Objection to Berkeley's Master Argument



The Master Argument (roughly) states that it is not possible for sensible objects to exist without a mind. Now part of Berkeley's Argument goes as follows:
Suppose something exists without being conceived of by anyone, in thinking of such an such an object you are actually conceiving of it and that is a Contradiction with it not being conceived of.
Now my problem with this is that there is no way of formalizing this. Since any such formalization must include a premise Q(x) which states that x is inconceived of. Now for Berkeley's Statement to hold that "in thinking of such an such an object you are actually conceiving of it", there must be a proposition such as Q(x) => ~Q(x) as a premise,so he can then arrive at the contradiction, given the semantics of Q that is simply not true.
Now my question is this:
Whether the Argument I provided is correct and whether it's the same as Russell's objection, concerning mixing the Representation with the Representative.


Posted 2016-06-30T09:52:28.153

Reputation: 181

Do you have a reference for Russell's objection? – None – 2016-06-30T12:59:23.740



You seem to be wanting to argue that Berkeley's argument isn't even valid. I don't think that's right. Berkeley's point could be made formal as so:

  1. There is an object o such that nobody conceives of it. (Premise)
  2. If (1) is true, then o is such that somebody conceives it. (Premise)
  3. o is such that somebody conceives it. (1, 2 modus ponens).
  4. Therefore, o is such that nobody conceives it and somebody conceives it. (By conjunction introduction on 1, 4)
  5. Contradiction.

The argument given is valid--if the premises were true, the conclusion would be true. But since no contradictions are ever true, then we have to give up either (1) or (2). Giving up (1) is what Berkeley wants us to do. To force us on to this desperate path though Berkeley needs to give us reason to think (2) is true.

That's what I interpret him to be doing in the snatch of quoted dialogue. Presumably he'd have to say something like, "You have to recognize (2) is true, because just by reading and understanding (1) you have begun to think about the object o!" But that doesn't sound right, and Russell gives the reason why: It confuses the properties of our mental representation of an object for properties of the object so represented. Russell's theory of knowledge by description (as opposed to acquaintance) in Problems of Philosophy pp. 42-45 seems to explain how this works.

It is like Berkeley is saying that just by reading premise (1) I have direct acquaintance with o, and in that sense if I do have acquaintance with it, I am able to "conceive" of it, perhaps, as (2) seems to say. However, if we take (1) to give me merely knowledge by description instead, then (2) looks clearly false since I can know of the existence of o by description without being myself acquainted with it, which is what (2) required.


Posted 2016-06-30T09:52:28.153


2As presented, 2 is not a premise, it is a deduction from the (to my mind incorrect) definition of conceiving of something. – None – 2016-07-01T02:49:06.413

2@jobermark, I don't follow. What's the definition at stake and how does it differ from what is expressed by (2)? – None – 2016-07-01T11:54:31.093

I think in 4 you mean 'on 1,3' – Eliran – 2016-07-01T13:02:11.977

He does not presuppose that someone conceives of 'o', he claims it is inherent in the meaning of 'conceives of'. Applying a definition to a case is not a premise, it is a 'cut' where an ongoing thread of argument from elsewhere (the discourse providing the definition) enters your argument and allows a deduction. He uses a lemma "what is thought of has already been conceived of" consequent to the unstated definition of conceives of". I contend that that definition is malformed, in that he sees the term as an idempotent reference (i.e. f(f(x)) = f(x) for all x) and it is not one. – None – 2016-07-01T13:50:09.993

You are claiming the argument is valid but not sound. By more modern standards it simply is not valid. – None – 2016-07-01T13:54:48.623

@EliranH Yep, good catch. That's right, (4) follows from (1) and (3). – None – 2016-07-01T13:56:40.933

2@jobermark If (1)-(5) actually are B's argument, then B's argument is provably valid, since the conclusion follows from the premises via the truth-functional laws of first-order logic. As far as the logic is concerned, you can make anything a premise you want to be; the logic constrains your conclusions, not your choice of premises. – None – 2016-07-01T13:58:14.417


