This is an excerpt from my book-in-progress on diverse elementary topics in logic, from the chapter on model theory. My view is that Boolean-valued models should be elevated to the status of a standard core topic of elementary model theory, adjacent to the ultrapower construction. The theory of Boolean-valued models is both beautiful and accessible, both generalizing and explaining the ultrapower method, while also partaking of deeper philosophical issues concerning multi-valued truth.

Boolean-valued models

Let us extend our familiar model concept from classical predicate logic to the comparatively fantastic realm of multi-valued predicate logic. We seek a multi-valued-truth concept of model, with a domain of individuals of some kind and interpretations for the relations, functions, and constants from a first-order language, but in which the truths of the model are not necessarily black and white, but exhibit their truth values in a given multi-valued logic. Let us focus especially on the case of Boolean-valued models, using a complete Boolean algebra $\newcommand\B{\mathbb{B}}\B$, since this case has been particularly well developed and successful; the set-theoretic method of forcing, for example, amounts to using $\B$-valued models of set theory with carefully chosen complete Boolean algebras $\B$ and has been used impressively to establish a sweeping set-theoretic independence phenomenon, showing that numerous set-theoretic principles such as the axiom of choice and the continuum hypothesis are independent of the other axioms of set theory.

The main idea of Boolean-valued models, however, is applicable with any theory, not just set theory, and there are Boolean-valued graphs, Boolean-valued orders, Boolean-valued groups, rings, and fields. So let us develop a little of this theory, keeping in mind that the basic construction and ideas also often work fruitfully with other multi-valued logics, not just Boolean algebras.

Definition of Boolean-valued models

We defined in an earlier section what it means to be a model in classical first-order predicate logic—one specifies the domain of the model, a set of individuals that will constitute the universe of the model over which all the quantifiers will range, and then provides interpretations on that domain for each relation symbol, function symbol, and constant symbol in the language. For each relation symbol $R$, for example, and any suitable tuple of individuals $a,b$ from the domain, one specifies whether $R(a,b)$ is to hold or fail in the model.

The main idea for defining what it means to be a model in multi-valued predicate logic is to replace these classical yes-or-no atomic truths with multi-valued atomic truth assertions. Specifically, for any Boolean algebra $\B$, a $\B$-valued model $\mathcal{M}$ in the language $\mathcal{L}$ consists of a domain $M$, whose elements are called names, and an assignment of $\B$-valued truth values for all the simple atomic formulas using parameters from that domain:$\newcommand\boolval[1]{[\![#1]\!]}$\begin{eqnarray*} % \nonumber % Remove numbering (before each equation) s=t&\mapsto& \boolval{s=t} \\ R(s_0,\ldots,s_n) &\mapsto& \boolval{R(s_0,\ldots,s_n)} \\ y=f(s_0,\ldots,s_n) &\mapsto& \boolval{y=f(s_0,\ldots,s_n)} \end{eqnarray*} The double-bracketed expressions here $\boolval{\varphi}$ denote an element of the Boolean algebra $\B$—one thinks of $\boolval{\varphi}$ as the $\B$-truth value of $\varphi$ in the model, and the model is defined by specifying these values for the simple atomic formulas.

We shall insist in any Boolean-valued model that the atomic truth values obey the laws of equality: $$\begin{array}{rcl} \boolval{s=s} &=& 1\\ \boolval{s=t} &=& \boolval{t=s}\\ \boolval{s=t}\wedge\boolval{t=u} &\leq& \boolval{s=u}\\ \bigwedge_{i<n}\boolval{s_i=t_i}\wedge\boolval{R(\vec s)} &\leq & \boolval{R(\vec t)}.\end{array}$$And if the language includes functions symbols, then we also insist on the functionality axioms: $$\begin{array}{rcl} \bigwedge_{i<n}\boolval{s_i=t_i}\wedge\boolval{y=f(\vec s)} &\leq& \boolval{y=f(\vec t)}\\ \bigvee_{t\in M}\boolval{t=f(\vec s)} &=& 1\\ \boolval{t_0=f(\vec s)}\wedge\boolval{t_1=f(\vec s)} &\leq& \boolval{t_0=t_1}.\\ \end{array}$$ These requirements assert respectively in the Boolean-valued context that the equality axiom holds for function application, that the function takes on some value, and that the function takes on at most one value. In effect, the Boolean-valued model treatment of functions regards them as as defined by their graph relations, and these functionality axioms ensure that the relation has the features that would be expected of such a graph relation.

In summary, a $\B$-valued model is a domain of names together with $\B$-valued truth assignments for the simple atomic assertions about those names, which obey the equality and functionality axioms.

