Elementary theory of the category of groupoids?

Question

One axiomatisation of set theory, the Elementary Theory of the Category of Sets, or ETCS for short, comes from category theory and states that sets and functions form a locally cartesian-closed, finitely complete and co-complete Heyting pretopos with a subobject classifier and a natural numbers object, whose generating object is the terminal object and whose epimorphisms are split.

Is there a corresponding axiomatic charactersation of the category of groupoids, in which one could do groupoid theory in, that does not involve first defining the concept of $\infty$-groupoid or homotopy types or other infinity categorical structures, and if so, what are the axioms?

And just as ETCS could be considered as a foundation of mathematics that only require sets and propositions, could this Elementary Theory of the Category of Groupoids serve as a more general foundation of mathematics that includes everything that could be done in ETCS as well as providing proper foundations for $1$-category theory? The collection of objects $Ob(C)$ in a category $C$ tend to be a groupoid; when $Ob(C)$ is a set, the category is called strict, so category theory as defined in ETCS or another set theory like ZFC could only speak of strict categories rather than general categories.

Edit: In a finitely complete category, finite limits are saturated under the terminal object and pullbacks. Does this still remain true when one moves to (2,1)-terminal objects and (2,1)-pullbacks and (2,1)-limits in finitely (co)complete (2,1)-categories? For ($\infty$,1)-categories, it doesn't seem to be the case that finite ($\infty$,1)-limits are saturated under ($\infty$,1)-terminal objects and ($\infty$,1)-pullbacks, if I am reading the nLab article on Lurie's ($\infty$,1)-pretopos correctly. In a (2,2)-category, (2,2)-terminal objects and (2,2)-pullbacks are known not to be enough for all finite (2,2)-limits; (2,2)-powers with the interval category are also needed.

Edit 2: I commented somewhere below that this theory as a foundational theory should be expressed in first order logic with isomorphisms, rather than first order logic with equality. I don't think this is the case anymore; ETCS should be the theory expressed in first order logic with isomorphism, as sets in ETCS are only isomorphic rather than equal. Rather, ETCG should be expressed in first order logic with equivalence of groupoids. It is only models of ETCG internal to ETCG that are expressed in first order logic with isomorphism, in the same way that models of ETCS internal to ETCS are expeessed in first order logic with equality.

Answer 1

I'm showing up a bit late to this party, but maybe I still have something to add. As Andrej and others have pointed out, one can obtain a type theory for 1-groupoids by starting with any form of HoTT and adding a 1-truncation axiom. (It's amusing (and perhaps deep) that in the type-theoretic context, it's easier to start from a type theory for $\infty$-groupoids and then assert an extra axiom that cuts down to 1-groupoids.) The paper of Harper and Licata is about a version of such a theory that satisfies the nice technical property of "canonicity", but if all you want is a formal system you can just add the 1-truncation axiom to Book HoTT --- and modify the univalence axiom so it doesn't contradict that. Basically you should assert only that there is a univalent universe of sets (0-truncated types).

Now, it sounds from the question like you're more interested in a "category-theoretic" phrasing of this, looking more like the phrasing of ETCS as "the category of sets is an elementary topos such that blah". I don't think this has been written down as such; for higher values of $n$ it's more common to discuss Grothendieck $n$-toposes than elementary ones. But similar to the type-theoretic case, one could start from a proposed definition of elementary $(\infty,1)$-topos and cut it back down to a (2,1)-category. This would produce something like: an elementary (2,1)-topos is a (2,1)-category (i.e. a category enriched in groupoids, or a 2-category or bicategory whose hom-categories are groupoids) such that

1. It has finite limits and colimits, in the bicategorical sense.
2. It is locally cartesian closed, in a bicategorical sense: pullback along any morphism has a right bicategorical adjoint.
3. It has a subobject classifier, i.e. the functor sending each object to the poset of fully faithful inclusions into it is representable.
4. For any faithful morphism, there is a generic faithful morphism classifying it and such that the morphisms it classifies are closed under composition, finite fiberwise limits and colimits, and dependent products.

The last axiom is the "object classifier", and there's some room for discussion in how it should be phrased. The above phrasing will allow you to use it to reason about arbitrary 0-truncated objects, but it requires some large cardinals to model.

I expect that with the object classifier, one can prove that the category is "(2,1)-exact" in the sense that any "internal groupoid" in a suitable sense is the "kernel" of some quotient. A while ago I made some study of exactness conditions for 2-categories, the remnants of which can be found here. If one wants to avoid any universes, one could take this exactness as part of the definition instead, although exactness isn't the only purpose of the universe so this would impoverish the definition a bit.

Finally one would want to add some axioms analogous to the blah in ETCS. A (2,1)-category should be "well-pointed" if the terminal object is a "generator" in a suitable sense. Certainly the functor $E(1,-) : E \to Gpd$ should be locally faithful. Probably it's too much to ask it to be locally fully faithful. One might need to add some more "projectivity and indecomposablity" conditions analogous to the constructive notion of well-pointedness for 1-categories, if they don't follow automatically the way they do for classical 1-categories; I haven't thought too much about this.

