Yale Philosophy Grad Program Replaces Logic with Formal Methods in Its Requirements


“Yale philosophy has officially replaced the grad program ‘logic requirement’ with a broader ‘formal methods requirement.’ Students can choose which course to take (logic, probability, stats, game theory, etc.).”

That’s Joshua Knobe (Yale) on Twitter yesterday. He adds: “Feels like a symptom of a much larger change in the discipline.”

The change follows years of discussion in the philosophy profession about the extent to which most philosophers make use of more than just basic logic in their work, and the extent to which training in statistics, probability, and other methods might be more useful in some philosophical subfields. In a 2016 post here, for example, Professor Knobe wrote:

It is widely agreed that graduate students in philosophy can benefit from training in formal methods. In our present system, we do at least a relatively good job of providing training in (certain areas of) symbolic logic, but I worry that we sometimes aren’t helping students to learn the methods that will prove most helpful to them in their actual philosophical work.

To give just one obvious example, people working in numerous areas of philosophy need to read and understand papers that make use of statistics. This sort of work has become increasingly important in philosophy of mind, moral psychology, experimental philosophy, feminist philosophy, global justice, and numerous other areas. But there is sometimes a curious mismatch between the training people in these areas receive and the actual work they go on to do. As a result, many philosophers are in the difficult position of doing research that relies very heavily on empirical findings while having a background that gave them knowledge of the Löwenheim–Skolem theorem but no understanding of what a correlation coefficient is.

In a post here in 2019, Liam Kofi Bright (LSE) wrote:

With all such purported pragmatic benefits of logic, I see them more often asserted than proven; but even setting that aside, it is not clear why an emphasis on usefulness picks out logic in particular, for either philosophical specifically or broader marketability. Actual arguments one encounters in both philosophy and broader life, and the sort of mathematical skills which it is useful to have both as citizen and employee, are frequently probabilistic or inductive rather than strictly deductive. It is very rare thing indeed that philosophers establish arguments which are sufficiently complex that formal logical training might be useful for understanding them, yet which at the same time even purport to be deductively valid arguments. Much more often we are in the business of offering suggestive considerations, or amalgamating assorted facts which together render plausible one’s conclusion. These are more naturally read as probabilistic or inductive arguments than deductive ones, and where they suggest some practical course of action they are perhaps most naturally analyzed with the tools of decision theory. In life beyond academic philosophy, it will frequently be claims couched in statistical garb that one will encounter in the newspapers and the workplace. If we want to help students understand and construct the sort of arguments we actually make and encounter, they need to know more than just deductive logic.

Yale Philosophy actually started allowing students to take a variety of formal methods courses instead of logic back in 2017. Knobe’s announcement is simply about a change in the official graduate program requirements. As he noted then, Yale was only one of many departments to make such changes.

It has been a while since that informal survey of programs, and no doubt other programs have made similar changes since then. It would be useful to hear more about what various departments are doing in this regard, so please let us know.

Discussion welcome.


Use innovative tools to teach clear and courageous thinking
Subscribe
Notify of
guest

47 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Syed
Syed
10 months ago

Not sure what to think about this trend. I was one of those people who thought logic wasn’t very helpful for doing most philosophy. However, as I learned more logic (different systems) I noticed a difference in how I understood philosophical writings. I also noticed a difference in the structure of my arguments. Of course, it might not be very helpful for ALL philosophy, but it seems like it would be an overall disadvantage, if one didn’t take any advanced logic?

Keli Birchfield
Reply to  Syed
10 months ago

Why not keep the logic requirement and provide students working in applied/interdisciplinary philosophy with opportunities to take a formal methods course in their field of application?

Syed
Syed
Reply to  Keli Birchfield
10 months ago

Agreed! I think a hybrid approach like this is probably best!

Kenny Easwaran
Reply to  Keli Birchfield
10 months ago

The general issue is that this type of thinking easily slides into a graduate program where you have to take three history classes, one seminar each in metaphysics, ethics, epistemology, philosophy of mind, philosophy of language, philosophy of science, a logic class, a statistical reasoning class, two foreign languages, and a class in continental and/or non-western philosophy.

