Gila Sherinterviewed by Richard Marshall.

Gila Sher has written a book on logic, The Bounds of Logic: A Generalized Viewpoint, and co-edited another one: Between Logic and Intuition: Essays in Honour of Charles Parsons. She is always brooding on epistemology, metaphysics and the philosophy of logic. She looks to fatten up truth in a groovy way. What gives this an extra zing is that she’s a top woman in an area stuffed with men.

3:AM:In 1999 you wrote a really interesting question: "is there a place for philosophy in Quine’s theory?" The threat was that the extreme naturalistic turn that you thought went even deeper than that proposed by logical positivists. You argued that Quine’s extreme naturalism - that denied the analytic-synthetic distinction, that argued for the interconnectedness of knowledge, universal revisability, methodological pragmatism and realism - was compatible with substantive philosophy. Can you say something about both what you argued in the paper and the general point about what you meant by philosophical naturalism?

Gila Sher:I am glad that you are starting with this paper, because this specific paper, more than anything else I’ve written, goes back to my earliest experience with philosophy. In my sophomore year as an undergraduate at the Hebrew University of Jerusalem two of my professors were engaged in a very lively and passionate debate about how philosophy should be done: should it follow the traditional, especially Kantian, template or should it follow the contemporary analytic template? We, the students, had a special place in this debate: each side wanted to pull us in its direction, and we played the role of both interrogators and jury.

So from the very start I was taught to have an active attitude to the question “What is philosophy?” Philosophy was not something you just learn; it was something you always approach with questions like “what philosophical method is used here? is it problematic? can it be improved? is there an idea for a new philosophical method here?” in the back of your mind. It was also in Israel that I became interested in Quine. I found Quine both extremely exciting and terribly frustrating. I spent a full year trying to sort out the “good” and “bad” Quine in a paper, without success. It was only when I became a graduate student at Columbia University that I finally managed to write a paper on Quine, 'Knowledge without Dogmas', that was the precursor of my 1999 paper.

In my 1999 paper I argue that Quine’s model of knowledge is a mixture of a few unusually important and innovative theses and a few traditional and quite reactionary theses. The first lead to a new, very promising way of doing philosophy, the second bring us back to an old empiricist way of doing philosophy, with its well-known problems. In short, there is a problem with Quine: what he gives us with one hand he takes away with the other. I approach this problem through an inner tension, first pointed out by Dummett, between Quine’s rejection of the analytic-synthetic bifurcation and his endorsement of a center-periphery bifurcation.

On my construal, the tension is primarily epistemic. The analytic-synthetic bifurcation, though a linguistic bifurcation, is associated with a traditional epistemic methodology, namely one that divides the different branches of knowledge into two groups: a group of disciplines for which the possibility of a conflict with reality does not arise and a group for which it does. Accordingly, with respect to the second group we develop an elaborate method of factual testing and confirmation, but with respect to the first group there is no need to develop such a method. I suggest that the traditional methodology associated with the analytic-synthetic bifurcation creates a Maginot line of epistemic defense. While all our defenses are concentrated in the synthetic zone, we might be vulnerable in the analytic zone. I conclude that, in the absence of a conclusive proof of the non-factuality of logic and mathematics, it is unwarranted - and a bad strategy - to treat them as analytic.

So Quine is right in rejecting the analytic-synthetic bifurcation. His new, non-analytic-synthetic model presents an opportunity for epistemologists to embark on a completely new methodology, one that has never been tried before. But Quine’s center-periphery thesis leads us back to a traditional methodology. This is because CP is essentially a traditional empiricist bifurcation. It divides the various branches of knowledge into two groups: logic and mathematics, located in the center, and the natural and social sciences, located in the periphery and intermediate zone. Although Quine is more flexible than his predecessors in allowing continuity between center and periphery, still, as an empiricist he doesn’t, and cannot, allow logic and mathematics to lie in the periphery (or experimental science to lie in the center).

As an empiricist, he is obliged to limit the interface between human knowledge and reality to sensory perception. That is, he must construe the periphery as purely experiential. But it’s quite clear that if logic and mathematics are factual, their factuality is not experiential, or at least is not exhausted by sensory connections with the world. So logic and mathematics must be left in the vicinity of the center, that is, be guided largely by the pragmatic, linguistic, conventional standards traditionally assigned to them by empiricists, and the natural and social sciences must be left in the periphery and intermediate zone, being guided by their own traditional standards.

My proposed solution to the inner tension in Quine’s model is to reconfigure the center-periphery duality, that is, make it non-traditional, so it further contributes to, rather than undermines, the promise of the non-analytic-synthetic model. First, I expand the conception of center and periphery so they are not bound by empiricist strictures. Periphery now represents not only sensory but also intellect-based interface between units of knowledge and reality, and the center is broadened to include all connections with the mind, not just pragmatic principles. Second, I reconfigure the model as a highly dynamic model - one in which every theory and discipline is, and must be, located in some respects and during some periods in the periphery and in others in the center. The idea is that each and every discipline is, and must be, subject both to veridical standards of adequacy (correspondence truth, factual evidence and justification) and to pragmatic and conceptual standards of adequacy. I conclude the paper with a lengthy sketch of the dynamic structure of the new model and its potential for solving philosophical problems and explaining philosophical phenomena.

Now, one result of this solution is that philosophy, as a discipline, is subject both to veridical and to pragmatic standards. Furthermore, because the revised model preserves all the Quinean principles you mentioned in your question, philosophy is interconnected with other fields of knowledge and is free to use experiential resources if and when they are of help. So philosophy is no longer either “traditional metaphysics” or “first philosophy”. But philosophy is not limited to the experiential methods of the empirical sciences either. (The revised model, with its broad periphery, allows intellect-based interface with reality.) That is, philosophy is not naturalistic in the radical sense of being part of empirical psychology. Instead, philosophy, like all other disciplines, is a largely independent discipline, with its own substantive questions, which it is free to pursue in whichever way turns out to be most fruitful.

