that our own experienced (and measured) time is divorced from the 'time' that marks the progress of the simulation (e.g. the discrete time steps in the Game of Life). It is only to the latter that the 'experiences' of the simulated 'creatures' are pegged. But this also highlights the fact that whatever those 'creatures' 'experience' (or 'think') doesn't even depend on our running the simulation at all. Destroying the computer doesn't destroy their 'world' any more than it would destroy the decimal development of the number Pi if this were what the computer was computing at the time when is was destroyed. Simulating a world just is exploring the features of a "logically possible world" (if that much) and falls short from creating one; or so I would be prepared to argue. From such considerations I had thought that I could easily infer that the simulated 'creatures' can't, by 'observational' and/or 'theoretical' means, or by any other means, discover the basic laws (i.e. the rules of the Game of Life) that govern the evolution of their 'world'. That's because those laws mirror the rules codified in the algorithm being executed by our computer, and this computer needn't even exist for their 'world' (and the 'objects' within it) to 'exist' for them.

But this inference involves a confusion. What depends contingently on the existence of the computer, the laws that govern its functioning, and the norms determining that it is functioning correctly (i.e. that it is running the algorithm that we want it to run) just is the simulation process, not the 'world' being simulated. Hence, there isn't any reason a priori, so far as I can see, why the 'creatures' that 'inhabit' some 'world' (which we may or may not want to, or be able to, run a simulation of) wouldn't 'discover' its 'fundamental laws' irrespective of the laws that govern the simulation process. It is those latter laws that are unknowable to them. But, if the simulation is running correctly, then the computer algorithm just happens to mirror the 'laws' of their world -- something that whey could come to 'know' as their 'fundamental physics'. (What I mean for them to be 'discovering' such 'laws' just is that such formal laws would be (part of) the content of the intentional states that we would ascribe them (or to those among them who are 'physicists') if we were to adopt the intentional stance toward them in the context of a simulation of their 'world' -- and of their 'scientific activity'.

My original question nevertheless remains, regarding the ontological status of the fundamental 'objects', the individual pixels in the Life Plane of Conway's Game of Life, from the pont of view of its 'inhabitants'. Assuming that they would correspond to theoretical entities knowable to potential 'inhabitants' of the Life World: are such 'objects' existing in themselves appart from the 'laws' governing them? (I conceive of them as simple objects that have the property of being either on or off, in addition to having relational topological and spatial properties) But I think what I originally intended to show can still be shown. It is only through adopting an intentional stance towards potential 'inhabitants' of the Life World that we have come to conceive of those pixels as objects at all, as opposed to them being uninterpreted symbols potentially churned out by a meaningless computer algorithm. Those 'objects' only have intelligible being qua theoretical entities that have some degree of explanatory value either for the of potential 'inhabitants' of the Life World, or, as re-identifiable patterns that are interesting to us (even in the case where no 'life' emerges in the Life World). Those objects, therefore, as Haugeland surmised, don't exist apart from the laws that govern them, and those laws only are intelligible as part of some set of constitutive standards that we (or Life World 'inhabitants') commit to when investigating them. The argument may not be quite complete yet, but I may have bitten more than I can easily chew.

I am wary of dwelling on the Game of Life, except as a rule-based game of something. Maybe its hubristic title puts me off. It's just a game people play on computers to me, of less interest than re-reading Wittgenstein's remarks about rules.

In Haugeland's paper I am struck by two things for starters. One is how to talk of patterns seems to mean we must talk of, just accept, 'elements' as primitives of some kind, in the frame we're discussing, to be able to talk of a pattern of elements. Thus in a pattern of dots a dot is primitive; by contrast in a land of the blind, for instance, only a raised, tactile dot would be primitive, and a visual dot would lack meaning. Perhaps this is related to Pierre-Normand's point about regarding something as 'objects' at all: what we call objects depends on our schema.

My second point is about rationality. 'Rationality is the mother of intention.' Haugeland quotes Dennett approvingly. I'm interested then to know what rationality is. It worries me that it's question-begging, if we look for something primary and call it patterning, but then it turns out it's grounded on this other even more primary thing. An example in my current studies is about the distinction between rationality of action and the rationality of action-explanation. There's a Davidsonian view of that, for instance, which turns every explanation of action into something propositional. There's a different view (there are lots, I don't mean it's one or the other) which argues that the grip of certain strong emotions explains some action, without need for a propositional attitude. The latter suggests we can find order even in those spheres of our activities where we are arational or irrational; the former seems to lean towards a notion that the observer and the actor must both be rational creatures in some way to start with. Well, I think this may be as muddled as I am but the muddle is about Haugeland's core principle: what is rationality when she's at home? Why does it have to be specified separately in thinking about patterning?