Boolean-valued equality

The nature of equality in the Boolean-valued setting offers an interesting departure from the usual classical treatment that is worth discussing explicitly. Namely, in classical predicate logic with equality, distinct elements of the domain are taken to represent distinct individuals in the model—the assertion $a=b$ is true in the model only when $a$ and $b$ are identical as elements of the domain. In the Boolean-valued setting, however, we want to relax this in order to allow equality assertions to have an intermediate Boolean value. For this reason, distinct elements of the domain can no longer be taken to represent definitely distinct individuals, since the equality assertion $\boolval{a=b}$ might have some nonzero or intermediate degree of truth. The elements of the domain of a Boolean-valued model are allowed in effect a bit of indecision or indeterminacy about who they are and whether they might be identical. This is why we think of the elements of the domain as names for individuals, rather than as the individuals themselves. Two different names can have an intermediate truth-value possibility of naming the same or different individuals.

Extending to Boolean-valued truth

By definition, every $\B$-valued model provides $\B$-valued truth assertions for the simple atomic formulas. We now extend these Boolean-valued truth assignments to all assertions of the language through the following natural recursion: $$\begin{array}{rcl} \boolval{\varphi\wedge\psi} &=& \boolval{\varphi}\wedge\boolval{\psi}\\ \boolval{\neg\varphi} &=& \neg\boolval{\varphi}\\ \boolval{\exists x\,\varphi(x,\vec s)} &=& \bigvee_{t\in M}\boolval{\varphi(t,\vec s)},\text{ if this supremum exists in }\B\\ \end{array}$$ On the right-hand side, we make the logical calculations in each case using the operations of the Boolean algebra $\B$. In the existential case, the supremum may range over an infinite set of Boolean values, if the domain of the model is infinite, and so this supremum might not be realized as an element of $\B$, if it is not complete. This is precisely why one commonly considers $\B$-valued models especially in the case of a complete Boolean algebra $\B$—the completeness of $\B$ ensures that this supremum exists and so the recursion has a solution.

The reader may check by induction on $\varphi$ that the general equality axiom now has Boolean value one. $$\boolval{\vec s=\vec t\wedge \varphi(\vec s)\implies\varphi(\vec t)}=1$$ The Boolean-valued structure $\mathcal{M}$ is full if for every formula $\varphi(x,\vec x)$ in $\mathcal{L}$ and $\vec s$ in $M$, there is some $t\in M$ such that $\boolval{\exists x\,\varphi(x,\vec s)}=\boolval{\varphi(t,\vec s)}$. That is, a model is full when it is rich enough with names that the Boolean value of every existential statement is realized by a particular witness, reducing the infinitary supremum in the recursive definition to a single largest element.

A simple example

For every natural number $n\geq 3$ let $C_n$ be the graph on $n$ vertices forming an $n$-cycle with edge relation $x\sim y$. So we have a triangle $C_3$, a square $C_4$, a pentagon $C_5$, and so on. Let $\B=P(\set{n\in\newcommand\N{\mathbb{N}}\N\mid n\geq 3})$ be the power set of these indices, which forms a complete Boolean algebra. We define a $\B$-valued graph model $\mathcal{C}$ as follows. We take as names the sequences $a=\langle a_n\rangle_{n\geq 3}$ with $a_n\in C_n$ for every $n$.

We define the equality and $\sim$ edge relations on these names as follows: \begin{eqnarray*} \boolval{a=b}&=&\set{n\mid a_n=b_n} \\ \boolval{a\sim b}&=&\set{n\mid C_n\newcommand\satisfies{\models}\satisfies a_n\sim b_n}. \end{eqnarray*} Thus, two names are equal exactly to extent that they have the same coordinates, and two names are connected by the edge relation $\sim$ exactly to the extent to that this is true on the coordinates. It is now easy to verify the equality axioms, since $a\sim b$ is true to at least the extent that $a’\sim b’$, $a=a’$, and $b=b’$ are true, since if $a_n=a’_n$, $b_n=b’_n$, and $a’_n\sim b’_n$ in $C_n$, then also $a_n\sim b_n$. So this is a $\B$-valued graph model.

So we have a Boolean-valued graph. Is it connected? Does that question make sense? Of course, we don’t have an actual graph here, but only a $\B$-valued graph, and so in principle we only know how to compute the Boolean value of statements that are expressible in the language of graph theory. Since connectivity is not formally expressible, except in the bounded finite cases, this question might not be sensible.