  • is in no way equivalent to the statement "Having thought of x, you have already conceived of it" first of all, it lacks reference to thought or reference. He is not assuming that statement, he is observing something like "Any variable in a proof, including this proof, is a reference." And assuming "anything that has a reference is conceived of, by defintion of 'conceived of'", and combining those into a deduction. So 1-5 is not his argument. It omits the step you then go on to analyze.
  • < – None – 2016-07-01T14:04:10.323

    The problem is with the definition of 'conceived of', which does not enter into your supposed expression of the argument (and which he does not state.) – None – 2016-07-01T14:08:30.563

    If you allow the technique you used here to mean something, then you can move any argument into validity by excising the invalid part out into another argument and framing it as a premise. So the distinction between sound and valid loses force. You need the formalization whose validity you are checking to include all the relevant parts of the argument. – None – 2016-07-01T16:47:41.320


    Berkeley's argument might be formalized as follows:

    • Ax = x is an assumption
    • Cx = x is conceived of
    {1}      1.   ∀x[Ax → Cx]                  Prem.
    {2}      2.   ~Ca                          Assum.
    {3}      3.   Aa                           Assum.
    {1}      4.   Aa → Ca                      1 UE
    {1,3}    5.   Ca                           3,4 MP
    {1,2,3}  6.   Ca & ~Ca                     2,5 &I
    {1,3}    7.   Ca                           2,6 RAA
    {1,3}    8.   ∀x[Cx]                       7 UI - INVALID
    {1,3}    9.   ~~∀x[Cx]                     8 DNI - INVALID
    {1,3}    10.  ~Ǝx[~Cx]                     9 QI - INVALID

    First of all, line 3 involves a meta argument which I believe might be formalized in a better way than how I did it. Even so, I don't find it very controversial because it only asserts that the assumption of a on line 2 implies that a is an assumption. With that, I invoke the premise which I added that everything which is assumed is conceived of. Even though there might be a better way to formalize it, the main point is that it can be formalized. Even so, I believe his argument has weakness for other reasons.

    What I find most questionable is the application of a rule of inference in a way for which it wasn't intended. It's common in logic to assume a typical instance (as on line 2) in order to infer a general principle (as on line 8). However in this case the general principle that it being inferred involves the act itself of assuming. I believe it's clearly fallacious to conclude that the ability to make an assumption implies that all such assumptions are actually made; it only implies that they can be made. Unless it can be shown that all things are actually conceived of, the argument doesn't hold.

    It might be argued that this weakness only has to do with the particular way that I represented it, but Berkeley's argument is really asserting that very thing. He is not allowing for the possibility of something to exist without an actual instance of it being conceived. The potential to conceive is not the same as actually conceiving, so no valid inference can be drawn. For that reason, Berkeley's argument doesn't hold.


    The problem with the proof as I originally presented it is that line 3 should have been entered as an assumption and not as a consequence of line 2. (I've edited it to show the correction.) Because of that, a is a free variable, so the application of Universal Introduction is not valid:

    "If we are tempted to generalise universally on a formula containing a particular name we can always look back at previous lines of proof, checking carefully that there is no line which contains a formula telling us something special about that named individual, e.g. that it has a particular property or properties. In other words, we must always take care to ensure that the particular name we use in such a context refers to an element of the domain which is genuinely typical in the context of that use." (Paul Tomassi, Logic, p. 277)


    Posted 2016-06-30T09:52:28.153


    " (1):However in this case the general principle that it being inferred involves the act itself of assuming. (2) I believe it's clearly fallacious to conclude that the ability to make an assumption implies that all such assumptions are actually made." could you elaborate a little more on both of these statements, and how they affect the formalized proof (which parts they discredit etc.)? – user2268997 – 2016-06-30T15:56:37.160

    What I feel is wrong is wrong with this, is that line two should actually use an existential quantifier. Since you're using existential quantification it cannot be stated that Aa, since a is not actually assumed. Hence you cannot derive 3 from 2. – user2268997 – 2016-06-30T16:05:40.180

    I actually did have an existential quantifier but realized it was superfluous because it requires a second assumption to instantiate a, which is exactly what line 2 currently does. From any formula Fa, one can conclude Ǝx[Fx], so the presence of the quantifier doesn't change the fact that its existence is being assumed. – None – 2016-06-30T17:25:13.317