If you want to encourage students to take more classes that are relevant to their research, then (at least at some margin) you have to allow them to substitute some of these classes for things that are currently requirements.

Eric Bennett Rasmusen
Eric Bennett Rasmusen
Reply to  Kenny Easwaran
10 months ago

Would 14 required courses plus 2 electives be a bad thing for the first 2 years of a PhD program?

Chris
Chris
Reply to  Eric Bennett Rasmusen
10 months ago

It would be a lot. A full load here is typically three courses a term, which would only get you 12 total courses after two years. (All of our students work as Teaching Assistants, in case this is different from econ. departments).

cecil burrow
Reply to  Eric Bennett Rasmusen
10 months ago

I don’t think you can expect grad students to write 16 non-awful papers in 2 years. Professional philosophers certainly can’t do that.

Eric Bennett Rasmusen
Eric Bennett Rasmusen
Reply to  cecil burrow
10 months ago

In econ, almost all courses use tests, not papers (tho maybe with a couple required papers quite aside from coursework– we had to do a stats paper and an econ history paper). In philosophy, is there a reason to use papers instead of tests for 1st and 2nd year courses, where students are learning the basics? Does PhD Logic use a paper instead of tests?

Phil grad student
Reply to  Eric Bennett Rasmusen
10 months ago

With all due respect, to my mind this would be rather terrible. This wouldn’t allow PhD students almost any flexibility to tailor their coursework to their research interests—which would undermine what I think the function of course work in most PhD programs should be: to help you develop yourself as a researcher and / or teacher with a number of strong AOSs / AOCs. Thankfully it is not something PhD programs currently implement, especially given how common it is that people come into a PhD program with a MA these days. At some point added breadth has no added value.

Eric Bennett Rasmusen
Eric Bennett Rasmusen
Reply to  Phil grad student
9 months ago

You might like to do as in econ. Our coursework is mostly for methods and things every economist should know. You read the current literature in a subfield the third year on your own, after you finish coursework.

cecul burrow
cecul burrow
Reply to  Eric Bennett Rasmusen
9 months ago

So what should every philosopher know? Good luck finding any sort of agreement on that.

M S
M S
10 months ago

Princeton is considering doing the same!

Caligula's Goat
10 months ago

This is good news! I think more flexibility should be offered for logic and language requirements. In truth, unless you’re focusing on history of philosophy and/or Non-Western, I’m not sure I see the point of having language requirements in graduate school.

A lot of philosophy is interdiscplinary now and it makes perfect sense for departments to incentive developing interdsicplinary skills by swapping coursework like logic/language (whose usefulness in an academic philosopher’s career is more situational and niche) for course competency in other discplines.

Rob
Rob
Reply to  Caligula's Goat
10 months ago

One might think that better philosophy would be done in many areas where people are writing in English if they had the language skills to read the relevant material being written in other languages. Unless, of course, only the stuff being written in English is of value…

David Wallace
Reply to  Rob
10 months ago

I think it’s the same principle. It’s valuable to know how to read texts in philosophy that are written in languages other than English. But it’s also valuable to know how to read texts in physics, or medical statistics, or linguistics, where the words are in English but the content isn’t intelligible unless you know the relevant formal topic. It’s going to vary from student to student, and perhaps from program to program, what the right way is to use very limited time resources.

Bryan Frances
10 months ago

I think it’s a good thing. But … just about the only useful thing I learned in my classes was logic, both first-order and modal. The other useful stuff came from reading groups. The other courses? Pffft.

Branden Fitelson
10 months ago

no brainer.

J.A.M.
10 months ago

I don’t think logic is what I would drop from degree requirements, even though adding more formal methods is a good thing.

In fact I think that’s the last thing I would sacrifice. I can’t think of anything of more value (logic is tied with inductive reasoning and game theory) that I picked up in my undergrad, so it would make more sense to de-prioritize something less rigorous (dare I say ethics).