Let me argue that it is indeterminate whether this graph is a complete graph with every pair of distinct nodes connected by an edge. After all, $C_3$ is the complete graph on three vertices, and it will follow from the theorem below that the Boolean value of the statement $\forall x,y\,(x=y\vee x\sim y)$ contains the element $3$ and therefore this Boolean value is not zero. Meanwhile, the assertion that the graph is not complete, however, also gets a nonzero Boolean value, since every $C_n$ except $C_3$ has distinct nodes with no edge between them. In a robust sense, the graph-theoretic truths of $\mathcal{C}$ combine the truths of all the various graphs $C_n$.

Note also that as $n$ increases, the graphs $C_n$ have nodes that are increasingly far apart. Fix any name $\langle a\rangle_{n\geq 3}$ and choose $\langle b\rangle_{n\geq 3}$ such that the distance between $a_n$ and $b_n$ in $C_n$ is not bounded by any particular uniform bound. In $\mathcal{C}$, it follows that $a$ and $b$ have no particular definite finite distance, and this can be viewed as a sense in which $\mathcal{C}$ is not connected.

A combination of many models

Let us flesh out the previous example with a more general analysis. Suppose we have a family of models $\set{M_i\mid i\in I}$ in a common language $\mathcal{L}$. Let $\B=P(I)$ be the power set of the set of indices $I$, a complete Boolean algebra. We define a $\B$-valued model $\mathcal{M}$ as follows. The set of names will be precisely the product of the models $M=\prod_i M_i$, which is to say, the $I$-tuples $s=\langle s_i\mid i\in I\rangle$ with $s_i\in M_i$ for every $i\in I$, and the simple atomic truth values are defined like this: \begin{eqnarray*} \boolval{s=t}&=&\set{i\in I\mid s_i=t_i} \\ \boolval{R(s,\ldots,t)}&=&\set{i\in I\mid M_i\satisfies R(s_i,\dots,t_i)}\\ \boolval{u=f(s,\ldots,t)} &=& \set{i\in I\mid M_i\satisfies u_i=f(s_i,\dots,t_i)}. \end{eqnarray*} One can now prove the equality and functionality axioms, and so this is a $\B$-valued model.

Theorem. The Boolean-valued model $\mathcal{M}$ described above is full, and Boolean-valued truth can be computed coordinatewise: $$\boolval{\varphi(s,\dots,t)}=\set{i\in I\mid M_i\satisfies\varphi(s_i,\dots,t_i)}.$$

Proof. We prove this by induction on $\varphi$, for assertions using only simple atomic formulas, $\wedge$, $\neg$, and $\exists$. The simple atomic case is part of what it means to be a Boolean-valued model. If the theorem is true for $\varphi$, then it will be true for $\neg\varphi$, since negation in $\B$ is complementation in $I$, as follows: $$\boolval{\neg\varphi}=\neg\boolval{\varphi}=I-\set{i\in I\mid M_i\satisfies\varphi}=\set{i\in I\mid M_i\satisfies\neg\varphi}.$$ If the theorem is true for $\varphi$ and $\psi$, then it will be true for $\varphi\wedge\psi$ as follows: \begin{eqnarray*} \boolval{\varphi\wedge\psi} &=& \boolval{\varphi}\wedge\boolval{\psi} \\ &=& \set{i\in I\mid M_i\satisfies\varphi}\cap\set{i\in I\mid M_i\satisfies\psi}\\ &=& \set{i\in I\mid M_i\satisfies\varphi\wedge\psi}. \end{eqnarray*} For the quantifier case $\exists x\, \varphi(x,s,\dots,t)$, choose $u_i\in M_i$ for which $M_i\satisfies\varphi(u_i,s_i,\dots,t_i)$, if possible. It follows that $$\set{i\in I\mid M_i\satisfies\exists x\,\varphi(x,s_i,\dots,t_i)}=\set{i\in I\mid M_i\satisfies\varphi(u_i,s_i,\dots,t_i)},$$ and from this it follows that $\boolval{\varphi(v,s,\dots,t)}\leq\boolval{\varphi(u,s,\dots,t)}$, since whenever $v_i$ succeeds as a witness then so also will $u_i$. Consequently, the model is full, and we see that \begin{eqnarray*} \boolval{\exists x\,\varphi(x,s,\dots,t)} &=& \bigvee_{v\in M}\boolval{\varphi(v,s,\dots,t)}\\ &=& \boolval{\varphi(u,s,\dots,t)}\\ &=& \set{i\in I\mid M_i\satisfies\exists x\,\varphi(x,s_i,\dots,t_i)}, \end{eqnarray*} as desired. $\Box$

Boolean ultrapowers

Every Boolean-valued model can be transformed into a classical model, a $2$-valued model, by means of a simple quotient construction. Suppose that $\mathcal{M}$ is a $\B$-valued model for some Boolean algebra $\B$ and that $\mu\newcommand\of{\subseteq}\of\B$ is an ultrafilter on the Boolean algebra.