3:AM:One of your interests is logic. And one of the issues you pursue it seems is whether logic is naturalistically explained. You look at this question from more than one angle. So one way in is to ask whether logic is obvious? At the bottom of that question is the thought that logic is so obvious, too obvious I guess, that it doesn’t need an explanation. Is that right? You totally disagree with that view don’t you?

GS:Yes, you are absolutely right. I totally disagree with this view. I think that logic requires a substantive (though not necessarily a naturalistic) explanation and that it’s one of the important tasks of philosophy to give it such an explanation. Let me expand a bit. Almost everyone agrees that logic plays a crucial role in knowledge: every branch of knowledge is bound by the norms of logic, and each can be undermined by a logical inconsistency. So if we want to have a veritable understanding of knowledge we need to have a veritable understanding of the logical norms: what they are, why they are so important, why (whether, how) they work in the world, why (whether, how) they guarantee valid inference, why (whether, how) they can be used to unearth systemic falsehoods (logical inconsistencies), whether there are ways to improve our current logical theories, etc. All these questions call for a serious answer. And to say that logic is simply obvious is to run away from these questions.

Moreover, it’s not clear to me at all that logic is obvious. If logic is obvious, then it should be quite obvious whether any given theory is logically consistent or not. But in many cases this is far from obvious. Some logical inconsistencies we now know of were far from being obvious before they were discovered (think, for example, of Russell’s paradox). And even today, very few mathematicians feel certain, let alone think it’s obvious, that contemporary mathematics (say, set theory) is logically consistent. Furthermore, even the simplest logical laws, like the law of excluded middle, are far from obvious. Some of these laws are quite controversial, and many (for example intuitionists) not only think they are not obvious but downright reject them.

So the view that logic is obvious may very well be false. Indeed, even if it’s true, it’s useless for philosophers, since it’s far from clear what kind of obviousness is characteristic of logic and why. Clearly, obviousness by itself is not a criterion of logicality. To borrow a famous example from G.E. Moore, it’s quite obvious to me that I have two hands, but this does not make the claim “I have two hands” logical. So what kind of obviousness is a sign of logicality? Under what conditions is it a sign of logicality? Why is it a sign of logicality? - To answer these questions is to give a substantive explanation of logic. This is what I claimed in my paper “Is logic a theory of the obvious?”.

In recent work, I’ve returned to this question from a different direction. Connecting explanation with justification, I asked myself why, if everyone appreciates the importance of logic for knowledge, are there so few attempts to provide a substantive explanation/justification of logic in the philosophical literature, past and present? Take Kant, for example. Kant constructed a new, detailed foundation for mathematics and natural science, but not for logic. Why? Why did he think it was essential to launch a revolution (his Copernican revolution) in the foundation of science and mathematics but not in the foundation of logic? Even in the late 19th and early 20th century, when distinguished mathematicians were engaged in investigations called “the foundations of logic and mathematics”, their main goal was to establish a foundation for mathematics and to the extent that they engaged in a foundational investigation of logic it was largely as a means to this end. (The logicists, for example, offered a detailed foundation for mathematics by reducing it to logic, but logic itself they left without a detailed foundation.) Why? And what about philosophers? There is of course Carnap, with his explanation of logic as conventional; there is Pen Maddy, with her naturalistic explanation of logic in her recent book, Second Philosophyand a few others. But Carnap explains logicby saying that there is nothing to explain - logic is merely conventional, and Maddy passes the burden of explaining logic largely to psychology. In any case, compared with the number of philosophers who attempted a substantive explanation/justification of science and mathematics, very few have attempted a substantive explanation/justification of logic. Why?

I think the root of the problem is methodological. For centuries most philosophers associated the foundational (explanatory + justificatory) projects of epistemology with a very specific methodology: the so-called foundationalist methodology. This led to conflation of the foundational project with the foundationalist methodology. Now, the foundationalist methodology is a very rigid methodology - requiring a strict hierarchy of units of knowledge, allowing an explanation/justification of any unit only in terms of lower units in the hierarchy, and banning all cases of circularity and infinite regress. This makes the task of providing a foundation for so-called “basic” units, including logic, impossible to carry out (unless you completely trivialize the foundational task).

So it’s not surprising that philosophers who are used to thinking of the explanation/justification of logic in foundationalist terms conclude that there is no point in trying to pursue this project. At the same time, philosophers who are aware of the serious problems with foundationalism and have renounced this methodology altogether also find themselves in a situation that discourages a foundational study of logic. Due to the common conflation of the foundational project with foundationalism many of these philosophers identify their renouncement of foundationalism with the renouncement of the foundational project itself. And for that reason they are reluctant to engage in a foundational projects of any kind, including the logical project. So it’s easy to understand why most philosophers are reluctant to engage in a serious foundational study of logic. (One popular way out is to opt for all-encompassing pragmatism and use this as a reason for not offering a detailed explanation/justification of logic.)

In my view, this reluctance is unwarranted. Today, due to various developments, the road is open for constructing a new non-foundationalist yet foundational methodology that would enable us to provide a substantive explanation/justification for all branches of knowledge, including logic. Indeed, the idea of a non-foundationalist yet foundational methodology has been recently pursued by a number of philosophers. For example, Susan Haackin her book Evidence & Inquiryhas developed a new methodology of this kind, a synthesis of coherentism and foundationalism, which she calls “foundherentism”. I think her work is an important step in the right direction, but it doesn’t go far enough. In particular, it is limited to the foundational project of the empirical sciences and leaves logic and mathematics outside the foundational project.

In my own recent work I have been engaged in developing a post-foundationalist methodology that extends to all branches of knowledge, including logic. This methodology naturally arises from the dynamic model of knowledge I talked about in my response to your first question, and it is based on a non-coherentist version of holism that I call “foundational holism”. The idea is to put the holistic methodology at the service of the foundational project.

