thoughts

Quasi Realism = Fictionalism?

Tue, 13 May 2008 23:26:37 +0100

A central problem in contemporary meta-ethics is the debate surrounding the object of ethical discourse. If we are realists, what in the world gives us reference for ethical truth? If we are not, then what is the notion of ethical truth to us? Moral facts are not visible to us like physical facts, and thus the moral realist has a grave problem to address: that of the location of ethical facts and truth-makers. Therefore, it may seem tempting to adopt an anti-realist stance with regard to moral facts. One such stance is expressivism, e.g. the work of Ayer. This position claims that the propositions of moral discourse are not descriptions of the state of affairs, but rather expression of attitude with regard to states of affairs, and thus are not truth-apt. When I say ‘murder is wrong’, I am just expressing my disdain for murder, and the core semantic role of this sentence is only to do that, rather than be true of false. This position offers us an account of moral discourse in which we do not need to worry about ontology, thus casting away one of the larger problems for the moral realist.

However, this position has its own problems. One such problem is the Frege-Geach problem:
(1) If lying is wrong, getting your little brother to lie is wrong.
(2) Lying is wrong.
Therefore, getting your little brother to lie is wrong.
According to the expressivist, (2) has no truth-value. However, the antecedent of one is unasserted, and is part of a conditional. Thus to obtain the conclusions, ‘lying is wrong’ must be truth-apt in (1). ‘Lying is wrong’ therefore has a different semantic role in (1) and (2). Equivocation fallacy: expressivism is wrong.

Blackburn proposes a revision/expansion of expressivism in an attempt to preserve the benefits of expressivism, while reconciling it with our logical practises. Do do so, we must restore the surface logical form, in order to explain how such instances of modus ponens do not threaten the expressivist’s position. He calls this position quasi-realism. It is a combination of Humean projectivism with expressivism, stating that when we say ‘murder is wrong’, we are projecting our attitude onto the world, i.e. acting as if something in the world made our attitude such. Thus we may postulate a ‘logic of attitudes’ allowing us to express modus ponens and other instances of logical reasoning as if our expressions were truth-apt, while preserving the expressivist core.

The Frege-Geach modus ponens can be re-formulated in Blackburn’s expressivist language Eex:
(1) H!(|B!(Lying);B!(Little brother lying)|)
(2) B!(Lying)
Therefore B!(Little brother lying)
Here H! and B! are the approval and disapproval operators, and |...;...| allows for second-order approval/disapproval of a series of attitudes. So here, to express approval of your little brother’s lies (negation of conclusion) while disapproving of lying (2) conflicts with the approval of joint disapprovals (expressed in 1). Blackburn calls this ‘fractured sensibilities’. We thus get something which works ‘like’ truth-apt logic, and thus reconciles our expressivist discourse with its surface logical form, as if it had truth apt-ness ‘all the way down’.

Lewis claims that Blackburn’s quasi-realism is, in fact, a form of fictionalism. A form of error-theory---the cognitivist view that the propositions of a particular discourse are systematically false, fictionalism can be summed up as a philosophical doctrine stating that some areas of discourse are truth-apt, but the our motivation for these discourses is not their truth-possibility in the actual world, but rather the practical impact of acting as if they were true. Quasi-realism, for Lewis, seeks to avoid the ‘fundamental error’ moral realists commit in stating that there are moral properties in the world and that we form moral judgement by examination of their instances, by its adoption of projectivism, which Lewis takes to be the same ‘as if’ involved in fictionalism.

Lewis then defends the merits of such an error-theory in morality. He argues that, while there are fundamental errors underlying moral discourse, the discourse can be corrected so as to leave ‘’most of the working of the theory intact’’. This is precisely the job of moral fictionalism: we talk as if moral realism is true, although it isn’t. We therefore get all the benefits of moral discourse without having to worry about the associated ontology. Our only concern is such fictionalism provide us with some practical result, but then again if there wasn’t a practical reason for preserving moral discourse as is, we wouldn’t need to consider moral fictionalism.