Define an equivalence relation $=_\mu$ on the set of names by $$a=_\mu b\quad\text{ if and only if }\quad \boolval{a=b}\in\mu.$$ The quotient structure $\mathcal{M}/\mu$ will consist of equivalence classes $[a]_\mu$ of names by this equivalence relation. We define the atomic truths of the quotient structure similarly by reference to whether these truths hold with a value in $\mu$. Namely, $$R^{\mathcal{M}/\mu}([a]_\mu,\ldots,[b]_\mu)\quad\text{ if and only if }\quad \boolval{R(a,\dots,b)}\in\mu.$$ For function symbols $f$, we define $$[c]_\mu=f([a]_\mu,\dots,[b]_\mu)\quad\text{ if and only if }\quad \boolval{c=f(a,\dots,b)}\in\mu.$$ These definitions are well-defined modulo $=_\mu$ precisely because of the equality axiom properties of the Boolean-valued model $\mathcal{M}$. For example, if $a=_\mu a’$, then $\boolval{a=a’}\in\mu$, but $\boolval{a=a’}\wedge\boolval{R(a)}\leq \boolval{R(a’)}$ by the equality axiom, and so if $\boolval{R(a)}$ is in $\mu$, then so will be $\boolval{R(a’)}$.

We defined atomic truth in the quotient structure by reference to the truth value being $\mu$-large. In fact, this reduction will continue for all truth assertions in the quotient structure, which we prove as follows.

Theorem. (Łoś theorem for Boolean ultrapowers) Suppose that $\mathcal{M}$ is a full $\B$-valued model for a Boolean algebra $\B$, and that $\mu\of\B$ is an ultrafilter. Then a statement is true in the Boolean quotient structure $\mathcal{M}/\mu$ if and only if the Boolean value of the statement is in $\mu$. $$\mathcal{M}/\mu\satisfies\varphi([a]_\mu,\dots,[b]_\mu)\quad\text{ if and only if }\quad\boolval{\varphi(a,\dots,b)}\in\mu.$$

Proof. We prove this by induction on $\varphi$. The simple atomic case holds by the definition of the quotient structure. Boolean connectives $\wedge$, $\neg$ follow easily using that $\mu$ is a filter and that it is an ultrafilter. Consider the quantifier case $\exists x\, \varphi(a,\dots,b,x)$, where by induction the theorem holds for $\varphi$ itself. If the quotient structure satisfies the existential $\mathcal{M}/\mu\satisfies\exists x\,\varphi([a],\dots,[b],x)$, then there is a name $c$ for which it satisfies $\varphi([a],\dots,[b],[c])$, and so by induction $\boolval{\varphi(a,\dots,b,c)}\in\mu$, in which case also $\boolval{\exists x\,\varphi(a,\dots,b,x)}\in\mu$, since this latter Boolean value is at least as great as the former. Conversely, if $\boolval{\exists x\,\varphi(a,\dots,b,x)}\in\mu$, then by fullness this existential Boolean value is realized by some name $c$, and so $\boolval{\varphi(a,\dots,b,c)}\in\mu$, which by induction implies that the quotient satisfies $\varphi([a],\dots,[b],[c])$ and consequently also $\exists x\,\varphi([a],\dots,[b],x)$, as desired. $\Box$

Note how fullness was used in the existential case of the inductive argument.

This will be a talk for the Department Colloquium of the Philosophy Department of the University of Notre Dame, 26 March 12 pm EST (4pm GMT).

Abstract: Potentialism is the view, originating in the classical dispute between actual and potential infinity, that one’s mathematical universe is never fully completed, but rather unfolds gradually as new parts of it increasingly come into existence or become accessible or known to us. Recent work emphasizes the modal aspect of potentialism, while decoupling it from arithmetic and from infinity: the essence of potentialism is about approximating a larger universe by means of universe fragments, an idea that applies to set-theoretic as well as arithmetic foundations. The modal language and perspective allows one precisely to distinguish various natural potentialist conceptions in the foundations of mathematics, whose exact modal validities are now known. Ultimately, this analysis suggests a refocusing of potentialism on the issue of convergent inevitability in comparison with radical branching. I shall defend the theses, first, that convergent potentialism is implicitly actualist, and second, that we should understand ultrafinitism in modal terms as a form of potentialism, one with surprising parallels to the case of arithmetic potentialism.