Very briefly, holism is used to license a rich network of resources, both internal and external, for providing a foundation for logic, free of rigid hierarchical requirements. Logic is given a dual grounding both in the mind and in the world, and the grounding itself is an ongoing, step-by-step process, involving loops and shifts in perspective, as in Quine’s metaphor of Neurath’s boat, or some variant thereof. (To patch a hole in the boat we find a temporary foothold within the boat and fix the hole using any resources available to us, either internal or external – that is, either created on the boat or taken from the sea. Once the hole is patched, we can use it as a standpoint for creating new resources, using them either to strengthen the original patch or create an improved patch to replace it with.) One distinctive feature of this method is its treatment of circularity. Not all circularity is precluded, and some uses of circularity are viewed as advantageous (“constructive circularity”). This is especially important for foundational projects involving basic branches of knowledge, like logic. (I should add that other philosophers, too - for example, Ernest Sosa- have recently written about the importance, for epistemology, of legitimizing non-vicious circularity.)


3:AM:In the course of that paper ['Is there a place ...'] you note Quine’s dismissing of a Carnapian linguistic explanation of logic. Dave Chalmershas resurrected Carnap’s approach in the Locke Lectureshe did in Oxford a few years ago with the cosmoscope and the idea of a bedrock of core concepts that allow us to explain everything. What do you make of this?

GS:Chalmers’ Locke Lectures ('Constructing the World') are very interesting and I like many things about them. For example, I like their emphasis on intellect (reason) as a central constituent of knowledge, I like the foundational nature of the project, I share Chalmers’ admiration of Carnap’s Aufbauand I like the constructive approach in general, I share Chalmers’ worry about “trivializing mechanisms” in philosophy, and so on. But there are also points of difference.

For example, Chalmers’ approach to his project appears to me to be foundationalist. Also: Chalmers accepts the traditional analytic-synthetic and apriori-aposteriori bifurcations while I reject them (for reasons which are different from those addressed by Chalmers in his lectures). In particular, I think one doesn’t have to countenance apriority in order to value, emphasize, and study the extremely important contributions of the intellect to knowledge. In a sense, the focus on apriority appears to me misleading. Why should we require the intellect to work in complete isolationfrom other sources of knowledge? The intellect doesn’t have to be separatefrom sensory experience in order to play a dominant role in knowledge.

In the extreme case, all knowledge might require some experience while reason is still the more important factor. Analogy: we may need something like small twigs to start a big fire, but once we start the fire, what really nourishes it and keeps it going are massive pieces of wood rather than small twigs. This is in a certain sense a Kantian view, but it rejects the Kantian dichotomies. The point is that both reason and sensation are enormously valuable for knowledge and we should investigate their roles in an open minded manner, without imposing unnecessary dichotomies.

As for logic, Chalmers says very little about it in his lectures (he notes that logic plays a less central role for him than for Carnap), and what he does say is very tentative. But the question you raised, whether logic is grounded in language/convention, is an important question, so let me say a few words about it. There are several well-known arguments in the philosophical literature against the view that logic is grounded in language/convention, some very good, others less so.

For example, I find two of Quine’s arguments compelling. One is that if we allow that logic is true by convention we should allow that any discipline is true by convention. But that’s absurd. The other is that being admitted into our corpus of knowledge based on convention is a passing trait of theories, significant historically but not normatively. Normatively, every theory must be established as true, regardless of whether we first created it with the help of conventions. In contrast, I find Quine’s argument from circularity less compelling. This argument says that the view that logic is language-based is circular, since you need logic in order to derive logic from conventions. This argument may be compelling for foundationalists, who shun all circularity, but not for holists.

However, the most important reasons for rejecting the linguistic/conventional conception of logic, in my view, are different from all of these. They have to do with the fact that knowledge, qua knowledge, must be grounded in reality, and that logic is included in this category. This involves both general considerations and considerations specific to logic. I’ve discussed the first part in my paper 'Epistemic Friction: Reflections on Knowledge, Truth, and Logic' and the second in papers like 'Is Logic in the Mind or in the World?'. As you probably know, traditionally most philosophers think of logic as grounded only in the mind (if it is grounded in anything at all).

Among the few who thought that logic is anchored in reality is Wittgensteinin the Tractatus. But Wittgenstein thought it was impossible to rationally talk about this matter, since to speak about logic we must go outside logic, but there’s nowhere to go beyond logic. This problem, however, is a foundationalist problem that does not arise for foundational holists like me. To speak rationally about logic we ascend to a Tarskian meta-language, or we move to a region on Neurath’s boat from which we have a relatively wide view both of the sea (world) and of the boat, or we shift back and forth within the dynamic model of knowledge I spoke about above, so we can view logic critically, from a point of view of “immanent transcendence”.

To explain logic’s grounding in reality I bring a few considerations. One is based on the common-sense observation that logic has to “work in the world” (that in the same way that an airplane might not get off the ground if its designers rely on an incorrect aerodynamical theory so it might not get off the ground if they rely on an incorrect logical theory). A second consideration is based on historical and counterfactual examples of cases in which logic fails due to factual errors. A third is a theoretical consideration that connects logic to reality through its close connection with truth.

Take a simple sentence, S1, which is true if and only if certain conditions are satisfied in the world. Suppose these conditions are such that they rule out the satisfaction of other conditions that have to be satisfied if another sentence, S2, is to be true. And suppose S1 is true. Then if your logical theory tells you that S1 logically implies S2, your logical theory is wrong. For your logical theory to be true, it cannot conflict with the world. And in this way our logical theories is constrainedby the world. But the connection is even stronger. Not only must logic not conflict in the world, but we can find a grounding for logic in the world.

There are laws in the world - laws that govern objects and properties in the world - that, due to their special features, have the ability to ground logical implications, and that, due to their connection with logical form, indeed do so. I call these laws “formal laws”. To see how they ground logical implications consider, once again, two sentences, say S3 and S4. When the conditions that must hold in the world for S3 to be true are connected to the conditions that have to hold in the world for S4 to be true by a formal law (of the appropriate kind, given the logical forms of S3 and S4), the truth of S3 guarantees the truth of S4. And this is the ground of the claim (by a correct logical theory) that S3 logically implies S4. There is much more to be said, but that’s the gist of my claim that logic is grounded in reality. (Of course, this type of grounding involves some circularity, but this is just the type of circularity that is sanctioned by the holistic methodology we talked about earlier.)

