Progress in Mathematics and Logic — IV

paradoxes[for a brief explanation of this ongoing series, as well as a full table of contents, go here]

A panoply of logics

The striking thing about contemporary logic is that it is plural. Indeed, Logics (Nolt 1996) was the title of the book I used as a graduate student at the University of Tennessee, and that in itself was a surprise for me, since I had naively always thought of logic as a single, monolithic discipline. (But why, really? We have different ways of doing geometry and mathematics, and certainly a plethora of natural sciences!). The following brief look at modern logic should be enough to convince readers that the field is both vibrant and progressive, in the sense discussed above.

Consider, for instance, modal logic (Garson 2009), which deals with the behavior of sentences that include modal qualifiers, such as “necessarily,” “possibly,” and so on. It is actually a huge field within logic, as it includes deontic (“obligatory,” “permitted,” etc.: Hilpinen 1971), temporal (“always,” “never”), and doxastic (“believes that”) logics. Arguably the most familiar type of modal logic is the so-called K-logic (named after influential 20th century philosopher Saul Kripke), which is a compound logical system made of propositional logic, a necessitation rule (using the operator “it is necessary that”) and a distribution axiom. A stronger system, called M, is built by adding the axiom that whatever is necessary is the case, and this in turn yields an entire family of modal logics with different characteristics. Starting again with K-logic, one can also begin to build a deontic logic by adding an axiom that states that if x is obligatory, then x is permissible. Similarly, one can build temporal logics (with the details depending on one’s assumptions about the structure of time) and other types of modal logics. Indeed, Garson (2009) presents an elegant “map” of the relationships among a number of modal logics, demonstrating how they can all be built from K by adding different axioms. Furthermore, Lemmon and Scott (1977) have shown that there is a general parameter (G) from the specific values of which one can derive many (though not all) the axioms of modal logic. Garson’s connectivity map  and Lemmon and Scott’s G parameter are both, well, highly logical, and aesthetically very pleasing, at the least if you happen to have a developed aesthetic sensibility about logical matters. (Notice, once again, that these are all nice examples of non-teleonomic progress propelled by internally generated problems and consisting in exploring additional possibilities in a broad conceptual space that is evoked once one adds axioms or assumptions.)

A major issue with modal logic is that — unlike classical logic (Benthem 1983) — it is not possible to use truth tables to check the validity of an argument, for the simple reason that nobody has been able to come up with truth tables for expressions of the type “it is necessary that” and the like. The accepted solution to this problem (Garson 2009) is the use of possible worlds semantics (Copeland 2002), where truth values are assigned for each possible world in a given set W. Which implies that the same proposition may be true in world W1, say, but false in world W2. Of course one then has to specify whether W2, in this example, is correctly related to W1 and why. This is the sort of problem that has kept modal logicians occupied for some time, as you might imagine. Modal logic has a number of direct philosophical applications, as in the case of deontic logic (McNamara 2010), which deals with crucial notions in moral reasoning, such as permissible / impermissible, obligatory / optional, and so forth. Deontic logic has roots that go all the way back to the 14th century, although it became a formal branch of symbolic logic in the 20th century. As mentioned, deontic logic can be described in terms of Kripke-style possible worlds semantics, which allows formalized reasoning in metaethics (Fisher 2011).

Of particular interest, but for different reasons, are also the next two entries in our little catalog: many-valued and “fuzzy” logics. The term many-valued logic actually refers to a group of non-classical logics that does not restrict truth values to two possibilities (true or false; see Gottwald 2009). There are several types of many-valued logics, including perhaps most famously Łukasiewicz’s and Gödel’s. Some of these admit of a finite number of truth values, others of an infinite one. The Dunn/Belnap’s 4-valued system, for instance, has applications in both computer science and relevance logic. Multi-valued logic also presents aspects that reflect back on philosophical problems, as in the area of concepts of truth (Blackburn and Simmons 1999), or in the treatment of certain paradoxes (Beall 2003), like the heap and bald man ones. More practical applications are found in areas such as linguistics, hardware design and artificial intelligence (e.g., for the development of expert systems), as well as in mathematics. Fuzzy logic is nested within many-valued logics (Zadeh 1988; Hajek 2010), consisting of an approach that makes possible to analyze vagueness in both natural language (about degrees of beauty, age, etc.) and mathematics. The basic idea is that acceptable truth values range across the real interval [0,1], rather than lying only at the extremes (0 or 1) of that interval, as in classical logic. Fuzzy logic deals better than two-valued logic with the sort of problems raised by Sorites paradox, since these problems are generated in situations in which small/large, or many/few quantifiers are used, rather than simple binary choices. Therefore, fuzzy logic admits of things being almost true, rather than true, and it is for this reason that some authors have proposed that fuzzy logic can be thought of as a logic of vague notions.