Well, I think this may be as muddled as I am but the muddle is about Haugeland's core principle: what is rationality when she's at home? Why does it have to be specified separately in thinking about patterning? — mcdoodle

If anything ought to be called Haugelans' core principle, in relation to the thesis defended in Pattern and Being, it seems to me that it might be the idea that recognizing objects in an empirical domain requires existential commitments directed simultaneously to those objects (i.e. sub-patterns) and to the normative standards that govern their behaviors (i.e. super-ordinate patterns) -- and this dual commitment both enables and presupposes an understanding of those objects. Surely, such an understanding of objects also presupposes that whoever understands them (and their empirical domain) is rational. But rationality as such isn't quite thematized as itself being the constitutive principle of objective perception. (Though such a more demanding thesis may be defended in the more explicitly developed neo-Kantian account of objective truth telling found in Haugeland's later Truth and Rule Following paper.

Rather, in Pattern and Being, Haugeland merely invokes the specific constitutive standard of rationality, and the corresponding empirical domain disclosed by intentional state ascriptions to rational beings (i.e. to Dennett's "intentional systems"), as just one example among many potentially discloseable empirical domains of objects. Being able to recognize super-ordinate "patterns" of rationality in the behavior of "intentional systems", and being able to objectively ascribe to them beliefs, desires and intentions, etc., go hand in hand. The recognitions of the former (i.e. of the constitutive standards of rationality) and recognition of the latter (i.e. particular beliefs and desires ascribable to individuals, etc.) make up a unitary understanding of two levels of patterns that are co-constituted -- that is, such that each one of them can be intelligibly recognized to exist only in relations to the other. The level of objects (sub-patterns) isn't thus basic in relation to the super-ordinate level of constitutive standards (higher level patterns) governing them. Both levels are equally basic.

It is a bit of a distraction that the only sort of entity that can understand an empirical domain of "intentional systems" itself must belong to such a domain and therefore be rational. Nevertheless, the subset of our "actions" that aren't fully, or not at all, "rational", and that may more closely resemble the instinctual, learned or conditioned behaviors of most of our animal relatives, also are phenomena that belong to a specific empirical domain (a proper object of study for ethology, perhaps) that would have its own constitutive standards, according to Haugeland, and those standard -- e.g. the demand that behaviors be seen as adaptive, make ecological sense, etc. -- will no doubt fall short from demanding the subsumption of those behaviors under demanding norms of rationality.

Can there be true and meaningful synthetic a priori statements?

Haugeland's view about objective perception illuminates this question, I think. There was an interesting thread three months ago about Kant's Prolegomena To Any Future Metaphysics. Let me quote this paragraph from Jamalrob:

A few words about a priori and a posteriori. These are about justification, i.e., how we come to know things, so they are epistemological concepts. In the CPR Kant says that “There can be no doubt that all knowledge begins with experience”, but that “although our knowledge begins with experience, it does not follow that it arises from experience.” What this means is that it is experience that calls forth knowledge, not that it is the source. For example, it is in experience that we come to know about cause and effect in the first place, but only because events must be experienced in terms of a prior, independent (pure) concept of the understanding. — jamalrob

This distinction between two ways in which knowledge relates to experience, here marked with the phrases "begins with" and "arises from" is interesting but puzzling. It seems clear that the first way for knowledge to relate to experience ("i.e. "begins with") characterizes this form of knowledge (expressible with synthetic a priori statements) as a condition for the very possibility of empirical (i.e. a posteriori) knowledge, and hence for meaningful and contentful "intuitions". Once the conditions for the possibility of such experiences are satisfied, then it is empirical knowledge of particular facts (and particular objects) that can "arise from" experience. This empirical knowledge is expressed with synthetic a posteriori statements.

The sorts of synthetic a priori statements that were centrally at issue, in the passages from the Critique or Pure Reason quoted by Jamalrob above, were statements that express the content of the pure concepts of the understanding, that is, the Kantian categories. Those are a priori concepts that purport to represent the form thought has in virtue of the fact that it relates essentially to intuition while abstracting from the specific empirical contents those intuitions provide. Whatever one might think of the feasibility of such an inquiry into the form of the pure understanding -- and for a masterful defense and realization, one can refer to Sebastian Rödl's Categories of the Temporal: An Inquiry into the Forms of the Finite Intellect, HUP, 2012 -- it may still sound incredible that there might be synthetic a priori statements that aren't pure in this sense, and that therefore express genuine knowledge about the empirical world, or, more precisely, about determinate domains of empirical objects.