3:AM:One way the naturalist argument for logic might go is to conclude that there is universality of logic and truth that can be explained by our natures, perhaps reducible to brain science eventually. A thought experiment would be to ask whether had we different logic would we be able to understand certain things that at present we can’t? What do you make of that thought?

GS:This is an interesting question that concerns the connection between logic’s grounding in the world and its grounding in the mind. But before I try to answer this question, I would like to say something about naturalism, since you have mentioned it a few times in your questions. (I should point out, though, that thinking about naturalism is still a work-in-progress for me, so I won’t say much about it beyond what I said in my Quine paper.) Naturalism is quite popular today among English speaking philosophers, and there is a large variety of strands whose relations are not always clear. So, to avoid misinterpreting people, I will speak schematically and limit myself to two quite radically different strands.

One strand of naturalism, which I support, says that traditional philosophers are not justified in requiring philosophical theories to be apriori. In constructing philosophical theories we should be free to help ourselves to any available type of knowledge, including empirical knowledge, if and when we deem it fruitful for answering the philosophical questions we are investigating. A different type of naturalism, and one that I oppose, says that philosophers should limit themselves to empirical inquiry or to inquiry that serves the needs of the natural and social sciences. I like naturalism inasmuch as it frees us from a rigid philosophical methodology focused on the traditional dogma of apriority; I dislike it inasmuch is it replaces one shackle by another and inasmuch as it tells us that there are no worthwhile philosophical questions that are distinct from those of the natural and social sciences.

Turning to the thought experiment you suggested, it seems right that to a considerable extent we are conditioned by our mind, including biological, psychological and other aspects of our mind, and that some aspects of our biological make-up limit our cognitive abilities in significant ways, and that if we had different mental/biological features we might have been able to understand things that, given our present limitations, we are unable to understand. This seems right both with respect to our logical capacities and with respect to our cognitive capacities in other areas.

At the same time, it’s important to realize that this doesn’t mean that logic (or physics, or biology) is grounded only in our brain wiring, innate concepts, etc. After all, whether or not our brain-wiring is sufficient for us to cognitively flourish in the world (or even to survive in the world) largely depends on the world itself - on whether its features and structure are such that our specific wiring enables us to cognitively access it. So in the end, all our knowledge, including our logical knowledge, in significantly dependent both on our mind and on the world.

3:AM:The existence of a plurality of logics seems to weaken the argument that logic is a given bedrock of all minds. So we have intuitionist logic in Dummett’s Oxford, we have dialetheistic and paraconsistent logics in Australia and dialectics in continental Europe and Chinese philosophy. Classical logic is strong in the USA. Put like that, it might look like logic is geographically particular. And that might suggest it is reliant on cultures. What do you make of the distribution of logics and the argument that they are tied to culture? And in particular, what do make of Graham Priest's contention that it is due to the influence of Quine that the USA is so unreceptive to alternatives to classical logic?

GS:I think that there are several levels on which we can ask whether logic is relative to culture. First, we can ask whether people in general, not specifically philosophers, actually use different patterns of logical reasoning in different cultures. Whether, say, Americans rely on the law of excluded middle more often than the British. Second, we can ask whether philosophers with different cultural background within the academia tend to subscribe to different philosophical views of logic. For example, we can ask whether British philosophers are more inclined to subscribe to intuitionistic philosophies of logic than American philosophers. I am not sure, but I tend to think that statistically cultural background might affect philosophical sympathies and dispositions.

Since Quine was an extremely influential philosopher of logic who was unreceptive to non-classical logics, I think that Priest’s conjecture that he is partly responsible to American philosophers’ unreceptiveness to such logics might be right. Third, and most importantly in my view, there is the question whether we, in constructing our own philosophical theories of logic, should limit ourselves to theories that appeal to philosophers of our own cultural background. Here, my answer is “No”. In developing our theories we should, putting it crudely, aim at the truth, and hope that our colleagues all over the world judge our theories based on equally universal norms. I think the same holds for our theories of truth.

3:AM:Susan Haack wrote about deviant logics and pointed out that deviant logics are successful only if they add to classical logic. This suggests that classical logic is fundamental. What do you think?

GS:This is an issue on which I suspend judgment. It’s an important issue, but not an issue I have investigated in sufficient depth to have a clear view about. I should add, perhaps, that in my work in the philosophy of logic I use classical logic as a default logic and this has led some people to think that I am committed to this logic. This is not the case. I use a classical framework because it’s more convenient for me to work within this framework than others (possibly because of the academic culture I grew up in). This, however, does not imply commitment to classical logic. Usually, I like to formulate my ideas first in general terms, which are applicable to numerous frameworks, and then in precise terms. Formulating ideas in precise terms, however, requires choosing a framework that has sufficient resources for a precise formulation. But that’s not commitment. The goal is to show readers how the general idea can be expressed in a precise manner using some framework so they can figure out how to express it in a precise manner within whatever framework they think is the right framework or is their preferred framework.


3:AM:You argue that knowledge requires both freedom and friction. Friction is the idea that theories need resistence so that, in Kant’s words, they don’t ‘hover idly in thin air’. Kant has a dove metaphor to present the problem: ‘the light dove, cleaving the air in her free flight, and feeling its resistance, might imagine that her flight would be still easier in empty space.’ This is a special problem for science and maths where friction isn’t easy to find. Could you say something about your thinking here and why some of Quine’s ideas help?

GS:I think of epistemic freedom and friction as a pair of complementary ideas that are central for understanding the nature of human knowledge and for attempts to construct a theoretical foundation for it. Knowledge without epistemic friction or constraints is not different from fantasy, but constraints by themselves are not sufficient for knowledge. They can limit and guide us in developing our theories, but they don’t generate knowledge. The generation of knowledge, and in particular theoretical knowledge, requires an active cognitive act on the part of the knower/theorist. Such an act is an act of epistemic freedom.