Blackburn disagrees with Lewis’s interpretation of quasi-realism. Quasi-realism, while adopting projectivism, is not an ‘as-if’ position in that it does not accept, at its lowest level, that moral propositions could be true (but really aren’t). Instead, it seeks to draw a ‘’functional distinction between belief and attitude’’, and to allow us to talk as if the propositions of moral discourse had truth-value. At its core, it still remains a non-cognitivist theory, in contrast with the cognitive error-theory underlying fictionalism. In short, while fictionalism seeks to preserve moral discourse despite a factual error, quasi-realism seeks to preserve moral discourse despite a methodological error. The former claims that the objects of moral discourse aren’t really there, but we should act as if they are; the latter claims that the expressions of moral discourse aren’t truth-apt, but we should act as if they are. Both theories share the same pragmatic goal, but remain fundamentally different in their evaluation of moral discourse’s fundamental error.

If the broad claim that there is something wrong with moral realism is accepted, we find ourself with a choice to make between two theories which offer to supplant it while preserving moral discourse as is. Which one to choose? The advantages of fictionalism are fairly clear. It does not require defense against the Frege-Geach problem, as it de facto accounts for the surface (and other) logical form of moral discourse, and all that matters for its adoption is that:
1. Moral realism is wrong
2. There could be moral facts, i.e. there is some possible world in which some moral propositions are true
3. Conserving moral discourse as is gives us some benefit over full rejection
Quasi-realism accepts (1) and (3), but rejects (2). For us to choose quasi-realism, we must show that even the possible truth of moral propositions is absurd, while to choose fictionalism, we must show that there is a problem with the quasi-realist’s rejection.

Off the bat, I find it hard to dismiss fictionalism in a way the quasi-realist needs to. I can imagine a world in which a certain class of acts G creates a positive hormonal reaction in all observers of my species, and another, B, creates a feeling of physical revulsion, because of some physical property. It is plausible to ascribe the term ‘moral property’ to the physical property which dictates our attitude to these things, and we can therefore use moral discourse properly in such a world.

On the other hand, quasi-realism may have more of a problem. The quasi-realist seems to have successfully accounted for the surface logical form of our moral discourse, but the mechanics underneath the surface remain purely expressivist. This is to say the substantial meaning of the expressions does no follow from the descriptive component of discourse, but only from the expression of an attitude. Blackburn has shown us how, in an expressivist language Eex we can perform an operation that behaves like modus ponens, through use of second-order approval of joint attitudes. But unlike true logic, where we take there to be some normativity underlying correct inference, and identifying incorrect inference, what serves as a norm here? Hale argues that this is where quasi-realism fails, for to explain why the conclusion B!(Little brother lying) should follow from H!(|B!(Lying);B!(Little brother lying)|) and B!(Lying), i.e. to explain why it is wrong to, from these premises, conclude H!(Little brother lying), will require the sort of truth-dependent normativity used in truth-apt logic, at some level. In admitting this, the quasi-realist alienates not the logicians, but the expressivists he seeks to save: the quasi-realist cannot have his expressivist cake and eat it logically.

So I’ve discussed how both fictionalism and quasi-realism are tempting solutions to the ontological problems faced by moral realists, we have discussed the differences between both theories, and the grounds on which to choose one or the other. At this stage, I see no clear objections that the quasi-realist could present to the fictionalist, while on the other hand, quasi-realism arguably has difficulties with preserving its expressivist core. Perhaps the quasi-realist has a sophisticated response to Hale’s objections, and perhaps moral realism is, after all tenable; but as things stand, if we admit that moral realism is in trouble, fictionalism seems like a likely candidate for preserving our moral discourse, if there is any practical reason to do so.

On Poincaré's Petitio Argument

Fri, 2 May 2008 17:04:04 +0100

Over the course of the last month, I have begun work on what will be my last bit of strictly philosophical work for at least another year, if not the last piece of philosophical work I produce while studying in a philosophy department. The issue at hand is the relation between logic and mathematics–and, to a certain extent, the case for logical pluralism.

Poincaré presented a two-part thesis at the turn of the 20th century. The negative component consisted in a series of arguments against foundationalism and mathematics, with the goal of rejecting views supporting that mathematics is an an analytic apriori subject matter. The positive component goes one step further, in claiming that arithmetic has epistemic priority over logic, and that, in fact, logic is simply a part of mathematics that has been re-carved and re-named by logicians.