A whole different way of thinking is afforded by so-called intuitionistic logic (Moschovakis 2010), which treats logic as a part of mathematics (as opposed to being foundational to it, as in the old Russell-Whitehead approach that we have discussed above), and — unlike mathematical Platonism (Linnebo 2011) — sees mathematical objects as mind-dependent constructs. Important steps in the development of intuitionistic logic were Gödel’s (who, ironically, was a mathematical Platonist!) proof (in 1933) that it is as consistent as classical logic, and Kripke’s formulation (in 1965) of a version of possible-worlds semantics that makes intuitionistic logic both complete and correct. Essentially, though, intuitionistic logic is Aristotelian logic without the (much contested) law of the excluded middle, which was developed for finite sets but was then extended without argument to the case of infinities.

Finally, a couple of words on what some consider the cutting edge, and some a wrong turn, in contemporary logic scholarship: paraconsistent and relevance logics. Paraconsistent logic is designed to deal with what in the context of classical logic are regarded as paradoxes (B. Brown 2002; Priest 2009). A paraconsistent logic is defined as one whose logical relations are not “explosive,” [12] with the classical candidate for treatment being the liar paradox: “This sentence is not true.” The paraconsistent approach is tightly connected to a general view known as dialethism (Priest et al. 2004), which is the idea that — contra popular wisdom — there are true contradictions, as oxymoronic as the phrase may sound. Paraconsistent logic, perhaps surprisingly, is not just of theoretical interest, as it turns out to have applications in automated reasoning: since computer databases will inevitably include inconsistent information (for instance because of error inputs), paraconsistent logic can be deployed to avoid wrong answers based on hidden contradictions. Similarly, paraconsistent logic can be deployed in the (not infrequent) cases in which people hold to inconsistent sets of beliefs, sometimes even rationally so (in the sense of instrumental rationality). More controversially, proponents of paraconsistent logic argue that it may allow us to bypass the constraints on arithmetic imposed by Gödel’s theorems. How does this work? One approach is known as “adaptive logic,” which begins with the idea that consistency is the norm, unless proven otherwise, or alternatively that consistency should be the first approach to a given sentence, with the alternative (inconsistency) being left as a last resort. That is to say, classical logic should be respected whenever possible. Paraconsistent logics can be generated using many-valued logic, as shown by Asenjo (1966), and the simplest way to do this is to allow a third truth value (besides true and false), referred to as “indeterminate” (i.e., neither true nor false).

Yet another approach within the increasingly large family of paraconsistent logics it to adopt a form of relevance logic (Mares 2012), whereby one stipulates that the conclusion of a given instance of reasoning must be relevant to the premises, which is one way to block possible logical explosions. According to relevance logicians what generates apparent paradoxes is that the antecedent is irrelevant to the consequent, as in: “The moon is made of green cheese. Therefore, either it is raining in Ecuador now or it is not,” which is a valid inference in classical logic (I know, it takes a minute to get used to this). The typical objection to relevance logic is that logic is supposed to be about the form, not the content, of reasoning — as we have seen when briefly examining the history of both Western and Eastern logic — and that by invoking the notion of “relevance” (which is surprisingly hard to cash out, incidentally) one is shifting the focus at least in part to content. Mares (1997), however, suggests that a better way to think about relevance logic comes with the realization that a given world X (within the context of Kripe-style many worlds) contains informational links, such as laws of nature, causal principles, etc. It is these causal links, then, that are deployed by relevance logicians in order to flesh out the notion of relevance (in a given world or set of possible worlds) — perhaps reminiscent of Dharmakīrti’s invoking of causal relations to assure the truth of a first premiss, as we have discussed. Interestingly, some approaches to relevance logic (e.g., Priest 2008) can be deployed to describe the difference between a logic that applies to our world vs a logic that applies to a science fictional world (where, for instance, the laws of nature might be different), but relevance logic has a number of more practical applications too, including in mathematics, where it is used in attempts at developing mathematical approaches that are not set theoretical, as well as in the derivation of deontic logics, and in computer science (development of linear logic).