Both epistemic friction and epistemic freedom, however, are multi-dimensional. Take friction, for example. Our knowledge is constrained by variety of factors. It is constrained by the world as the object of our theories, but it is also constrained by the mind as the source of norms we impose on our theories. These norms themselves make up a wide and diverse array of demands and desiderata, from the demand for veridicality (truth, evidence, factual justification) to pragmatic and possibly even aesthetic desiderata. The mind’s involvement in knowledge, however, is not limited to the active generation of norms. We are also constrained by other aspects of “mind”, for example by our biological and cognitive limitations. Nor is the mind’s role in knowledge exhausted by “constraints”. Despite having significant cognitive limitations, we do have considerable cognitive powers, and these are crucial sources of epistemic enablement, which also falls under freedom.

As for science and mathematics, two familiar forms of friction are experiment and proof. But friction goes beyond those, and here some of Quine’s ideas do help. For example, by rejecting the analytic-synthetic dichotomy Quine allows mathematics to be grounded both in reality and in the mind, and by rejecting the apriori-empirical dichotomy he opens the door to various combinations of intellect and experience that together can do the job. An especially welcome feature of these new options is that they avoid the well-known pitfalls of Platonism, empiricism, pure pragmatism, and conventionalism. But other Quinean ideas stand in the way. For example, by limiting the interface of our system of knowledge with reality to sensory perception Quine significantly limits our chances of accounting for the veridicality of abstract knowledge. Part of my motivation for developing a broader and more dynamic alternative to Quine’s model is just this difficulty.

3:AM:Your position has developed a theory of truth that suggests that because our minds are intricate we have intricate ways of making our thoughts correspond to reality. This is original because usually correspondence theories that tie meaning to correspondence with reality tend not to authorize different ways in which the correspondence can be delivered. It also seems properly naturalistic in that it takes seriously evidence of how people actually think. You label the position ‘Pluralism within the bounds of correspondence’, which is not as funky as it ought to be (‘Correpluralism’, ‘Intricate Correspondence’, heck, there must be something!) but this is really important. Can you say something about how you propose this works?

GS:Correspondence is commonly viewed as a single relation between language and reality, a relation that is relatively simple and straightforward, and one that holds in all fields of correspondence alike, regardless of their differences. This way of looking at correspondence, however, is not very fruitful, as can be seen from the fact that it has led many philosophers to either give up the correspondence theory as overly simplistic, or treat correspondence as a trivial, uninteresting notion, or regard correspondence (and even truth) as limited to few, relatively simple contexts or fields. My own view is that truth and correspondence are epistemically universal in the sense that all fields of knowledge are governed by a correspondence standard of truth, and that this standard might in principle be intricate, and its intricacy might vary - both in pattern and in magnitude - between (relevantly) different fields.

To understand how intricate correspondence works, I think it’s best to start by trying to understand why truth is, or is likely to be, intricate correspondence. My starting point is a very general question about truth: Why do humans have a standard or a norm of truth in the first place? In raising the same question about normativity in general, Christine Korsgaardsays that we need normative standards because we are creatures who look critically at ourselves and question our actions and beliefs. In the case of truth we ask: Are our beliefs and theories correct? We say: our theories say the world is thus and so, but are they right? And we devise a standard of truth to answer these questions and guide us in forming beliefs and developing theories of all aspects of the world.

I think three interrelated factors make such a standard especially important for us yet at the same time highly problematic, and a fourth factor makes it feasible but at the cost of complexity. The first three factors are: (i) the limitations of our cognitive capacities, (ii) the complexity of the world (relative to these capacities), and (iii) our considerable epistemic ambitions: we want to know the world in all its complexity in spite of our limitations. (We are willing to make some compromises, but we clearly want to know more than we can easily or “naturally” know.) The fourth, positive factor is our intricate mind: the powerful cognitive resources we do have. We are limited in some ways, but we are quite capable in others: we are not bad in combinatorics, abstraction, or generalization, we are good improvisers, we have an impetus to try different things (create different routes to reality), we have a rich imagination, and so on. All these make our epistemic ambitions realizable at least to some extent. But if our hope for knowledge depends on our ability to forge intricate routes to reality, then to be useful, our standard of truth must take this into account. Our standard of truth must allow that our theories correspond to reality in complex, “composite”, ways, and that this complexity varies according to the aspect of reality our theories aim at and the cognitive resources available to us.

Let me give you an example. Suppose that there are no mathematical individuals in the world, say, no numbers: 1,2,3, ... . But at the same time there are cardinalities in the world in the sense that properties of objects in the world, or collections of objects in the world, have cardinalities. In this case, cardinalities are real, but they are 2nd-level properties rather than individuals. Suppose, too, that for one reason or another humans work better with individuals than with properties of properties. For example, we, humans, are better at figuring out laws of cardinality when we think of them as laws of individual numbers than as laws of properties of properties.

As a result, our theories of cardinality are 1st-order rather than 2nd-order theories. In that case, what we would need is a “composite” standard of truth (correspondence) rather than a “simple” standard. Our standard would have to take into account the way our theory of cardinality actually works. Suppose we study cardinality by positing a level of numerical individuals (level 0) which are systematically connected to 2nd-level cardinalities, so that our 1st-order laws (of posited numbers) are systematically connected to higher-level laws governing (real cardinalities in) the world. Then the relevant correspondence proceeds in two steps: one connecting 1st-order laws to laws on the level of posits, and another connecting the latter to real laws of 2nd-level cardinalities. This 2-step correspondence is “composite”. It is not reducible to a 1-step correspondence: 1st-order cardinality theory has a life of its own. This is reflected in such differences between 1st- and 2nd-order arithmetic as differences in categoricity (all models of 2nd-order arithmetic are isomorphic, but this is not the case with 1st-order arithmetic). All the same, 1st-order arithmetic is systematically connected to the reality of 2nd-level cardinalities.

3:AM:Deflationism is another area of truth that you explore and it has been thought to offer significant opportunities for a naturalist approach to truth. Roy Sorensenonce hoped that deflationary accounts of truth a laHorwichwould become well embedded but they weren’t and so correspondence theories predominate. But it’s important to consider this theory. Can you say what you take the position to be and what its significance is? I think when people brood on logic and truth it’s sometimes difficult to see what is at stake when different theories are presented. Could you say why different theories of logic and truth are not frictionless but have effects and so why these arguments are crucial?