In a recent paper written for the department, I have argued that his negative thesis can, with some (non-trivial) revision, be stood up against critic’s objections and cause worry for certain types of foundationalism (namely Fregean logicism and its neo-logicist successor), while not affecting others (notably Dedekind & Peano-type axiomatic foundationalism). His positive thesis, on the other hand, can be dismissed by turning his own petitio argument ‘on its head’ and applying to Poincaré’s own view. The result, I claim, is that there is no intrinsic connection between mathematics and logic, and that the adoption of logical pluralism dispels some of the main worries put forth by my revised petitio argument.

In my dissertation, I plan to expand this conclusion and examine possible responses the logical monist might give, as well as possible counter-arguments the neo-logicist might provide against my revised petitio argument. I also wish to investigate the link between Poincaré’s view and modern theses of what Stewart Shapiro calls the “mathematics first” school of thought, and attempt to determine whether the success of the revised petitio argument would support this approach to philosophy of mathematics, or whether or failure of the revised petitio argument would weaken its foundations.

Results of this work to be announced in September.

Against a "proof" of the existence of god

Sun, 28 Oct 2007 11:31:28 +0000

Normally, I don’t go out of my way to present arguments against evangelical theses, especially those which present themselves as “logical”. If a mathematician were to spend his time pointing out to every fool who bought the follow joke proof that 1 = -1 that is wasn’t so, he wouldn’t get much done.

(Deductions: 6 from 5 & 4, 7 from 6 & 1, 8 from 7 & 2, 10 from 9 & 3)
The error is obviously with (1), which only holds for positive a and b, resulting in the falsity of the step from (6) to (7).

Although hopefully, one by one, the rogues would shift their position, the mathematician would be doing nothing more with his time than counter-propaganda. If it weren’t for the fact that people actually vote and make ethical decisions based on the type of error presented in the above video, it would be very tempting to go the way of the mathematician, and – if after argument they refused to at least act towards the world and ontology with a certain prudence and modesty – leave them to their affairs.

After a few years of talking to evangelicals and dealing with this sort of rubbish conclusion seeking overconfident from of reasoning, I usually try to steer clear of any such polemic. However, after watching the above video, I was so stricken with the immodest attitude this Sean guy presented his argument with I simply couldn’t resist the temptation to defer my reading of Frege for another hour and write up a response (which I then taped up, so I could post it on Youtube). The video and text of the response are below...

Download video response (.mov)

Normally, I don't bother with religion much. I think people have the right to be irrational, or – in more generous terms – to fill explanatory gaps with whatever makes them comfortable, as long as they don't derive a wide cosmological view from it, or allow those components of their epistemology to take precedence over evidential components.

But here, I have to step in. On one hand, I really do appreciate this Sean fellow's attempt to approach the question of the existence of God in an argumentative and – dare I say – logical way. On the other hand, I'm not sure I can really endorse his rather immodest conclusion that he's not only come up with a more straightforward and simple answer to this historically dominant question than some of the greatest theologians and philosophers of the past (then again, this is not to say that such an event is impossible), but also that his arguments are bullet-proof. Unfortunately for him, it's already quite full of holes, as I will now demonstrate.

Sean contends that it is impossible that there could be an infinite amount of time before us. He attempts a reductio ad absurdem by stating that "If at a time t, I were to say x would happen at t + INFINITY, then x would never happen". This is correct. He then applies modus tollens of the consequent of this conditional, to obtain what he thinks is the negation of the antecedent, namely that "there is no infinite time before us (as demonstrated by our present existence and the previous conditional)". The conclusion of the modus tollens is in fact "it is not the case that at time t, x would happen at t + INFINITY". What does this say? Only that it is impossible that someone could have created the universe at a time NOW - INFINITY. Well, where does that leave us...
Either:
a) there was no beginning of the universe, in which case the could have been an infinite amount of time before us. This sort of idea, for example, can be found in Einstein's cyclical model hypothesis, or more recently, in the work of Neil Turok, from the University of Cambridge.
OR
b) the universe came into existence an infinite amount of time ago.

Well, the game's already up, since Sean has to assume B to continue his argument. This, of course, is a nice example of question begging since his assumption will then be contained in the conclusion (namely that an intelligent designer created the universe a finite amount of time ago). So for the rest of his argument to even start to make sense, he needs to reject a pretty big hypothesis of physics.