This will be an event for the $\Phi$-Math Reading Group at the Institute for Logic, Language, and Computation (ILLC) at the University of Amsterdam, 19 March 2021 6pm CET (5pm GMT). Zoom access here.

I shall make a brief presentation of the overall contents of the book, including a discussion of my perspective on the subject, and then get into some of the detailed issues with which the book engages. After this, we shall open up for discussion and comments.

I was so glad to be involved with this project of Hannah Hoffman. She had inquired on Twitter whether mathematicians could provide a proof of the irrationality of root two that rhymes. I set off immediately, of course, to answer the challenge. My wife Barbara Gail Montero and our daughter Hypatia and I spent a day thinking, writing, revising, rewriting, rethinking, rewriting, and eventually we had a set lyrics providing the proof, in rhyme and meter. We had wanted specifically to highlight not only the logic of the proof, but also to tell the fateful story of Hippasus, credited with the discovery.

Hannah proceeded to create the amazing musical version:

The diagonal of a square is incommensurable with its side
an astounding fact the Pythagoreans did hide

but Hippasus rebelled and spoke the truth
making his point with irrefutable proof

it’s absurd to suppose that the root of two
is rational, namely, p over q

square both sides and you will see
that twice q squared is the square of p

since p squared is even, then p is as well
now, if p as 2k you alternately spell

2q squared will to 4k squared equate
revealing, when halved, q’s even fate

thus, root two as fraction, p over q
must have numerator and denomerator with factors of two

to lowest terms, therefore, it can’t be reduced
root two is irrational, Hippasus deduced

as retribution for revealing this irrationality
Hippasus, it is said, was drowned in the sea

but his proof live on for the whole world to admire
a truth of elegance that will ever inspire.

This was a talk for the Kobe Set Theory Workshop, held on the occasion of Sakaé Fuchino’s retirement, 9-11 March 2021.

Abstract. I shall discuss senses in which set-theoretic forcing can be seen as a computational process on the models of set theory. Given an oracle for the atomic or elementary diagram of a model of set theory $\langle M,\in^M\rangle$, for example, one may in various senses compute $M$-generic filters $G\subset P\in M$ and the corresponding forcing extensions $M[G]$. Meanwhile, no such computational process is functorial, for there must always be isomorphic alternative presentations of the same model of set theory $M$ that lead by the computational process to non-isomorphic forcing extensions $M[G]\not\cong M[G’]$. Indeed, there is no Borel function providing generic filters that is functorial in this sense.

This is joint work with Russell Miller and Kameryn Williams.

Abstract. The principle of open determinacy for class games — two-player games of perfect information with plays of length ω, where the moves are chosen from a possibly proper class, such as games on the ordinals — is not provable in Zermelo-Fraenkel set theory ZFC or Gödel-Bernays set theory GBC, if these theories are consistent, because provably in ZFC there is a definable open proper class game with no definable winning strategy. In fact, the principle of open determinacy and even merely clopen determinacy for class games implies Con(ZFC) and iterated instances Con(Con(ZFC)) and more, because it implies that there is a satisfaction class for first-order truth, and indeed a transfinite tower of truth predicates $\text{Tr}_\alpha$ for iterated truth-about-truth, relative to any class parameter. This is perhaps explained, in light of the Tarskian recursive definition of truth, by the more general fact that the principle of clopen determinacy is exactly equivalent over GBC to the principle of elementary transfinite recursion ETR over well-founded class relations. Meanwhile, the principle of open determinacy for class games is strictly stronger, although it is provable in the stronger theory GBC+$\Pi^1_1$-comprehension, a proper fragment of Kelley-Morse set theory KM.

This is currently a draft version only of my article-in-progress on the topic of linearity in the hierarchy of consistency strength, especially with large cardinals. Comments are very welcome, since I am still writing the article. Please kindly send me comments by email or just post here.

This article will be the basis of the Weeks 7 & 8 discussion in the Graduate Philosophy of Logic seminar I am currently running with Volker Halbach at Oxford in Hilary term 2021.

I present instances of nonlinearity and illfoundedness in the hierarchy of large cardinal consistency strength—as natural or as nearly natural as I can make them—and consider philosophical aspects of the question of naturality with regard to this phenomenon.

It is a mystery often mentioned in the foundations of mathematics, a fundamental phenomenon to be explained, that our best and strongest mathematical theories seem to be linearly ordered and indeed well-ordered by consistency strength. Given any two of the familiar large cardinal hypotheses, for example, generally one of them will prove the consistency of the other.

Why should it be linear? Why should the large cardinal notions line up like this, when they often arise from completely different mathematical matters? Measurable cardinals arise from set-theoretic issues in measure theory; Ramsey cardinals generalize ideas in graph coloring combinatorics; compact cardinals arise with compactness properties of infinitary logic. Why should these disparate considerations lead to principles that are linearly related by direct implication and consistency strength?