Yet that is exactly what Haugeland's account of objectivity enables us to better understand. Wilfrid Sellars -- whose Concepts as Involving Laws and Inconceivable Without Them gets credited by Haugeland as anticipatory of some of his own ideas (further developed in his Truth and Rule Following) -- also defended a broader conception of (inpure) synthetic a priori truths.

Here is a quote from Willem A. DeVries' book on Sellars:

"Sellars believes that any conceptual framework, necessarily, includes valid forms of material inferences. But synthetic a priori truths are framework relative; there is no set of framework-independent synthetic a priori truths. The structure of our conceptual framework, which is responsible for our a priori knowledge and which we often take to reflect articulation of reality itself, is in fact mind-dependent to a significant degree."

The denial of framework independent synthetic a priori truths is the denial of what Kant argues for, and Rödl seconds, but this isn't my concern here. What is rather interesting, for my present purpose, is the assimilation of framework dependent synthetic a priori truths with the expressions of valid forms of material inferences.

Valid forms of material inferences are inferences that are warranted by the conceptual content of terms involved in the premises and conclusion, rather than being warranted by the (deductive) logical form of those statements alone. Hence, for instance, the truth that

(A) Montreal is north of New York City

can be validly (materially) inferred from the truth that

(B) New York City is south of Montreal.

This inference is materially rather than deductively valid. The inference is valid in virtue of a form of inference that can be made explicit with the use of the synthetic a priori statement:

(M) "X is south of Y if and only if Y is north of X".

This synthetic a priori statement is partially constitutive of the meanings (i.e. the conceptual contents) of the relational predicates "... is south of..." and "... is north of...". It may be worth noticing, also, that when this synthetic a priori statement is furnished as an additional premise, then one can logically (i.e. deductively) infer (B) from (A) and (M).

To be continued in my next post...

(Continued from my previous post)

Sellars wasn't alone in defending such a notion of synthetic a priori truths that aren't pure, in the Kantian sense, but rather must be relativized to specific conceptual schemes. Peter Strawson and Paul Grice, responding to Quine, in their In Defence of a Dogma, also defended the idea of synthetic a priori statements that are distinguished from other empirical judgments about particulars by making explicit necessary conceptual connections between the concepts used to make such judgments.

(At this early stage of the debate, and some 25 years before Kripke's Naming and Necessity was published, the analytic/synthetic distinction wasn't neatly demarcated from the necessary/contingent and a priori/a posteriori distinctions; and, indeed, Kripke's demarcations may have obscured some connections between those allegedly purely logical, modal and epistemological distinctions. But this would be a topic for another thread).

Wittgenstein's "grammatical remarks" also can be construed as the expression of synthetic a priori truths, in this sense. On that view, logical grammar, as opposed to deductive logic (first order predicate logic) aims at making explicit conceptual truths that are tacitly understood by language users, and compliance with which is exhibited in their practice, while also manifesting their understandings of the objects and concepts talked about.

This brings us back to Haugeland's claim about the constitutive inter-dependence between two levels of objective patterns. Inasmuch as patterns at both levels can be disclosed by means of empirical inquiry, they both are, indeed empirical. Patterns at the higher level, though, concern norms, laws and regularities. They are what Haugeland calls constitutive standards. Objects (i.e. patterns at the "lower level") can't be identified, or indeed, thought about, independently from the constitutive standards that they are subsumed under. One can't identify, or make sense of, a bishop, in the game of chess, that wouldn't be properly moved along diagonals on the chess board. Likewise, one can't identify, or make sense of an object's having a determinate mass that wouldn't weigh anything in the Earth gravitational field and/or wouldn't accelerate when subjected to a net force in accordance with Newton's second law.

Yet, it is one thing to recognize a chess piece to be a bishop, in a particular case, and another to acknowledge that, as such, it must move along diagonals (i.e. to acknowledge a constitutive rule of chess). Likewise, it is one thing to recognize that an object has a mass of one kilogram and another to recognize (or insist) that it must obey Newton's third law in all circumstances. The latter sort of necessary truths, expressed by synthetic a priori statements relativized to specific empirical domains, make up sets of constitutive standards for the domain. They collectively express an understanding of the objects that belong to this domain. Having such an (at least rudimentary) understanding is a requirement for those objects to be recognizable as what they are.