    What I meant with your points, (1) and (2), is that the assumption of something only implies that any such assumption can be made, but Berkeley needs to show that all sensible things are conceived. You can't infer an are from a can, so line 8 is an invalid generalization. – None – 2016-06-30T17:31:30.063

    The invalidity of 8 that you're pointing out, depends on the semantics of the Relations A and C. Now a Formal Proof given a set of premises is either correct or incorrect. Seeing that the proof you provided is sound, then either Berkeley's Argument is in fact correct(2 is refuted), or something must be wrong with the other premises(1 or 3?) – user2268997 – 2016-07-01T08:13:49.733

    I've edited my answer to show why it was invalid. – None – 2016-07-01T12:13:41.337


    I don't follow your exact objection. It seems like taken seriously, your objection would rule out proofs by contradiction altogether. From Q(x) -> ~Q(x), we do not deduce Q(x) is not a meaningful statement, only that the statement is not true. And that is exactly what Berkeley is after, for your Q not to be true of any x.

    But back off and look at examples. By Berkeley's argument, as you have presented it, the bicycle had been conceived of before fire was discovered. Once any caveman reached the concept that there were things of which he had not yet conceived, every one of those things was immediately conceived of, at least in the sense Berkeley seems to be using here. So this first human to doubt his own omniscience, at that very time, conceived of bicycles and microwave ovens? That is clearly nonsense.

    The problem is that conceiving of something is not transitive (or more formally 'idempotent'): conceiving of conceiving of something is not the same thing as conceiving of it. I can conceive of conceiving of the most important technological innovation for the next thousand years. But I cannot conceive of the innovation itself, or I would be able to start development on it right now. I would at least be able to tell you the basic idea. And I can't.

    Any attempt to formalize the argument is going to show the telescoping of two nested "failures to conceive" into one, and point up the flaw in the argument.

    An attempt at a formalism:

    C = 'concepts of'

    1. Choose y in {x: C(x) = {}}
    2. The reference to y in 'Choose y' in 1 is in C(y)

      (** This is erroneous)

    3. So C(y) and not C(y)
    4. So there is no y
    5. Therefore for no x is C(x) = {}

    Unfortunately the reference to y in 'Choose y' is not a concept of a given thing, it is a concept of a thing lacking a concept.

    This involves the concept of the thing having a concept, so we are close. But we can close the gap only if C(C(x)) = C(x). By some slippery English, you can make it seem like the concept of the concept of something implies there being a concept of it. But it is not so.

    It falls to the bicycles-and-cavemen example. The concept of that which I have not yet imagined does not provide the concept of a bicycle just because bicycles are something I have not yet imagined. The technical reason is that intensionally defined sets need not have extension (evaluation of an iterator can be lazy, so the constructors implicit in the enumeration may remain uncalled, for the functional programmers out there).


    Posted 2016-06-30T09:52:28.153


    What I'm Saying is that it presupposes the statement Q(x) => ~Q(x) , which given the semantics of Q, is not valid. – user2268997 – 2016-07-01T08:02:09.230

    I understand what the problem with his argument is intuitively, however I want to know what goes wrong when trying to formalize it. So could you demonstrate what your last statement would like? – user2268997 – 2016-07-01T08:04:18.540

    His argument does not presuppose this, it deduces it from a misunderstanding of the meaning of 'conceive of'. "By thinking of this you have already conceived of it" is not a premise, it is application of a malformed pair of definitions. – None – 2016-07-01T13:40:41.087

    It does presuppose it, since it is not inferred from any of the other statements. – user2268997 – 2016-07-03T13:23:17.840

    When you say "This involves the concept of the thing having a concept", doesn't this imply the thing has a concept(which has a concept of it's own) to begin with? – user2268997 – 2016-07-03T13:55:50.897

    And on the Definition of Concept: isn't the existence of y in C(y)? – user2268997 – 2016-07-03T14:27:44.457

    Is the whole 'bicycles' example meaningless to you? Do you not see that conceiving of "that of which I have not conceived" does not actually conceive of the individual things of which you have not conceived? So NO, the concept of the thing having a concept does not involve an actual concept of the thing itself. Otherwise, deal with the example and whether cavemen had the concept of a bicycle. – None – 2016-07-04T23:23:52.360