Let's be generous, and imagine Sean manages to do this. Does his argument still make sense? Well first of all, our second option only states that the universe started some finite amount of time ago. He also correctly points out that according to modern physics matter cannot be created or destroyed. Sean wishes to reject this piece of physics with more physics: from general relativity, he then reminds the viewer that time and space are intricately linked, and that space and therefore time do not exist without matter. At this point, I'm rather impressed, and waiting for the other shoe to drop. It does so when Sean calmly explains that therefore, since everything started a finite amount of time ago, linearly speaking, then therefore there was a time before that when matter didn't exist, therefore before space and thus... time... existed... err, wait, so there's time before time? Well, there are so many things Sean is doing wrong here that it could take a while to go through them. First of all, his talk of existence of anything before the existence of the universe, which he must commit to since he's discussing a causal agent that precedes space-time, requires a notion of absolute time. Of course, the whole basis of his statement that matter "wasn't always there" was based on the Einsteinian notion of space time, which itself postulates a relative, non-absolute time. So to have his cake and eat it, Sean is both relying upon an absolute notion of time to say that could be a creator, and a relative notion of time to explain why there must have been a creator. This is, of course, a direct inconsistency.

Okay, let's now be extra generous, and overlook that previous argument, and state that there was a creator, or creative principle. Sean takes the current complex state of existence to indicate that the creator must be highly intelligent, since everything has such complexity. Well, although it's not like everything popped into place as is, let's approach this question with patience. I don't wish to rule out the possibility that there is an intelligent creator, just like I don't want to rule out the possibility that I'll get a full PhD scholarship, even though both seem equally unlikely given the evidence I have (what does that say about the AHRC?), but I do want to quickly demonstrate the lack of necessity of this claim. Let's imagine I have a cup of six dice, and tip it over, and something fairly improbable happens: all the dice roughly fall into a line, with the up-side ordered from 1 to 6. Let's imagine now I'm carrying a tray with a dozen of these cups of dice, and trip and fall, and the same thing happens where each cup has fallen. I'm now amazed... the probability of such an event happening is low, but non-zero. Must I conclude that the cup, or something else, has intelligently designed this outcome? The non-zero probability guarantees the physical possibility of such an occurrence amongst a series of random events involving dice and cups, so while I have every right to be surprised, I would probably not start worshipping the cups.

So let's go through Sean's failings once more:
First, he has faultily argues that there must be a beginning of time. In fact, he assumes this from his misinterpretation of a modus tollens, and therefore is guilty of question begging, or vicious circularity.

Second, if we allow him to overlook the first objection, he uses a conjunction of contradictory theories to demonstrate that the universe can have a creator, and that it must have a creator. This conclusion only stems from the conjunction, which – being contradictory – doesn't make much sense. So his attempt to use physics to argue against physics fails too.

Third, if we allow him to overlook the first two objections, he still fumbles through a poor argument for intuition for the existence of a necessarily intelligent creator.

To sum up, because I don't think the first objection is damning for the possibility of a creative principle (although it is damning for Sean's argument), and fortunately one does not need to accept Sean's ludicrous second argument for one to still believe in a creative principle, it is still reasonable to believe (or perhaps observe) a god like Spinoza's God (although perhaps we can drop the loaded terminology and just call it Nature or Existence). However, even if someone were to argue the case for this a bit better than Sean, this God still does not correspond in any way to the personified God in the Bible. Sean's grasping at straws to throw intelligence somewhere in there is a pretty good pointer at the weakness of arguments for this, and only after one has considered that the whole set-up for the speculation he offers at the end fails anyway.

Indexicals and object-oriented programming I

Mon, 22 Oct 2007 20:20:31 +0100

Indexicals offer an interesting problem, both in language, and in discussion of subjectivity and objectivity. These problems are discussed – amongst others – in Perry (1979), Stalnaker (1981), who mostly talk about the linguistic aspects, and Nagel (1979), who mostly talks about the issue of subjective-objective aspects. I'm going to focus mostly on the latter sort of aspect in this post. The issue is as such: indexicals (e.g. "this", "now", and especially "I") provide us with a particular sort of knowledge that the same phrase, with a coreferring expression swapped in for the indexical, does not. For example, while "I am sitting here right now" will provide you with the exact same information as "Edward Grefenstette is sitting at a table in St Andrews at 8.17pm on October 27th 2007", provided you know my name, the time and where I am writing this post from, it is a different matter altogether for me.