GS:Deflationism in philosophy is an approach I oppose. Like naturalism, it’s held by philosophers who differ so much from each other that it’s hard to give a single characterization of it. So let me tell you, instead, what the deflationist view I oppose is. In the case of truth, I oppose the view that truth is a shallow, thin, uninteresting, trivial subject-matter and that the reason we have a word for it in our language is to enable us to deal with certain expressive complications. (Deflationists say that the task of “true” is to enable people to overcome such difficulties as this: you want to express your endorsement of everything X said but you are unable to actually assert everything he or she said. “True” enables you to express your endorsement by simply saying “Everything X said is true”.)

My own view is completely different. I think truth is a deep, rich, very interesting, and substantive subject matter, and the reason we need a word for it in our language is its great importance for many areas of our life, from the intellectual to the moral and practical. I can understand it if someone says: “Look, I am investigating Y (where Y is something different from truth), and as far as my investigation is concerned it’s sufficient to treat ‘truth’ as a thin notion”. But I find it hard to understand why someone who investigates truth itself would say that. If truth is a trivial subject-matter why devote time and energy to studying it?

In his paper on Tarski’s theory of truth Hartry Fieldessentially says: the scientist will never be satisfied with a non-substantive theory of her subject-matter; why should the philosopher? (I think he actually uses the expression “it would be crazy” for the philosopher to do so.) I share Hartry’s view in this paper. From a slightly different perspective, what bothers me about deflationism in philosophy is that to be a deflationist with respect to X is, in effect, to decide that I am not going to ask hard, deep, critical questions about X. But what is the point of intellectual research if not doing just that? Deflationists sometimes say that their theory is important since it shows that truth is not a mysteryas philosophers used to think. I myself like mysteries. We need mysteries in order to have something big to explore. Mysteries are what make philosophers (and scientists, and mathematicians) into explorers, and what’s more satisfying than that? There are two ways to deal with mysteries: solve them, and dissolve them. Some mysteries do need dissolving. But it’s very unfortunate when a mystery that requires a solution is “dissolved” instead.

Ironically, for me it was by thinking about deflationism that I began to see how a substantive theory of truth could be possible and how it should proceed. In 'On the Possibility of a Substantive Theory of Truth' and later in 'In Search of a Substantive Theory of Truth' I asked myself: what is it about truth on the one hand and our way of thinking about it on the other that leads us to think that deflationism is the only viable option? I wasn’t looking for the arguments deflationists actually give for their view, but for something that might underliephilosophers feeling that there may not be another choice.

My conclusion was that the root of the problem lies in two dissonant circumstances: on the one hand the subject of truth itself is especially rich, broad, diverse, complex, and multi-faceted; on the other hand we philosophers somehow assume that a substantive account of truth would take the form of a single and simple definition (common denominator). This, I thought, is a recipe for defeat. I then compared truth to other broad and diverse subject-matters, like nature, and I noted that scientists would think it’s absurd to expect that our knowledge of nature could consist in a single and simple definition of the concept “nature”. Of course, in some ways understanding truth and knowing its principles is not the same as knowing nature and the principles governing it. But in other ways the two cases are not so different. In both cases we have a serious problem of unity and disunity(similarity and diversity, generality and particularity, abstraction and attention to detail, etc.).

Now, if you read the literature on unity and disunity in the philosophy of science you see that the problem itself has nothing to do with the empirical method (although its specific manifestations in empirical science of course do). The problem arises for any broad and diverse subject matter, one with deep commonalities but also deep differences. And truth, like some other philosophical subject-matters, seems to fall under this category. On the one hand all truths have to do with the relation between theory (language, belief, thought) and the world. On the other hand, there are significant differences between truth in logic and physics, ethics and chemistry, and so on. According to Freeman Dyson, the solution to this problem in science is a fruitful balancebetween unity and disunity.

And I suggested that exactly the same solution is right for philosophy. This led me to propose that we think of truth as a family of theories, and eventually to the more unified proposal that we think of truth as a family of correspondence principles. A third step was thinking of truth as composite correspondence, where the structure of correspondence might differ from field to field (as I explained in my answer to your last question), though the different structures are systematically interconnected. You may ask: How do we know what the structure of truth is in different fields? My answer is: we “look and see” (as Wittgenstein liked to say). We do what the scientists are doing, though in ways that fit our special subject-matter. (This conception of truth has some similarities with those of Crispin Wright, Michael Lynch, and Terrence Horgan.)

You also asked why I think the argument between deflationism and substantivism is important, and you mentioned friction and logic. I think the argument is partly an argument about whether we do or don’t need substantial friction in philosophy. Kant thought that Plato’s philosophy lacked friction because there was no way of testing its truth. But there are also other reasons a theory can lack friction. One such reason is non-substantiveness or triviality. Since deflationists advocate a non-substantive theory of truth, they advocate giving up an important source of friction: substantiveness. So it makes a difference with respect to friction whether you are a deflationist or a substantivist. Clearly, it makes a difference to our understanding of truth, and it makes a difference to our understanding of other things which are significantly connected to truth, like logic.

3:AM:Josh Knobeand the experimental philosophers have a substantial research programme that takes even the most abstract-seeming concerns of philosophy into the domain of empirical experiments. So they will say to you: have you tested out your theoretical claims with what the folk actually think? Are they confirmed? Is this an approach that helps your position about questions of truth and logic? And are there experiments that do support what you are claiming or is this something that needs to be done? So, for example, one area where the experimental philosopher might look is when considering the dispute you discussed a minute ago between deflationism and substantive correspondence theories of truth. Where they would presumably move in would be when they read someone like Michael Dummett claiming that ‘The layman ... expects philosophers to answer deep questions of great import for an understanding of the world ... And the layman is quite right. ... Yet he finds most writings by philosophers of the analytical school disconcertingly remote from these concern. The complaint ... is ... understandable. ...’ Josh Knobe and co. worry that the opinions of the folk are often used like this to justify their own intuitions about an argument. Can you say something about this?