So, what do we get from the above historical and contemporary overviews of logic, with respect to how the field makes progress? The answer is, I maintain, significantly different from what we saw for the natural sciences in Chapter 4, but not too dissimilar from the one that emerged in the first part of this chapter in the case of mathematics. Logicians explore more and more areas of the conceptual space of their own discipline. Beginning, for instance, with classical two-valued logic it was only a matter of time before people started to consider three-valued and then multi-valued, and finally infinitely-valued (such as fuzzy) types of logical systems. Naturally enough, this progress was far from linear, with some historians of logic identifying three moments of vigorous activity during the history of Western logic, with the much maligned Middle Ages not devoid of interesting developments in the field. Other signs of progress can readily be seen in the expansion of the concerns of logicians, beginning with Aristotelian syllogisms or similar constructs (as in the parallel developments of the Stoics) and eventually exploding in the variety of contemporary approaches, including deontic, temporal, doxastic logics and the like. Even dialethism and the accompanying paraconsistent and relevance logics can be taken as further explorations of a broad territory that began to be mapped by the ancient Greeks: once you chew for a while (in this case, a very long while!) on the fact that classical logic can yield explosions and paradoxes, you might try somewhat radical alternatives, like biting the bullet and treating some paradoxes as “true,” or pushing for the need for a three-valued approach when considering paradoxes. As in the case of mathematics, there have been plenty of practical (i.e., empirical) applications of logic, which in themselves would justify the notion that the field has progressed. But in both mathematics and logic I don’t think the empirical aspect is quite as crucial as in the natural sciences. In science there is a good argument to be made that if theory loses contact with the empirical world (Baggott 2013) then it is essentially not science any longer. But in mathematics and logic that contact is entirely optional as far as any assessment of whether those fields have been making progress by the light of their own internal standards and how well they have tackled the problems that their own practitioners set out to resolve.

What about philosophy, then? It is finally to that field that I turn next, examining a number of examples of what I think clearly constitutes progress in philosophical inquiry. As we shall see, however, the terrain is more complex and perilous. More complex because while philosophy is very much a type of conceptual activity, sharing in this with mathematics and logic, it also depends on input from the empirical world, both in terms of commonsense and of scientific knowledge. More perilous because a relatively easy case could be made that a significant portion of philosophical meanderings aren’t really progressive, and some even smack of mystical nonsense. Unfortunately, this sort of pseudo-philosophy does appear in the philosophical literature, side by side with the serious stuff, and occasionally is even the result of the writings of the same authors! Nothing like that appears to be happening either in science or in mathematics and logic, which I believe is a major reason why philosophy keeps struggling to be taken seriously in the modern academic world.


[12] A logical explosion is a situation where everything follows from a contradiction, which is possible within the standard setup of classical logic.


Asenjo, F.G. (1966) A Calculus of Antinomies. Notre Dame Journal of Formal Logic 7:103-105.

Baggott, J. (2013) Farewell to Reality: How Modern Physics Has Betrayed the Search for Scientific Truth. Pegasus.

Beall, J.C. (ed.) (2003) Liars and Heaps. Clarendon Press.

Benthem, J.F. van (1983) Modal Logic and Classical Logic Bibliopolis.

Blackburn, S. and Simmons, K. (eds.) (1999) Truth. Oxford University Press.

Brown, B. (2002) On Paraconsistency. In: D. Jacquette (ed.), A Companion to Philosophical Logic, Blackwell, pp. 628-650.

Copeland, B.J. (2002) The genesis of possible worlds semantics. Journal of Philosophical Logic 31:99-137.

Fisher, A. (2011) Metaethics: An Introduction. Acumen Publishing.

Garson, J. (2009) Modal Logic. Stanford Encyclopedia of Philosophy (accessed on 31 May 2013).

Gottwald, S. (2009) Many-valued logic. Stanford Encyclopedia of Philosophy (accessed on 22 August 2013).

Hajek, P. (2010) Fuzzy logic. Stanford Encyclopedia of Philosophy (accessed on 30 March 2011).

Hilpinen, R. (1971) Deontic Logic: Introductory and Systematic Readings D. Reidel.

Lemmon, E. and Scott, D. (1977) An Introduction to Modal Logic. Blackwell.

Linnebo, Ø. (2011) Platonism in the philosophy of mathematics. Stanford Encyclopedia of Philosophy  (accessed on 11 October 2012).

Mares, E.D. (1997) Relevant Logic and the Theory of Information. Synthese 109:345–360.

Mares, E. (2012) Relevance logic. Stanford Encyclopedia of Philosophy (accessed on 28 August 2013).

McNamara, P. (2010) Deontic logic. Stanford Encyclopedia of Philosophy (accessed on 22 August 2013).