    If a programming analogy would help, in a lazy object-oriented language a function that would return a list of elements of a given class, were it called, does not imply the class exists. The class's definition may be compiled out of your code completely, if that function is never called, and it may never have existed to begin with. Having a name or a reference is not the same thing as being defined, especially if the name or reference would point at the 'null' version of the thing named or referred to. – None – 2016-07-04T23:39:12.210

    Informally and intuitively, it is (and was to begin with) quiet clear. Saying (thinking) there is something that no one has conceived of, doesn't say anything about it (it is still unthought of). However that's not the matter on question. The point is to pinpoint the exact mistake formally. And then try to generalize and apply it to other problems. – user2268997 – 2016-07-05T07:37:57.453

    And There is something wrong with caveman analogy. The Argument never asserts the existence of conceived things, only the non-existence of unconceived ones. – user2268997 – 2016-07-05T07:49:36.113

    The caveman argument has no bearing or dependence on the existence of the conceived things. So this objection makes no sense. You can replace 'bicycle' with 'warp drive' if you want to, so that the thing in question is a fiction. The argument has to do with the existence of the concept of those things only. The point is that they would not have that concept of a bicycle, even though they had the concept of such a concept (the kind of thing they could not yet conceive of). – None – 2016-07-05T18:11:09.990

    It is absolutely impossible to accurately formalize something that simply makes no sense. I have formalized the erroneous argument as stated and pointed out the error in the formal argument. I could make the formalization tighter by quoting and involving a type language, but it would be pointless nitpicking. The error is in passing over from expressing the possibility of a reference to actually making a reference. Only if you are a modal realist or other deep Platonist do all things potentially referenced actually have references. And even then time or mode intervenes. – None – 2016-07-05T18:22:47.960

    That was a temporary misunderstanding on my part about the caveman analogy. – user2268997 – 2016-07-06T01:06:42.463

    Well the problem is, an actual reference is inferred from a potential reference (existential quantification) through existential instantiation in a formal argument. – user2268997 – 2016-07-06T01:14:35.710

    Only if you are a deep, deep, Platonist, and even then time and order matter. As the caveman analogy points out, even if the thing exists in principle, (there is a concept of a bicycle, cavemen just don't have it yet) it does not yet exist at the time it is being required by the proof. The argument does not contain the concept -- IT CONTAINS ONLY THE CONCEPT OF SAID CONCEPT. It is not instantiated. Period. – None – 2016-07-06T02:27:08.723

    I am done repeating myself. Go find someone else to ignore? – None – 2016-07-06T02:32:22.563

    I am not ignoring you. Sorry for the frustrating conversation. Though I may have found the culprit: Here is my definition for a concept of a thing: any Idea that can divide things into two categories of {having that concept} or {not having that concept} so that complete knowledge of something allows one to decide which category the object belongs to. now according to this definition, 'Having no Concept of something' is indeed a concept of that thing (though the former category will be empty).Now B's Argument will be valid with this definition,however it will practically have no consequences. – user2268997 – 2016-07-06T05:14:10.973

    Now I understand what this is about.

    – user2268997 – 2016-07-06T05:14:28.630

    I will make one more try before throwing in the towel. Your definition fails to accord with usage. Is 'Being a square circle' the concept of a thing? It is a set selector. But isn't a square circle inconceivable? So 'Being a square circle' is only the potential concept, it cannot be a real concept, because clarifying the potential idea into an actual one is beyond the mind. How about the element in Russel's paradox? Do we understand it? – None – 2016-07-06T13:38:55.017

    So these are inconceivable things, and we have considered them. Do we have concepts of them, or not? In the sense of the argument, one would have to say that we do. But they are exactly those things of which we cannot form concepts. Your approach of saying the set is just empty does not work, because we listed elements. The reference to a generic element satisfying a rule is not a reference to a real instance. (We do not have concepts of thing just because we imagine their possibility, we only have concepts of them if we can imagine their actuality or we realize their possibility.) – None – 2016-07-06T14:38:52.840