The phenomenon is viewed by many in the philosophy of mathematics as significant in our quest for mathematical truth. In light of Gödel incompleteness, after all, we must eternally seek to strengthen even our best and strongest theories. Is the linear hierarchy of consistency strength directing us along the elusive path, the “one road upward” as John Steel describes it, toward the final, ultimate mathematical truth? That is the tantalizing possibility.

Meanwhile, we do know as a purely formal matter that the hierarchy of consistency strength is not actually well-ordered—it is ill-founded, densely ordered, and nonlinear. The statements usually used to illustrate these features, however, are weird self-referential assertions constructed in the Gödelian manner via the fixed-point lemma—logic-game trickery, often dismissed as unnatural.

Many set theorists claim that amongst the natural assertions, consistency strengths remain linearly ordered and indeed well ordered. H. Friedman refers to “the apparent comparability of naturally occurring logical strengths as one of the great mysteries of [the foundations of mathematics].” Andrés Caicedo says,

It is a remarkable empirical phenomenon that we indeed have comparability for natural theories. We expect this to always be the case, and a significant amount of work in inner model theory is guided by this belief.

Stephen G. Simpson writes:

It is striking that a great many foundational theories are linearly ordered by <. Of course it is possible to construct pairs of artificial theories which are incomparable under <. However, this is not the case for the “natural” or non-artificial theories which are usually regarded as significant in the foundations of mathematics. The problem of explaining this observed regularity is a challenge for future foundational research.

John Steel writes “The large cardinal hypotheses [the ones we know] are themselves wellordered by consistency strength,” and he formulates what he calls the “vague conjecture” asserting that

If T is a natural extension of ZFC, then there is an extension H axiomatized by large cardinal hypotheses such that T ≡ Con H. Moreover, ≤ Con is a prewellorder of the natural extensions of ZFC. In particular, if T and U are natural extensions of ZFC, then either T ≤ Con U or U ≤ Con T.

Peter Koellner writes

Remarkably, it turns out that when one restricts to those theories that “arise in nature” the interpretability ordering is quite simple: There are no descending chains and there are no incomparable elements—the interpretability ordering on theories that “arise in nature” is a wellordering.

Let me refer to this position as the natural linearity position, the assertion that all natural assertions of mathematics are linearly ordered by consistency strength. The strong form of the position, asserted by some of those whom I have cited above, asserts that the natural assertions of mathematics are indeed well-ordered by consistency strength. By all accounts, this view appears to be widely held in large cardinal set theory and the philosophy of set theory.

Despite the popularity of this position, I should like in this article to explore the contrary view and directly to challenge the natural linearity position.

Main Question. Can we find natural instances of nonlinearity and illfoundedness in the hierarchy of consistency strength?

I shall try my best.

You have to download the article to see my candidates for natural instances of nonlinearity in the hierarchy of large cardinal consistency strength, but I can tease you a little by mentioning that there are various cautious enumerations of the ZFC axioms which actually succeed in enumerating all the ZFC axioms, but with a strictly weaker consistency strength than the usual (incautious) enumeration. And similarly there are various cautious versions of the large cardinal hypothesis, which are natural, but also incomparable in consistency strength.

(Please note that it was Uri Andrews, rather than Uri Abraham, who settled question 16 with the result of theorem 17. I have corrected this from an earlier draft.)

This bried unpublished note (11 pages) contains an overview of the Gödel fixed-point lemma, along with several generalizations and applications, written for use in the Week 3 lecture of the Graduate Philosophy of Logic seminar that I co-taught with Volker Halbach at Oxford in Hilary term 2021. The theme of the seminar was self-reference, truth, and consistency strengths, and in this lecture we discussed the nature of Gödel’s fixed-point lemma and generalizations, with various applications in logic.

Gödel’s fixed-point lemma An application to the Gödel incompleteness theorem

Finite self-referential schemes An application to nonindependent disjunctions of independent sentences

Gödel-Carnap fixed point lemma Deriving the double fixed-point lemma as a consequence An application to the provability version of Yablo’s paradox

Kleene recursion theorem An application involving computable numbers An application involving the universal algorithm An application to Quine programs and Ouroborous chains

This is a graduate seminar in the Philosophy of Logic at the University of Oxford, run jointly by myself and Volker Halbach in Hilary Term 2021.

The theme will be self-reference, truth, and the hierarchy of consistency strength.

A detailed schedule, including the list of topics and readings is available on Volker’s web site.