Moschovakis, J. (2010) Intuitionistic logic. Stanford Encyclopedia of Philosophy (accessed on 22 August 2013).

Nolt, J. (1996) Logics. Cengage Learning.

Priest, G. (2008) An Introduction to Non-Classical Logic: From If to Is. University of Cambridge Press.

Priest, G. (2009) Paraconsistent logic. Stanford Encyclopedia of Philosophy (accessed on 14 September 2012).

Priest, G., Beall, J.C. and Armour-Garb, B. (eds.) (2004) The Law of Non-Contradiction. Oxford University Press.

Zadeh, L.A. (1988) Fuzzy logic. Computer 21:83-93.


135 thoughts on “Progress in Mathematics and Logic — IV

  1. Hi Robin

    “Really, they believe they can make a reasoned argument without being able to separate reason and desire at all?”

    They are not making an argument or doing philosophy. They ran a science experiment. Their desire to do science and to investigate human nature influenced their decision to run an experiment. After that the science took over. Science is designed to block human bias. Controls, double blind studies and all of that. The information from the science, the test data is not an opinion so it is not influenced by any desire other than the desire of the scientists to do science in the first place.

    “Will you grant me that the rest of us might also be able to somewhat separate reason and desire to the extent that they do?”

    As I said we all try to separate them as much as possible. Studies seem to indicate it’s virtually impossible. But having this information is enlightening in such a way as to enable us to notice our intuitional bias more than we ever did thus giving us the best possible chance to be as reasoned as possible. That’s all any of us can hope for. But Kant thought he could completely sever the two. We know better than that now. For now. We know know what seems to be so as far as we know.

    Liked by 2 people

  2. Hi garthdaisy,

    “They are not making an argument or doing philosophy. They ran a science experiment. Their desire to do science and to investigate human nature influenced their decision to run an experiment. After that the science took over. Science is designed to block human bias. Controls, double blind studies and all of that. The information from the science, the test data is not an opinion so it is not influenced by any desire other than the desire of the scientists to do science in the first place.”

    Sorry, but this is simply not true. The scientific method leaves plenty of room for the data to be influenced by desire, the starting assumptions, the way the information is collected, the way it is analysed and collated, the way it is reported, and in so many other things.

    The “spotlight among the voxels” method, so popular in neuroscience, is particularly well suited to wish fulfilment.

    If you begin science with the idea that the data cannot be influenced by any other desire than the desire of scientists to do science, then you are almost guaranteed to be about to do bad science.

    Liked by 1 person

  3. Robin,

    “The scientific method leaves plenty of room for the data to be influenced by desire,”

    Well not on purpose. It does everything it can, which is pretty good. It does better than pure human reasoning alone. But yes, desire will affect any human reasoning including that which is done in science. I’m not sure why you think this means we must disregard or reject the findings. We just factor in that pesky human desire variable. What else can we do? Chuck science and reason? Deny desire?

    Liked by 1 person

  4. Synred,

    Yes, I know Bob Crease, he was Chair of the Philosophy Department at Stony Brook while I was a faculty in the Ecology & Evolution Department there, in the early ’00’s. Nice, reasonable guy.


    # They are not making an argument or doing philosophy. They ran a science experiment #

    Boy, where is Coel when one needs him? There is no sharp distinction between science experiments and arguments, in that science is not neutrally descriptive, one doesn’t just look at tables of data, one interprets them and presents conclusions. I.e., scientists make arguments.

    # Science is designed to block human bias #

    But it doesn’t. At best it reduces it.

    Liked by 1 person

  5. Boy, where is Coel when one needs him?

    Having marmalade on toast for breakfast! 🙂

    Hi Robin,

    If you begin science with the idea that the data cannot be influenced by any other desire than the desire of scientists to do science, then you are almost guaranteed to be about to do bad science.

    Science proceeds by assuming that all humans are horribly biased. A double-blind controlled trial says to a doctor that they are so biased that they can’t even be trusted with knowing whether the phial in their hand is the drug or a placebo.

    In an example from my own field, take developing instruments for a new astrophysics satellite. Data-processing software designed to detect certain signals is developed by one university group. To test it, fake signals are injected into data by a second university group. Then a third group in the collaboration uses the software to try to find the signals.

    All of that is an attempt to reduce bias. A recent outcome was that software that the first group had reckoned to be good for finding 80% of the signals was found by the third group to only find 30%.

    Of course one can never eliminate bias, but by accepting that all scientists are prone to such bias one can try to use methods that reduce it as much as possible.