We may say, following both Haugeland (in Truth and Rule Following) and Sellars (in Concepts as Involving Laws) that understanding an empirical domain consists in having an (at least) tacit grasp of its constitutive standards. This understanding can be made explicit, and expressed with the use of synthetic a priori statements. Those statements are a priori because the objects that belong to the domain can't exist without obeying the standards. In fact, their obeying the constitutive standards (norms, laws of nature, etc.) is the specific manner in which objects exist in their respective domains.

This also brings us back to the distinction made by Kant, as quoted by Jamalrob in the beginning of my previous post, between knowledge that begins with experience and knowledge that arises from experience.

Laws of Newtonian mechanics, or the rules of chess (to stick with those two simple examples) can't be discovered a priori (in an epistemological sense). They either must be disclosed empirically (in the first case) or stipulated (in the second case). Once this has been done, however, the synthetic a priori (in a metaphysical sense) statements that express them determine what must necessarily occur in experience. Synthetic a priori statements therefore determine the range of possible (intelligible) experiences within a given empirical domain. Synthetic a posteriori statements express the content of actual experiences -- that is, experiences of actually existing objects that have specific determinations.

However, there can be no objective perception, and no objective judgment, when there is no prior understanding of what is possibly experienced (that is, what are possible and intelligible syntheses of various sensible determinations in a single object). Hence, knowledge of synthetic a priori truths begins with experience. This synthetic a priori knowledge makes possible objective judgments about actual objects; judgments, that is, that arise from experiences (synthetic a posteriori judgments).

Those two levels of empirical inquiry correspond to the two levels of co-constituted patterns discussed by Haugeland in Pattern and Being. The super-ordinate level of laws and/or norms is discernible as a set of consistent patterns that emerges from a wide and systematic range of experiences (experimentations, we may rather say), which delimits possible further experiences, while the sub-ordinate level is populated with patterns of unified determinations that make up the actual objects that figure in experience and that accord with the laws disclosed at the higher level. They are the two levels of understanding (meaningful thoughts) and knowledge (contentful thoughts), that can only be actualized jointly.

Finally started on this, and my first, quick impression is that Haugeland is trying to perform a phenomenological reduction of patterns, suspending the 'natural attitude' - which he refers to as the 'mathematical definition' as opposed to the 'operational definition'. No wonder this bloke wrote a book on Heidegger.

From Section 7:

Obviously, what keeps recognition from being thus vacuous is its being beholden somehow to what is ostensibly being recognized, yet in such a way that the criteria of correctness are induced from above. — Haugeland

So I can't plan out a chess match in my head, and then say that, every time my phone rings, another move in that match has taken place, and then say that my phone is playing chess against itself, because my supposed pattern recognition here is not at all beholden to the phone; the game would proceed the same way if I applied this method to anything else that happens repeatedly. A little further on in the same section, we have this:

[actual cases of pattern recognition are] the concrete way in which recognition holds itself to its object.

Anyone else smell Wittgenstein here?

...there is a way of grasping a rule which is not an interpretation, but which is exhibited in what we call "obeying the rule" and "going against it" in actual cases." — Wittgenstein

Haugeland is saying, I think, that a pattern exists when there is a set of rules that govern what something should do, and how we're supposed to respond when it does (or doesn't) do what it's supposed to do. We follow these rules by responding the way we're supposed to.

The whole thing is a bit fishy, though. I get the same vague sense from this paper as I do from Dennett, Heidegger, and, to a lesser extent, Wittgenstein - namely, the sense that I've been hoodwinked somehow. It's very subtle, but there's a gap in it somewhere.

I think that the sleight-of-hand here is in the whole assumption of pragmatism - namely, the idea that the ultimate test of something's validity is whether or not it's useful. It seems to underlie everything in this paper (and most of Dennett's work), but I never see it satisfactorily defended. I can't point exactly to where the chink in the armor is here, not yet anyway - I will have to study this some more.

Haugeland is saying, I think, that a pattern exists when there is a set of rules that govern what something should do, and how we're supposed to respond when it does (or doesn't) do what it's supposed to do. We follow these rules by responding the way we're supposed to. — Pneumenon

There is one (possibly two) distinction that Haugeland makes and that you possibly missed, and another one that is merely implicit in Haugeland's discussion, and that he makes more explicitly in a subsequent paper: (Truth and Rule Following).

The first (explicit) distinction to be made distinguishes two levels of patterns exemplified, respectively, in the constitutive standards that objects obey, and, secondly, in the objects themselves. For instance, to recognize something as a bishop in the game of chess is to recognize it as the object that it is, and to recognize that bishops, in the general case, only move along diagonals, corresponds to the recognition of (or insistence on, or institution of) a higher level pattern -- i.e. the existence of a constitutive standard.

