Holism and the Underlying Structure of Our Acceptance Criteria
Holism and the Underlying Structure of Our Acceptance Criteria


Quine, in his Two Dogmas of Empiricism, rejects the analytic/synthetic distinction and recommends a position that has been being labeled as 'holism'. My interest in this paper is to present a version of holism that is in many respects similar to the Quinean version but is explicitly designed to shed some light on the way different regions of our thought relate. I try to derive this holism by the image of our practices that I believe Wittgenstein's argument concerning rule-following stimulates. The motivation for this more Wittgensteinian holism is to make explicit some features of our thinking that would otherwise be hidden. In particular, a holism in this spirit introduces an image where our different practices communicate much more than we would be prompt to admit at first sight. This, in turn, encourages us to suspect any attempt to take any of our practices in isolation from all the others.

Table of contents

    1. It is a common place that the way we live is founded largely on the way we choose to think. It is a significant element of our practices that there should be at least a specific relation connecting thought to each other. These relations are what enable one to speak about reasons to accept a given belief. This paper aims to gain some understanding of how our beliefs influence each other. Put in another way, my interest here is to improve our understanding of what is behind the acceptance criteria that make us inclined to hold or reject a belief.

    A possible way to handle this preoccupation is to accept the position that emerges from the last two sections of Quine's Two Dogmas of Empiricism. There he presents a point of view that tries to argue against the idea, common to several logical empiricists, that we have means to hold a distinction between analytic and synthetic statements. Quine aims to reject the idea that experience could on its own tell us something about how it should be incorporated in the web of our beliefs. He comments:

    No particular experiences are linked with any particular statements in the interior of the field, except indirectly through considerations of equilibrium affecting the field as a whole. (Quine 1953)

    Quine focuses on how a non-conceptual content1 – the given originated in our sensorial stimuli – can interfere in the articulation of our concepts. However, his way of thinking about how non-conceptual content relate to conceptual states could also be applied to the connections between different conceptual states.

    The argument in the Two Dogmas that I believe that leads to the above holism is the recognition that nothing prevents us to take as true any belief “if we make drastic enough adjustments elsewhere in the system”. In other words, we have indefinite ways to take seriously whatever is given through our senses; we can even face it as an episode of hallucination. My view is that this argument, as I commented before, would hold no matter if we are concerned with conceptual or non-conceptual contents. If I am right about this, it follows that if we accept, on this ground, that the given alone tells us nothing, it seems inevitable to accept also that a belief, by itself, tells us nothing. Accordingly, the argument of the Two Dogmas can be used even if we are inclined to reject what Davidson once called the third dogma.2

    In this form, holism tells us that non-conceptual contents can disturb, as much as conceptual content, any region of our thought. Therefore the Quinean argument tells us nothing about the very possibility to draw a line that bounds the influence of whatever content. Someone could accept his argument and say something like: «When we question whether the Newtonian theory is consistent to a certain experience of an inclined plane we place outside the sphere of doubt beliefs like ‘I like waterfalls more than beaches’ or ‘these computers are faster than those’. When we take seriously a certain belief we always place some set of beliefs outside the range of doubt. Perhaps we have problems deciding if some beliefs are outside the sphere of doubt but, certainly, we can do this to many others». (I will henceforth refer to this line of argument as the 'standard line').

    Thus it seems reasonable to affirm that the argument found in the Two Dogmas gives no argument against the idea of postulated analytic statements. The kind of analytic/synthetic distinction against which Quine argues is one that claims only that there are statements that suffer no influence by experience whatsoever. But once we can draw the line of influence that a thinkable content exerts in our thought, someone could accept the Quinean argument and, at the same time, defend the existence of analytic statements if we can postulate them to be so.

    Thus, the message that holism presented along these lines seems to offer is that when we takes a new content into consideration we can in principle change any belief but, in practice, we can have a very different picture. We can be presented with statements that are to be held as postulates. In other words, Quine's arguments have very little to prevent the possibility that we take some statements as not to be revised by fiat.

    2. If we are inclined to accept the Wittgensteinian argument concerning rule following, any talk of non-conceptual content makes no sense. A non-conceptual content is independent of whatever other content that could constitute a rule. I understand the rule following argument (RFA) as presenting reasons against the possibility of a thinkable content determined by anything other a public practice. That is to say, rules can only be understood in relation to a set of shared practices. A statement can therefore make sense only within a set of statements and those have to refer to its role within a set of practices – its role is context- dependent. Contexts, themselves, cannot be defined without a reference to conceptual practices. Non-conceptual content is therefore in direct opposition to the very message of Wittgenstein's RFA.

    Therefore it follows that first, the RFA gives us good reasons to abandon the third dogma that Quine is committed to and, second, the Quinean argument in favor of holism is independent of accepting or not the third dogma. Another sort of holism seems to come up from the RFA; a holism that incorporates the critique to the notion of non-conceptual content. I would like to further argue that accepting the RFA has more to contribute to a sound version of holism.

    3. I believe there is no inconsistency in rejecting the third dogma while accepting the first two. In other words, I believe we can abandon the non-conceptual content idea while taking the consequences of holding the analytic/synthetic distinction. I would like to examine now whether a holism along the lines I sketched above – one that is grounded on Wittgenstein's RFA – leads us to a rejection of the third dogma and further makes it harder that we take some analytic statements as postulates. The issue here is to compare the consequences of these two versions of holism so that we can gain insight about how they differ.