    Then, the “gold standard” test of whether a scientific idea is good is whether it can make predictions about things that the scientists didn’t already know. Again, that reduces the role of human judgement.

    Much of science is a set of quality-control measures intended to overcome human foibles, wishful thinking and biases. The evidence seems to be that by such methods science largely does succeed in generating sufficiently unbiased knowledge.

    But it’s then a fair question about other fields that are not so entwined with data about how they do their quality control.

    Liked by 1 person

  6. Massimo,

    “But it doesn’t. At best it reduces it.”

    As much as possible. Well beyond human reasoning alone. It’s the best method we have for reducing human bias, isn’t it?

    Liked by 1 person

  7. Couvant2104, I nor Popper nor any “critical rationalist” would disagree with what you said. Maybe check out the link I posted earlier to what Rafe Champion wrote. David Miller has two books, Critical Rationalism …, Out of Error …
    Popper was fully award of Duhem and addresses this in the 1930’s. He is not a naive falsificationist.


  8. Some fields are more biased than others. Drug trials are notoriously biased. The human-all-to-human biases are all too obvious. See, e.g., Bad Pharma

    Physics is somewhat less biased. but still has to be careful. Theories need to compete and experiments repeated. A looming problem is cost. As experiments become more expensive like the LHC and satieties, it becomes every more difficult to repeat them. At LHC there are several independent experiments, but at the next machine (if there is one), it may not be possible. To some extent there are internal checks (e.g., as Coel describes) within an experiment by different groups analyzing the same question and testing each other, but it’s not ideal. They’re using the same data and usually the same ‘reconstruction’ software.

    The essential assumption is reality. If we are all probing the same thing, we can hope to converge on something approximating it. It seems to work though sometimes it takes a long time.

    The post-modern perspective misses the point of science. it is because scientist suffer from greed, power hunger, vanity, etc. (‘human nature’) that we test against reality — to keep from wondering off into never-never land.

    Liked by 2 people

  9. Hi garthdaisy

    No I don’t think we should give up reasoning, but if it were actually impossible to separate reason and desire then that is just what we should do.

    But, as I said before, for any of these scientists who will tell us of the impossibility or extreme difficulty of doing this, then to the extent that they believe they have produced meaningful research, to the same extent they believe that reason and desire can be separated.

    It is like Alex Rosenberg saying that if linguistic meaning is anything at all, it sis nothing but a set of nested desires and beliefs the speaker has about his listeners and himself.

    I wonder if he believed that the essay he was busy writing was either meaningless of just a nested set of desires and beliefs about himself and his readers, and if so why he would even bother writing it in the first place..


  10. Also, I think that some here have unintentionally given the impression that Kant thought that we had a ready made faculty for reliable reasoning. He didn’t, he wrote at length about the pitfalls of reasoning and the many delusions to which we lead ourselves, and his ambition was to develop a way of avoiding them.

    He knew well about desires leading beliefs, he even pointed out that the desire for knowledge itself is likely to lead us to delusion.


  11. As well as the philosophical aspects, though they all inject various biases.
    To distill information from its context is to create bias.
    Turtles all the way down.


  12. Hi Coel,

    “Science proceeds by assuming that all humans are horribly biased.”

    That us indeed the ideal. But do you really think that happens in real life science even most of the time?


  13. Hi Robin,.

    But do you really think that happens in real life science even most of the time?

    To a large extent, yes. For example the need for experimental controls and rigorous statistical evaluation of results is accepted, and, further than that, plenty of scientists are continually checking and criticising what other scientists do. Of course all actual science falls short of the ideal, but plenty of it is good enough.

    (Some areas of science may, of course, be worse than others, but I’m talking here about how things seem in physics.)

    Hi Massimo,

    No, I don’t think science is a method to reduce bias. It’s a method to discover things about the world.

    I presume that you meant something along the lines that science is about discovering things about the world, and as part of that many of science’s methods are attempts to reduce human bias? (Rather than meaning that science’s methods do not reduce human bias.)


  14. Coel,

    # Rather than meaning that science’s methods do not reduce human bias #

    No, I meant to say that the sort of biases science is inherently equipped to deal with are the minor ones. Double blinds, for instance, take care of the obvious immediate bias of an investigator favoring a pet theory. But science as an empirical activity has no way to correct the broader biases, like those that enter into deciding what is worth investigating, funding, etc.. For those, Longino’s ideas of science as structured social knowledge are more illuminating.

    Liked by 1 person

Comments are closed.