Substituting "Edward Grefenstette" for "I" will not tell me that it is me who is sitting at this table writing this post, unless is know that I am the one sitting here. This, obviously, requires further use of an indexical. So is there a way to analyse the meaning of I without using an indexical? Can I talk about this “I” fellow in a way that fits into the objective world, and that isn’t intrinsically private (or at least, a component of it)? It seems, since all explanation of why beliefs involving “I” rather than plainly “Edward Grefenstette” (that is, without the conjoined belief that “I am Edward Grefenstette”) will explicitly require this indexical link between “I” and “Edward Grefenstette” (which may, however, be implicit in conversation), that there is indeed something special about the indexical which is subjective and irreducible to objective statements.

Naturally, there is an objective aspect to such indexicals, since I can understand John when he says say “I am John Doe”. I know that here “I” is being defined as a linguistic shortcut for “John Doe”. After I have acquired this knowledge, should John Doe tell me “I am hungry”, I should be allowed to rephrase this as “John Doe is hungry” without problem (note, however, that for John Doe the sentence would carry the same special weight as the use of indexicals referring to myself in previous examples).

Nagel (1979, chapter 4) uses the problem of indexicals as an illustration of – and in support of – his thesis that while the subjective world is part of the objective world and can be integrated into it, there are invariably cases where tension exists and such integration is not possible. Such links between indexicals and beliefs are a good example of this. If I cannot explain the effect an “I” has on my behaviour that “Edward Grefenstette” without the indexical link has, and if I cannot explain the indexical “I” (or any other) without an ad infinitum recursion of definitions involving indexicals, than indeed there is some aspect of subjectivity which is not fully integrable into the objective world, as demonstrated by our linguistic practises.

Perhaps Nagel is completely right here. It’s a bit hard to tell, since – true to the style of the previous chapters of his book – he does little more than wave his hand at vague intuitions and ill-defined problematic cases (although some may contend that there is nothing wrong with a “bit” of lack of clarity when dealing difficult conceptual problems – I’d be inclined to disagree on account of there being a difference between being unable to be clearer, in which case one at best presents a paper and works on the issue some more, and being unwilling to be clearer, in which case either one restrains oneself from publishing a book and hoping that one’s authority in the domain will cause readers to buy positions on face value rather on the basis of argumentative support, or one publishes the aforementioned book and does not complain about the incurred wrath of students forced to study the said book within the context of their masters degree). Nonetheless, this is not reason enough, or reason at all (as James Harris has rightfully pointed out to me in conversation) to reject Nagel’s position (should we generously choose attribute such a qualifier to the underlying content of this chapter).

However, before accepting this, I wish to do some hand-waving of my own in the opposite direction. The following is principally based on intuition in that it does not go “all the way” (this is, after all, possibly work in progress). I will, however, try to be as clear as possible in my argumentation. I wish to examine the self-reference indexical in analogy with how classes work in object-oriented programming (OOP). My experience of OOP is rather limited, so I apologise if I butcher my explanation, but I’ll hopefully get the bits which are relevant to my “argument” right.

In OOP, “objects” (e.g. an alien in space invaders) belong to a class (e.g. aliens in space invaders). The class, rather than a way of describing or classifying the aliens you see on screen, as the name might indicate, instead plays a generative role in that it serves as guidelines for the construction of new objects of the class (both in the comp-sci sense, and in the logical sense). It specifies attributes, variables, functions, etc... that the objects of that class will possess. Let us focus on variables. There are two kinds of variables: private and public. A public variable (the alien’s remaining hit-points before it is destroyed) is publicly available to the program, so that it may read it (and indicate the remaining hit-points on a player’s screen) or write over it (so that a function from the main block of the program may change the alien’s hit-points when it is struck by the player’s laser). A private variable, on the other hand, is protected from being read or written to by anything other than the object itself (e.g. the decision the alien “wants” to move in. Whether or not this is possible can be determined by something outside of the object, but the decision of which direction to attempt movement in is, in some sense, the alien’s own). Of course, private variables can be, with the aid of appropriate functions within the class of the object (and thus within the object itself), communicated to the external program (the alien object could have a “tell(x)” function that would return the value of one of its internal private variables x – which it can read) or even written to (using a similar internal write(x) function), although providing both methods (especially the latter) might defeat the whole purpose of having a public/private distinction.