The merely implicit distinction alluded to above is that between two sorts or "rule-following". When the objects (lower level patterns) obey the objective standards, they -- the objects themselves -- are "behaving" in accordance with rules, we may say. But our recognizing them the be the objects that they are, on the one hand, and our recognizing that they accord with the constitutive standards, on the other hand, both depend on our correctly exercising perceptual abilities, and hence, our correctly following rules. (More comments on that later, in a followup post).

Those distinctions become apparent when (and make it possible that) the objects that we perceive appear not to behave in accordance with the constitutive standards of the domains that they belong to. Those distinctions are reflected in the different ways in which we can be mistaken and the different ways in which, accordingly, we attempt to correct those mistakes. One source of error consists in our having misidentified an object (e.g. we thought it was a bishop when, in actuality, it was a rook). Another source of error occurs when we are mistaken about some constitutive standard. The second case, however, splits into two sub-cases. When an object has been correctly identified (as a bishop, say) and then seems to have moved in a way that doesn't accord with a constitutive standard, this could be because we have incorrectly judged it (in experience) to violate this constitutive standard, or we are mistaken about this standard being empirically valid. (For the standard to be empirically valid, here, isn't meant to be inconsistent with its being brought to bear to experience a priori. For more on that, see my two posts above about synthetic a priori propositions.)

This last distinction can be illustrated with numerous cases of apparent falsification of scientific theories when, historically, the theory was actually, and correctly, saved from falsification through the adjonction of an auxiliary hypothesis. Oftentimes, the ad hoc (so called) hypothesis was later empirically confirmed independently. One classical example is that of Uranus that was seen not to precisely follow the orbit prescribed by Newton's laws of motion and gravitation. Inasmuch as Le Verrier was justified to postulate the existence of another planet that had precisely the mass and orbit of Neptune, in order to explain this anomaly, then the apparent discrepancy exemplified an error in our bringing to bear the constitutive standards of Newtonian mechanics to the behavior of Uranus. But it sometimes also turns out (as it later came to be the case with the anomaly in the orbit of Mercury) that the standards themselves have to be adjusted or relinquished.

This brings me back to your claim quoted above regarding what it is that the rules say regarding what we are supposed to do when objects don't behave (or don't appear to behave) in the way they should. It is quite important to Haugeland that no rule dictates what it is that we should do in such circumstances. That is a matter of adjusting our knowledge of (objects in) the world with our understanding of (the constitutive standards of) the world. The adjustment is mutual, and, in some cases, the former may be at fault, and in other cases the latter may be at fault. But there is no general rule for deciding which. There only possibly could exist such a rule if empiricism were true, and the objects (lower level "patterns") could be singled out in experience independently of the standards that govern their behaviors. But Haugeland's insistance that the higher level patterns always are disclosed as constitutive standards for the objects that can populate some empirical domain precludes the very possibility of such an independent identification of objects as correlates of "raw" experience.

I think that the sleight-of-hand here is in the whole assumption of pragmatism - namely, the idea that the ultimate test of something's validity is whether or not it's useful. — Pneumenon

I think this overlooks an important difference. In another related paper, Haugeland explicitly disowns the label of "pragmatism". His account is meant to vindicate the notions of truth and objectivity in opposition with the pragmatism of Rorty that seeks to dispense entirely with both. But, at the same time, philosophers like Putnam and David Wiggins have endorsed pragmatism in a way that is consistent with Haugeland's efforts. His account also is sharply contrasted with Quine's pragmatic/holistic empiricism. Brandom and Davidson also want to recover objectivity within their own inferentialist and coherentist accounts, respectively, but while the latter struggles with the concept of empirical experience, the former seeks to dispense with it entirely. Haugeland offers a account of experience as the exercise of recognitional abilities that are constituted as such inseparably from the (understanding of) the constitutive standards that govern the objets thus recognized and that make them intelligible.

(Above edited to switch "former" and "latter".)

Late reply to you, but I do want to mention that this:

Haugeland offers a account of experience as the exercise of recognitional abilities that are constituted as such inseparably from the (understanding of) the constitutive standards that govern the objets thus recognized and that make them intelligible.

This smells very strongly of John McDowell, whom I understand to be wearing a fine misting of eau de Kant.

This smells very strongly of John McDowell, whom I understand to be wearing a fine misting of eau de Kant. — Pneumenon

That is, of course, no coincidence since Haugeland was McDowell's colleague at Pittsburgh and he offered two graduate courses on the philosophy of John McDowell (one of them specifically on Mind and World). It also happens that John McDowell is my favorite analytic philosopher.