The seminar will be held Fridays 9-11 am during term, online via Zoom at 812 2300 3837.

The final two sessions of term will be specifically on the hierarchy of consistency strength, based on my current article in progress concerning the possibility of natural instances of incomparability and ill-foundedness in the hierarchy of large cardinal consistency strength.

This will be a talk for a new mathematical logic seminar at the University of Warsaw in the Department of Hhilosophy, entitled Epistemic and Semantic Commitments of Foundational Theories, devoted to formal truth theories and implicit commitments of foundational theories as well as their conceptual surroundings.

Abstract. According to the math tea argument, perhaps heard at a good afternoon tea, there must be some real numbers that we can neither describe nor define, since there are uncountably many real numbers, but only countably many definitions. Is it correct? In this talk, I shall discuss the phenomenon of pointwise definable structures in mathematics, structures in which every object has a property that only it exhibits. A mathematical structure is Leibnizian, in contrast, if any pair of distinct objects in it exhibit different properties. Is there a Leibnizian structure with no definable elements? We shall discuss many interesting elementary examples, eventually working up to the proof that every countable model of set theory has a pointwise definable extension, in which every mathematical object is definable.

The talk will be held online via Zoom ID: 998 6013 7362.

Abstract. It is a mystery often mentioned in the foundations of mathematics that our best and strongest mathematical theories seem to be linearly ordered and indeed well-ordered by consistency strength. Given any two of the familiar large cardinal hypotheses, for example, generally one of them proves the consistency of the other. Why should this be? The phenomenon is seen as significant for the philosophy of mathematics, perhaps pointing us toward the ultimately correct mathematical theories. And yet, we know as a purely formal matter that the hierarchy of consistency strength is not well-ordered. It is ill-founded, densely ordered, and nonlinear. The statements usually used to illustrate these features are often dismissed as unnatural or as Gödelian trickery. In this talk, I aim to overcome that criticism—as well as I am able to—by presenting a variety of natural hypotheses that reveal ill-foundedness in consistency strength, density in the hierarchy of consistency strength, and incomparability in consistency strength.

The talk should be generally accessible to university logic students, requiring little beyond familiarity with the incompleteness theorem and some elementary ideas from computability theory.

The best way to learn mathematics is to dive in and do it. Don’t just listen passively to a lecture or read a book—you have got to take hold of the mathematical ideas yourself! Mount your own mathematical analysis. Formulate your own mathematical assertions. Consider your own mathematical examples. I recommend play—adopt an attitude of playful curiosity about mathematical ideas; grasp new concepts by exploring them in particular cases; try them out; understand how the mathematical constructions from your proofs manifest in your examples; explore all facets, going beyond whatever had been expected. You will find vast new lands of imagination. Let one example generalize to a whole class of examples; have favorite examples. Ask questions about the examples or about the mathematical idea you are investigating. Formulate conjectures and test them with your examples. Try to prove the conjectures—when you succeed, you will have proved a theorem. The essential mathematical activity is to make clear claims and provide sound reasons for them. Express your mathematical ideas to others, and practice the skill of stating matters well, succinctly, with accuracy and precision. Don’t be satisfied with your initial account, even when it is sound, but seek to improve it. Find alternative arguments, even when you already have a solid proof. In this way, you will come to a deeper understanding. Test the statements of others; ask for further explanation. Look into the corner cases of your results to probe the veracity of your claims. Set yourself the challenge either to prove or to refute a given statement. Aim to produce clear and correct mathematical arguments that logically establish their conclusions, with whatever insight and elegance you can muster.

This book is offered as a companion volume to my book Proof and the Art of Mathematics, which I have described as a mathematical coming-of-age book for students learning how to write mathematical proofs.

Spanning diverse topics from number theory and graph theory to game theory and real analysis, Proof and the Art shows how to prove a mathematical theorem, with advice and tips for sound mathematical habits and practice, as well as occasional reflective philosophical discussions about what it means to undertake mathematical proof. In Proof and the Art, I offer a few hundred mathematical exercises, challenges to the reader to prove a given mathematical statement, each a small puzzle to figure out; the intention is for students to develop their mathematical skills with these challenges of mathematical reasoning and proof.

Here in this companion volume, I provide fully worked-out solutions to all of the odd-numbered exercises, as well as a few of the even-numbered exercises. In many cases, the solutions here explore beyond the exercise question itself to natural extensions of the ideas. My attitude is that, once you have solved a problem, why not push the ideas harder to see what further you can prove with them? These solutions are examples of how one might write a mathematical proof. I hope that you will learn from them; let us go through them together. The mathematical development of this text follows the main book, with the same chapter topics in the same order, and all theorem and exercise numbers in this text refer to the corresponding statements of the main text.