This raises the question “what exactly is the purpose of such a distinction between private and public object variables?”. Hopefully quickly explaining this will dodge an important philosophical land mine, in addition to clarify the basis of my analogy a little further. “Private” here does not mean the same thing as “private” as in “private language”, or as in “externally inaccessible”. Invariably, if sufficient methods are provided by the object (ie. internal functions relaying the content of private variables) or by the environment (since the computer memory is physically accessible). The motivation between the drawing of this distinction is to offer the object enough protection so that it is “handled” in a correct (ie. predictable) way. In other words, it is a way of anchoring an essentially practical distinction between two sorts of data within a mechanically operational framework: on one hand you have public data, which is useful to others, and on the other hand you have private data, which is of limited or no use to others, and for which external intrusion into or interference with its internal mode of operation would cause confusion, unpredictability or simply produce meaningless behaviour on the part of the object.

One more thing I’d like to introduce, although it may well be sufficiently implied by the above, is that objects can have private functions as well. These are functions that seek to only manipulate data “fed” to the object and internal data, rather than communicate something to the external world, or to be used by the external world within the context of “handling” the object.

What I’d like to investigate, and where I’ll end today’s writing on the matter (since I have other things to deal with) is with the following intuition. Perhaps we could deal with our use of self-reference indexicals in a way which is similar to which an object OBJ would deal with a private constant (in programming terms, an unchanging variable, as its name indicates) named “SELF”. More than a variable, it could perhaps be a set of names or definite descriptions (if you choose to make a distinction), or perhaps some other elements (representations of the physical token we call ourselves, concepts about perspective, etc). And then there would be a function SELFREFERENCE(x), which would return a boolean “true” value if the evaluated element “x” corresponded to the contents of the object’s own SELF variable, and false if it did not. SELFREFERENCE could therefore be used in various other internal functions of the object, both private and public, as part of conditionals which could, down the functional chain of causation, cause external changes of behaviour or internal changes of “belief” (or both) based on whether or not the object in a particular proposition OBJ is dealing with qualifies as being equal to SELF (and thus consequently referring to OBJ). The beauty of separating elements and functions involved in self-reference and self-identification (in OBJ’s case, at least) like so is that it is perfectly compatible with a more general type of function we could call REFERENCESUB which would sub indexicals for non-indexical coreferring expressions, so as to account for OBJ’s ability to understand other people’s more trivial self-reference (ie. use of the indexical as a linguistic shortcut), while still maintaining the dual usage of self-reference in conditions where OBJ is the entity using self-reference, rather than listening to another entity self-refer, in which case REFERENCESUB does its usual job, but – presuming the designer has included special conditions using SELFREFERENCE in REFERENCESUB for cases where the self-reference indexical also denotes SELF – REFERENCESUB also could account for our change of behaviour and/or belief in cases where OBJ refers to itself.

My intuition, therefore, is that we are similar to OBJ, with two principle and related differences (which, as you will see, could be false without harming my intuition). The first, is that we do not have any such explicitly sharply defined functions. This, I believe, is because while our mind could be argued to be some sort of computer, it does not seem to operate in such a clean, sequential way. The second is the rough equivalent of these functions is not placed in our minds by a designer, but rather generated through the process of acquiring and developing our linguistic abilities. This, I believe, is because our linguistic abilities and our more complex cognitive mechanisms are intimately related. Someone (sympathetic to Chomsky or Pinker’s way of seeing things) may be tempted to reject both these differences between humans and OBJ, and may have good arguments for their rejection. I am not learned enough in philosophy of mind to respond to these immediately, and such a response would be beyond the scope of the exposition. However, in claiming this, I do not feel I am guilty of the same sort of academic crime that which I criticised in the beginning of this post, as such a rejection of these differences would not, as I have mentioned, be counter to my intuition, but rather to the broader theory of mind which I wish to integrate this intuition into. Returning to the actual topic of discussion, I conclude my exposition by stating that it is possible, under further refining of this intuition and its implications, that we may be able to demonstrate how the way we acquire the linguistic notion of self-reference and its proper application, both as a linguistic shortcut and as a belief-modifier when applied to ourselves (rather than observed) may prove to be close, in terms of general form, to the way the SELFREFERENCE function works and is integrated into more general functions such as REFERENCESUB in OBJ’s case.