    The talk on analytic statements as postulates bears on the assumption that the standard line holds. The standard line, in its turn, depends on an amount of control over the consequences of accepting some beliefs: that we could draw a line that would determine the reach of influence of these beliefs. The question here is to establish whether some criteria of acceptability of some statements, namely analytic statements that are postulate, are somehow influenced by empirical beliefs. I believe that accepting RFA leads to an image of our practices according to which our capacity to establish the consequences of what we think is far more limited than what the standard line seems to suggest.

    Moreover, those postulate analytic statements are understood differently against a background where the RFA holds and against a more strictly Quinean one. If we accept the RFA, a statement can only play an epistemic role within a specific context. As a consequence, to talk about analytic statements against this background is to imply that they act as such with respect to some contexts. (Surely, these contexts could involve several practices.) If any statement is to be taken as analytic within this RFA background, several others would have to follow it: no statement can be considered analytic in isolation.

    4. The RFA repels an image that seems prima facie attractive: the image that we can separate out our beliefs from the course of our practices. The image implies that we could have the same web of beliefs organized by our practices in different ways. The RFA goes against this idea in the sense that it faces this dualism as an instance of the dualism between the institution and the application of rules. If this latter dualism is rejected, the course of our practices cannot be distinguished from what we believe. Whatever makes us take a specific course of practices is determined by our beliefs.

    The alternative image that emerges from the RFA has that our criteria of acceptance are determined by our practices themselves and by nothing further. Changes in our ways of thinking are not independent from our practices; our beliefs do not constitute a separate realm isolated from our practices in general nor could they have a self-standing mechanism of evolution. It is therefore more appropriate to talk about relations among practices than about relations among beliefs for beliefs can only be understood if we can locate them in a practice.

    I have observed that against the background of the RFA no statement can be taken as analytic in isolation. This, in itself, constitutes no problem for the standard line. Within the standard line, we could argue that the statements that ground the one that is a candidate analytic statement are presuppositions behind this very statement and we have at least some capacity to make them explicit given a context. In a context of argument, we exercise precisely this capacity to recognize deductive relations among our beliefs in order to make sure that what we are putting forward is coherent. The RFA, on the other hand, entails that the statements that ground the candidate analytic statement are not only presuppositions.

    Accepting RFA leads us towards the idea that the use of any term must be reflected on previous practices where the term appears. A term has nothing else to it – our current use of the term can make reference to nothing but previous use. When we understand a term, we do not grasp a rule from a set of rules itself devoid of any practice. In fact, the rejection of a distinction between instances where the use is instituted and instances where it is applied leads us to take any new application of a concept as a slight change in the underlying meaning.

    When we use a term, we relate it to other terms, in a specific way (or, according to a rule). In any of our practices we make several connections among terms. Let us suppose, for example, that one attributes a specific image to the majority of children one sees. Suppose further that it is an image which discredits completely what these children say. It wouldn't be strange if, in the future, this person attributes this same image again to an unseen child. In this case, she would not do that because she knew how to understand a rule in a given moment and something else was added to her understanding in a later moment. The attribution of the image is not a decision taken after her understanding of the meaning of “child”. Nothing less than all our understanding is originated from our practices.

    However, the above example concerns a specific relation. Actually the use of terms within our practices assumes a more intricate structure. The more a vocabulary relates to a greater number of heterogeneous terms the more it is hard to draw a line of influence for this vocabulary. My claim is that the empirical beliefs are formed by a vocabulary of this kind. Empirical beliefs – and therefore empirical vocabularies – are present in a diversity of moments in our lives.3 The very capacity of detect all consequences of our empirical beliefs is not supported by argument.

    So I think that we have no more than a partial capacity to detect the impact of our empirical beliefs. Thus, if we accept the RFA an analytic statement could be considered as such only as a conjecture. However, someone could question if the RFA has any force against the idea of analytic statements as postulate if interpreted in a instrumentalist way. I believe that it has nothing to say about this. However, a Wittgensteinian route to holism, more than a Quinean one, seems to encourage a strong suspicion towards any attempt to split our thinking in compartments. Whatever our thinking does, it does because it establishes connections.


    1. Davidson, D. 1980 “On the Very Idea of a Conceptual Scheme” Inquires into Truth and Interpretation, Oxford: Oxford University Press, 183-198.
    2. Quine, W. V. O. 1953 “Two Dogmas of Empiricism” From a Logical Point of View, Cambridge: Cambridge University Press, 20-46.
    3. Wittgenstein, L. 1974 Philosophical Investigations, Oxford: Blackwell.
    4. Wittgenstein, L. 1969 On Certainty, Oxford: Blackwell.
    By ‘non-conceptual content’ I aimed to refer to thought contents that are expected to be intelligible independently of whatever other thought content. An empiricist like Quine, for example, holds a non-conceptual content in the sense that he believes that our senses inform us about the world independently of any other content we happen to possess – our senses themselves make a difference.
    That is, the idea that we can separate out conceptual scheme from empirical content (Davidson 1980)
    One way to recognize this is to think in the constant presence of a kind of empirical propositions what once Wittgenstein called ‘propositions that stand fast’ (Wittgenstein 1969).
    Makmiller Martins Pedroso, Brasília. Date: XML TEI markup by WAB (Rune J. Falch, Heinz W. Krüger, Alois Pichler, Deirdre C.P. Smith) 2011-13. Last change 18.12.2013.
    This page is made available under the Creative Commons General Public License "Attribution, Non-Commercial, Share-Alike", version 3.0 (CCPL BY-NC-SA)


    • There are currently no refbacks.