Welcome to Cantor’s Ice Cream Shoppe! A huge choice of flavors—pile your cone high with as many scoops as you want!

Have two scoops, or three, four, or more! Why not infinitely many? Would you like $\omega$ many scoops, or $\omega\cdot2+5$ many scoops? You can have any countable ordinal number of scoops on your cone.

And furthermore, after ordering your scoops, you can order more scoops to be placed on top—all I ask is that you let me know how many such extra orders you plan to make. Let’s simply proceed transfinitely. You can announce any countable ordinal $\eta$, which will be the number of successive orders you will make; each order is a countable ordinal number of ice cream scoops to be placed on top of whatever cone is being assembled.

In fact, I’ll even let you change your mind about $\eta$ as we proceed, so as to give you more orders to make a taller cone.

So the process is:

You pick a countable ordinal $\eta$, which is the number of orders you will make.

For each order, you can pick any countable ordinal number of scoops to be added to the top of your ice-cream cone.

After making your order, you can freely increase $\eta$ to any larger countable ordinal, giving you the chance to make as many additional orders as you like.

At each limit stage of the ordering process, the ice cream cone you are assembling has all the scoops you’ve ordered so far, and we set the current $\eta$ value to the supremum of the values you had chosen so far.

If at any stage, you’ve used up your $\eta$ many orders, then the process has completed, and I serve you your ice cream cone. Enjoy!

Question. Can you arrange to achieve uncountably many scoops on your cone?

Although at each stage we place only countably many ice cream scoops onto the cone, nevertheless we can keep giving ourselves extra stages, as many as we want, simply by increasing $\eta$. Can you describe a systematic process of increasing the number of steps that will enable you to make uncountably many orders? This would achieve an unountable ice cream cone.

What is your solution? Give it some thought before proceeding. My solution appears below.

Alas, I claim that at Cantor’s Ice Cream Shoppe you cannot make an ice cream cone with uncountably many scoops. Specifically, I claim that there will inevitably come a countable ordinal stage at which you have used up all your orders.

Suppose that you begin by ordering $\beta_0$ many scoops, and setting a large value $\eta_0$ for the number of orders you will make. You subsequently order $\beta_1$ many additional scoops, and then $\beta_2$ many on top of that, and so on. At each stage, you may also have increased the value of $\eta_0$ to $\eta_1$ and then $\eta_2$ and so on. Probably all of these are enormous countable ordinals, making a huge ice cream cone.

At each stage $\alpha$, provided $\alpha<\eta_\alpha$, then you can make an order of $\beta_\alpha$ many scoops on top of your cone, and increase $\eta_\alpha$ to $\eta_{\alpha+1}$, if desired, or keep it the same.

At a limit stage $\lambda$, your cone has $\sum_{\alpha<\lambda}\beta_\alpha$ many scoops, and we update the $\eta$ value to the supremum of your earlier declarations $\eta_\lambda=\sup_{\alpha<\lambda}\eta_\alpha$.

What I claim now is that there will inevitably come a countable stage $\lambda$ for which $\lambda=\eta_\lambda$, meaning that you have used up all your orders with no possibility to further increase $\eta$. To see this, consider the sequence $$\eta_0\leq \eta_{\eta_0}\leq \eta_{\eta_{\eta_0}}\leq\cdots$$ We can define the sequence recursively by $\lambda_0=\eta_0$ and $\lambda_{n+1}=\eta_{\lambda_n}$. Let $\lambda=\sup_{n<\omega}\lambda_n$, the limit of this sequence. This is a countable supremum of countable ordinals and hence countable. But notice that $$\eta_\lambda=\sup_{n<\omega}\eta_{\lambda_n}=\sup_{n<\omega}\lambda_{n+1}=\lambda.$$ That is, $\eta_\lambda=\lambda$ itself, and so your orders have run out at $\lambda$, with no possibility to add more scoops or to increase $\eta$. So your order process completed at a countable stage, and you have therefore altogether only a countable ordinal number of scoops of ice cream. I’m truly very sorry at your pitiable impoverishment.

Abstract. Recent years have seen a flurry of mathematical activity in set-theoretic and arithmetic potentialism, in which we investigate a collection of models under various natural extension concepts. These potentialist systems enable a modal perspective—a statement is possible in a model, if it is true in some extension, and necessary, if it is true in all extensions. We consider the models of ZFC set theory, for example, with respect to submodel extensions, rank-extensions, forcing extensions and others, and these various extension concepts exhibit different modal validities. In this talk, I shall describe the state of current developments, including the most recent tools and results.