GS:When Josh Knobe says in his interview with you that there is no complete separation between philosophy and the other branches of knowledge, of course I agree. Likewise, when he sets up an experiment, using questionnaires, to test his conjecture that people’s abstract theory is what determines their view on free will, I think this is an appropriate thing to do. Generally, I think, what tools you use depends on what you are investigating and what conjectures you are making. That is, what forms of friction are required, or are fruitful, for your philosophical project depends on what the project is. But when I, for example, investigate what is required for an inference to be truth-transmitting with a strong modal force, or whether it’s a sound methodology to treat logic and mathematics as completely immune to refutation by the world, or how one could/should go about developing a substantive theory of truth, then administering a questionnaire about these things to a sampling of people does not seem the most fruitful way to proceed, as I expect Knobe would agree.

This is not to say that empirical discoveries can’t be of great value for philosophy. For example, I think the study of consciousness calls for cooperation between theorists and experimentalists. I myself am especially interested in the way empirical discoveries can “point beyond themselves”, so to speak, to things like the status of geometry or the need for a revision in logic. But as far as I can see, this is not what experimental philosophy is about, at least for now.

As for your citation from Dummett, which I really like, at least on my own reading of it Dummett’s appeal to what the folk think is mainly a literary flourish. Dummett could just as well have said something like “When I look back at my young self and remember how I expected philosophy to answer deep questions of great import for an understanding of the world, I still think I was right. Analytical philosophers neglected this aspect of philosophy since they were engaged in a destructive project, but today we are once again ready to do constructive work in philosophy ...”. (Although the way he said this was more eloquent).


3:AM:Another aspect of this is the thought that philosophers have intuitions that are biased. So one experiment by Vaesenand Petersonthat the experimental philosophers discuss suggests that the intuition that Gettiercounter-examples to the idea that knowledge is justified true belief only seem intuitively right relative to certain languages. Philosophers say things like ‘not all intuitions are created equal. Philosophers have better intuitions than average students at a bus stop’ but these experiments suggest that philosophers are also susceptible to hidden and surprising bias. For someone like yourself, thinking about logic and truth, does this strike you as being important?

GS:I think the question of whether and how to use intuitions in philosophy is a very important question. Vaesen and Peterson claim that philosophers’ intuitions are not reliable since they’ve been shown to vary according to external parameters like language. Their experiment shows that American and German philosophers, for example, have different intuitions with respect to whether Gettier’s counter-examples are genuine counter-examples to the view that knowledge is justified true belief. As an English-speaking philosopher whose mother-tongue his Hebrew, I am not surprised by their result. The question, for me, is whether such results justify demanding/recommending that philosophers stop or at least minimize their appeal to intuitions.

My own approach to the use of intuition in philosophy is a bit broader. There are at least two questions here: (1) What is the value of particular philosophical intuitions, that is, intuitions concerning particular cases? (2) What is the value of general philosophical intuitions as a starting point of a philosophical investigation?

I myself am quite skeptical about the reliability of my own intuitions concerning particular cases, so I try to minimize my appeal to such intuitions, and I must say I am also hesitant to rely on other philosophers’ intuitions about such cases. But others find them useful, for example, in epistemology and moral theory. Intuitions of this kind, however, are not subject to substantial friction requirements. So they need to be used with great care. Using them as part of the reflective-equilibriummethod is better, since there we are required to check them against general principles (which themselves will be checked against particular cases, until a reflectiveequilibrium is achieved). Here, I think, Vaesen and Peterson’s experiments are also relevant. One important lesson from their experiments is that we should try to avoid the use of particular intuitions in cases that are especially susceptible to linguistic and other biases.

The use of general intuitions as a starting point for a philosophical investigation is different in kind. This use, too, requires care, but the holistic-foundational methodology and the dynamic model of knowledge I talked about earlier provide us with tools for handling such intuitions. Using the starting assumptions to make some progress in our knowledge, we can turn back and rationally examine them in light of our expanded knowledge, the new tools we have created based on this knowledge, and so on. At some point we will either discard the original assumptions or accept them - as they are or in a revised form - on grounds that exceed mere uncritical intuition.

3:AM:There’s a great deal of discussion about the sexism of some academic philosophy. As a top woman philosopher of logic and truth how aware have you been of this? There are various branches of feminist philosophy: is there feminist logic? I guess one argument from your intricate pluralism that might support this would be that gender might be a factor in correspondence routes. And again, there might be experimental data that could provide support for this. What are your thoughts about this?

GS:Concerning the status of women in philosophy, my sense is that there’s generally good will on most people’s part, but much more has to be done. There’s need for improvement in such areas as the number of female students and faculty, faculty rank and salary, participation in conferences, keynote addresses and other special addresses, representation in journal editorial boards, and more. It’s not simple to achieve positive results in this area, but we must keep trying.

As for feminist logic and your novel idea of gender-determined routes of correspondence in the context of pluralism, there are of course many interesting perspectives on logic and truth, and this might be one of them. From the perspective my own work has focused on, one central question these possibilities raise is the balance between unity and disunity.

As I see it, there are indeed genuine pressures for pluralism in the case of truth, but the case of logic is different. One way to see this is to think of logical structure. Logical structure is an important factor in truth, but it’s only one factor and a highly specific factor at that. There is so much more to truth than logical structure! It’s true that logical structure is a universal factor, in the sense that it’s present in all fields of truth. But its breadth is not an indication of pluralism. One reason logical structure is universal is its narrowness and one-dimensionality, which render it homogeneous. It’s completely oblivious to the diversity of subject-matters of statements and theories (it’s topic neutral), and that’s the reason it can be present in all types of truth and have the same impact on all, which makes it universal.