Tim
Tim
Reply to  J.A.M.
10 months ago

Or, perhaps, metaphysics

Parrot Bird
Parrot Bird
Reply to  J.A.M.
7 months ago

Ethics can be the most rigorous if we play the cards right. It’s literally part of practical philosophy that solves real-world problems–along with political philosophy and philosophy of law. The rest are more abstract in nature.

Phoenix, son of Amyntor
Phoenix, son of Amyntor
10 months ago

I had to take predicate logic both as an undergraduate and as part of my Ph.D. I hated it. However, I think there is one argument on its behalf: it is a foreign language that authors sometimes dip into in their own writings. E.g. you might encounter it in an essay in contemporary metaphysics or, say, an analysis of ancient arguments about self-refutation. Of course, there are plenty of genres where you won’t encounter it, but without some training in it, what is one to do when one does encounter it?

Kenny Easwaran
Reply to  Phoenix, son of Amyntor
10 months ago

I think it’s definitely important for anyone working in most areas of contemporary analytic philosophy to have enough familiarity with predicate logic to read a few sentences in it! But you can get this from an undergraduate class, that a lot of students would have taken before grad school anyway. There’s no need to take a full graduate logic class.

JDRox
JDRox
Reply to  Kenny Easwaran
10 months ago

Sure, but I don’t think we can rely on those undergraduate courses! I went to grad school with a fair number of people who had gone to “continental” undergrad programs and who really didn’t have the logic background they needed. Of course, a “full graduate logic class” was much more than they needed in a sense, but maybe worth it just because they had to practice/use the basic stuff while they were struggling with the advanced stuff.

newly tt
newly tt
Reply to  JDRox
10 months ago

My program had a way for grad students to take the upper-level undergrad logic course if they hadn’t had it in undergrad. This way we were prepared for the grad logic course. Presumably other programs could do something like that? I guess the justification might be harder if you don’t have to prepare for a grad course, though.

JDRox
JDRox
Reply to  newly tt
10 months ago

I’m happy with just requiring an upper-level undergrad logic course, or with testing out of the requirement, or whatever. I just think we should be very sure that grad students are reaching a certain level of logical competence.

Martin Lenz
10 months ago

Makes sense. I’d add another requirement: a translation course in which philosophical texts are to be translated from a foreign language. There is a great amount of skills coming with training to translate.

Matt L
Reply to  Martin Lenz
10 months ago

Isn’t this essentially what most language requirements are in grad programs? At Penn, at least, the exam for the “Language X for reading knowledge” was actually a translation exam – i.e., translate some pages of a philosophers in a set period of time. Of course, most people can’t really learn to do that well in a semester. (I remember hearing one of the German professors, who sometimes taught the course, having said, of some philosophy grad student’s translation of Kant, “Your translation doesn’t really make sense to me, but then, Kant doesn’t really make sense to me in German, either, so I guess it’s good enough to pass.”)

Sam Duncan
Sam Duncan
Reply to  Matt L
10 months ago

I’m all for learning languages. I’m very glad I speak and read German and would love to learn Greek one day. But like Matt L I have severe doubts about the usefulness of language requirements that could be knocked out in a semester or two (I have the same doubts about the usefulness of giving students a smattering of formal logic too). That seems more like a stupid human trick than it does anything useful. I feel like in an ideal world grad students would have to either really learn either a language or go in for some serious learning of relevant formal methods but would also be given a year or two to do just that. But that sort of suggestion seems to go against the currents of grad student training which seem to flow very heavily toward narrow specialization and getting students out quickly.

Patrick Mayer
Patrick Mayer
10 months ago

I would keep a logic requirement and add a second formal methods requirement that could take the form of further logic courses (like a course in modal logic – my grad level logic class did not cover modal logic) but also probability, stats, game theory etc. This would let people specialize a bit. Probably modal logic makes more sense for people doing metaphysics. A course on game theory would be more valuable for people doing political philosophy, probability would probably be useful for everyone but particularly for epistemology. But I would not get rid of the logic requirement. I have trouble seeing how someone is going to understand much 20th century Anglophone philosophy without an understanding of logic. If I were going to cut something it would probably be an M & E requirement, if only because you are going to get a fair bit of M & E in the history of philosophy courses that are probably required.