Then if one wants a "classical" version analogous to ETCS, one can add the axiom of choice. The only thing to be aware of here is that it can pertain only to 0-truncated objects, as discussed in type theory in the HoTT Book. In a (2,1)-category, one way to state the internal axiom of choice is that for any (not necessarily truncated) object $U$ and 0-truncated (i.e. representably faithful) morphisms $Y\to X\to U$ such that $Y\to X$ is surjective, there exists a surjective map $V\to U$ such that the pullback of $Y\to X$ to $V$ has a section. Note that $V$ need not be 0-truncated either. In the presence of well-pointedness, this might simplify to a more "external" axiom of choice analogous to ETCS's "all epimorphisms split", but I haven't thought much about that either.

Answer 2

Robert Harper and Dan Licata studied the topic under the name 2-dimensional type theory, see their Canonicity for 2-Dimensional Type Theory and possibly these slides. As they are computer scientists there is a lot of talk about the syntactic properties of such theories, but underneath it really is just groupoids (and they make the connection explicit). If you stare long enough at their rules, they can all be understood in terms of groupiod structures.

In the context of homotopy type theory one can formulate an axiom stating that all types are groupoids: $$\Pi(A : \mathsf{Type}) (x, y : A) (p, q : x = y) (\alpha, \beta : p = q)\,.\, \alpha = \beta.$$ It states that for any type $A$, points $x, y : A$, parallel paths $p, q$ from $x$ to $y$, any 2-cells between $p$ and $q$ are equal. This is a straightforward modification of Streicher's axiom $K$ which states that all types are sets. This may look simpler than Harper's and Licata's 2-dimensional type theory, but is actually not capturing groupoids directly – it's more like 2-truncated $\infty$-groupoids, where all the higher structure is still around, but is declared to be contractible.

Answer 3

My rambling comments are starting to cohere into a quasi-answer, though not a definitive one:

1. In addition to ETCS, Lawvere also formulated ETCC, the elementary theory of the category of categories.

One could very likely arrive at a similar axiomatization of groupiods, either by adapting Lawvere's ETCC axioms, or by directly using the fact that groupoids form a full subcategory of categories.

Of course, the way you formulate the axioms of ETCS reflects a great deal of conceptual development which has happened since Lawvere originally wrote down the theory. I don't know of a similar modernized, sleek way to package Lawvere's axioms for the category of categories, and I think that's partly because it would be harder to do / the necessary concepts may not have been studied (yet?). So if one were to groupoid-ify ETCC, there would still be more work to make the theory more "palatable".

2. You might also be interested in groupoid models of type theory, where groupoids are thought of in a similar way to what you suggest, lying somewhere between sets and homotopy types and undelrying a theory which can be used foundationally.

3. I'll also point out that your idea of thinking of the underlying groupoid of a category as a sort of "improved version" of the set of objects of the category does come up. For example, this way of thinking leads to the idea of a complete Segal space as opposed to a Segal category.

Because of the difficulties I alluded to above in groupoid-ifying ETCC, it may be better to return to the comparison to ETCS. The formulation you give, though it doesn't directly invoke the notion of an elementary topos, is closely related to it. A groupoid version of this axiomatization, then, would be talking about something close in spirit to a higher theory of elementary topoi. There has been a great deal of interest in formulating such a theory for a long time.

Going beyond ordinary topoi, the most famous development is Lurie's theory of (Grothendieck) $\infty$-topoi. Nima Rasekh has given a theory of elementary $\infty$-topoi, but the theory is very much in its infancy.

You want something in the middle -- a (2,1)-topos. (The "2" means that you want your topos to be a 2-category; the "1" means that only 1-morphisms and not 2-morphisms will be non-invertible in your category.) I believe that people like Mike Shulman have thought about notions in this area, but I don't know if anything is published.

I haven't thought carefully about this, but the main thing I'd think needs to be straightened out is the following. In ordinary topoi, the subobject classifier does a great deal of work, but in in $\infty$-topoi the concept needs to be strengthened to an object classifier, and size issues cause technical annoyances. The other way of thinking about what the object classifier encodes is in terms of descent (a scary category-theoretic word which means different things to different people, but which in each individual case turns out to be not scary). Thinking in these terms, the theory of $\infty$-topoi actually turns out to be in some sense cleaner than the theory of ordinary topoi. In fact the following is true in $\infty$-topoi but (its corresponding version) is false in ordinary topoi: if $\mathcal E$ is an $\infty$-topos, the "slice" functor $\mathcal E^{op} \to CAT_\infty$, $X \mapsto \mathcal E/X$ preserves limits. I'm not sure whether a (2,1)-topos is expected to be more like $\infty$-topoi or 1-topoi in this regard.