Now, there are many interesting ways to generalize and expand our conception of logical structure, and there are many disagreements among philosophers about its parameters (logical constants) and the principles governing them, but that’s not pluralism. So from the perspective from which I study logic there’s not much room for pluralism in logic, although from other perspectives there might be. The case of (general) truth and correspondence is more delicate. Here, too, I would seek as much unity as possible (though without sacrificing genuine differences), so a proposal for a new source of diversity would have to be carefully examined. Since I’ve not examined the gender possibility thoroughly, however, I can’t say much about it now.

3:AM:OK, so you’ve talked about many details about your theories of truth and logic. So if you were to give a general statement about what truth and logic are, and what you think will be the big developments in the next decade pertaining to these two things, what would you say? In particular, what do you think is at stake?

GS:It’s hard to say what truth and logic are in a few lines, so whatever I say is bound to be unsatisfactory. But I will try (without being too repetitive and with bringing in a few new things). Truth, as I see it, is first of all a normative standard for thoughts (statements, theories, etc.) and only secondarily a property of thoughts. For truth to emerge in human life, three conditions have to be satisfied. The first condition I call “immanence”. It says that for a standard of truth to be needed in our life, we have to be creatures who engage in cognitive activities that direct themselves at the world, as we do when we stand inside a theory. (Ironically, to stand inside a theory is to look at something external, its subject-matter, the world.)

The second condition is “transcendence”. We have to step outside our theories, or more generally our immanent thoughts, and find a standpoint from which we can see both these thoughts and the world they are directed at. This transcendence is human transcendence rather than a “God’s eyeview” transcendence. Transcending is something like ascending to a Tarskian meta-language or moving to a higher standpoint on Neurath’s boat. The third and last condition is normativity, or a special kind of normativity. Standing outside our theories we must be interested in examining them critically with respect to their correctness, that is, with respect to whether what they say about the world is the case. So truth requires a multi-dimensional cognitive life, involving immanence, transcendence, and critical normative questions.

Now, the normative question of truth concerns the relation between what a sentence says and the way the world is; as such it is a correspondence question. However, since the range of immanent thoughts is extremely broad and diverse, a correspondence standard of truth that would apply to all these thoughts is, or may in principle be, a family of standards. The different members of this family are connected to each other through their core principle of correspondence, but correspondence itself allows a family of routes from our thoughts to reality - some relatively direct, others composite.

We may say that the standard of truth is a dynamic standard with a fixed core. Its dynamic nature has two dimensions: change from field to field, and change in time. Our standard of truth changes over time in accordance with changes in our understanding of the world, of the conditions that have to hold in the world so that a given thought is correct. But the core of truth - immanence, transcendence, special type of normativity, correspondence, etc. - remains the same. This explains how our judgements of truth change in the course of history while still remaining judgments of truth (how sentences about, say, atoms that satisfied the ancient’s standard of truth don’t satisfy ours). This aspect is especially important for understanding the role truth plays in our changing body of knowledge. (Of course, much more has to be said.)

Turning to logic: logic, as I see it, is both a branch of knowledge and a powerful method of inference. As a method of inference logic exploits certain systematic connections between language and the world - connections between certain structural features of sentences and certain structural (formal) features of the world - to devise a method for an especially strong type of inference. As a branch of knowledge, logic studies the conditions under which such inference holds. Since these conditions are formal in nature, logic is grounded in formal facets of the world.

More specifically, logic is grounded in laws governing formal features of objects and properties in the world (like being a complement, being in an intersection, being identical to, being a non-empty property, having a certain cardinality, and so on), features that are represented in language by logical constants. But these features and their laws are the subject-matter of mathematics, so the formal can be seen to ground not only logic but also mathematics. So this conception of logic is connected with (or gives rise to) a new type of “logicism”: a joint foundation for logic and mathematics in the formal. (Of course, this foundation is holistic.) Much is missing in this brief outline, but in a nutshell this is what logic is, on my theory.

As for what will the big developments in the next decade be in the philosophy of logic and truth, it’s very hard to predict. From my perspective, I see opportunities in these fields that are connected to more general opportunities in philosophy. So let me say something about these. Philosophy in the second part of the 20th century was to a large extent shaped by Quine’s two revolutions: the revolution in mid-century that rejected the traditional philosophical bifurcations and replaced foundationalism by holism, and the naturalist revolution a decade or two later. While the naturalist revolution, which took hold of philosophy in recent decades, diverted philosophers’ attention from the classical questions of philosophy, my sense is that many in the profession would like to return to these questions, that is, ifthis could be done without returning to the old ways of pursuing them (which lead to serious problems). The possibility of renewing the foundational projects of philosophy on a new basis - a holistic basis, without the traditional bifurcations, but also without Quine’s empiricism and its inherent limitations - seems to me to hold a great promise. Some important steps in this direction have already been taken by various philosophers. Much is at stake - not in the long run (since if not now, philosophers might do this sometime in the future), but for us in the present, who care about how philosophy develops in our time.

3:AM:And finally, for the smart readers at 3:AM Magazine, are there five books you could recommend that you have found illuminating, not necessarily directly, for your thinking into these matters?

GS:Finding five non-philosophy books that illuminated my thinking on knowledge, truth, and logic is a difficult task. But let me mention two ways in which things outside philosophy, especially in the arts (novels, films, plays, dance, visual art), are important to my work in philosophy. First, novels, movies, etc., stimulate the imagination, which is very important for doing philosophy: for drawing new connections, for contemplating new concepts, for seeing things from new perspectives, and so on. Second, they give us a sense of a very deep kind of truth, different from the kinds of truth I have investigated so far, but one that I hope to incorporate into my study of truth in the future. Since recently these two types of stimulation have often come to me from the movies, I hope you don’t mind if I recommend to your readers five recent movies that I found extremely exciting in one or another of these respects (or both). It’s hard to choose, but here they are:

1. Waltz with Bashir(Ari Folman, 2008)
2. Tango(Carlos Saura, 1998)
3. Of Gods & Men(Xavier Beauvois, 2010)
4. The Diving Bell & The Butterfly(Julian Schnabel, 2007)
5. The Secret in Their Eyes(Juan Jose Campanella, 2009)

Richard Marshallis still biding his time.