Eric Bennett Rasmusen
Eric Bennett Rasmusen
Reply to  Patrick Mayer
10 months ago

I’m an economist with a game theory text suitable for philosophy PhD students. See http://www.rasmusen.org/GI/ . But I think logic would be much more important.
How about Logic plus a required Methods course taught by a philosophy professor and covering (a) game theory, (b) probability and stats, (c) how to write referee reports, (d) good citation practices, and (d) how to give a seminar? I taught these last to my first year PhD students in econ. https://rasmusen.org/GI/reader/writing.pdf

Philmath
10 months ago

Some Philosophy of Science have dropped their grad logic requirements since a while. Pitt HPS (https://www.hps.pitt.edu/graduate/graduate-student-handbook#PhD-Formal) for example requires an undergrad logic course+ further logic/math/programming/digital humanities. Yale (and other non philosophy of science programs) doing this is maybe a starker indication of shift in the discipline.

David Wallace
10 months ago

Pitt HPS has this requirement (Pitt Philosophy has a more traditional logic requirement):

Students must demonstrate proficiency in formal methods related to their research or career goals. These methods include both (a) and (b) listed below.
(a) proficiency in basic symbolic logic (at least first order predicate logic), and
(b) further coursework, at either the upper-level undergraduate or graduate level, involving formal methods in an area approved by the DGS. Such areas include a more advanced course in logic (such as logical theory, mathematical logic, or modal logic), higher mathematics (such as set theory, linear algebra, calculus, probability and statistics, or other specialized area of mathematics), or computational methods (ranging from computer programming to theory of computation and including suitable courses in digital humanities or information science).

FWIW I think this (which I wasn’t involved in drawing up) is about right: obviously everyone needs a bit of logic, but set theory and the metatheory of predicate calculus are quite specialized topics in modern philosophy and, outside philosophy of maths/logic itself and certain areas of philosophy of language, it’s hard to see them as sufficiently important to justify making them compulsory.

That’s not to say that students don’t get some benefit from doing higher logic. But Kenny Easwaran is exactly right: the question isn’t whether there’s some benefit, it’s whether the benefit is worth using up an extremely valuable chunk of space on the core curriculum. (I feel the same way about language requirements: they’re beneficial, but for many students something else will be more beneficial.)

All that said, I don’t think it would necessarily be wrong for a given program to have a formal logic requirement (or a language requirement, or a causal/statistical modelling requirement, or a science-course requirement) – provided they see it as a distinctive feature of their program in particular, which students can weigh up when applying, and not something that every self-respecting philosophy grad program requires.

JDRox
JDRox
Reply to  David Wallace
10 months ago

To add to what I said to Kenny above, I guess I mostly agree with this, but the bit of logic everybody needs is *really* needed. I’d be happy letting people test out of the requirement, but part of the reason a “whole” grad logic class is valuable is that you spend the whole semester practicing/using the basic stuff that you really need. Frankly, the logic most people really need could be taught in 2-4 weeks. But if that was it (the course was one credit, or included other topics, or whatever) most people would forget it all more or less right away.

Brian Weatherson
Brian Weatherson
Reply to  JDRox
10 months ago

But do you actually use more of the “basic stuff you really need” in a metatheory class than in a class that covers lambda calculus, accuracy measures, properties of expectations, social choice functions, etc? You spend quite a lot of time thinking about, and using, quantifiers and connectives in all of those units, so a good formal methods course (like the one Sarah Moss teaches here at UM) might be better by this standard than the old compulsory metatheory course.

JDRox
JDRox
Reply to  Brian Weatherson
10 months ago

Sure, I agree with this: I mean, in a way, it’s an argument for keeping a formal logic requirement, since basic formal logic is presupposed/used in using and/or theorizing about the lambda calculus, accuracy measures, properties of expectations, social choice functions, etc. Sorry if I’ve misunderstood the “other” position!