The above exposition may be fairly muddled at this stage. However because of the methodology underlying this approach – to give a sharper account of self-reference through analogy with formal systems – it, in return, opens itself to sharper criticism which may aid in demonstrating where error lies, or where further discussion is needed for further clarification. Here are a few questions I’ll be looking into:
Is this really addressing the core problem of (absence of) privacy in the causal relation between self-referent indexicals and belief/behaviour?
Is the vicious regress present in trying to analyse personal usage of indexicals like “I” really avoided, even in ‘ideal’ cases (ie. We operate exactly like OBJ)? This condition must be met for the intuition to be worth looking into in the first place, and I haven’t fully demonstrated that it is yet.
What supporting arguments in the literature can be found? I am interested in looking into how the (supposed) impossibility of Private Language would point towards the notion of self and mechanisms of self-reference similar to OBJ’s being developed when we learn language. So it will probably be worthwhile looking into seeing exactly how this intuition fits into specific theories of language.

Naturally, I am eager for your feedback on this matter, so please don’t hesitate to email me if you see anything wrong, weird or good. or if you have recommended reading.

The notion of self

Sat, 13 Oct 2007 10:35:57 +0100

The fourth chapter of Nagel’s (1979) book covers the notion of the objective self. It’s short, but quite frustrating, because the author writes a lot and doesn’t say much (I jest only partially... I am aware that Nagel’s reputation precedes him in matters of philosophy of law and politics, and his paper “What’s it like to be a bat” is a classic. However I remain thoroughly unimpressed with this book, but concede that it could be due to my lack of experience... or having read the works of far too many good academics. You know, the kind with actual arguments). Nagel asks what makes us ourselves and not someone else. He supports the view that there is something about self-reference that is unique, given particular the nature of indexicals, and asks how we come to know them, and how we could ground them and ourselves in objectivity.

I’m a bit confused as to what exactly the problem is. In typical Nagelian fashion, he dismisses this line of questioning, stating that (or at least implying that) those adopting it are ignoring the problem. However, he doesn’t explain why it is an unfair claim. In fact, one could respond that it is he who is ignoring the true nature of the problem, as there seems to be a conflation of the existential aspects of the self and the epistemic aspects of self-reference in the way he frames the problem.

There is no normative framework in which the notion of self can be discussed: I can perfectly well conceive of a state of the world where the token corresponding to what I now call myself could not having existed, and the rest of the world remain the same. My existence is purely contingent, a matter of fact. One could quickly object: “but you are simply making this claim relative to the particular token ‘Edward Grefenstette’! Your term ‘I’/’my’ could still refer to some other token”. Overlooking the possibility of implied substance dualism, the answer is fairly straightforward: one can conceive of a state of affairs in which no conscious beings exists, while the rest of the world remains the same, therefore my existence is still contingent.

On the other hand, my being myself is necessary. There are no two ways about it: asking myself “Could I have been someone else?” is precisely asking “Could the person that is me not be the person that is me?” Analysing the self-reference relation gives the logical structure of self-referential terms:

So answering the first question of this paragraph with “yes” or the second with “no” simply leads to a logical contradiction. To me, Nagel is making a big deal of a question which, for all I care, could be as interesting and intellectually enriching as asking:

Whoop-dee-doo for you, Thomas.

That said, the nature of indexicals and the issue of subjectivity in language (in particular with regard to its link to behaviour) is certainly an interesting one. I’m also reading through a nice paper by John Perry, entitled “The problem of the essential indexical” (1979), which is well written and presents the problem in a clear and structured way. A nice change from the above...

NB: The above position is bound to change sooner or later (probably the former) without my updating this post (that would be cheating...), so don’t come at me foaming in the mouth if I’ve completely missed the point of particular argument – I’ll hopefully have realised so myself. However, what is unlikely to change is my strong dislike for this book and its “methodology”, even if I come to agree with some of the positions found within its pages (whatever those might be...).