David Wallace
10 months ago

There was quite a good discussion of this (mostly focusing on language requirements rather than logic requirements) about ten years ago at the now-defunct NewAPPS blog:
https://www.newappsblog.com/2012/03/logic-and-languages.html

David Hyder
10 months ago

I fulfilled my logic requirement at Yale in the 1980s by taking computer science courses, although it did have to be formally approved.

But later, I came to regret having exercised that option. I had to spend a lot of time in grad school learning logic from a philosophical perspective, as well as modal logic, set theory, etc.

These logics have in the end proved more useful to me, though perhaps that’s simply due to having the computer background as well. So, whether or not it happens during the undergraduate phase, I do think they are essential.

Roy T Cook
Roy T Cook
10 months ago

This post is framed as if Yale is doing something radical, but I think that shifts in this direction have been becoming more common in departments that focus on things other than merely traditional analytic language/metaphysics/epistemology for a while. Although the University of Minnesota still has a two-course “logic” requirement, one of the four courses on the list for satisfying this requirement is a Formal Methods course (designed and taught, at present, by Samuel Fletcher) that covers set theory, statistics, probability, and other non-logic topics (all from both a technical and a philosophical perspective). This has been the situation at the U of M for more than half a decade. The idea of allowing such courses to satisfy the “formal” portion of PhD training was supported by pretty much everyone in the department (including the logicians, so long as we also continued to offer the advanced logic courses as options as well). The reason, briefly put, is that broadening the kinds of formal training that PhD students receive amounts to nothing more than correctly tracking the way that philosophical methodology is changing and becoming much more than merely “regiment in FOL and prove some theorems”.

Richard Zach
10 months ago

Already 20 years ago only about half of ranked US PhD programs had a requirement like the one Yale now got rid of. Many required just basic predicate logic or some kind of disjunctive requirement. https://richardzach.org/2004/10/formal-logic-and-philosophy-iii/

Richard Russell Wood
Richard Russell Wood
10 months ago

I guess that leaves CompSci and the more “technical” programs to take up the gauntlet.

sigh

Aeon J. Skoble
10 months ago

Several folks have mentioned that the really necessary stuff a grad student needs comes up in a good upper-level undergrad formal logic class. But not every BA program requires this. So how about keeping a formal logic requirement in the grad program but students can place out of it if they had the right sort of undergrad logic.

David Wallace
Reply to  Aeon J. Skoble
10 months ago

Pitt HPS already does that.

Andy
10 months ago

I could be convinced that logic has special relevance to philosophy, and should be upheld. But if the suggestion is that departments are relaxing their standards in the sense of making things easier or less rigorous, I’m not sure I buy that, because I don’t view logic as particularly more demanding than statistics or various other “formal methods” courses.

Francisco Díaz Montilla
10 months ago

It is not a bad idea. Deductive logic is a powerful tool, of course. But sometimes probability or other formal approache is needed.

Milton Ponson
10 months ago

As a mathematician I have always been fascinated by the many formal methods used in science for reasoning and conducting research, and in particular for empirical research. I currently focus on all formal methods that are useful for knowledge representation for open, ethical, inclusive and explainable artificial intelligence. Increasing use of psychology and philosophy in AI development necessitates the use of all manner of formal methods.
So the Yale broadening of requirements makes sense.

Christopher Gauker
10 months ago

I suppose that a lot of people who deny the importance of logic think that learning logic means becoming skilled in the use of some deductive calculus or that it is supposed to help us craft good arguments. I don’t think that is where the primary importance lies. In many fields of philosophy the concept of representation plays a central role. We want to know in what way a sentence, or the utterance of a sentence, correctly represents the world. That holds not only for philosophy of language and mind but also for philosophy of science and for metaethics. For better or worse, the concept of truth-in-a-model that one acquires in a logic course serves as a primary point of reference in all these discussions of representation. A philosopher who has not had enough logic to understand how the concept of truth-in-a-model is used, for instance, to define the logical validity of arguments, will not be in a position to understand contemporary discussions of the nature of representation.

Last edited 10 months ago by Christopher Gauker
David Hyder
Reply to  Christopher Gauker
10 months ago

Fully agree.